You are on page 1of 233

ISBN 961-6242-42-3 JANEZ ORENIK A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS PREDVIDLJIV VIDIK (OBLIKO)SKLADENJSKIH DVOJNIC

SLOVENSKA AKADEMIJA ZNANOSTI IN UMETNOSTI ACADEMIA SCIENTIARUM ET ARTIUM SLOVENICA


RAZRED ZA FILOLOKE IN LITERARNE VEDE CLASSIS II: PHILOLOGIA ET LITTERAE DELA OPERA 58

JANEZ ORENIK

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS


PREDVIDLJIV VIDIK (OBLIKO)SKLADENJSKIH DVOJNIC

LJUBLJANA 2001

SPREJETO NA SEJI RAZREDA ZA FILOLOKE IN LITERARNE VEDE SLOVENSKE AKADEMIJE ZNANOSTI IN UMETNOSTI DNE 24. OKTOBRA 2000 IN NA SEJI IZVRILNEGA ODBORA PREDSEDSTVA DNE 17. SEPTEMBRA 2001 Knjino izdajo je finanno omogoilo Ministrstvo za olstvo, znanost in port Republike Slovenije

CIP Kataloni zapis o publikaciji Narodna in univerzitetna knjinica 81367 ORENIK, Janez A predictable aspect of (morpho)syntactic variants = Predvidljiv vidik (obliko)skladenjskih dvojnic / Janez Orenik. Ljubljana : Slovenska akademija znanosti in umetnosti, 2001. (Dela / Slovenska akademija znanosti in umetnosti, Razred za filoloke in literarne vede = Opera / Academia scientiarum et artium slovenica, Classis II. Philologia et litterae ; 58) ISBN 961-6242-42-3 114588160

FOR VARJA AND KARIN

Table of Contents
Acknowledgements.................................................................................................................9 Introduction Also the Summary of the Monograph..........................................................11 Examples...............................................................................................................................21 Albanian p. 21, Arabic p. 25, Arawak p. 30, Bahasa Indonesia p. 30, Bambara p. 32, Breton p. 32, Carib p. 33, Catalan p. 35, Chinese p. 35, Croatian p. 38, Danish p. 40, English (Old) p. 42, Estonian p. 43, Finnish p. 43, French p. 46, German p. 60, Germanic languages (old) p. 73, Gothic p. 73, Greek (Ancient) p. 75, Guaran p. 79, Hausa p. 80, Hidatsa p. 81, Hungarian p. 82, Icelandic p. 84, Italian p. 86, Japanese p. 90, Kekchi p. 91, Kharia p. 92, KinyaRwanda & Nez Perce p. 93, Lakhota p. 93, Lappish p. 98, Latin p. 98, Macedonian p. 102, Miriam p. 103, Norwegian p. 104, Paiute p. 105, Persian p. 105, Quechua p. 106, Quich p. 107, Quileute p. 109, Romance languages p. 109, Romanian p. 110, Russian p. 115, Sanskrit p. 117, Slovenian p. 120, Spanish p. 178, Swedish p. 183, Tadzhik p. 191, Tagalog p. 192, Thai p. 193, Tiwi p. 194, Turkish p. 195, Walbiri p. 198, Welsh p. 199, Wikchamni p. 200, Miscellanea p. 200 The preterite and present perfect tenses in German...........................................................213 Slovenski povzetek Slovenian Summary.........................................................................223 References...........................................................................................................................231 Index....................................................................................................................................237

Acknowledgements
Many persons have helped me in various ways in the course of my writing this book, and I wish to thank in particular the following persons: an anonymous Slovenian lady from Brazil, John Ole Askedal, Wayles Browne, Igor Grdina, Franc Jakopin, Domen Kavi, Milena Milojevi-Sheppard, Tom Muhlstein, Albina Neak-Lk, Vladimir Poganik, Chikako Shigemori-Buar, Mitja Skubic, John Charles Smith, and Marina Zorman. Over the last few years I have been trying parts of the book on students in courses. The students provided improvements to the text, and I owe them particular thanks. I owe gratitude to Margaret G. Davis for the correction of my English. Ms. Davis has also suggested several improvements of the contents. I owe special and personal gratitude to my wife Varja Cvetko-Orenik and to Karin Cvetko-Rasmussen for putting up with me and encouraging me. I dedicate the book to them. I am deeply indebted to Slovenska akademija znanosti in umetnosti (The Slovenian Academy of Sciences and Arts) for publishing this monograph in one of its series. Two points concerning technical matters. Female authors are referred to with the initial of their first name followed by their surname, whereas male authors are mentioned by surname only. E.g. the fictitious "Richards & J. Smith" refers to a putative joint publication of Henry Richards and Janet Smith. Occasionally I mention an author without adding the year of publication to his/her name. This tells the reader that I have not exploited or even not consulted that text. Nor is it listed in the References or in the Index. The author's addresses: Office: Filozofska fakulteta, SI-1001 Ljubljana, Slovenia E-mail: janez.oresnik@uni-lj.si

Introduction Also the Summary of the Monograph


Key words: naturalness, morphology, morphosyntax, syntax, and the languages listed in the Table of Contents. The subject-matter of the present monograph is a (language-universal) theory developed in Slovenia by a small group of linguists (under my guidance), who mainly use English, German, and Slovenian language material as the base of verification. Our work owes much to, and exploits, the (linguistic) Naturalness Theory as elaborated especially at some Austrian and German universities; cf. Mayerthaler 1981, Wurzel 1984, Dressler et al. 1987, Stolz 1992, Dressler 2000. Naturalness Theory has also been applied to syntax, notably at the University of Klagenfurt; the basic references are Dotter 1990, Mayerthaler & Fliedl 1993, Mayerthaler et al. 1993, 1995, 1998. Within the natural syntax of the Klagenfurt brand, the Slovenian work group has built an extension, which will henceforth be referred to as "the Slovenian Theory." The Slovenian Theory studies the behaviour of (mostly syntactic and morphosyntactic, less often morphological) expressions here called variants. Whenever two variants are included in the same naturalness scale, and consequently one variant can be asserted to be more natural than the other, the Slovenian Theory has something to say about the grammatical properties of the two variants. Naturalness Theory operates with two basic predicates, "marked" and "natural." I can not see any reason to distinguish the two predicates within the Slovenian Theory, therefore I use throughout one predicate only, namely "natural." (This standpoint was implied as early as Mayerthaler 1987, 50.) Beside the technical terms "natural(ness)" and "naturalness scale," which have already been alluded to, the terms "sym-value" and "sem-value" (adopted from Mayerthaler 1981, 10 and passim) must be mentioned. The sym-value refers to the naturalness of an expression in terms of its coding properties. The sem-value refers to the naturalness of an expression in terms of its semantic complexity. Any naturalness scale assumes the format >sem (a, b) or the format >sym (c, d). The format >sem (a, b) reads: With respect to semantic complexity, a is more natural than b. For ease of computation, a is assigned a high sem-value (symbol: >sem "more natural with respect to semantic complexity"), and b is assigned a low sem-value (symbol: <sem "less natural with respect to semantic complexity"). The format >sym (c, d) reads: With respect to coding properties, c is more natural than d. For ease of computation, c is assigned a high sym-value (symbol: >sym "more natural with respect to coding properties"), and d is

11

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

assigned a low sym-value (symbol: <sym "less natural with respect to coding properties"). The assumptions of the Slovenian Theory (in my recently revised version) can be briefly stated as follows. In a pair of variants, within each variant, one of the following alternatives obtains: 1. at least one >sym-value tends to associate with at least one additional >sym-value and/or with at least one <sem-value; 2. at least one <sym-value tends to associate with at least one additional <sym-value and/or with at least one >sem-value; 3. at least one >sem-value tends to associate with at least one additional >sem-value and/or with at least one <sym-value; 4. at least one <sem-value tends to associate with at least one additional <sem-value and/or with at least one >sym-value. In a nutshell, all theoretically possible clusters of sem- and sym-values are allowed, except >sem combined with >sym, or <sem combined with <sym, or >sem combined with <sem, or >sym combined with <sym. For instance, it is not expected that an elliptical construction (= <sym) would be used in the non-third person (= <sem) only, or that an analytic construction (= >sym) would be used in the third person (= >sem) only, in case that there is a real choice among the grammatical persons. Furthermore, it is not expected that the non-third grammatical person (= <sem) would be used in the singular (= >sem) only, or that an analytic construction (= >sym) would invariably be elliptical (= <sym). In the above items (1-4) the object of the meta-verb "associate" refers to the interior of the unit under observation, OR to a part of the immediate environment of the unit under observation. The Slovenian Theory covers both cases. Forschungsgeschichtlich , the predecessor of the above assumptions (1-4) is the familiar principle of constructional iconicity as formulated in Natural Morphology. The principle runs as follows. Iff a semantically more marked category C j is encoded as 'more' featured than a less marked category C i, the encoding of Cj is said to be iconic (Mayerthaler 1987, 48-9). Using the predicate "natural," the principle can be briefly stated as follows: <sem in combination with >sym is iconic. In the Slovenian Theory, the principle has been extended to (morpho)syntax and expanded. The Slovenian Theory functions only ex post facto . An example of a deduction: x. English. The numerical indication of frequency normally consists of a cardinal number followed by the word times, e.g. four times, except that there are one-word expressions available for the lowest numbers: once, twice, and archaically thrice . (Collins Cobuild 1990, 270-1.) The two syntactic variants: the type once, and the type four times. x.1. The assumptions of Naturalness Theory: x.1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of x.1.1: x.1.1.1. >sym (the type four times, the type once) / in English I.e. with respect to coding properties, the type four times is more natural than the type once, in English.

12

INTRODUCTION ALSO THE SUMMARY OF THE MONOGRAPH

x.1.2. >sem (low number, non-low number) I.e. with respect to semantic complexity, any low number is more natural than any non-low number. (Mayerthaler 1981, 15.) x.2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: x.2.1. >sym tends to associate with <sem x.2.2. <sym tends to associate with >sem x.3. The consequences: From x.1.1.1, x.1.2 and x.2.1 it can be deduced: x.3.1. If there is any difference between low and non-low numbers in numerical indications of frequency, such that one kind of numbers uses the pattern four times, and the other kind of numbers uses the pattern once, it is the non-low numbers that tend to use the pattern of four times. Q.E.D. From x.1.1.1, x.1.2 and x.2.2 it can be deduced: x.3.2. If there is any difference between low and non-low numbers in numerical indications of frequency, such that one kind of numbers uses the pattern four times, and the other kind of numbers uses the pattern once, it is the low numbers that tend to use the pattern of once. Q.E.D. We are entering on the interpretation of deduction x. Reference will be made to the language properties a-d mentioned earlier. It can be seen in item x.1.1.1 that "the type four times" (let us say that it represents property a) has a high sym-value. It can further be seen in item x.1.1.1 that "the type once" (let us say that it represents property b) has a low sym-value. It can be seen in item x.1.2 that "low number" (let us say that it represents property c) has a high sem-value. It can further be seen in item x.1.2 that "non-low number" (let us say that it represents property d) has a low sem-value. According to item x.2.1 high sym-value (= property a in our case) tends to associate with low sem-value (= property d). According to item x.2.2 low sym-value (= property b) tends to associate with high sem-value (= property c). From this it can be deduced: Where property a (= "the type four times") obtains, property d (= non-low number) also obtains. Where property b (= "the type once") obtains, property c (= low number) also obtains. The underscored clauses represent predictions. The latter are meant to impart the information that it is natural for "the type once" to occur in low numbers, and for "the type four times" to occur in non-low numbers. (No prediction is being made as to the placement of the cut-off point between low and non-low numbers.) A question to a linguist: "If you should be told that some language has direct objects in the accusative and genitive cases, and that that language uses one kind of direct objects in affirmative clauses, and the other kind in negative clauses, where would you place the accusatives, and where the genitives, on the basis of your experience with languages?" The linguist's reply: "I would predict accusative direct objects in affirmative clauses, and genitive direct objects in negative clauses."It is this type of predictions that the Slovenian Theory is able to make. Only the reader who has noticed the narrowness of the question, and understood its limitations, will find the contents of this volume of some use. If a generative theory generates accusative and genitive direct objects, and stipulates that one kind is used in negative clauses, and the other kind elsewhere, it can be left to the Slovenian Theory to distribute the two kinds of direct objects: the accusative object in affirmative clauses, the genitive objects in negative clauses. Thus the Slovenian Theory

13

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

(instead of the generative theory) finishes the generation of some structures, notably of those here called variants. (This is a Lilliputian step for linguistics, and a large one for me.) In the bulk of this book, the Slovenian Theory is illustrated with selected examples, each of which is dealt with in a deduction whose format is as follows: y. Language name(s). Brief description of the language data to be considered. The two variants: V1 (containing the properties a and b) and V 2 (containing the properties c and d). y.1. The assumptions of Naturalness Theory: y.1.1. states a naturalness scale applicable to the language data considered. The natu ralness scale mentions the properties a and c, and allots them a high and a low sem/symvalue, respectively. y.1.2. states another naturalness scale applicable to the language data considered. The naturalness scale mentions the properties b and d, and allots them a high and a low sem/ sym-value, respectively. y.2. The assumptions of the Slovenian Theory, concerning any two variants: y.2.1. states an assumption applicable to y.1.1-2. y.2.2. states another assumption applicable to y.1.1-2. (The complete list of the assumptions of the Slovenian Theory was adduced above.) y.3. The consequences: From y.1.1, y.1.2 and y.2.1 it can be deduced: y.3.1. If there is any difference between properties a and b, such that one of them combines with property c, and the other of them combines with property d, it is property a that tends to combine with property c. Q.E.D. From y.1.1, y.1.2 and y.2.2 it can be deduced: y.3.2. If there is any difference between properties a and b, such that one of them combines with property c, and the other of them combines with property d, it is property b that tends to combine with property d. Q.E.D. y.4. Note(s). (Optional.) The remainder of this chapter provides a commentary on the sundry items in the format of the deductions. In the examples, I have endeavoured to use language data which are as simple, clear, and varied as possible, in addition to being, hopefully, undisputed. I have often chosen cases already mentioned and/or discussed in the technical literature. What I have circumvented, although it would have been necessary, is to exploit the language data of some linguistic monograph or grammar systematically, so as to register any counterexamples to the Slovenian Theory. I believe that such a move would be premature, in the light of the circumstance that so many naturalness scales of potential help are not yet available. Regrettably, there is no telling at the present time whether a putative counterexample is a true one. In the present monograph, 250-60 examples taken from more than 50 languages and dialects are considered. The reader will be astonished to notice the almost complete absence of English cases. While there are many in my records, they could not be included in the present monograph, for reasons unrelated to linguistics. I hope to publish them later. I have tried to keep the number of my data sources relatively low in order to avoid the

14

INTRODUCTION ALSO THE SUMMARY OF THE MONOGRAPH

dangerous mingling of too many terminology systems. On the same account I seldom quote from my sources' references to even more basic works. In spite of these conscious efforts, my meta-theoretical apparatus is not fully consistent. In the present monograph, one of the crucial notions is that of (morphological, morphosyntactic, syntactic) variants. They almost always appear in pairs. In the early stages of our research it was assumed that the members of each pair contract the relation of (near-)synonymity. However, in the type of linguistics that I am familiar with, there is no exact measure of (near-)synonymity for any two variants (with some exceptions). Moreover, recent investigations have shown that the Slovenian Theory remains valid even if the notion of variant is broadened to subsume any (morphological, morphosyntactic, syntactic) units that belong to the same (morphological, morphosyntactic, syntactic) superordinate unit or category. Thus not only are the two English indefinite articles, a(n) and sm (e.g. a book, sm milk ), variants, but also the English definite and indefinite articles can constitute a pair of variants (because they are subsumed under "article"). As this notion of variant is somewhat indeterminate, I expressly enumerate, in each deduction, the two (seldom more than two) variants considered. (Morphosyntactic and syntactic variants are simplistically classed together as "syntactic variants" in the remainder of the book.) The present monograph is not an introduction to Naturalness Theory. Concerning its principles and other teachings, the reader is referred to the basic works listed at the beginning of this chapter. (An understanding of written German, besides English, is essential.) Naturalness Theory cannot be furthered in my scholarly environment, for lack of pertinent resources (though this may change, given the rich data which will slowly become available on the internet, see Plank & E. Filimonova 2000). But I have ventured to suggest a new naturalness scale here and there. My guesses, even if found plausible, deserve the reader's scepticism, in spite of any argumentation that I adduce in support of such attempts. I am, however, offering two kinds of scales which seem to me to be rooted in the logic of language relations, rather than in the language dataalthough I do believe that the suggested scales are compatible with the data: (a) >sem (x, y), where y does not constitute a "natural class." In such a case, y is less sem-natural than x. Cf. deductions 46, 210 and 213. (b) >sem (A + B, A), i.e. any phenomenon is more sem-natural than just a part of that phenomenon. Cf. deductions 33, 62 and 93. The fact that a given scale is either of the form >sem (x, y) (where y is not a "natural class") or of the form >sem (A + B, A) will be marshalled in this volume as support for the plausibility of that scale. I am not exactly proud of these two scale types, for they increase the power of Naturalness Theory considerably, and so are a step in the wrong direction. However, quite a few language phenomena would remain outside the scope of the theory if these scale types were not postulated. Rather, they are a necessary evil. Any scale of the format >sem (A + B, A) in combination with another scale reflects the well-known observation that what is more sem-natural (less marked) is more varied. The scale format >sem (A + B, A) has a systematic weakness, which becomes apparent if the scale format is rewritten as >sem (A + B, X), where X can be either A or B; hence the choice between A and B (in X) is arbitrary. Thus no predictions can be based on the Xpart of the scale format. Therefore, the scales of the scale format >sem (A + B, A) can

15

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

only be used in the consequence section of each deduction in such a way that they are fully presupposed, i.e. they must precede the first mention of the words "such as." In some cases, this results in awkward formulations of the consequence sections. Regrettably, a scale of the type >sem (A + B, A) can also contain nonsense, e.g. >sem (perfective and imperfective aspects, perfective aspect) / noun phrase. On the whole it is not likely that noun phrases would show aspectual distinctions; under such a presup position, the example should not pass as well-formed. However, this is hardly a specific problem of my monograph: the difficulty has not been disposed of in any language theory, to the best of my knowledge. I do hope that the format >sem (A + B, A) can soon be replaced by the significantly less powerful and more manageable format >sem (+A and -A, +A or -A). Indeed most of my examples conform to the latter formatbut not all. A theoretically possible instance of >sem (A + B, A) is this: >sem (singular & plural & dual, singular & plural) / noun phrase, i.e. a noun phrase that distinguishes singular, plural and dual is more sem-natural than a noun phrase that distinguishes just singular and plural. The scale is not meant to imply that noun phrases which distinguish singular, plural and dual are more common cross-linguistically than noun phrases which distinguish just singular and plural. What the scale does imply is that, in a language whose noun phrases partly distinguish singular, plural and dual, and partly just singular and plural, the noun phrases of the former kind are more sem-natural than the noun phrases of the latter kind. For instance, the noun phrases of the former kind would comprise personal pronouns, and the noun phrases of the latter kind would comprise all the remaining noun phrases of the language.Notice that a language phenomenon which is common cross-linguistically can be assigned a high sem-value. But the fact that a phenomenon has a high sem-value does not automatically warrant the reverse conclusion that that phenomenon is common cross-linguistically. As is well known, numerous naturalness scales are due to Mayerthaler, therefore his work is quoted time and again in the present monograph. Even when I could have referred the reader to another author, I normally opted for Mayerthaler's version ceteris paribus, for consistency. My records contain about 460 deductions at the time of this writing. In my collection, a number of naturalness scales are employed with some persistence. For instance, the transparency scale is used 229 times and the frequency scale 74 times. If anything, this indicates that the tendency towards an upper limit on the quantity of the scales is a realistic expectation. Furthermore, taking into consideration the fact that the transparency scale is the most often utilized, and with success, it can be surmised that the language faculty strongly favours parameters that are simple indeed. (The remaining scales each appear about 20 times at best, and their contents are much less general than in the more common scales just mentioned.) Certain naturalness scales present real or apparent difficulties: (a) A notorious case of conflicting naturalness "laws" are the existing scales of gram matical person. At least two traditions can be distinguished. One begins with Jakobson 1932, if not earlier, and was elaborated especially in Greenberg's work: >sem (third, first, second) / grammatical person. The other tradition is usually ascribed to Bhler (for instance, Bhler 1934): >sem (first, second, third) / grammatical person. (The formulation of the two scales just adduced has been "modernized" here.) From work with language material it emerges that Jakobson's scale can be of avail in most cases. The German present perfect and preterite tenses (see the chapter treating them) crucially involve the

16

INTRODUCTION ALSO THE SUMMARY OF THE MONOGRAPH

grammatical person, and they convincingly support the Jakobson-Greenberg tradition. In a conspicuously smaller number of cases, Bhler's version assures a "result." However, insofar as Bhler's scale finds application in my examples, it can be reduced to >sem (first, non-first) / grammatical person. The "non-first grammatical person" of this version can be interpreted as a case of non-"natural class:" the Jakobson-Greenberg tradition suggests that the first and the second persons constitute a "natural class;" if so, the third and the second persons of Bhler's scale cannot constitute a "natural class." Hence Bhler's scale boils down to an instance of the new kind of scales that I have suggested above, and is no true alternative to the Jakobson-Greenberg tradition.The coordination of two grammatical persons of the type 'you and X' (yielding the second-person anaphora) can be handled according to Jakobson-Greenberg, see deduction 252. I hold a comparable view of the scale >sem (second, non-second) / grammatical person, suitable for the subsystem of appeal, notably in connection with the imperative and with the pronouns of address (e.g. German du vs. Sie). The non-second person does not constitute a "natural class," and is therefore less sem-natural than the second person. This scale can be applied to a small number of cases only, like the Bhler scale, and unlike the Jakobson-Greenberg scale. It does not seem necessary to postulate some fourth variant of the grammatical-person scale, although I invite the reader to consult Croft 1990 passim, e.g. 111-3, with references. (b) Ferguson 1996, 243 has drawn attention to the paradigms of the Old English and Old Swedish present indicatives. Old English distinguishes the grammatical persons in the singular only, Old Swedish in the plural only. Both situations can be accounted for by the Slovenian Theory, see deductions 20 and 222. The two examples show that some naturalness scales contradict each other, which is understandable in the light of the familiar assumption that our brain works with partially conflicting parameters. Occasionally it is possible to treat the same language data in two different deductions, and yet achieve the same predictions (based, however, on unequal assumptions!). This is a realistic situation: not all speakers of a language have internalized the same grammar. It is possible that the alignment rules of the Slovenian Theory can be simplified, as suggested by the following consideration. It follows from the pair of scales >sem (a, b) and >sym (c, d) that a tends to associate with d, and b tends to associate with c. The same result is obtained in the pair of scales >sem (a, b) and >sem (d, c): a tends to associate with d, and b tends to associate with c. This leads to the equation: >sem (a, b) = >sym (b, a). The equation makes it possible to transform any >sem-scale into a corresponding >sym-scale, and v.v. Thus it suffices to postulate only sem- or only sym-values. Consequently the distinction between sem- and sym-values is not necessary, and we can "return" to (bare) markedness. Markedness (m) is restored with the aid of the following equations: >sem (a, b) = <m (a, b), i.e. a is less marked than b; >sym (a, b) = <m (b, a), i.e. b is less marked than a. The only remaining alignment rules are, >m tends to associate with at least one another >m; <m tends to associate with at least one another <m. This has actually been suggestedindependentlyby Andersen 2001 (following Andersen 1972 and even earlier Mare 1952), where the phenomenon covered by the two alignment rules is called markedness agreement. The simplification just described is not implemented in the continuation supported as it is only with data about morphological and (morpho)syntactic variants, which might be too weak a basis for such a far-reaching step.

17

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

Most deductions yield two consequences each. However, in a minority of deductions only one consequence can be stated. The latter kind are here called asymmetrical. The reason for the existence of asymmetrical deductions is to be sought in the circumstance that naturalness scales of a certain type, or of certain types, have so far not been formulated, to the best of my knowledge. However, the division of the deductions into symmetrical and asymmetrical ones is of no theoretical importance. Contrariwise, crucial importance is attached to the contents of the consequences (regardless of their number) stated in the deductions. Each consequence assumes the shape of a conditional. Its protasis enumerates the conditions under which the apodosis is valid. The apodosis predicts the presence or absence of some language property in one of the variants considered. Predicting X amounts to asserting that X cannot be otherwise. Hence in the apodosis, the behaviour of that language property is accounted for; the behaviour of that language property is explained (synchronically speaking, of course). The success of each prediction depends on the reliability of the assumptions on which that prediction rests.Historical explanations are not the subject-matter of the present volume. This approach guarantees that the synchronic explanations of data are couched within a theoryin contradistinction to the practice in many grammars, which allow pretheoretical explanations. The dilemma will be especially acute in the coming years, with the publication of grammars based on corpus analysis (an outstanding specimen already on the market is Biber et al. 1999) and tempted to pour life into their statistical data by trying to explain them. To be sure, any explanations must take many aspects of the data into ac count. Suffice it to mention that the sundry registers require widely divergent registerinternal explanations, as well as inter-register explanations. Non-linguistic factors have to be considered. Therefore a unitary theory of synchronic explanations is out of the question, at least at the present state of the art. However, some segments of the data, notably the register-independent, are susceptible to synchronic explanations incorporated within a theory such as ours. Although the reader may find my monograph convincing in limited ways only, it does suggest that naturalness scales are worthy of producing and exploiting. The following deduction z applying an example taken from real lifeand meant as entertainmentillustrates my conviction (which I do not propose to elaborate on) that the combination of naturalness scales with the tenets of the Slovenian Theory can even predict aspects of non-linguistic behaviour: z. Dog's reaction. When our dog Zen meets any person carrying a conspicuously large burden, he barks at that person. The two variants: person carrying (not carrying) a conspicuously large burden. z.1. The assumptions of Naturalness Theory: z.1.1. >sym (more transparent, less transparent) / unit I.e. with respect to coding properties, a unit of greater transparency is more natural than a corresponding unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of z.1.1: z.1.1.1. >sym (person carrying a large burden, person not carrying a large burden) I.e. with respect to coding properties, a person carrying a large burden is more natural than a person not carrying a large burden. z.1.1.2. >sym (dog barking, dog not barking) I.e. with respect to coding properties, a barking dog is more natural than a dog not barking. z.2. The assumptions of the Slovenian Theory, concerning any two variants: 18

INTRODUCTION ALSO THE SUMMARY OF THE MONOGRAPH

z.2.1. >sym tends to associate with another >sym z.2.2. <sym tends to associate with another <sym z.3. The consequences: From z.1.1.1-2 and z.2.1 it can be deduced: z.3.1. If there is any difference between a person carrying a large burden and a person not carrying a large burden, such that in one case our dog barks at the person, and in the other case our dog does not bark at the person, it is the person carrying a large burden that tends to get barked at by our dog. Q.E.D. From z.1.1.1-2 and z.2.2 it can be deduced: z.3.2. If there is any difference between a person carrying a large burden and a person not carrying a large burden, such that in one case our dog barks at the person, and in the other case our dog does not bark at the person, it is the person not carrying a large burden that tends not to get barked at by our dog. Q.E.D.

19

Examples
1. Albanian. Pronominal possessor, first and second person, alienable and inalienable possession. In inalienable possession, the possessor precedes the possessum, and there is no definite article, e.g. im at 'my father.' In alienable possession, the possessor follows the possessum, and the latter attaches the definite article, e.g. tren-i im 'my train.' (Lyons 1999, 129.) The present deduction is continued in deductions 2 and 3. The two syntactic variants: the type im at , and the type tren-i im. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type tren-i im, the type im at) / in Albanian I.e. with respect to coding properties, the type tren-i im is more natural than the type im at , in Albanian. 1.2. >sym (+postposed, -postposed) / adjective in Albanian I.e. with respect to coding properties, a postposed adjective is more natural than a preposed adjective, in Albanian.In this respect, Albanian resembles the Romance langu ages. 1.3. >sem (-alienable, +alienable) / possession I.e. with respect to semantic complexity, inalienable possession is more natural than alienable possession. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 2.3. >sym tends to associate with <sem 2.4. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2, 1.3, 2.1 and 2.3.it can be deduced: 3.1. If there is any difference between the type im at and the type tren-i im, such that one expresses alienable possession, and the other expresses inalienable possession, and such that in one the possessor precedes the possessum, and in the other the possessor follows the possessum, it is the type tren-i im that tends to express alienable possession, and its possessor tends to follow the possessum. Q.E.D. From 1.1.1, 1.2, 1.3, 2.2 and 2.4 it can be deduced: 21

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.2. If there is any difference between the type im at and the type tren-i im, such that one expresses alienable possession, and the other expresses inalienable possession, and such that in one the possessor precedes the possessum, and in the other the possessor follows the possessum, it is the type im at that tends to express inalienable possession, and its possessor tends to precede the possessum. Q.E.D. 2. Albanian. Pronominal possessor, third person, alienable and inalienable possession. In inalienable possession, a particle precedes the possessum, to which the definite article attaches, e.g. e m-a 'his/her/their mother;' the possessor is not expressed. In alienable possession, the possessor is followed by a particle and by the clitic dative of the possessor, e.g. shoq-ja e tij 'his female colleague.' (Lyons 1999, 129; supplemented with Snoj 1991, 53 ff.) The present deduction continues deduction 1, and is continued in deduction 3. The two syntactic variants: the type e m-a, and the type shoq-ja e tij . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type shoq-ja e tij , the type e m-a) / in Albanian I.e. with respect to coding properties, the type shoq-ja e tij is more natural than the type e m-a, in Albanian. 1.2. >sem (-alienable, +alienable) / possession I.e. with respect to semantic complexity, inalienable possession is more natural than alienable possession. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type shoq-ja e tij and the type e m-a, such that one type expresses alienable possession, and the other type expresses inalienable possession, it is the type shoq-ja e tij that tends to express alienable possession. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type shoq-ja e tij and the type e m-a, such that one type expresses alienable possession, and the other type expresses inalienable possession, it is the type e m-a that tends to express inalienable possession. Q.E.D. 3. Albanian. Pronominal possessor, third person, alienable and inalienable possession. In inalienable possession, the same form is used for the singular and for the plural, e.g. e m-a 'his/her/their mother;' the possessor is not expressed. In alienable possession, a separate form is used for the masculine and feminine singular, and a separate for the plural, thus shoq-ja e tij 'his female colleague,' shoq-ja e saj 'her female colleague,' shoqja e tyre 'their female colleague;' the possessor is expressed. (Lyons 1999, 129; supplemented with Snoj 1991, 53 ff.) The present deduction continues deductions 1 and 2. The two syntactic variants: the type e m-a, and the type shoq-ja e tij . 1. The assumptions of Naturalness Theory:

22

EXAMPLES

1.1. >sem ('his/her/their;' 'his,' 'her,' 'their') / pronominal possessor in Albanian I.e. with respect to semantic complexity, the expression of 'his,' 'her' and 'their' with one form is more natural than the expression of these three separately, in Albanian.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 1.2. >sem (-alienable, +alienable) / possession I.e. with respect to semantic complexity, inalienable possession is more natural than alienable possession. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the type e m-a and the type shoq-ja e tij , such that one expresses three meanings of the possessor, and the other expresses just one meaning of the possessor, and such that one expresses alienable possession, and the other expresses inalienable possession, it is the type e m-a that tends to express inalienable possession and several meanings of the possessor. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the type e m-a and the type shoq-ja e tij , such that one expresses three meanings of the possessor, and the other expresses just one meaning of the possessor, and such that one expresses alienable possession, and the other expresses inalienable possession, it is the type shoq-ja e tij that tends to express alienable possession and just one meaning of the possessor. Q.E.D. 4. Albanian. Proper nouns normally occur in definite form, e.g. Agim-i 'Agim (man's name).' They occur without the mark of definiteness (a) after certain, mostly locative, prepositions, e.g. n Tiran 'in Tirana,' (b) in apposition to a definite noun phrase, e.g. n fshatin Dushk 'in the village of Dushk,' (c) in vocative use, e.g. Lumt! 'Oh, Lumta.' (Lyons 1999, 121-2.) The present deduction deals with cases (a-b). Case (c) is considered in deduction 5. The two syntactic variants: proper name in definite form, and proper name without definite form. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type n Tiran, the type Tirana) / in Albanian I.e. with respect to coding properties, the type n Tiran is more natural than the type Tirana, in Albanian. 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than -presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite)

23

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the proper name and the preposition + proper name, such that one form is +definite, and the other -definite, it is the preposition + proper name that tends to lack the definite form. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the proper name and the preposition + proper name, such that one form is +definite, and the other -definite, it is the proper name without the preposition that tends to show the definite form. Q.E.D. 4. Note. The present deduction also accounts for case (b), mutatis mutandis. 5. Albanian. Proper nouns normally occur in definite form, e.g. Agim-i 'Agim (man's name).' They occur without the mark of definiteness (a) after certain, mostly locative, prepositions, e.g. n Tiran 'in Tirana,' (b) in apposition to a definite noun phrase, e.g. n fshatin Dushk 'in the village of Dushk,' (c) in vocative use, e.g. Lumt! 'Oh, Lumta.' (Lyons 1999, 121-2.) The present deduction deals with case (c). Cases (a-b) are considered in deduction 4. The two syntactic variants: person name having and lacking the definite article. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (+definite article, -definite article) / person name I.e. with respect to coding properties, a person name having the definite article is more natural than a person name lacking the definite article. 1.2. >sem (+vocative, -vocative) / person name I.e. with respect to semantic complexity, the vocative use of person names is more natural than other uses of person names.The vocative use of person names is primordial. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between a person name having the definite article and a person name lacking the definite article, such that one of them is used as vocative, and the other not, it is the person name having the definite article that tends to lack the vocative use. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between a person name having the definite article and a person name lacking the definite article, such that one of them is used as vocative, and the

24

EXAMPLES

other not, it is the person name lacking the definite article that tends to show the vocative use. Q.E.D. 6. Arabic. The first persons of the verb do not distinguish gender, whereas the remaining persons do distinguish gender. (Grande 1972, 391.) The two syntactic variants: the first persons, and the remaining persons of the verb. 1. The assumptions of Naturalness Theory: 1.1. >sem (genders conflated, genders separated) / the verbal persons of Arabic I.e. with respect to semantic complexity, the conflation of genders in a verbal person is more natural than the separation of genders in a verbal person, in Arabic.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 1.2. >sem (+first, -first) / grammatical person in the verb of Arabic I.e. with respect to semantic complexity, the first person is more natural than the nonfirst persons, in the verb of Arabic.The non-first persons do not constitute a "natural class," therefore their sem-naturalness is low. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the first and the non-first verbal persons, such that one kind of persons distinguishes genders, and the other kind does not, it is the first persons that tend not to distinguish genders. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the first and the non-first verbal persons, such that one kind of persons distinguishes genders, and the other kind does not, it is the nonfirst persons that tend to distinguish genders. Q.E.D. 4. Note. In the personal pronouns of Arabic, masculine and feminine are distinguished, except in the first person (Lyons 1999, 137). The present deduction applies, mutatis mutandis. 7. Arabic, Egyptian. The pronominal possessives are realized as affixes on the head noun = the possessum. The affixes are phonologically identical to personal inflections of the verb representing the object argument of the verb. E.g. arabiyyit-na 'our car' and fihmuu-na 'they understand us.' (Lyons 1999, 126.) The two syntactic variants: affix as object argument of the verb, and affix as possessive pronominal on the noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (verb, noun) I.e. with respect to semantic complexity, the verb is more natural than the noun.The morphology of the verb is mostly better developed than the morphology of the noun. Such a state of affairs is expected with the less marked (the verb) and the corresponding more marked (the noun) entities. 1.2. >sem (accusative, genitive) / in nom.-acc. languages I.e. with respect to semantic complexity, the accusative is more natural than the genitive, in nominative-accusative languages. (Mayerthaler et al. 1998, 167.) A special case of 1.2: 1.2.1. >sem (accusative, genitive) / affix as argument in nom.-acc. languages

25

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to semantic complexity, the accusative affix is more natural than the genitive affix, as argument in nominative-accusative languages. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes the accusative affix, and the other part of speech takes the genitive affix, it is the verb that tends to take the accusative affix. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes the accusative affix, and the other part of speech takes the genitive affix, it is the noun that tends to take the genitive affix. Q.E.D. 4. Note. Cf. deductions 164 and 231. 8. Arabic. The definite article is a prefix, the indefinite article is a suffix, e.g. alxdim-u 'the servant,' xdim-u-n 'a servant.' (Lyons 1999, 93.) The two syntactic variants: the definite and the indefinite articles. 1. The assumptions of Naturalness Theory: 1.1. >sym (suffix, prefix) I.e. with respect to coding properties, a suffix is more natural than a prefix. (In the spirit of Mayerthaler et al. 1998, 34.) 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than -presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the definite and indefinite articles, such that one is a prefix and the other a suffix, it is the indefinite article that tends to be a suffix. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the definite and indefinite articles, such that one is a prefix and the other a suffix, it is the definite article that tends to be a prefix. Q.E.D. 9. Arabic. The dual of nouns is formed with a suffix, the plural of nouns is often "broken," e.g. walad 'boy,' dual walad-ayn, plural awld. (Grande 1972, 387-8. The examples taken from Iraqi Arabic, Cruse 1999, 269.) The two morphological variants: the noun in the dual, and the noun in the plural. 1. The assumptions of Naturalness Theory: 1.1. >sym (suffixal formation, "broken" formation) / of grammatical number in Arabic I.e. with respect to coding properties, suffixal formation is more natural than "broken"

26

EXAMPLES

formation of grammatical number, in Arabic. (Cf. Mayerthaler et al. 1998, 273.) 1.2. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the formation of the dual and the plural of nouns, such that one formation is suffixal and the other "broken," it is the dual that tends to show the suffixal formation. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference in the formation of the dual and the plural of nouns, such that one formation is suffixal and the other "broken," it is the plural that tends to show the "broken" formation. Q.E.D. 10. Arabic, Iraqi. The numeral ithnayn 'two' is excluded before duals, although higher numerals take a plural form. (Cruse 1999, 269.) The two syntactic variants: the dual without the overt numeral 'two,' and the plural with overt numerals. 1. The assumptions of Naturalness Theory: 1.1. >sem (-redundant, +redundant) / cardinal numeral in expressions of exact quantity I.e. with respect to semantic complexity, non-redundant cardinal numeral is more natural than redundant cardinal numeral, in expressions of exact quantity.This is based on the circumstance that cardinal numerals are the prototypical expressions of exact quantity. A special case of 1.1: 1.1.1. >sem (present, absent) / cardinal numeral in expressions of exact quantity, in Iraqi Arabic I.e. with respect to semantic complexity, a present cardinal numeral is more natural than a lacking cardinal numeral, in expressions of exact quantity, in Iraqi Arabic. 1.2. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the dual and the plural of nouns concerning the presence or absence of the corresponding cardinal numerals, it is the plural that tends to allow the presence of the cardinal numeral. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the dual and the plural of nouns concerning the presence or absence of the corresponding cardinal numerals, it is the dual that tends to

27

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

disallow the presence of the cardinal numeral. Q.E.D. 11. Arabic, Iraqi. Adjectives agree with their head nouns, in respect of number. However, the construction adjective + human head noun is treated differently from the construction adjective + non-human head noun. Human duals are treated as plurals for the purposes of concord: walad jaml 'a handsome boy,' waladayn jamln 'two handsome boys,' awld jamln '(more than two) handsome boys.' Plural and dual non-human nouns are treated like feminine singular nouns for the purposes of concord: al-bayt al-jaml 'the beautiful house,' al-baytayn al-jamla 'the two beautiful houses,' al-byt al-jamla 'the (more than two) beautiful houses.' (Cruse 1999, 268.) These phenomena are dealt with in the present deduction and in deduction 12. The two syntactic variants: part-of-speech units which have, outside the singular, plural and dual forms, and part-of-speech units which have, outside the singular, plural forms only. 1. The assumptions of Naturalness Theory: 1.1. >sem (noun, adjective) I.e. with respect to semantic complexity, a noun is more natural than an adjective. The adjective is not a universal category (Mayerthaler et al. 1998, 24). 1.2. >sem (plural and dual forms, plural forms only) / outside the singular of nouns and adjectives, in Iraqi Arabic I.e. with respect to semantic complexity, nouns and adjectives that have, outside the singular, plural and dual forms are more natural than nouns and adjectives that have, outside the singular, plural forms only, in Iraqi Arabic.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between part-of-speech units which have, outside the singular, plural and dual forms, and part-of-speech units which have, outside the singular, plural forms only, such that one class of units comprises human head nouns, and the other class of units comprises adjectives dependent on those human head nouns, it is the part-ofspeech units which have, outside the singular, plural and dual forms that tend to be human head nouns. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between part-of-speech units which have, outside the singular, plural and dual forms, and part-of-speech units which have, outside the singular, plural forms only, such that one class of units comprises human head nouns, and the other class of units comprises adjectives dependent on those human head nouns, it is the part-ofspeech units which have, outside the singular, plural forms only that tend to be adjectives dependent on human head nouns. Q.E.D. 12. Arabic, Iraqi. Adjectives agree with their head nouns, in respect of number. However, the construction adjective + human head noun is treated differently from the construction adjective + non-human head noun. Human duals are treated as plurals for the purposes of concord: walad jaml 'a handsome boy,' waladayn jamln 'two handsome boys,' awld jamln '(more than two) handsome boys.' Plural and dual non-human nouns

28

EXAMPLES

are treated like feminine singular nouns for the purposes of concord: al-bayt al-jaml 'the beautiful house,' al-baytayn al-jamla 'the two beautiful houses,' al-byt al-jamla 'the (more than two) beautiful houses.' (Cruse 1999, 268.) These phenomena are dealt with in the present deduction and in deduction 11. The two syntactic variants: part-of-speech units which have singular, plural and dual forms, and part-of-speech units which have singular forms only. 1. The assumptions of Naturalness Theory: 1.1. >sem (noun, adjective) I.e. with respect to semantic complexity, a noun is more natural than an adjective. The adjective is not a universal category (Mayerthaler et al. 1998, 24). 1.2. >sem (singular, plural and dual forms, singular forms only) / outside the singular of non-human nouns and adjectives, in Iraqi Arabic I.e. with respect to semantic complexity, nouns and adjectives that have singular, plural and dual forms are more natural than nouns and adjectives that have singular forms only, in Iraqi Arabic.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between part-of-speech units which have singular, plural and dual forms, and part-of-speech units which have singular forms only, such that one class of units comprises non-human head nouns, and the other class of units comprises adjectives dependent on those non-human head nouns, it is the part-of-speech units which have singular, plural and dual forms that tend to be non-human head nouns. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between part-of-speech units which have singular, plural and dual forms, and part-of-speech units which have singular forms only, such that one class of units comprises non-human head nouns, and the other class of units comprises adjectives dependent on those non-human head nouns, it is the part-of-speech units which have singular forms only that tend to be adjectives dependent on non-human head nouns. Q.E.D. 4. Note. This deduction does not account for the fact that that singular of the adjective which accompanies the non-singular head noun is of the feminine gender. 13. Arawak (a language family spoken in large parts of South America). Number marking is obligatory in nouns referring to people, but optional otherwise. (Cruse 1999, 269.) The two syntactic variants: +human noun, and -human noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 1.2. >sem (obligatory, optional) / number marking in nouns of Arawak I.e. with respect to semantic complexity, obligatory forms are more natural than optional forms, for number marking in nouns of Arawak.It is much more common crosslinguistically that number marking is obligatory than optional. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants:

29

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between human and non-human nouns, such that in one kind of nouns number marking is obligatory, and in the other kind of nouns number marking is optional, it is the human nouns that tend to display obligatory number marking. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between human and non-human nouns, such that in one kind of nouns number marking is obligatory, and in the other kind of nouns number marking is optional, it is the non-human nouns that tend to display optional number marking. Q.E.D. 14. Bahasa Indonesia. (a) An interrogative clause is recognized from its sentence intonation or from its question words. (b) Only if the finite verb is an auxiliary does the inverted order of the subject and that verb obtain. (Atmosumarto 1994, 259-60.) This deduction deals with case (b). Case (a) is considered in deduction 14a. The two syntactic variants: declarative clause, and interrogative clause characterized by the element order verb - subject. 1. The assumptions of Naturalness Theory: 1.1. >sem (declarative, interrogative) / sentential mode I.e. with respect to semantic complexity, the declarative sentential mode is more natural than the interrogative sentential mode. (Mayerthaler et al. 1998, 326.) 1.2. >sem (SV, VS) / in SVO-languages I.e. with respect to semantic complexity, the element order subject - verb is more natural than the element order verb - subject, in SVO-languages.This follows from the basicness of SVO in SVO-languages (Mayerthaler et al. 1998, 349). Indonesian languages are SVO (S. Steele 1978, 590). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between declarative and interrogative clauses, such that one kind of clauses has the element order SV, and the other kind of clauses has the element order VS, it is the declarative clauses that tend to have the element order SV. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between declarative and interrogative clauses, such that one kind of clauses has the element order SV, and the other kind of clauses has the element order VS, it is the interrogative clauses that tend to have the element order VS. Q.E.D. 14a. Bahasa Indonesia. (a) An interrogative clause is recognized from its sentence intonation or from its question words. (b) Only if the finite verb is an auxiliary does the inverted order of the subject and that verb obtain. (Atmosumarto 1994, 259-60.) This deduction deals with case (a). Case (b) is considered in deduction 14. The two syntactic variants: interrogative clause characterized by its sentence intonation or by its question word, and interrogative clause characterized by the element 30

EXAMPLES

order verb - subject. 1. The assumptions of Naturalness Theory: 1.1. >sem (VS, SV) / in interrogative clauses of Bahasa Indonesia I.e. with respect to semantic complexity, the element order verb - subject is more natural than the element order subject - verb, in interrogative clauses of Bahasa Indonesia.This follows from deduction 14. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem (auxiliaries, most other verbal lexemes) I.e. with respect to semantic complexity, the auxiliaries are more natural than most other verbal lexemes. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between auxiliaries and other verbs, such that one kind is characterized, under interrogation, by the element order SV, and the other kind by the element order VS, it is the "other" verbs that tend to show the element order SV. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between auxiliaries and other verbs, such that one kind is characterized, under interrogation, by the element order SV, and the other kind by the element order VS, it is the auxiliaries that tend to show the element order VS. Q.E.D. 15. Bambara (a Congo-Cordofanian language or dialect spoken especially in Mali). The past tense is expressed with an auxiliary ( ye) with transitive verbs but with a suffix (la) with intransitive verbs. (Schachter 1985, 10-1.) The two syntactic variants: the past tense of transitive verbs, and the past tense of intransitive verbs. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (transitivity, intransitivity) I.e. with respect to coding properties, transitivity is more natural than intransitivity. 1.1.2. >sym (auxiliary, suffix) / as means of expressing past tense in Bambara I.e. with respect to coding properties, an auxiliary is more natural than a suffix, as a means of expressing the past tense in Bambara. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with another <sym 2.2. >sym tends to associate with another >sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the past tense of transitive and intransitive

31

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

verbs, such that one kind of verbs uses an auxiliary, and the other kind of verbs uses a suffix, it is the intransitive verbs that tend to use a suffix. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the past tense of transitive and intransitive verbs, such that one kind of verbs uses an auxiliary, and the other kind of verbs uses a suffix, it is the transitive verbs that tend to use an auxiliary. Q.E.D. 16. Breton. There is subject agreement, except in affirmative clauses, where the verb assumes the default value = the third person singular. E.g. an dud ne lavarent netra met a chome sioul 'the people said nothing but stayed quiet.' (The first verb in the third person plural, the other verb in the third person singular.) (J. B. Johannessen 1998, 42, referring to Borsley & Stephens and to a personal communication by Frederik Otto Lindeman.) The two syntactic variants: the verb in affirmative clauses, and the verb in nonaffirmative clauses. 1. The assumptions of Naturalness Theory: 1.1. >sem (+affirmative, -affirmative) / clause I.e. with respect to semantic complexity, the affirmative clause is more natural than the non-affirmative clause. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (full inflection, default form) / verb I.e. with respect to coding properties, a verb showing full inflection is more natural than a verb showing the default form only.Full inflection is easier for the hearer to process than the default form. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the verb in affirmative clauses and the verb in non-affirmative clauses, such that in one kind the verb has full inflection, and in the other kind the verb takes the default form, it is the verb in affirmative clauses that tends to take the default form. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the verb in affirmative clauses and the verb in non-affirmative clauses, such that in one kind the verb has full inflection, and in the other kind the verb takes the default form, it is the verb in non-affirmative clauses that tends to have full inflection. Q.E.D. 17. Carib (an Amerindian language family in the Guayanas). The notion 'I and you' is expressed with the dual: kxko . The notion 'I and he' is expressed with the exclusive plural: ana. (Lyons 1999, 136.) The two syntactic variants: 'I and you,' and 'I and he.' 1. The assumptions of Naturalness Theory:

32

EXAMPLES

1.1. >sem ('I and he,' 'I and you') I.e. with respect to semantic complexity, 'I and he' is more natural than 'I and you.' This is based on the admittedly slim evidence of Menomini: the pronoun for the exclusive plural ('I and he') is derived from the pronoun [nenah] 'I;' the pronoun for the inclusive plural is derived from the pronoun [kenah] 'thou.' In both cases, the change to the plural consists in the replacement of the word-final [h] with the glottal stop. (Data from Bloomfield 1933, 256.) The presupposition is that the pronoun 'I' is more sem-natural than the pronoun 'thou.' 1.2. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between 'I and he' and 'I and you,' such that one is expressed with the plural, and the other with the dual, it is 'I and he' that tends to be expressed with the plural. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between 'I and he' and 'I and you,' such that one is expressed with the plural, and the other with the dual, it is 'I and you' that tends to be expressed with the dual. Q.E.D. 18. Carib (an Amerindian language family in the Guayanas). Possession is indicated with personal prefixes on the noun. If the noun has an extended form beside a nonextended one, the personal affixes attach to the extended form. E.g. 'stone:' non-extended form to:pu; extended form to:puru; with personal affix a-to:puru 'your stone.' (Lyons 1999, 126.) The two morphological variants: the extended and the non-extended form of the same noun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (+extended, -extended) / word in Carib I.e. with respect to coding properties, an extended word is more natural than a corresponding non-extended word, in Carib. 1.1.2. >sym (noun with affix, noun without affix) I.e. with respect to coding properties, a noun with affix attached is more natural than the same noun without an affix. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced:

33

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.1. If there is any difference between an extended and a non-extended form of one and the same noun, such that one form attaches personal affixes, and the other not, it is the extended form that tends to attach personal affixes. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between an extended and a non-extended form of one and the same noun, such that one form attaches personal affixes, and the other not, it is the non-extended form that tends not to attach personal affixes. Q.E.D. 19. Catalan. Personal names are used with the definite article, e.g. la Maria. Names of men use l' before any vowel, e.g. l'Eduard, and en elsewhere, e.g. en Joan. (Hualde 1992, 282.) The two syntactic variants: personal names having two variants of the definite article, and personal names having only one variant of the definite article. 1. The assumptions of Naturalness Theory: 1.1. >sem (two variants, one variant) / of the definite article, with Catalan personal names I.e. with respect to semantic complexity, personal names admitting two variants of the definite article are more natural than personal names admitting only one variant of the definite article, in Catalan.The scale has the format >sem (A + B, A). 1.2. >sem (masculine, feminine) I.e. with respect to semantic complexity, masculine is more natural than feminine. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between personal names having two variants of the definite article and personal names having only one variant of the definite article, such that one kind of personal names is masculine, and the other kind of personal names is feminine, it is the personal names having two variants of the definite article that tend to be masculine. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between personal names having two variants of the definite article and personal names having only one variant of the definite article, such that one kind of personal names is masculine, and the other kind of personal names is feminine, it is the personal names having only one variant of the definite article that tend to be feminine. Q.E.D. 20. Chinese, Mandarin. With unaccusative verbs, the position of the subject is VS, and the subject is interpreted as -definite, e.g. li-le rn le come person 'a person came.' If the subject is moved into the SV-position, it is interpreted as +definite, e.g. rn li le person come 'the person came.' (Lyons 1999, 88-9, with data from Li & S. Thompson.) The two syntactic variants: VS and SV of unaccusative verbs. 1. The assumptions of Naturalness Theory: 1.1>sem (SV, VS) / in SVO-languages I.e. with respect to semantic complexity, the element order subject - verb is more natural than the element order verb - subject, in SVO-languages.This follows from the basicness of SVO in SVO-languages (Mayerthaler et al. 1998, 349). 1.2. >sem (+presupposed, -presupposed) 34

EXAMPLES

I.e. with respect to semantic complexity, +presupposed is more natural than presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the element orders SV and VS of unaccusative verbs, such that under one order the subject is interpreted as +definite, and under the other order as -definite, it is under VS that the subject is interpreted as -definite. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the element orders SV and VS of unaccusative verbs, such that under one order the subject is interpreted as +definite, and under the other order as -definite, it is under SV that the subject is interpreted as +definite. Q.E.D. 4. Note. Similarly in Cantonese (Lyons 1999, 88-9, with data from Matthews & V. Yip). 21. Chinese, Mandarin. Pronouns have a number distinction, common nouns do not. E.g. t 'he, she, it,' tmen 'they,' vs. sh 'book(s).' (Croft 1990, 111, referring to Li & S. Thompson.) The two syntactic variants: two number forms and one number forms. 1. The assumptions of Naturalness Theory: 1.1. >sem (two number forms, one number form only) I.e. with respect to semantic complexity, two number forms are more natural than one number form only.The scale has the format >sem (A + B, A). 1.2. >sem (personal pronoun, noun) I.e. with respect to semantic complexity, a personal pronoun is more natural than a noun. (In the spirit of Croft 1990, 112-3 [with references] on the animacy hierarchy.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between two number forms and one number forms, such that one kind of forms are common nouns, and the other kind of forms are pronouns, it is the two number forms that tend to be pronouns. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between two number forms and one number forms, such that one kind of forms are common nouns, and the other kind of forms are pronouns, it is the one number forms that tend to be common nouns. Q.E.D. 22. Chinese, Mandarin. The affirmative answer to a yes-no question is expressed by whatever verb occurred in the question. The negative answer is bu 'not' optionally followed by this verb. (Schachter 1985, 32.) The two syntactic variants: affirmative answer to yes-no question, and negative answer to yes-no question. 1. The assumptions of Naturalness Theory: 35

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.1. >sem (incomplete repetition of question, other means) / used in answering a yesno question I.e. with respect to semantic complexity, an incomplete repetition of the material of the question is more natural than other means, in answers to yes-no questions.This is based on the circumstance that answering a yes-no question by repeating some of the material of the question incompletely is more widespread than any other means of answering a yes-no question. Even languages that use 'yes' and 'no' as answers employ the said repetition as an alternative or enlarging means of answering, e.g. Are you hungry? (Yes,) I am. 1.2. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the positive answer to a yes-no question and the negative answer to such a question, such that one kind of answer is a partial repetition of the question, and the other kind of answer is something else, the affirmative answer tends to contain an incomplete repetition of the material in the question. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the positive answer to a yes-no question and the negative answer to such a question, such that one kind of answer is a partial repetition of the question, and the other kind of answer is something else, the negative answer tends to contain something else than an incomplete repetition of the material in the question. Q.E.D.Namely, the negative answer contains bu 'no.' 4. Note. Cf. deduction 23. 23. Chinese, Mandarin. The affirmative answer to a yes-no question is expressed by whatever verb occurred in the question. The negative answer is bu 'not' optionally followed by this verb. (Schachter 1985, 32.) The two syntactic variants: affirmative answer to yes-no question, and negative answer to yes-no question. 1. The assumptions of Naturalness Theory: 1.1. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit.The ability to expand syntactically involves the tendency to enhance transparency.

36

EXAMPLES

2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the positive and the negative answer to a yes-no question, such that one kind of answer can expand syntactically, and the other kind cannot, it is the positive answer that does not tend to expand syntactically. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the positive and the negative answer to a yes-no question, such that one kind of answer can expand expand syntactically, and the other kind cannot, it is the negative answer that tends to expand syntactically. Q.E.D.Namely, the negative answer contains bu 'no.' 4. Note. Cf. deduction 22. 24. Croatian. The auxiliary of the future tense is clitic if the clause is affirmative; nonclitic (accented) if the clause is not affirmative. E.g. ja u pisati 'I will write,' neu pisati 'I will not write,' hou li pisati 'will I write;' only the first example contains a clitic auxiliary. (Vaillant 1966, 109.) The two syntactic variants: the future tense with the clitic auxiliary, and the future tense with the non-clitic auxiliary. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (accented auxiliary + infinitive, clitic auxiliary + infinitive) / the future tense in Croatian I.e. with respect to coding properties, the accented auxiliary + infinitive is more natural than the corresponding clitic auxiliary + infinitive.All other things equal, an accented word is easier for the addressee to decode than the corresponding clitic. 1.2. >sem (affirmation, negation/interrogation) I.e. with respect to semantic complexity, affirmation is more natural than negation or interrogation.This is based on the assumption that negation and interrogation do not constitute a "natural class." Negation and interrogation are not even corresponding categories. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the form of the affirmed and of the non-affirmed future tense, such that the auxiliary is accented in one kind of future tense, and not accented in the other kind of future tense, it is the non-affirmed future tense that tends to contain an accented auxiliary. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced:

37

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.2. If there is any difference in the form of the affirmed and of the non-affirmed future tense, such that the auxiliary is accented in one kind of future tense, and not accented in the other kind of future tense, it is the affirmed form that tends to contain a clitic auxiliary. Q.E.D. 25. Croatian. There is a tendency to express future time with the (periphrastic) future tense in main clauses, and with the (non-periphrastic) present tense in associated dependent clauses, e.g. kad doe, vidjee 'when you come, you will see.' (Vaillant 1966, 93, 109.) The two syntactic variants: the periphrastic future tense and the non-periphrastic present tense as expressions of future time. 1. The assumptions of Naturalness Theory: 1.1. >sem (future tense, present tense) / as expression of future time in languages that have a future tense I.e. with respect to semantic complexity, the future tense is more natural than the present tense, as an expression of future time in languages that have a future tense. 1.2. >sem (main, dependent) / clause I.e. with respect to semantic complexity, a main clause is more natural than a dependent clause.I consider cleft sentences such as it was Ted who broke the news to me as prototypical combinations of a main and a dependent clauses. It can be seen from the example that it is the main clause that shows high sem-properties. Phylogenetically, main clauses are earlier than dependent clauses. 1.3. >sem (auxiliary, other verb) / as finite verb I.e. with respect to semantic complexity, an auxiliary is more natural than other verbs, as finite verb. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-3 and 2.1 it can be deduced: 3.1. If there is any difference between the use of the future tense and of the present tense as expressions of future time, such that one expression of future time is used in main clauses, and the other expression of future time is used in dependent clauses, and such that in one expression the finite verb is an auxiliary, and in the other expression the finite verb is a full verb, it is the future tense that tends to be used in main clauses and to contain an auxiliary as finite verb. Q.E.D. From 1.1-3 and 2.2 it can be deduced: 3.2. If there is any difference between the use of the future tense and of the present tense as expressions of future time, such that one expression of future time is used in main clauses, and the other expression of future time is used in dependent clauses, and such that in one expression the finite verb is an auxiliary, and in the other expression the finite verb is a full verb, it is the present tense that tends to be used in dependent clauses and to contain a full finite verb. Q.E.D. 4. Note. Similarly in English: if you come, you will see . 26. Danish. The definite article is a suffix if the noun phrase consists of the head noun only. Otherwise the definite article is mostly a word. E.g. mand-en 'the man,' den gode mand 'the good man.' In the noun phrase consisting of hele, selve, al(le), begge + head

38

EXAMPLES

noun the article is appended to the head noun, e.g. hele dag-en 'the whole day,' selve pave-n 'the Pope himself,' alle drenge-ne 'all the boys,' begge piger-ne 'both girls.' (Diderichsen 1957, 103.) The two syntactic variants: adjective + head noun, and (hele/selve/al(le)/begge) + head noun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (word, morpheme) I.e. with respect to coding properties, a word is more natural than a morpheme. 1.1.2. >sym (adjective + head noun, [hele/selve/al(le)/begge ] + head noun) / in Danish I.e. with respect to coding properties, adjective + head noun is more natural than a noun phrase consisting of just the head noun, perhaps preceded by hele/selve/al(le)/begge , in Danish. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the adjective + head noun and the head noun optionally preceded by hele/selve/al(le)/begge , such that the definite article is a word in one construction, and a suffix in the other construction, it is the definite article in adjective + head noun that tends to be expressed as a word. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the adjective + head noun and the head noun optionally preceded by hele/selve/al(le)/begge , such that the definite article is a word in one construction, and a suffix in the other construction, it is the definite article in bare head nouns (perhaps preceded by hele/selve/al(le)/begge ) that tends to be expressed as a suffix. Q.E.D. 27. Danish. The definite article is a suffix on the noun, e.g. mand-en 'the man.' (See note 4 for a qualification.) The indefinite article is a word, e.g. en mand 'a man.' (Diderichsen 1957, 53.) The two syntactic variants: the suffixal definite article, and the indefinite article. 1. The assumptions of Naturalness Theory: 1.1. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than presupposed. (Mayerthaler 1981, 14.) A special case of 1.1: 1.1.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency

39

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.2.1. >sym (word, morpheme) I.e. with respect to coding properties, a word is more natural than a morpheme. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the definite and indefinite articles, such that one is a suffix and the other a word, it is the definite article that tends to be a suffix. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the definite and indefinite articles, such that one is a suffix and the other a word, it is the indefinite article that tends to be a word. Q.E.D. 4. Note. The definite article of Danish can also be a word. See deduction 26. 28. Danish. The present tense as expression of future time is especially popular with perfective verbs. (Diderichsen 1957, 137.) The two syntactic variants: the (synthetic) present tense as expression of future time, and the (analytic) construction vil/skal + infinitive as expression of future time. 1. The assumptions of Naturalness Theory: 1.1. >sem (future tense, praesens pro futuro) / as expression of future time I.e. with respect to semantic complexity, the future tense is more natural than praesens pro futuro, as expression of future time.A unit (here: future tense) specialized for a category (here: future time) is more sem-natural than a corresponding unit (here: praesens pro futuro) not specialized for that category (here: future time). Similarly, the reflexive pronoun is more sem-natural than the corresponding non-reflexive pronoun as expression of reflexivity. 1.2. >sem (perfective & imperfective, perfective only) / verbal aspect I.e. with respect to semantic complexity, indifference as to the choice of the verbal aspect is more natural than using the perfective aspect only.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the use of the perfective & the imperfective verbal aspect and the exclusive use of the perfective verbal aspect, such that one use is found in the future tense, and the other use is found in the praesens pro futuro, it is the use of both the perfective and the imperfective aspect that tends to be found in the future tense. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the use of the perfective & the imperfective verbal aspect and the exclusive use of the perfective verbal aspect, such that one use is found in the future tense, and the other use is found in the praesens pro futuro, it is the exclusive use of the perfective aspect that tends to be found in the praesens pro futuro.

40

EXAMPLES

Q.E.D. 29. English, Old. In the conjugation of the present indicative of most verbs, the grammatical persons were distinguished in the singular only. (Ferguson 1996, 243). E.g. sg. bere, bir(e)st, bir(e), pl. bera 'bear' (Wright & E. Wright 1925, 253). My attention was drawn to this matter by John C. Smith viva voce, 2000. The two morphological variants: the singular present indicative, and the plural present indicative. 1. The assumptions of Naturalness Theory: 1.1. >sem (three desinences, one desinence) / per number in the present indicative of most Old English verbs I.e. with respect to semantic complexity, three desinences per number is more natural than one desinence per number, in the present indicative of most Old English verbs.The scale has the format >sem (A + B, A). 1.2. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a number showing three desinences and a number showing only one desinence, such that one kind of number is used in the singular, and the other kind of number is used in the plural, it is the number showing three desinences that tends to be used in the singular. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a number showing three desinences and a number showing only one desinence, such that one kind of number is used in the singular, and the other kind of number is used in the plural, it is the number showing only one desinence that tends to be used in the plural. Q.E.D. 4. Note. Cf. deduction 222. 30. Estonian. A predicative noun can be expressed with the nominative or with the essive. (After change-of-state verbs with the translative; this case is not treated here.) If the nominative is used, it implies permanency (even inalienable capacities), e.g. mees on meie saadik Londonis 'the man is our [permanent] ambassador in London.' If the essive is employed, it implies contingency, e.g. mees on meie saadikuna Londonis 'the man is our [not necessarily permanent] ambassador in London.' Similarly in predicative nouns and adjectives of Finnish. (Anderson 1999, 63.) The two syntactic variants: the predicative noun in the nominative, and the predicative noun in the essive. 1. The assumptions of Naturalness Theory: 1.1. >sem (nominative, other cases) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than other cases, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. (Mayerthaler et al. 1998, 241.) 1.2. >sem (-alienable, +alienable) / property I.e. with respect to semantic complexity, an inalienable property is more natural than an alienable property. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 41

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

A special case of 1.2: 1.2.1. >sem (+permanent, -permanent) / property I.e. with respect to semantic complexity, a permanent property is more natural than a non-permanent property. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the predicative noun in the nominative and in the essive, such that one case expresses permanent property, and the other case expresses non-permanent property, it is the nominative that tends to express the permanent property. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the predicative noun in the nominative and in the essive, such that one case expresses permanent property, and the other case expresses non-permanent property, it is the essive that tends to express the non-permanent property. Q.E.D. 31. Finnish. (a) Possessive suffixes may be accompanied by a genitive personal pro noun for emphasis, e.g. non-emphatic hattu-nsa 'his hat,' emphatic hn-en hattu-nsa same meaning. (b) The plural pronouns in the genitive can occur without the corresponding personal suffix if the possessive structure expresses some relationship other than ownership, e.g. meidn kyl 'our village.' (Lyons 1999, 127 fn.) The present deduction deals with case (a). Case (b) is considered in deduction 32. The limitation of case (b) to the plural is treated in deduction 33. The two syntactic variants: the type hattu-nsa, and the type hn-en hattu-nsa. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type hn-en hattu-nsa, the type hattu-nsa) / in Finnish I.e. with respect to coding properties, the type hn-en hattu-nsa is more natural than the type hattu-nsa, in Finnish. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, -emphasis is more natural than +emphasis. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type hattu-nsa and the type hn-en hattunsa, such that one is emphatic, and the other not, it is the type hn-en hattu-nsa that tends to be emphatic. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced:

42

EXAMPLES

3.2. If there is any difference between the type hattu-nsa and the type hn-en hattunsa, such that one is emphatic, and the other not, it is the type hattu-nsa that tends not to be emphatic. Q.E.D. 32. Finnish. (a) Possessive suffixes may be accompanied by a genitive personal pro noun for emphasis, e.g. non-emphatic hattu-nsa 'his hat,' emphatic hn-en hattu-nsa same meaning. (b) The plural pronouns in the genitive can occur without the corresponding personal suffix if the possessive structure expresses some relationship other than ownership, e.g. meidn kyl 'our village.' (Lyons 1999, 127 fn.) The present deduction deals with case (b). The limitation of case (b) to the plural is treated in deduction 33. Case (a) is considered in deduction 31. The two syntactic variants: the type meidn kyl 'our village' (expressing possession other than ownership), and the type kyl-mme 'our village' (expressing ownership). 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type meidn kyl , the type kyl-mme) / in Finnish I.e. with respect to coding properties, the type meidn kyl is more natural than the type kyl-mme, in Finnish. 1.2. >sem (+ownership, -ownership) / possession I.e. with respect to semantic complexity, ownership is more natural than other kinds of possession.Ownership is the prototypical possession. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type meidn kyl and the type kyl-mme, such that one expresses ownership, and the other a different type of possession, it is the type meidn kyl that tends to express a kind of possession different from ownership. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type meidn kyl and the type kyl-mme, such that one expresses ownership, and the other a different type of possession, it is the type kyl-mme that tends to express ownership. Q.E.D. 33. Finnish. (a) Possessive suffixes may be accompanied by a genitive personal pronoun for emphasis, e.g. non-emphatic hattu-nsa 'his hat,' emphatic hn-en hattu-nsa same meaning. (b) The plural pronouns in the genitive can occur without the corresponding personal suffix if the possessive structure expresses some relationship other than ownership, e.g. meidn kyl 'our village.' (Lyons 1999, 127 fn.) Case (a) is considered in deduction 31. Case (b) is dealt with in deduction 32. The present deduction treats the limitation of case (b) to the plural. The two semantic variants: possession which does not specifically mark ownership, and possession which does not express ownership. 1. The assumptions of Naturalness Theory: 43

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.1. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) A special case of 1.1: 1.1.1. >sem (the type hattu-nsa, the type meidn kyl ) / in Finnish I.e. with respect to semantic complexity, the type hattu-nsa is more natural than the type meidn kyl , in Finnish.The type hattu-nsa represents the singular only and expresses +/-ownership, within possession. The type meidn kyl represents the plural only and expresses -ownership, within possession. 1.2. >sem (+/-ownership, -ownership) / possession I.e. with respect to semantic complexity, possession which does not specifically mark ownership is more natural than possession which does not express ownership.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.1. <sem tends to associate with another <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between possession which does not specifically mark ownership and possession which does not express ownership, such that one kind of possession is used with the type hattu-nsa, and the other kind of possession is used with the type meidn kyl , it is the possession which does not specifically mark ownership that tends to occur with the type hattu-nsa. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.1. If there is any difference between possession which does not specifically mark ownership and possession which does not express ownership, such that one kind of possession is used with the type hattu-nsa, and the other kind of possession is used with the type meidn kyl , it is the possession which does not express ownership that tends to occur with the type meidn kyl . Q.E.D. 34. French. In the names of streets and squares, the preposition de is not used if the name contains the surname of a person, e.g. rue Pasteur. If however the name of the person is preceded by a title, the preposition de is used, e.g. place du Roi Albert . If the name contains a place name, the preposition de is likewise used, e.g. avenue de Versailles . (Grevisse 1993, 536.) The two syntactic variants: the type rue Pasteur, and the type avenue de Versailles . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type avenue de Versailles , the type rue Pasteur) / in French I.e. with respect to coding properties, the type avenue de Versailles is more natural than the type rue Pasteur, in French. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a

44

EXAMPLES

less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem (name of person, name of place) / names of streets, squares I.e. with respect to semantic complexity, the name of a person is more natural than the name of a place, as a constituent of a street name or a square name. 2. The assumptions of the Slovenian Theory: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type rue Pasteur and the type avenue de Versailles, such that in one type the constituent name is subordinated to the preposition de, and in the other that preposition is lacking, it is the type avenue de Versailles that tends to contain the preposition de. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type rue Pasteur and the type avenue de Versailles, such that in one type the constituent name is subordinated to the preposition de, and in the other that preposition is lacking, it is the type rue Pasteur that tends to lack the preposition de. Q.E.D. 4. Note. The deduction opposing the type rue Pasteur to the type place du Roi Albert would be very similar to the present one. 35. French. Until the 16th century, the possessive genitive could lack the preposition de, especially the genitive of person, e.g. le rei gunfanuner 'the king's standard-bearer,' li chevaus le rei 'the king's horse.' (Grevisse 1993, 536.) The two syntactic variants: de + noun phrase, and bare noun phrase, both as an expression of the possessive genitive. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (de + noun phrase, bare noun phrase) / possessive genitive in French I.e. with respect to coding properties, de + noun phrase is more natural than the bare noun phrase, in French (until the 16th century). 1.2. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between de + noun phrase and the bare noun phrase, as an expression of the possessive genitive, such that one expression mostly denotes persons, and the other not, it is de + noun phrase that tends not to denote persons. Q.E.D.

45

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between de + noun phrase and the bare noun phrase, as an expression of the possessive genitive, such that one expression mostly denotes persons, and the other not, it is the bare noun phrase that tends to denote persons. Q.E.D. 36. French. In the standard language, outside formulaic expressions and POUVOIR etc., the normal sentence negation is ne pas, e.g. il ne vient pas 'he is not coming.' If the sentence negation associates with an infinitive, it is ne pas, e.g. ne pas se pencher au dehors 'do not lean out' = 'it is dangerous to lean out of the window.' (Grevisse 1993, 1458.) The two syntactic variants: the sentence negation ne pas, and the sentence negation ne pas. 1. The assumptions of Naturalness Theory: 1.1. >sym (ne pas, ne pas) / in French I.e. with respect to coding properties, the negation ne pas is more natural than the negation ne pas.Behaghel's Law (Behaghel 1932, 4). 1.2. >sem (+finite, -finite) / verb I.e. with respect to semantic complexity, a finite verb is more natural than an infinite verb. (Mayerthaler et al. 1993, 145.) A special case of 1.2: 1.2.1. >sem (+finite, infinitive) / verb I.e. with respect to semantic complexity, a finite verb is more natural than an infinitive. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the sentence negation ne pas and the sentence negation ne pas, such that one of them is used with infinitives, and the other elsewhere, it is the sentence negation ne pas that tends to be used with infinitives. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the sentence negation ne pas and the sentence negation ne pas, such that one of them is used with infinitives, and the other elsewhere, it is the sentence negation ne pas that tends to be used elsewhere. Q.E.D. 4. Note. Cf. deductions 37-41. 37. French. If the sentence negation negates an infinitive, it is ne pas, e.g. ne pas se pencher au dehors 'do not lean out' = 'it is dangerous to lean out of the window.' However, if the infinitive is 'be' + subject complement, or 'have to' + another infinitive, or 'have' in formulas, the negation is ne pas, e.g. pour n'tre pas trop long 'so as not to be too long,' pour n'avoir pas prononcer un nom aussi clinquant 'so as not to have to pronounce a name so tinselly,' elle est convenue de n'en avoir pas envie 'she agreed not to have envy because of it.' (Grevisse 1993, 1458-9.) The two syntactic variants: the sentence negation ne pas, and the sentence negation ne pas, as negation of infinitive. 1. The assumptions of Naturalness Theory: 1.1. >sym (ne pas, ne pas) / in French

46

EXAMPLES

I.e. with respect to coding properties, the negation ne pas is more natural than the negation ne pas, in French.Behaghel's Law (Behaghel 1932, 4). 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem ('be' as copula, 'have' as modal and in formulas; most other verbal lexemes) / in French I.e. with respect to semantic complexity, 'be' as copula, 'have' as modal and in formulas are more natural than most other verbal lexemes, in French. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the sentence negation ne pas and the sentence negation ne pas, within the infinitive, such that one negation is used with 'be' as copula and 'have' as modal & 'have' in formulas, and the other negation is used with "other" verbal lexemes, it is the negation ne pas that tends to be used with "other" verbal lexemes as infinitives. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the sentence negation ne pas and the sentence negation ne pas, within the infinitive, such that one negation is used with 'be' as copula and 'have' as modal & 'have' in formulas, and the other negation is used with "other" verbal lexemes, it is the negation ne pas that tends to be used with the following infinitives: 'be' as copula and 'have' as modal & 'have' in formulas. Q.E.D. 4. Note. Cf. deductions 36 and 38-41. 38. French. In the standard language, the normal sentence negation is ne pas, e.g. il ne vient pas 'he is not coming.' In formulaic expressions (including proverbs), the sentence negation is just ne, e.g. il n'en a cure 'he does not care about it,' il n'a que deux fils 'he only has two sons.' (Examples from Hock 1991, 360-1. Cf. Grevisse 1993, 1445 ff.) The two syntactic variants: the sentence negation ne pas, and the sentence negation ne. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (ne pas, ne) / sentence negation in French I.e. with respect to coding properties, the sentence negation ne pas is more natural than the sentence negation ne, in French. 1.1.2. >sym (-formula, +formula) I.e. with respect to coding properties, a non-formula is more natural than a formula. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym

47

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the sentence negation ne pas and the sentence negation ne, such that one is used in formulaic expressions, and the other outside them, it is the sentence negation ne pas that tends not to be used in formulaic expressions. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the sentence negation ne pas and the sentence negation ne, such that one is used in formulaic expressions, and the other outside them, it is the sentence negation ne that tends to be used in formulaic expressions. Q.E.D. 4. Note. Cf. deductions 36-7 and 39-41. 39. French. In the standard language, the normal sentence negation is ne pas, e.g. il ne vient pas 'he is not coming.' Associated with POUVOIR 'can,' SAVOIR 'know,' CESSER 'stop,' and OSER 'dare,' the sentence negation is just ne (pas), e.g. il ne cesse (pas) de parler 'he does not stop speaking.' (Examples from Hock 1991, 360-1. Cf. Grevisse 1993, 1445 ff.) The two syntactic variants: the sentence negation ne pas, and the sentence negation ne (pas). 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ne pas, ne (pas)) / sentence negation in French I.e. with respect to coding properties, the sentence negation ne pas is more natural than the sentence negation ne (pas), in French. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem (POUVOIR, SAVOIR, CESSER, OSER ; most other verbal lexemes) / in French I.e. with respect to semantic complexity, the verbs POUVOIR, SAVOIR, CESSER, OSER are more natural than most other verbal lexemes, in French. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the sentence negation ne pas and the sentence negation ne (pas), such that one negates POUVOIR, SAVOIR, CESSER, OSER, and the other negates "other" verbs, it is the sentence negation ne pas that tends to negate "other" verbs. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced:

48

EXAMPLES

3.2. If there is any difference between the sentence negation ne pas and the sentence negation ne (pas), such that one negates POUVOIR, SAVOIR, CESSER, OSER, and the other negates "other" verbs, it is the sentence negation ne (pas) that tends to negate the verbs POUVOIR, SAVOIR, CESSER, OSER . Q.E.D. 4. Note. Cf. deductions 36-8 and 40-1. 40. French. In the standard language, outside formulaic expressions and POUVOIR etc., the normal sentence negation is ne pas, e.g. il ne vient pas 'he is not coming.' Under inversion of the subject and the verb, the sentence negation is weakened to (ne) pas, e.g. (ne) vient-il pas 'is he not coming.' (Examples from Hock 1991, 360-1. Cf. Grevisse 1993, 1445 ff.) The two syntactic variants: the sentence negation ne pas, and the sentence negation (ne) pas. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ne pas, (ne) pas) / sentence negation in French I.e. with respect to coding properties, the sentence negation ne pas is more natural than the sentence negation (ne) pas, in French. 1.2. >sem (VS, SV) / interrogative clause in French I.e. with respect to semantic complexity, the element order verb - subject is more natural than the element order subject - verb, in French interrogative clauses.See item 4.1 of this deduction. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the sentence negation ne pas and the sentence negation (ne) pas, such that one is used when the element order is SV, and the other is used when the element order is VS, it is the sentence negation ne pas that tends to be used when the element order is SV. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the sentence negation ne pas and the sentence negation (ne) pas, such that one is used when the element order is SV, and the other is used when the element order is VS, it is the sentence negation (ne) pas that tends to be used when the element order is VS. Q.E.D. 4. Notes. 4.1. The scale of item 1.2 is derived from the following two more basic scales: >sem (declarative, interrogative) / sentential mode (Mayerthaler et al. 1998, 326) and >sem (SV, VS) / SVO-languages. According to these two scales, the normal element order of interrogative clauses is VS. 4.2. Cf. deductions 36-9 and 41. 41. French. In the standard language, outside formulaic expressions and POUVOIR etc., the normal sentence negation is ne pas, e.g. il ne vient pas 'he is not coming.' If the sentence negation associates with a non-verb, it is just pas, e.g. pas de bruit 'no noise, please.' (Examples from Hock 1991, 360-1. Cf. Grevisse 1993, 1446-7.) 49

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

The two syntactic variants: the clause negation ne pas, and the phrase negation pas. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (ne pas, pas) / negation in French I.e. with respect to coding properties, the negation ne pas is more natural than the negation pas, in French. 1.1.2. >sym (clause, phrase) / negated I.e. with respect to coding properties, a negated clause is more natural than a negated phrase. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the negation ne pas and the negation pas, such that one negates clauses, and the other negates phrases, it is the negation ne pas that tends to negate clauses. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the negation ne pas and the negation pas, such that one negates clauses, and the other negates phrases, it is the negation pas that tends to negate phrases. Q.E.D. 4. Note. Cf. deductions 36-40. 42. French. Proper names used as vocative are bare. Common nouns used as vocative take a determiner, e.g. salut, les mecs 'hello, chaps,' bonjour, mon gnral 'good morning, general.' (Lyons 1999, 152.) The two syntactic variants: proper name and common noun as vocatives. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type les mecs, the type Jean) / vocative in French I.e. with respect to coding properties, the type les mecs is more natural than the type Jean , as vocative in French. 1.2. >sem (proper noun, common noun) I.e. with respect to semantic complexity, a proper name is more natural than a common noun.According to the animacy hierarchy (Croft 1990, 112, with references). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem

50

EXAMPLES

3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type Jean (proper noun) and the type les mecs (common noun) as vocatives, such that one type contains a determiner, and the other type not, it is the type les mecs that tends to contain a determiner. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type Jean (proper noun) and the type les mecs (common noun) as vocatives, such that one type contains a determiner, and the other type not, it is the type Jean that tends not to contain a determiner. Q.E.D. 43. French. Personal pronouns. Gender is indicated (a) in third person singular only (English, German), (b) in third person only (French), (c) in third person and plural generally (Spanish), (d) in all persons, singular and plural, except first (Arabic). (Lyons 1999, 137.) Case (b) is treated in the present deduction. Case (a) is dealt with in deduction 57, case (c) in deduction 210. Case (d) is considered in deduction 6. The two syntactic variants: personal pronoun of the third person, and the personal pronouns of the non-third person. 1. The assumptions of Naturalness Theory: 1.1. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 1.2. >sem (+visible, -visible) / gender in languages that show gender distinctions I.e. with respect to semantic complexity, visible gender is more natural than nonvisible gender, in languages that show gender distinctions.If a language has gender distinctions, the most natural option is for them to be visible. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the personal pronouns of the third person and the personal pronouns of the non-third person, such that one kind is differentiated as to gender, and the other kind not, it is the personal pronouns of the third person that tend to be differentiated. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the personal pronouns of the third person and the personal pronouns of the non-third person, such that one kind is differentiated as to gender, and the other kind not, it is the personal pronouns of the non-third person that tend to lack differentiation. Q.E.D. 44. French. If the possessum is definite and the possessor pronominal, the latter is expressed with a possessive adjective, e.g. ma bicyclette 'my bicycle.' In other cases of possession, a prepositional phrase is used after the possessum, e.g. la bicyclette de Jeanne 'Jeanne's bicycle,' une bicyclette Jeanne, moi 'a bicycle of Jeanne's, of mine.' (Lyons 1999, 130.) The present deduction is continued in deduction 45. The two syntactic variants: the type ma bicyclette , and the type une bicyclette moi . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit

51

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type une bicyclette moi , the type ma bicyclette ) / in French I.e. with respect to coding properties, the type une bicyclette moi is more natural than the type ma bicyclette , in French. 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than -presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type ma bicyclette and the type une bicyclette moi, such that in one type the possessum is definite, and in the other type indefinite, it is in the type une bicyclette moi that the possessum tends to be indefinite. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type ma bicyclette and the type une bicyclette moi , such that in one type the possessum is definite, and in the other type indefinite, it is in the type ma bicyclette that the possessum tends to be definite. Q.E.D. 45. French. If the possessum is definite and the possessor pronominal, the latter is expressed with a possessive adjective, e.g. ma bicyclette 'my bicycle.' In other cases of possession, a prepositional phrase is used after the possessum, e.g. la bicyclette de Jeanne 'Jeanne's bicycle,' une bicyclette Jeanne, moi 'a bicycle of Jeanne's, of mine.' (Lyons 1999, 130.) The present deduction continues deduction 44. The two syntactic variants: the type la bicyclette de Jeanne , and the type une bicyclette Jeanne . 1. The assumptions of Naturalness Theory: 1.1. >sem (genitive, other cases) / case dependent on a head noun, in nom.-acc. languages I.e. with respect to semantic complexity, the genitive is more natural than other cases dependent on a head noun, in nominative-accusative languages.The genitive is the adnominal case par excellence . 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than -presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite.

52

EXAMPLES

Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type la bicyclette de Jeanne and the type une bicyclette Jeanne , such that one type is definite, and the other indefinite, and such that the possessor is in the genitive or in the dative case, it is the type la bicyclette de Jeanne that tends to be definite and to contain a genitive. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type la bicyclette de Jeanne and the type une bicyclette Jeanne , such that one type is definite, and the other indefinite, and such that the possessor is in the genitive or in the dative case, it is the type une bicyclette Jeanne that tends to be indefinite and to contain a dative. Q.E.D. 46. French. The partitive indefinite consists of the preposition de 'of' + definite article + the rest of the noun phrase, e.g. de la bire 'sm beer.' This construction omits the definite article in negative contexts, e.g. je n'ai pas de bire 'I do not have any beer.' (Lyons 1999, 100-1.) The two syntactic variants: the type de la bire , and the type (pas) de bire. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (de la bire, de bire ) / in French I.e. with respect to coding properties, the type de la bire is more natural than the type de bire , in French. 1.2. >sem (negative context, any other context) I.e. with respect to semantic complexity, the negative context is more natural than any other context.This is based on the fact that the other contexts do not constitute a "natural class;" they include affirmative declarative and affirmative interrogative contexts, e.g. tu as de la bire 'you have sm beer,' as-tu de la bire 'have you got any beer.' 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type de la bire and the type (pas) de bire , such that one is used in negative contexts, and the other elsewhere, it is the type de la bire that tends to be used in "other" contexts. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type de la bire and the type (pas) de bire , such that one is used in negative contexts, and the other elsewhere, it is the type de bire that tends to be used in negative contexts. Q.E.D.

53

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

4. Note. Mutatis mutandis, the deduction also applies to the Spanish no es problema 'it is not a problem,' with omitted indefinite article, compared to es un problema 'it is a problem,' Es un problema? 'is it a problem.' (Lyons 1999, 105, partly referring to Butt & C. Benjamin.) 47. French. In the example-sentence nous avons rapproch la montagne 'we brought the mountain close,' the object la montagne 'the mountain' is totally affected. In the example-sentence nous nous sommes rapprochs de la montagne 'we approached the mountain,' the object de la montagne is not affected. (Hopper & S. Thompson 1980, 262.) The two syntactic variants: the affected object la montagne, and the non-affected object de la montagne. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (de la montagne, la montagne ) / object in French I.e. with respect to coding properties, the object de la montagne is more natural than the object la montagne , in French. 1.2. >sem (+affected, -affected) / object I.e. with respect to semantic complexity, an affected object is more natural than a nonaffected object. (In the spirit of Hopper & S. Thompson 1980 passim.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the objects la montagne and de la montagne , such that one is affected and the other non-affected, it is de la montagne that tends not to be affected. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the objects la montagne and de la montagne , such that one is affected and the other non-affected, it is la montagne that tends to be affected. Q.E.D. 48. French. Of the two affirmative replies, oui and si, it is the latter that is more likely to be expanded with a partial repetition of the material in the corresponding question, e.g. Il ne vient pas? Si(, il vient). 'Isn't he coming? Yes(, he's coming).' (Data from Schachter 1985, 32.) The two syntactic variants: the type oui(, il vient), and the type si(, il vient). 1. The assumptions of Naturalness Theory: 1.1. >sem (oui, si) / in French I.e. with respect to semantic complexity, the yes-word oui is more natural than the yes-word si, in French.This is based on the circumstance that si is used in answers to negative yes-no questions, which lack a special yes-word in many languages. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency

54

EXAMPLES

is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit. - The ability to expand involves the tendency to enhance transparency. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the yes-words oui and si concerning their ability to expand syntactically, it is oui that is less likely to expand than si. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the yes-words oui and si concerning their ability to expand syntactically, it is si that is more likely to expand than oui. Q.E.D. 49. French. While French has distinct reflexive and non-reflexive third person pronouns, it does not make such a distinction for the other persons, but instead uses the same first and second person pronouns both reflexively and non-reflexively, e.g. je me vois 'I see me = myself,' as opposed to ils les/se voient 'they see them/themselves.' (Schachter 1985, 27-8.) Similarly in some other languages (Lichtenberk 1999, 314). The two syntactic variants: the reflexive pronoun, and the personal pronoun, both expressing reflexivity. 1. The assumptions of Naturalness Theory: 1.1. >sem (reflexive, personal) / pronoun expressing reflexivity I.e. with respect to semantic complexity, a reflexive pronoun is more natural than a personal pronoun, as an expression of reflexivity. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between personal and reflexive pronouns as an expression of reflexivity, such that one kind of pronouns is used in the third person, and the other kind of pronouns in the non-third person, it is the reflexive pronouns that tend to be used in the third person. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between personal and reflexive pronouns as an expression of reflexivity, such that one kind of pronouns is used in the third person, and the other kind of pronouns in the non-third person, it is the personal pronouns that tend to be used in the non-third person. Q.E.D. 50. French. The same pronoun is used to express reflexivity and reciprocity, e.g. ils se 55

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

flattent 'they flatter themselves' and 'they flatter each other.' If disambiguation is necessary, the reciprocal pronoun is expanded, e.g. ils se flattent l'un l'autre 'they flatter one another.' (Schachter 1985, 29-30.)Similarly in Slovenian (Milena MilojeviSheppard viva voce, 1999). The two syntactic variants: the reflexive type ils se flattent , and the reciprocal type ils se flattent (l'un l'autre). 1. The assumptions of Naturalness Theory: 1.1. >sem (reflexive use, reciprocal use) / of the same pronoun I.e. with respect to semantic complexity, the reflexive use of a pronoun is more natural than the reciprocal use of the same pronoun.This is based on the admittedly slim evidence that the reciprocal pronoun of Akan (a Congo-Kordofanian language in Ghana) is a doubling of the reflexive pronoun. (The Akan data from Schachter 1985, 29.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit.The ability to expand involves the tendency towards transparency. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the reflexive pronoun expressing reflexivity and the reflexive pronoun expressing reciprocity, concerning their ability to expand syntactically, it is the reflexive pronoun expressing reflexivity which tends not to expand. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the reflexive pronoun expressing reflexivity and the reflexive pronoun expressing reciprocity, concerning their ability to expand syntactically, it is the reflexive pronoun expressing reciprocity which tends to expand. Q.E.D. 51. French. In the non-standard conjugation, the subject pronoun is accented, and strengthened with the corresponding proclitic pronoun, e.g. moi j'aime 'I love.' In the third persons, the strengthening is optional, e.g. lui (il) aime 'he loves.' (Mayerthaler 1981, 2930.) The two syntactic variants: a subject which can be either accented pronoun + clitic pronoun or accented pronoun, and a subject which can be only accented pronoun + clitic pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sem (accented + clitic & accented, only accented + clitic) / subject pronoun in French I.e. with respect to semantic complexity, a subject pronoun which can be either accented + clitic or accented is more natural than a subject pronoun which can be only

56

EXAMPLES

accented + clitic, in French.The scale has the format >sem (A + B, A). 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a subject which can be either accented pronoun + clitic pronoun or accented pronoun and a subject which can be only accented pronoun + clitic pronoun, such that one kind of subject is used in the third person, and the other kind of subject is used in the non-third person, it is the subject which can be either accented pronoun + clitic pronoun or accented pronoun that tends to be used in the third person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a subject which can be either accented pronoun + clitic pronoun or accented pronoun and a subject which can be only accented pronoun + clitic pronoun, such that one kind of subject is used in the third person, and the other kind of subject is used in the non-third person, it is the subject which can be only accented pronoun + clitic pronoun that tends to be used in the non-third person. Q.E.D. 52. French. In the imperative of transitive verbs, the full form of any accusative pro noun is used, e.g. salue-moi 'greet me,' except in the third persons, where the clitic form of any accusative pronoun is used, e.g. salue-le 'greet him.' (Mayerthaler 1981, 30.) The two syntactic variants: imperative + accented form of accusative pronoun, and imperative + clitic form of accusative pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a unit of greater morphological transparency is more natural than a corresponding unit of lesser morphological transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (accented form, clitic form) / personal pronoun I.e. with respect to coding properties, any accented form of the personal pronoun is more natural than the corresponding clitic form. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the use of the accented accusative pronoun and the use of the clitic accusative pronoun in the imperative clause, such that one kind of

57

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

pronoun is used in the third person, and the other kind of pronoun is used in the non-third person, it is in the non-third person that the accented form tends to be used. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the use of the accented accusative pronoun and the use of the clitic accusative pronoun in the imperative clause, such that one kind of pronoun is used in the third person, and the other kind of pronoun is used in the non-third person, it is in the third person that the clitic form tends to be used. Q.E.D. 53. French. The equivalent of English question tags (e.g. You haven't eaten, have you?) is a single formulaic expression: n'est-ce pas is that not. Similarly in German: nicht wahr not true. Similarly in many other languages. (Schachter 1985, 33.) In Slovenian: kajne and some other alternatives (Toporii 1976, 385; 2000, 445; Herrity 2000, 323-4). In non-standard British English: innit, etymologically connected with isn't it and possibly with ain't it (Biber et al. 1999, 1122-3). The two syntactic variants: the clause preceding the question tag, and the question tag viewed as a partial repetition of the clause. 1. The assumptions of Naturalness Theory: 1.1. >sem (repetition, its original) I.e. with respect to semantic complexity, repetition is more natural than its original.

58

EXAMPLES

This is based on the fact that a repetition is easily retrievable from memory, and on the circumstance that repetition is imitation, which is speakers' innate ability (Li 1986, 401). 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.2: 1.2.1. >sym (-formula, +formula) I.e. with respect to coding properties, a non-formula is more natural than a formula. 1.2.2. >sym (-short, +short) / syntactic unit I.e. with respect to coding properties, a non-short syntactic unit is more natural than a short syntactic unit. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the clause preceding the question tag, and the question tag itself, such that one construction is a formulaic expression, and the other construction is not a formulaic expression, and such that one construction is short, and the other construction is not short, it is the question tag that tends to be a formulaic expression, and tends to be short. Q.E.D. From 1.1, 1.2.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the clause preceding the question tag, and the question tag itself, such that one construction is a formulaic expression, and the other construction is not a formulaic expression, and such that one construction is short, and the other construction is not short, it is the clause preceding the question tag that tends not to be a formulaic expression, and tends not to be short. Q.E.D. 54. French. In the early history of French nouns, only those nominatives singular sur vived that had a desinence; the others were replaced by the accusative singular. (Mayerthaler 1981, 65 ff.) The two morphological variants: nominative singular containing a desinence, and nominative singular lacking any desinence. 1. The assumptions of Naturalness Theory: 1.1. >sem (nominative, accusative) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative case is more natural than the accusative case, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. 1.2. >sym (nominative singular lacking desinence, nominative singular containing desinence) I.e. with respect to coding properties, the nominative singular lacking any desinence is more natural than the nominative singular containing a desinence. (Mayerthaler 1981, 28.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: 59

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the nominative singular with and without desi nence, such that one kind of nominative keeps the form of the nominative, and the other kind of nominative assumes the form of the accusative, it is the nominative singular without desinence that tends to assume the form of the corresponding accusative. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the nominative singular with and without desi nence, such that one kind of nominative keeps the form of the nominative, and the other kind of nominative assumes the form of the accusative, it is the nominative singular with desinence that tends to keep the form of the old nominative singular. Q.E.D. 55. German. De-verbal nouns as subjects associate with conspicuously few verbs only, e.g. Apollo-Flug 'flight of Apollo' with erfolgen 'take place, transpire.' (B. Sandig 1971, 41. Her list of such verbs contains no more than five almost synonymous items, but is marked as incomplete.) The two syntactic variants: subject associating with many verbs, and subject associating with few verbs. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (de-verbal, underived) / noun in German I.e. with respect to coding properties, de-verbal nouns are more natural than underived nouns, in German. 1.2. >sem (many verbs, few verbs) / dependent on a type of subject I.e. with respect to semantic complexity, many verbs dependent on a type of subject are more natural than few such verbs.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between a subject associating with many verbs and a subject associating with few verbs, such that one kind of subject comprises de-verbal nouns, and the other kind of subject comprises underived nouns, it is the subject associating with many verbs that tends to comprise underived nouns. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between a subject associating with many verbs and a subject associating with few verbs, such that one kind of subject comprises de-verbal nouns, and the other kind of subject comprises underived nouns, it is the subject associating with few verbs that tends to comprise de-verbal nouns. Q.E.D. 56. German. The genitivus explicativus is (morphologically) a true genitive in the singular, e.g. die Mglichkeiten der Entspannung 'the chances for dtente,' and mostly a prepositional phrase in the plural, e.g. die Mglichkeiten von Unruhen 'the chances for distur-

60

EXAMPLES

bances.' (Engel 1988, 621.) The two syntactic variants: the type (die Mglichkeiten) der Entspannung , and the type (die Mglichkeiten) von Unruhen . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type von Unruhen, the type der Entspannung) / in German I.e. with respect to coding properties, the type von Unruhen is more natural than the type der Entspannung), in German. 1.2. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the genitivus explicativus in the singular and plural, such that one kind is a true genitive, and the other kind is a prepositional phrase, it is the genitivus explicativus in the singular that tends to be a true genitive. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the genitivus explicativus in the singular and plural, such that one kind is a true genitive, and the other kind is a prepositional phrase, it is the genitivus explicativus in the plural that tends to be a prepositional phrase. Q.E.D. 57. German. Personal pronouns. Gender is indicated (a) in third person singular only (English, German), (b) in third person only (French), (c) in third person and plural gene rally (Spanish), (d) in all persons, singular and plural, except first (Arabic). (Lyons 1999, 137.) Case (a) is treated in the present deduction. Cases (b-c) are dealt with in deductions 43 and 210. Case (d) is considered in deduction 6. The two syntactic variants: personal pronoun of the third person singular, and the remaining personal pronouns of the singular and plural. 1. The assumptions of Naturalness Theory: 1.1. >sem (third person singular, the remaining) / personal pronoun I.e. with respect to semantic complexity, the personal pronoun of the third person singular is more natural than the remaining personal pronouns.The remaining personal pronouns (of both numbers) do not constitute a "natural class." Also, the third person is more sem-natural than the non-third person (according to the linguistic tradition beginning with Jakobson 1932); the singular is more sem-natural than the plural (Mayerthaler 1981, 15). 1.2. >sem (+visible, -visible) / gender in languages which show gender distinctions I.e. with respect to semantic complexity, visible gender is more natural than nonvisible gender, in languages that show gender distinctions.If a language has gender distinctions, the most natural option is for them to be visible.

61

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the personal pronouns of the third person singular and the remaining personal pronouns, such that one kind is differentiated as to gender, and the other kind not, it is the personal pronouns of the third person singular that tend to be differentiated. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the personal pronouns of the third person singular and the remaining personal pronouns, such that one kind is differentiated as to gender, and the other kind not, it is the "remaining" personal pronouns that tend to lack differentiation. Q.E.D. 58. German. In the colloquial language, first names are used with the article, e.g. die Claudia, der Hans. (Lyons 1999, 122.) The two syntactic variants: the types Claudia & die Claudia, and the type Claudia. 1. The assumptions of Naturalness Theory: 1.1. >sem (conversation, other registers) I.e. with respect to semantic complexity, conversation is more natural than the other registers. (Dotter 1990, 228.) 1.2. >sem (the type Claudia & the type die Claudia, only the type Claudia) / in German I.e. with respect to semantic complexity, using the types Claudia and die Claudia is more natural than using only the type Claudia.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between using the types Claudia & die Claudia and using only the type Claudia, such that one usage occurs in conversation, and the other usage occurs in the remaining registers, it is the types Claudia & die Claudia that tend to be used in conversation. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between using the types Claudia & die Claudia and using only the type Claudia, such that one usage occurs in conversation, and the other usage occurs in the remaining registers, it is the type Claudia that tends to be used in the remaining registers. Q.E.D. 59. German. With surnames of female celebrities, the definite article is used, e.g. die Dietrich . (Lyons 1999, 122.) The two syntactic variants: surname of female celebrity, and surname of male celebrity. 1. The assumptions of Naturalness Theory: 1.1. >sem (masculine, feminine) I.e. with respect to semantic complexity, masculine is more natural than feminine.

62

EXAMPLES

(Mayerthaler 1981, 15.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (the type die Dietrich, the type Goethe ) / in German I.e. with respect to coding properties, the type die Dietrich is more natural than the type Goethe, in German. 2. The assumptions of the Slovenian Theory: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the surname of a female celebrity and the surname of a male celebrity, such that one kind of surname is used with the definite article, and the other kind of surname is not used with the definite article, it is the surname of a female celebrity that tends to be used with a definite article. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the surname of a female celebrity and the surname of a male celebrity, such that one kind of surname is used with the definite article, and the other kind of surname is not used with the definite article, it is the surname of a male celebrity that tends not to be used with a definite article. Q.E.D. 60. German. The colloquial use of the definite article with surnames carries a connotation of assumed familiarity, e.g. der Brandt ist ein ganz kluger Kopf 'Brandt is very clever.' (Lyons 1999, 122.) The two syntactic variants: (colloquial neutral) Brandt, and (colloquial familiar) der Brandt. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type der Brandt, the type Brandt) / in German I.e. with respect to coding properties, the type der Brandt is more natural than the type Brandt, in German. 1.2. >sem (speaker's neutral use, speaker's air of familiarity) I.e. with respect to semantic complexity, a speaker's neutral use is more natural than a speaker expressing familiarity. (In the spirit of Dotter 1990, 228.) 2. The assumptions of the Slovenian Theory: 2.1. >sym tends to associate with <sem 2.2. >sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced:

63

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.1. If there is any difference between the colloquial type Brandt and the colloquial type der Brandt, such that one type expresses familiarity, and the other type does not, it is the type der Brandt that tends to be used when the speaker expresses assumed familiarity. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the colloquial type Brandt and the colloquial type der Brandt, such that one type expresses familiarity, and the other type does not, it is the type Brandt that tends to be used when the speaker expresses himself neutrally. Q.E.D. 61. German. A plurale tantum sometimes has a counterpart in the singular containing more sound body and structure than the plurale tantum, e.g. singular Elternteil 'parent,' plurale tantum Eltern 'parents.' (Mayerthaler 1981, 51.) The two morphological variants: the type Elternteil , and the type Eltern. 1. The assumptions of Naturalness Theory: 1.1. >sem (plural, singular) / plurale tantum I.e. with respect to semantic complexity, the plural is more natural than the singular, in pluralia tantum.Pluralia tantum do not usually employ corresponding notions in the singular. 1.2. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a unit of greater morphological transparency is more natural than a corresponding unit of lesser morphological transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (the type Elternteil , the type Eltern) / in German I.e. with respect to coding properties, the type Elternteil is more natural than the type Eltern, in German. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the singular and the plural of a plurale tantum, concerning how elaborate their morphological structure is, it is the plural that tends to be encoded more simply. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the singular and the plural of a plurale tantum, concerning how elaborate their morphological structure is, it is the singular that tends to be encoded in a more complex way. Q.E.D. 62. German. Non-standard German uses the periphrastic type er tut rauchen beside the non-periphrastic er raucht, both meaning 'he smokes.' Basically, the periphrasis is acceptable with all verbal lexemes except the frequent ones, such as SEIN, HABEN , and the modal verbs. (Eroms 1984.) The two syntactic variants: all verbal lexemes, and all verbal lexemes except SEIN, HABEN and the modal verbs. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit

64

EXAMPLES

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type er tut rauchen , the type er raucht) / non-standard German I.e. with respect to coding properties, the type er tut rauchen is more natural than the type er raucht, in non-standard German. 1.2. >sem (all verbal lexemes; verbal lexemes except SEIN, HABEN , modal verbs) / in German I.e. with respect to semantic complexity, all verbal lexemes are more natural than the same verbal lexemes except SEIN, HABEN and the modal verbs, in German.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between using all verbal lexemes and using all verbal lexemes except SEIN, HABEN and the modal verbs, such that one use occurs in the type er raucht, and the other use occurs in the type er tut rauchen , it is the use of all verbal lexemes that tends to occur in the type er raucht. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between using all verbal lexemes and using all verbal lexemes except SEIN, HABEN and the modal verbs, such that one use occurs in the type er raucht, and the other use occurs in the type er tut rauchen , it is the use of all verbal lexemes except SEIN, HABEN and the modal verbs that tends to occur in the type er tut rauchen . Q.E.D. 63. German. Main and dependent clauses. Main clauses show the basic word orders SV and VS, whereas dependent clauses show only the basic order SV. In main clauses S and V (= the finite verbal form) are near to each other, whereas in dependent clauses S and V are maximally far from each other. Examples: ich kaufte heute ein Buch 'I bought a book today', heute kaufte ich ein Buch 'today, I bought a book', da ich heute ein Buch kaufte 'that I bought a book today'. The two syntactic variants: the element orders SV & VS, and only the element order SV. 1. The assumptions of Naturalness Theory: 1.1. >sem (main, dependent) / clause I.e. with respect to semantic complexity, a main clause is more natural than a dependent clause.Main clauses are phylogenetically earlier than dependent clauses. 1.2. >sem (SV & VS, only SV) / in German I.e. with respect to semantic complexity, admitting the element orders SV and VS is more natural than admitting only the element order SV, in German.The scale has the format >sem (A + B, A). 1.3. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.3: 65

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.3.1. >sym (S and V separated, S and V together) / in German finite clauses I.e. with respect to coding properties, a finite clause in which S and V are separated is more natural than a finite clause in which S and V are together, in German.Having S and V separated is more conspicuous than having S and V together. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem and with <sym 2.2. <sem tends to associate with another <sem and with >sym 3. The consequences: From 1.1-2, 1.3.1 and 2.1 it can be deduced: 3.1. If there is any difference between using the element orders SV & VS and using only the element order SV, such that one usage occurs in main clauses, and the other usage occurs in dependent clauses, and such that in one usage S and V are separated, and in the other usage S and V are together, it is the element orders SV & VS that tend to occur in main clauses, whereby S and V are together. Q.E.D. From 1.1-2, 1.3.1 and 2.2 it can be deduced: 3.2. If there is any difference between using the element orders SV & VS and using only the element order SV, such that one usage occurs in main clauses, and the other usage occurs in dependent clauses, and such that in one usage S and V are separated, and in the other usage S and V are together, it is only the element order SV that tends to occur in dependent clauses, whereby S and V are separated. Q.E.D. 4. Note. Similarly in Old English, Dutch, and Frisian. 64. German. Verb + particle (the so-called "trennbar zusammengesetzte Verben"). In the main clause, the finite verb occupies the second syntactic slot of the clause, and the particle is clause-final, e.g. er stand frh auf 'he got up early.' In the dependent clause introduced by a subordinator, both the particle and the finite verb are clause-final, in that order, e.g. da er auf-stand 'that he got up.' (Schachter 1985, 45-6.) The two syntactic variants: the main clause and the dependent clause. 1. The assumptions of Naturalness Theory: 1.1. >sym (+adjacency, -adjacency) / of units that belong together I.e. with respect to coding properties, adjacency of units that belong together is more natural than their non-adjacency.Behaghel's Law (Behaghel 1932, 4). 1.2. >sem (main clause, dependent clause) I.e. with respect to semantic complexity, a main clause is more natural than a depen dent clause.Phylogenetically, main clauses are of earlier origin than dependent clauses. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between main and dependent clauses, such that in one kind of clauses the finite verb and its particle are separated, and in the other kind of clauses the finite verb and its particle are united, it is in the main clause that the finite verb and the particle tend to be separated. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between main and dependent clauses, such that in one kind of clauses the finite verb and its particle are separated, and in the other kind of clauses the finite verb and its particle are united, it is in the dependent clause that the

66

EXAMPLES

finite verb and the particle tend to be united. Q.E.D. 4. Note. Mutatis mutandis, the same deduction can account for the contrast ich habe das Buch gekauft 'I have bought the book' v. weil ich das Buch gekauft habe 'because I have bought the book.' 65. German. Element order in clause-final verb chains. The prevailing order is left branching, weil ihm das Auto gestohlen worden sein soll 'because his car is said to have been stolen'. Right branching occurs if the clause-final chain contains at least two bare infinitives. In that case right branching involves any form of HABEN (obligatorily) and the finite forms of WERDEN (frequently), and some other common finite forms (less frequently). Examples: weil er dies nicht hat tun wollen, er wird wegen seiner Verpflichtungen nicht haben kommen knnen , als ich die Vgel im Garten habe zwitschern hren, weil sie sich das Paket wird schicken lassen , weil er wegen seines Urlaubs nicht wird haben kommen knnen . (Askedal 1991; Duden Grammatik 1998, 8167; Askedal 2000, 1137.) The two syntactic variants: right branching and left branching in clause-final verb chains containing at least two bare infinitives. 1. The assumptions of Naturalness Theory: 1.1. >sem (right, left) / branching of elements I.e. with respect to semantic complexity, right branching of elements is more natural than left branching of elements.This is supported by the circumstance that in German main clauses right branching prevails, and in German dependent clauses left branching prevails (Mayerthaler et al. 1998, 312, 315-6). 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem (HABEN, finite forms of WERDEN, and finite forms of some further verbal lexemes; most other verbal lexemes) / in German I.e. with respect to semantic complexity, HABEN , finite forms of WERDEN, and finite forms of some further verbal lexemes are more natural than most other verbal lexemes, in German. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between right branching and left branching in clausefinal verb chains containing at least two bare infinitives, such that one kind of branching involves HABEN , the finite forms of WERDEN and of some further verbal lexemes, and the other kind of branching involves most remaining verbal lexemes, it is the right branching that tends to involve HABEN , the finite forms of WERDEN and of some further verbal lexemes. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between right branching and left branching in clausefinal verb chains containing at least two bare infinitives, such that one kind of branching involves HABEN , the finite forms of WERDEN and of some further verbal lexemes, and the other kind of branching involves most remaining verbal lexemes, it is the left branching that tends to involve the remaining verbal lexemes. QED. 4. Notes. 67

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

4.1. The falling frequency of HABEN, finite forms of WERDEN, and finite forms of some further verbal lexemes reflects the falling tendency of those lexemes/forms to participate in right branching. 4.2. In the construction bare infinitive + bare infinitive, the modal verbal lexeme BRAUCHEN 'need' can be preceded by the infinitie particle zu. 4.3. The situation is similar in Dutch (Posten 1973), only the cut-off point between the verbal lexemes participating in right branching and other verbal lexemes is placed in a different place than in German. 4.4. The Slovenian Theory cannot predict the placement of the cut-off point between the verbal lexemes participating in right branching and other verbal lexemes. 66. German. The attributive adjective is inflected, the predicative adjective is not in flected, e.g. ein guter Mann 'a good man,' as opposed to der Mann ist gut 'the man is good.' (Duden Grammatik 1998, 280-1.) 1. The assumptions of Naturalness Theory: 1.1. >sem (several case forms, one case form) / of the same adjectival lexical item, in German. I.e. with respect to semantic complexity, several case forms of the same adjectival lexical item are more natural than just one such case form, in German.The scale has the format >sem (A + B, A). 1.2. >sem (attributive, predicative) / adjective I.e. with respect to semantic complexity, the attributive use of adjectives is more natural than the predicative use of adjectives. (Mayerthaler et al. 1998, 20.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the inflected and the non-inflected use of adjectives, such that one occurs in attributive position and the other in predicative position, it is the inflected use that tends to prevail in attributive position. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the inflected and the non-inflected use of adjectives, such that one occurs in attributive position and the other in predicative position, it is the non-inflected use that tends to prevail in predicative position. Q.E.D. 67. German. The negation of clauses containing transitive verbs. With an indefinite object, the majority type is as exemplified in ich habe kein Geld 'I have no money.' With a definite object, the majority type is as exemplified in ich sehe den Wagen nicht 'I cannot see the car.' (Duden Grammatik 1998, 722-3.) The two syntactic variants: negation of verb + indefinite object, and negation of verb + definite object. 1. The assumptions of Naturalness Theory: 1.1. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, the indefinite direct object is more natural than the definite direct object.In the spirit of Mayerthaler 1981, 14 and 1987, 42, concerning the markedness relations of the object as being the opposite of those typical of the subject. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency 68

EXAMPLES

is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (nicht + determiner of direct object, negative determiner of direct object) / clause negation in German I.e. with respect to coding properties, nicht + determiner of direct object is more natural than a bare negative determiner of the direct object, in German clause negation. Nicht occupies an extra syntactic slot. Any slot for the determiner of the direct object is present regardless of negation. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the negation of verb + indefinite object and the negation of verb + definite object, such that the clause negation nicht is used with one construction, and the negative determiners are used in the other construction, it is verb + indefinite object that tends to be negated with a negative determiner before the direct object. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the negation of verb + indefinite object and the negation of verb + definite object, such that the clause negation nicht is used with one construction, and the negative determiners are used in the other construction, it is verb + definite object that tends to be negated with nicht . Q.E.D. 68. German. The following example contains the conjunction of two time clauses (in square brackets): wenn [jemand nach Hause kommt und da steht der Gerichtsvollzieher vor der Tr] 'when someone comes home and there stands the bailiff in front of the door.' The first conjunct shows theexpectedtypical element order of dependent clauses (i.e. the finite verb is clause-final), whereas the other conjunct displayscontrary to expectations the element order of main clauses. Such syntax of the second conjunct is optional. (J. B. Johannessen 1998, 40, referring to Hhle.) The two syntactic variants: a main & a dependent clause, and only a dependent clause. 1. The assumptions of Naturalness Theory: 1.1. >sym (first, second) / conjunct in coordination I.e. with respect to coding properties, the first conjunct is more natural than the second conjunct.The first conjunct is structurally richer than the second conjunct, in coordination (speaking of averages). The first conjunct is more conspicuous than the second conjunct. 1.2. >sem (main clause & dependent clause, dependent clause) I.e. with respect to semantic complexity, contracting a main or dependent clause is more natural than contracting just a dependent clause.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences:

69

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between contracting a main or a dependent clause and contracting only a dependent clause, such that one option is the first conjunct in coordination, and the other option is the second conjunct in coordination, it is contracting a main or dependent clause that tends to occur in the second conjunct in coordination. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between contracting a main or a dependent clause and contracting only a dependent clause, such that one option is the first conjunct in coordination, and the other option is the second conjunct in coordination, it is contracting a dependent clause that tends to occur in the first conjunct in coordination. Q.E.D. 4. Note. J. B. Johannessen 1998, 39-42 adduces similar data from Dutch and compa rable data from Norwegian. 69. Germanic languages, old. The verb of the relative clause is in the indicative mood if the superordinate clause is affirmative. The verb of the relative clause is in the subjunctive mood if the superordinate clause is negative. (M. Kozianka 1999.) The two syntactic variants: the relative clause containing the finite verb in the indica tive mood, and the relative clause containing the finite verb in the subjunctive mood. 1. The assumptions of Naturalness Theory: 1.1. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 1.2. >sem (+indicative, -indicative) I.e. with respect to semantic complexity, the indicative is more natural than nonindicative. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (indicative, subjunctive) / in old Germanic languages I.e. with respect to semantic complexity, the indicative is more natural than the subjunctive, in the old Germanic languages. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between affirmative superordinate clause + relative clause and negative superordinate clause + relative clause, such that the mood of the relative clause is indicative or subjunctive, the relative clause's verb tends to be in the indicative mood if the superordinate clause is affirmative. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between affirmative superordinate clause + relative clause and negative superordinate clause + relative clause, such that the mood of the relative clause is indicative or subjunctive, the relative clause's verb tends to be in the subjunctive mood if the superordinate clause is negated. Q.E.D. 70. Gothic. The dual lacks the third verbal person (in the verb and in the personal pronoun). (Braune & Ebbinghaus 1966, 98.) The two morphological variants: finite verb showing all three persons, and finite verb showing only the non-third persons. 1. The assumptions of Naturalness Theory:

70

EXAMPLES

1.1. >sem (singular, plural; dual) I.e. with respect to semantic complexity, the singular is more natural than the plural, and the latter is more natural than the dual. (Greenberg 1966, 31-7.) 1.2. >sem (all three persons, only non-third person) / in Gothic finite verb I.e. with respect to semantic complexity, a finite verb showing all three persons is more natural than a finite verb showing only the non-third person, in Gothic.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a finite verb showing all three persons and a finite verb showing only the non-third persons, such that one kind of finite verb is used in the singular and plural, and the other kind of finite verb is used in the dual, it is the finite verb showing all three persons that tends to be used in the singular and plural. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a finite verb showing all three persons and a finite verb showing only the non-third persons, such that one kind of finite verb is used in the singular and plural, and the other kind of finite verb is used in the dual, it is the finite verb showing only the non-third persons that tends to be used in the dual. Q.E.D. 4. Notes. 4.1. Cf. deduction 71. 4.2. In item 1.1, the cut-off point between the plural and the dual has been placed arbitrarily. 71. Gothic. The dual lacks the third verbal person (in the verb and in the personal pronoun). (Braune & Ebbinghaus 1966, 98.) The third verbal person of the dual is replaced by the third verbal person of the plural. (Hirt 1932, 139; 1934, 10, with references.) The two morphological variants: the third person dual, and the non-third person dual. 1. The assumptions of Naturalness Theory: 1.1. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) A special case of 1.2: 1.2.1. >sem (+third, -third) / grammatical person of the dual I.e. with respect to semantic complexity, the third person is more natural than the non-third person, in the dual. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 71

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.1. If there is any difference between the third person dual and the non-third person dual, such that one set of persons uses special dual forms, and the other set of persons uses plural forms, it is the third person dual that tends to use plural forms. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the third person dual and the non-third person dual, such that one set of persons uses special dual forms, and the other set of persons uses plural forms, it is the non-third person dual that tends to use special dual forms. Q.E.D. 4. Note. Cf. deduction 70. 72. Greek, Ancient. In comedy and tragedy, female characters often speak about themselves in the first person plural masculine. (Schwyzer 1950, 243.) The two syntactic variants: a woman speaking about herself in the first person singular feminine, and a woman speaking about herself in the first person plural masculine. 1. The assumptions of Naturalness Theory: 1.1. >sem (everyday life, on stage) I.e. with respect to semantic complexity, everyday life is more natural than life on stage. 1.2. >sem (singular, plural) / of the subject pronoun I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 1.3. >sem (expected gender agreement, default gender agreement) / of subject complement I.e. with respect to semantic complexity, expected gender agreement is more natural than default gender agreement.The scale has the format >sem (A + B, A). The default gender agreement is masculine for animate nouns (Humbert 1972, 78). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-3 and 2.1 it can be deduced: 3.1. If there is any difference between expected gender agreement and default gender agreement, such that one kind of agreement is found in everyday life, and the other kind of agreement is found on stage, and such that the subject pronoun is singular with one kind of agreement, and plural with the other kind of agreement, it is the expected gender agreement that tends to be found in everyday life, and with the subject pronoun in the singular. Q.E.D. From 1.1-3 and 2.2 it can be deduced: 3.2. If there is any difference between expected gender agreement and default gender agreement, such that one kind of agreement is found in everyday life, and the other kind of agreement is found on stage, and such that the subject pronoun is singular with one kind of agreement, and plural with the other kind of agreement, it is the default gender agreement that tends to be found on stage, and with the subject pronoun in the plural. Q.E.D. 4. Note. Cf. deductions 136-7. 73. Greek, Ancient. In the subjunctive and optative of the perfect and pluperfect, compound tense forms prevail (in the active) or are the only ones used (in the mediopassive). Thus the subjunctive of the active perfect is prevalently replaced by the corresponding compound forms, e.g. pepaideuk of paideuein 'educate' by pepaideuks . The subjunctive of the mediopassive perfect can only be expressed with compound forms, e.g. with 72

EXAMPLES

pepaideumenos , etc. (Bornemann & Risch 1978, 90-3.) This deduction is continued in deduction 73a. The two syntactic variants: the perfect indicative, and the perfect non-indicative. 1. The assumptions of Naturalness Theory: 1.1. >sem (+indicative, -indicative) / mood I.e. with respect to semantic complexity, the indicative is more natural than the nonindicative, within moods. (Mayerthaler 1981, 14.) A special case of 1.1: 1.1.1. >sem (indicative, subjunctive/optative) / in Ancient Greek I.e. with respect to semantic complexity, the indicative is more natural than the sub junctive and the optative, in Ancient Greek. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (the compound tenses, the simple tenses) / the perfect and pluperfect of Ancient Greek I.e. with respect to coding properties, the compound tenses of the (plu)perfect are more natural than the corresponding simple tenses, in Ancient Greek. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the simple and the compound tenses of the perfect and pluperfect, such that one kind is used in the indicative, and the other kind in the subjunctive/optative, it is the compound tenses that tend to be used in the subjunctive/ optative. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the simple and the compound tenses of the perfect and pluperfect, such that one kind is used in the indicative, and the other kind in the subjunctive/optative, it is the simple tenses that tend to be used in the indicative. Q.E.D. 4. Note. The exclusive use of the compound forms in the subjunctive/optative of the perfect mediopassive (in contradistinction to the less restricted use of the compound forms in the subjunctive/optative of the perfect active) is considered in deduction 73a. 73a. Greek, Ancient. In the subjunctive and optative of the perfect and pluperfect, compound tense forms prevail (in the active) or are the only ones used (in the medio passive). Thus the subjunctive of the active perfect is prevalently replaced by the corre sponding compound forms, e.g. pepaideuk of paideuein 'educate' by pepaideuks . The subjunctive of the mediopassive perfect can only be expressed with compound forms, e.g. with pepaideumenos , etc. (Bornemann & Risch 1978, 90-3.) This deduction continues deduction 73. The two syntactic variants: simple & compound tenses v. only compound tenses, within the subjunctive and optative of the (plu)perfect. 1. The assumptions of Naturalness Theory:

73

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.1. >sem (active, mediopassive) I.e. with respect to semantic complexity, the active is more natural than the mediopassive. (In the spirit of Mayerthaler 1981, 15.) 1.2. >sem (simple & compound, only compound) / tenses of subjunctive and optative (plu)perfect in Ancient Greek I.e. with respect to semantic complexity, using simple and compound tenses is more natural than using only compound tenses, in Ancient Greek subjunctive and optative (plu)perfect. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the subjunctive and optative of the (plu)perfect, between using simple & compound tenses and using only compound tenses, such that one usage occurs in the active, and the other usage occurs in the mediopassive, it is the use of simple & compound tenses that tends to occur in the active. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within the subjunctive and optative of the (plu)perfect, between using simple & compound tenses and using only compound tenses, such that one usage occurs in the active, and the other usage occurs in the mediopassive, it is the exclusive use of compound tenses that tends to occur in the mediopassive. Q.E.D. 74. Greek, Ancient. The possessives are of two kinds: pronominal, e.g. emos adjective, mou clitic genitive, both 'my,' and full noun phrases, e.g. tou adelphou genitive 'the brother's.' (Lyons 1999, 124-5.) The two syntactic variants: the nominal and the pronominal possessives. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (genitive of full noun phrase; pronominal clitic, clitic genitive) / in Ancient Greek I.e. with respect to coding properties, the genitive of a full noun phrase is more natural than a pronominal adjective or a clitic genitive, in Ancient Greek. 1.1.2. >sym (nominal, pronominal) / possessive I.e. with respect to coding properties, a nominal possessive is more natural than a pronominal possessive. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the nominal and the pronominal possessives, such that one of them is the genitive of a full noun phrase, and the other is a clitic genitive

74

EXAMPLES

or a pronominal adjective, it is the nominal possessive that tends to be the genitive of a full noun phrase. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the nominal and the pronominal possessives, such that one of them is the genitive of a full noun phrase, and the other is a clitic genitive or a pronominal adjective, it is the pronominal possessive that tends to be a clitic genitive or a pronominal adjective. Q.E.D. 4. Note. This deduction is almost vacuous, yet valid. 75. Greek, Ancient (Attic). The singular, plural and dual of nouns are distinguished. While the singular and the plural display practically the same cases, the dual has only two (one represents the nominative, accusative and vocative, the other represents the genitive and dative). (Croft 1990, 99-100.) The two syntactic variants: the noun in the singular & plural, and the noun in the dual. 1. The assumptions of Naturalness Theory: 1.1. >sem (singular, plural; dual) I.e. with respect to semantic complexity, the singular is more natural than the plural; the plural is more natural than the dual. (Greenberg 1966, 31-37.) 1.2. >sem (relatively many case forms of the same number, relatively few case forms of the same number) I.e. with respect to semantic complexity, relatively many case forms of the same number are more natural than relatively few forms of the same number.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between having relatively many case forms of the same number and having relatively few case forms of the same number, such that one option is used in the singular & plural, and the other option is used in the dual, it is the option of relatively many case forms in the same number that tends to be used in the singular & plural. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between having relatively many case forms of the same number and having relatively few case forms of the same number, such that one option is used in the singular & plural, and the other option is used in the dual, it is the option of relatively few case forms in the same number that tends to be used in the dual. Q.E.D. 4. Note. In item 1.1, the cut-off point between the plural and the dual has been placed arbitrarily. 76. Greek, Ancient. In Homer, the use of preposition + case is less frequent when the case is an actant, than when the case is a circumstant. (L. Conti Jimnez 1999.) The two syntactic variants: preposition + case, and bare case. 1. The assumptions of Naturalness Theory: 1.1. >sem (+actant, -actant) I.e. with respect to semantic complexity, an actant is more natural than a non-actant

75

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

(= circumstant). (Mayerthaler 1981, 14.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (preposition + case, bare case) I.e. with respect to coding properties, preposition + case is more natural than bare case. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between preposition + case and bare case, such that one tends to be reserved for actants and the other for circumstants, it is the bare case that tends to associate with actants. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between preposition + case and bare case, such that one tends to be reserved for actants and the other for circumstants, it is preposition + case that tends to associate with circumstants. Q.E.D. 77. Guaran (an Amerindian language spoken in Paraguay). Only first- and secondperson pronouns show number marking. (Lyons 1999, 214.) The two syntactic variants: non-third person pronouns, and other noun phrases. 1. The assumptions of Naturalness Theory: 1.1. >sem (+realization of number marking, -realization of number marking) / noun phrases of languages which do have number marking I.e. with respect to semantic complexity, the realization of number marking is more natural than the non-realization of number marking, in noun phrases of languages which do have number marking.If a language has number marking, the most natural option is its realization. 1.2. >sem (non-third person pronouns; other noun phrases) I.e. with respect to semantic complexity, the non-third person pronouns are more natural than other noun phrases.Other noun phrases (comprising third-person pronouns and full noun phrases) do not constitute a "natural class." 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the non-third person pronouns and other noun phrases, such that one kind shows number marking, and the other not, it is the non-third person pronouns that tend to show number marking. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the non-third person pronouns and other noun phrases, such that one kind shows number marking, and the other not, it is the other noun

76

EXAMPLES

phrases that tend to lack number marking. Q.E.D. 78. Hausa (Chadic, Afroasiatic languages). There is a suffixed definite article, e.g. yaro-n 'the boy,' and a free-form definite article, e.g. yaro in same meaning. The suffixed article is optional, mostly used anaphorically (and cataphorically) when the referent is not easy to access. There is not much difference in use between the suffixed and the free-form articles, except that the referent of the noun with the free-form article has to be mentioned previously. (Lyons 1999, 52-4, referring to Kraft & Kirk-Greene and to Jaggar.) This deduction is continued in deduction 78a. The two syntactic variants: the definite noun yaro-n & yaro in, and the definite noun yaro. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (yaro in & yaro-n, yaro ) / definite noun in Hausa I.e. with respect to coding properties, the type yaro din & yaro-n is more natural than the type yaro, as the definite noun in Hausa. 1.2. >sem (+easy to access, -easy to access) / referent I.e. with respect to semantic complexity, a referent easy to access is more natural than a referent difficult to access.Something that is easy to access is easily retrievable from memory. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the variants of the definite noun, yaro in & yaro-n and yaro, such that one is easy to access, and the other not, it is the type yaro in & yaro-n that is difficult to access. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the variants of the definite noun, yaro in & yaro-n and yaro, such that one is easy to access, and the other not, it is the type yaro that is easy to access. Q.E.D. 78a. Hausa (Chadic, Afroasiatic languages). There is a suffixed definite article, e.g. yaro-n 'the boy,' and a free-form definite article, e.g. yaro in same meaning. The suffixed article is optional, mostly used anaphorically (and cataphorically) when the referent is not easy to access. There is not much difference in use between the suffixed and the free-form articles, except that the referent of the noun with the free-form article has to be mentioned previously. (Lyons 1999, 52-4, referring to Kraft & Kirk-Greene and to Jaggar.) This deduction continues deduction 78. The two syntactic variants: optional previous mention of a definite noun, and obli gatory previous mention of a definite noun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit

77

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (yaro in, yaro-n & yaro) / definite noun in Hausa I.e. with respect to coding properties, the type yaro in is more natural than the type yaro(-n), as the definite noun in Hausa. 1.2. >sem (+/-mentioned previously, +mentioned previously) / definite noun in Hausa I.e. with respect to semantic complexity, a definite noun whose referent has been optionally mentioned previously is more natural than a definite noun whose referent has obligatorily been mentioned previously.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between optional previous mention of a definite noun and obligatory previous mention of a definite noun, such that one kind of mention is connected with the type yaro in, and the other kind of mention is connected with the type yaro(-n), it is the optional previous mention of a definite noun that tends to be connected with the type yaro(-n). Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between optional previous mention of a definite noun and obligatory previous mention of a definite noun, such that one kind of mention is connected with the type yaro in, and the other kind of mention is connected with the type yaro(-n), it is the obligatory previous mention of a definite noun that tends to be connected with the type yaro in. Q.E.D. 79. Hidatsa (one of the Siouan languages). There is a suffixed definite article - s, restricted to anaphoric use. Elsewhere definiteness is not marked. (Lyons 1999, 53, referring to Matthews.) The two syntactic variants: the suffixed definite article - s, and the definiteness not marked. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (the definite article - s, the definite article ) / with definite nouns of Hidatsa I.e. with respect to coding properties, the definite article - s is more natural than the lack of any definite article, with definite nouns of Hidatsa. 1.1.2. >sym (+anaphoric, -anaphoric) / use of the definite noun in Hidatsa I.e. with respect to coding properties, the anaphoric use is more natural than the nonanaphoric use of the definite noun in Hidatsa.The anaphoric use is easier for the hearer to

78

EXAMPLES

process than non-anaphoric use. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the definite article - s and the lack of definite article with definite nouns, such that one expression is used anaphorically, and the other elsewhere, it is the definite article - s that is used anaphorically. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the definite article - s and the lack of definite article with definite nouns, such that one expression is used anaphorically, and the other elsewhere, it is the lack of the definite article that is used non-anaphorically. Q.E.D. 80. Hungarian. The basic element order is SVO. However, objects and other arguments with no article are placed in front of the (nonprefixed) verb; they are nonspecific and often form a semantic unit with the verb up to idiomaticity, e.g. Pter knyv-et olvas Peter book read 'Peter is reading books / a book.' Thus the element order is SOV. (Kenesei et al. 1998, 73-4.) The two syntactic variants: the basic element order SVO, and the basic element order SOV. 1. The assumptions of Naturalness Theory: 1.1. >sem (SOV, SVO) / basic element order I.e. with respect to semantic complexity, the basic element order SOV is more natural than the basic element order SVO. (Mayerthaler et al. 1998, 309. Cf. the discussion in A. Siewierska 1999, 412-3.) 1.2. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, the indefinite direct object is more natural than the definite direct object.In the spirit of Mayerthaler 1981, 14 and 1987, 42, con cerning the markedness relations of the object as being the opposite of those typical of the subject. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the basic element orders SVO and SOV, such that one is used with +definite objects, and the other is used with -definite objects, it is SOV that tends to be used with -definite objects. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the basic element orders SVO and SOV, such that one is used with +definite objects, and the other is used with -definite objects, it is SVO that tends to be used with +definite objects. Q.E.D. 81. Hungarian. The so-called objective conjugation is used in combination with a definite direct object. The so-called subjective conjugation is used in combination with an indefinite direct object and elsewhere. (More detailed data in Hopper & S. Thompson 1980, 258.) The two syntactic variants: the objective and the subjective conjugations. 1. The assumptions of Naturalness Theory: 79

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (objective, subjective) / conjugation in Hungarian I.e. with respect to coding properties, the objective conjugation is more natural than the subjective conjugation, in Hungarian.The desinences of the objective conjugation have more sound body than the desinences of the subjective conjugation. 1.2. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, the indefinite direct object is more natural than the definite direct object.In the spirit of Mayerthaler 1981, 14 and 1987, 42, con cerning the markedness relations of the object as being the opposite of those typical of the subject. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the subjective and the objective conjugations, such that one is used with definite direct objects and the other with indefinite direct objects, it is the subjective conjugation that tends to be used with indefinite direct objects. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the subjective and the objective conjugations, such that one is used with definite direct objects and the other with indefinite direct objects, it is the objective conjugation that tends to be used with definite direct objects. Q.E.D. 82. Icelandic. The nominative singular of masculine strong nouns mostly has a desinence, e.g. hest-ur 'horse.' Outside the masculine nouns such a desinence is exceptional, within strong declensions. Likewise the nominative singular of strong masculine adjectives mostly has a desinence, e.g. falleg-ur 'beautiful.' (Einarsson [1945] 1967, 32 ff., 50 ff.) The two syntactic variants: the nominative singular of masculine strong nouns and adjectives, and the nominative singular of non-masculine strong nouns and adjectives. 1. The assumptions of Naturalness Theory: 1.1. >sym (+, -) / desinence in the nominative singular I.e. with respect to coding properties, lack of desinence in the nominative singular is more natural than its presence. (Mayerthaler 1981, 28.) 1.2. >sem (+masculine, -masculine) I.e. with respect to semantic complexity, the masculine is more natural than other genders. (Greenberg 1966, 34-40.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the nominative singular of masculine strong nouns & adjectives and the nominative singular of non-masculine strong nouns & adjectives, such that one kind of nominatives has a desinence, and the other kind of nominatives 80

EXAMPLES

lacks any desinence, it is the nominative singular of the masculine strong nouns & adjec tives that tends to have a desinence. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the nominative singular of masculine strong nouns & adjectives and the nominative singular of non-masculine strong nouns & ad jectves, such that one kind of nominatives has a desinence, and the other kind of nomina tives lacks any desinence, it is the nominative singular of the non-masculine strong nouns & adjectives that tends to lack a desinence. Q.E.D. 83. Icelandic. The anaphoric expression for two coordinated genders G1 + G1 also contains gender G1. E.g. the anaphora of the coordination Sigga og Anna 'Sigga (fem.) and Anna (fem.)' contains feminine gender: r 'they (fem.)' The anaphoric expression for two coordinated genders G1 + G2 contains neuter gender (= the least sem-natural gender). E.g. the anaphora of the coordination Sigga og Jn 'Sigga (fem.) and Jn (masc.)' contains neuter gender: au 'they (neuter).' (Einarsson [1945] 1967, 133.) The two syntactic variants: the anaphoric expression of G1 + G1, and the anaphoric expression of G1 + G2, where G1 and G2 are genders. 1. The assumptions of Naturalness Theory: 1.1. >sem (G1 + G1, G1 + G2) / coordination of two genders I.e. with respect to semantic complexity, the coordination of two instances of the same gender is more natural than the coordination of two different genders.The coordination of two instances of the same gender is cognitively simpler than the coordination of two different genders. 1.2. >sem (expected gender, default gender) / anaphora of two coordinated noun phrases I.e. with respect to semantic complexity, the expected gender is more natural than the default gender, of the anaphora of two coordinated noun phrases.The expected gender is the one used in G1 + G1, thus any gender. The default gender is neuter in Icelandic. The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the anaphoric expression for the coordination of two different genders and the anaphoric expression for the coordination of two instances of the same gender G1, whereby the choice is between the neuter gender and the gender G1, it is the anaphoric expression for the coordination of two instances of the same gender G1 that tends to contain the gender G1. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the anaphoric expression for the coordination of two different genders and the anaphoric expression for the coordination of two instances of the same gender G1, whereby the choice is between the neuter gender and the gender G1, it is the anaphoric expression for the coordination of two different genders that tends to contain the neuter gender. Q.E.D. 4. Notes. 4.1. This deduction does not address the question as to why the neuter gender is the default gender in Icelandic. 4.2. Cf. deduction 185.

81

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

84. Icelandic. In the conjugation of the present indicative, most verbs have desinences in all the forms but the first person singular, e.g. g dvel, dvel-ur, hann dvel-ur, vi dvel-jum, i dvel-ji, eir dvel-ja 'dwell.' (Einarsson [1945] 1967, 90-1.) The two morphological variants: the first person singular and the other persons, in the present indicative. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (+desinence, -desinence) / the inflection I.e. with respect to coding properties, the presence of desinence is more natural than lack of desinence. 1.2. >sem (first person singular, the other persons) / the present indicative of Icelandic verbs I.e. with respect to semantic complexity, the first person singular is more natural than the other persons, in the present indicative of Icelandic verbs.The other persons (of both numbers) have a low sem-value because they do not constitute a "natural class." 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the first person singular and the other persons of the conjugational paradigm, such that one kind lacks any desinence, and the other has them, it is the other persons that tend to have desinences. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the first person singular and the other persons of the conjugational paradigm, such that one kind lacks any desinence, and the other has them, it is the first person singular that tends to lack any desinence. Q.E.D. 85. Italian. There is a contrast between la mia casa 'my house' and mia madre 'my mother.' The latter example is a case of inalienable possession, but limited to a small num ber of kinship terms. (Lyons 1999, 128-9.) The two syntactic variants: the type la mia casa (includes some kinship terms), and the type mia madre (i.e. a subset of kinship terms). 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type la mia casa, the type mia madre) / in Italian I.e. with respect to coding properties, the type la mia casa is more natural than the

82

EXAMPLES

type mia madre, in Italian. 1.2. >sem (a subset of kinship terms, a subset of kinship terms & other types of possession) / in Italian I.e. with respect to semantic complexity, a subset of kinship terms is more natural than the combination of another subset of kinship terms & other types of possession, in Italian. The element "a subset of kinship terms and other types of possession" does not form a "natural class." 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type la mia casa and the type mia madre, such that one expresses kinship terms only, and the other expresses a subset of kinship terms and other types of possession, it is the type la mia casa that tends to express a subset of kinship terms and other types of possession. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type la mia casa and the type mia madre, such that one expresses kinship terms only, and the other expresses a subset of kinship terms and other types of possession, it is the type mia madre that tends to express a subset of kinship terms only. Q.E.D. 86. Italian. Colloquially, the definite article is used with female first names, e.g. la Maria. (Lyons 1999, 122.) The two syntactic variants: both the type Maria and the type la Maria v. only the type Maria. 1. The assumptions of Naturalness Theory: 1.1. >sem (the types (la) Maria, only the type Maria) / first name in Italian I.e. with respect to semantic complexity, using the types (la) Maria is more natural than using only the type Maria, in Italian first names.The scale has the format >sem (A + B, A). 1.2. >sem (conversation, other registers) I.e. with respect to semantic complexity, conversation is more natural than other registers. (Dotter 1990, 228.) 2. The assumptions of the Slovenian Theory: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between using both the type Maria and the type la Maria and using only the type Maria, such that one use occurs in conversation, and the other use occurs in other registers, it is the use of the types (la) Maria that tends to occur in conversation. Q.E.D. From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between using both the type Maria and the type la Maria and using only the type Maria, such that one use occurs in conversation, and the other use occurs in other registers, it is the exclusive use of the type Maria that tends to occur in other registers. Q.E.D. 87. Italian. With surnames the definite article is used for prominent male and female 83

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

persons. (Lyons 1999, 122.) E.g. il Pavarotti, la Callas. The two syntactic variants: non-prominent Pavarotti, Callas, and prominent il Pavarotti, la Callas. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type la Callas, the type Callas) / in Italian I.e. with respect to coding properties, the type la Callas is more natural than the type Callas, in Italian. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, -emphasis is more natural than +emphasis. (Mayerthaler 1981, 15, 159.) A special case of 1.2: 1.2.1. >sem (-prominent, +prominent) I.e. with respect to semantic complexity, -prominent is more natural than +prominent. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the surnames of +/-prominent persons, such that one set of surnames is used with the definite article, and the other not, it is the prominent persons that tend to take the definite article. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the surnames of +/-prominent persons, such that one set of surnames is used with the definite article, and the other not, it is the nonprominent persons that tend not to take the definite article. Q.E.D. 88-9. Italian. In subordinate clauses containing the finite verb in the conjunctive mood, the subordinator can be omitted in case that the element order is verb - subject, e.g. (se) tornasse Mario 'if Mario should return.' (Rizzi 1982, 84-5.) The two syntactic variants are: the construction admitting the types se Mario tornasse & se tornasse Mario , and the construction tornasse Mario. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type se Mario tornasse, the type (se) tornasse Mario ) / in Italian I.e. with respect to coding properties, the type se Mario tornasse is more natural than the type (se) tornasse Mario, in Italian. 1.2. >sem (VS, SV) / element order if the subject is displaced, in SVO-languages I.e. with respect to semantic complexity, the element order VS is more natural than

84

EXAMPLES

the element order SV if the subject is displaced, in SVO-languages.See item 4 of this deduction. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the types se Mario tornasse (subject not displaced) and (se) tornasse Mario (subject displaced), such that in one type the element order is SV, and in the other type the element order is VS, it is in the type se Mario tornasse that the element order tends to be SV. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the types se Mario tornasse (subject not displaced) and (se) tornasse Mario (subject displaced), such that in one type the element order is SV, and in the other type the element order is VS, it is in the type (se) tornasse Mario that the element order tends to be VS. Q.E.D. 4. Note. The scale of item 1.2 is derived from the following two more basic scales: >sym (+displaced, -displaced) / subject and >sem (SV, VS) / SVO-languages. According to these two scales, the normal element order in clauses containing a displaced subject is VS. 90. Italian. The non-honorific personal pronoun of address uses the second person singular: tu 'you.' The corresponding honorific pronoun uses the third person singular: Lei 'you.' Similarly in a number of other languages. The two syntactic variants: honorific 'you,' and non-honorific 'you.' 1. The assumptions of Naturalness Theory: 1.1. >sem (second person, third person) / of the personal pronoun within the subsystem of appeal I.e. with respect to semantic complexity, the second person is more natural than the third person, in the personal pronoun within the subsystem of appeal.The second person is the prototypical person of address ('appeal'). It is also basic in the imperative. 1.2. >sem (-honorific, +honorific) / personal pronoun of address I.e. with respect to semantic complexity, the non-honorific personal pronoun of address is more natural than the corresponding honorific pronoun.Primitive and early civilizations employ non-honorific pronouns of address only. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the honorific and the non-honorific personal pronouns of address, such that one takes the second grammatical person, and the other takes the third grammatical person, it is the non-honorific pronoun that tends to use the second person. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the honorific and the non-honorific personal pronouns of address, such that one takes the second grammatical person, and the other takes the third grammatical person, it is the honorific pronoun that tends to use the third person. Q.E.D. 4. Note. The same deduction would apply to the pronouns serving to address more than one person, namely the non-honorific voi 'you, 2nd person plural' as against the 85

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

honorific Loro 'you, 3rd person plural.' 91. Italian. The affirmative imperative of the second person singular is a finite form, e.g. ridi 'laugh.' The corresponding negative imperative consists of the negative word non + infinitive, e.g. non ridere 'do not laugh.' The two syntactic variants: the affirmative imperative of the second person singular, and the corresponding negative imperative. 1. The assumptions of Naturalness Theory: 1.1. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 1.2. >sem (+finite, -finite) / verbal form I.e. with respect to semantic complexity, any finite form of the verb is more natural than any infinite form of the verb. (Mayerthaler et al. 1993, 145.) A special case of 1.2: 1.2.1. >sem (finite form, infinitive) / in Italian I.e. with respect to semantic complexity, any finite form is more natural than the infinitive, in Italian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the affirmative and negative imperatives concerning their formal structure, such that one is a finite verb, and the other an infinitive, it is the affirmative imperative that tends to assume the form of a finite verb. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the affirmative and negative imperatives concerning their formal structure, such that one is a finite verb, and the other an infinitive, it is the negative imperative that tends to be expressed with the infinitive. Q.E.D. 92. Japanese. Personal pronouns: 1SG watasi, 2SG anata, 3SG kono/sono/ano hito, 1PL watasitati, 2PL anatatati, 3PL kono/sono/ano hitotati . (Lyons 1999, 146.) The two syntactic variants: three forms v. one form within any person in the personal pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sem (three forms, one form) / of any person in the personal pronoun I.e. with respect to semantic complexity, three forms are more natural than one form, in any person of the personal pronoun.The scale has the format >sem (A + B, A) with respect to gender. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within any person in the personal pronoun, between 86

EXAMPLES

three forms and one form, such that one option is used in the third person, and the other option is used in the non-third person, it is three forms that tend to be used in the third person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within any person in the personal pronoun, between three forms and one form, such that one option is used in the third person, and the other option is used in the non-third person, it is one form that tends to be used in the non-third person. Q.E.D. 93. Japanese. Transitive clauses may express volitionality. The dative-nominative frame is used with verbs which describe ability, possession, emotion or the likeall nonvolitional by nature, e.g. 'Taroo understands English.' When these verbs express attempt, intention, wish or the like (with certain morphological and other changes), they can no longer take the dative-nominative frame. They take the nominative-accusative transitive frame, e.g. 'Taroo tried to understand English.' (Tsunoda 1999, 390.) The two syntactic variants: +/-volitionality and -volitionality. 1. The assumptions of Naturalness Theory: 1.1. >sem (nominative-accusative frame, dative-nominative frame) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative-accusative frame is more natural than the dative-nominative frame, in nominative-accusative languages.This follows from the basic properties of the nominative-accusative languages. 1.2. >sem (+/-volitionality, -volitionality) / transitive clauses I.e. with respect to semantic complexity, indifference to volitionality is more natural than non-volitionality, in transitive clauses.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between +/-volitionality and -volitionality, such that one option requires the nominative-accusative frame, and the other option requires the dativenominative frame, it is +/-volitionality that tends to require the nominative-accusative frame. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between +/-volitionality and -volitionality, such that one option requires the nominative-accusative frame, and the other option requires the dativenominative frame, it is -volitionality that tends to require the dative-nominative frame. Q.E.D. 94. Kekchi (a Mayan language of Guatemala). With common nouns, the definite article is li, e.g. li ixk 'the woman.' There are distinct definite-article forms with personal proper names, e.g. laj Manu 'Manuel,' lix Rosa 'Rosa.' (Lyons 1999, 123, referring to F. Eachus & R. Carlson.) The two syntactic variants: two variants v. one variant of the definite article. 1. The assumptions of Naturalness Theory: 1.1. >sem (two variants, one variant) / definite article in Kekchi I.e. with respect to semantic complexity, two variants of the definite article are more natural than only one variant of the definite article, in Kekchi.The scale has the format 87

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

>sem (A + B, A). 1.2. >sem (proper noun, common noun) I.e. with respect to semantic complexity, a proper noun is more natural than a common noun.This follows from the animacy hierarchy (Lyons 1999, 214, with references). A special case of 1.2: 1.2.1. >sem (personal proper name, common noun) I.e. with respect to semantic complexity, a personal proper name is more natural than a common noun. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the definite articles li and laj/lix , such that one kind is used with common nouns, and the other kind is used with personal proper names, it is the article laj/lix that tends to be used with personal proper names. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the definite articles li and laj/lix , such that one kind is used with common nouns, and the other kind is used with personal proper names, it is the article li that tends to be used with common nouns. Q.E.D. 95. Kharia (an Austroasiatic Munda language). Pronouns and +animate nouns have a number distinction, the remaining nouns do not. E.g. biloi 'cat,' biloiki 'cats' vs. sore 'stone(s).' (Croft 1990, 112, referring to Biligiri.) The two syntactic variants: pronoun & +animate noun, and -animate noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (+number distinction, -number distinction) / nouns and pronouns I.e. with respect to semantic complexity, nouns and pronouns showing the number distinction are more natural than nouns and pronouns lacking the number distinction.If a language has number distinction, its implementation is more natural than its lack. 1.2. >sem (personal pronoun & +animate noun, -animate noun) I.e. with respect to semantic complexity, a personal pronoun and a +animate noun are more natural than a -animate noun.Croft 1990, 112-3 [with references] on the animacy hierarchy. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between personal pronouns and +animate nouns on the one hand, and -animate nouns on the other, such that one kind expresses grammatical number, and the other kind does not express grammatical number, it is the pronouns and the +animate nouns that tend to express the number. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between personal pronouns and +animate nouns on the one hand, and -animate nouns on the other, such that one kind expresses grammatical number, and the other kind does not express grammatical number, it is the -animate nouns that tend not to express the number. Q.E.D.

88

EXAMPLES

96. KinyaRwanda (Bantu, Congo-Kordofanian language) and Nez Perce (one of the Penutian languages). Direct objects have a desinence. If the direct object is non-patient, even the verb obtains a desinence. (Givn 1995, 233.) The two syntactic variants: patient and non-patient as direct objects. 1. The assumptions of Naturalness Theory: 1.1. >sem (+patient, -patient) / direct object I.e. with respect to semantic complexity, the direct object as patient is more natural than the direct object as non-patient.The prototypical direct object is a patient. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (two desinences, one desinence) / marking direct object in KinyaRwanda and Nez Perce I.e. with respect to coding properties, two desinences marking the direct object are more natural than one desinence marking the direct object, in KynyaRwanda and Nez Perce. 2. The assumptions of the Slovenian Theory: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the patient and the non-patient (expressed in the direct object), such that one of them is marked with one desinence, and the other is marked with two desinences, it is the patient that tends to be marked with one desinence. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the patient and the non-patient (expressed in the direct object), such that one of them is marked with one desinence, and the other is marked with two desinences, it is the non-patient that tends to be marked with two desinences. Q.E.D. 97. Lakhota (a dialect of Dakota, one of the Siouan languages). The possessive prefixes attach directly to the noun only in expressions of inalienable possession, e.g. ni-ciye 'your (sg.) elder brother.' Otherwise the possessive prefixes are attached to a particle ta, and this is prefixed to the noun, e.g. mi-ta-ka k 'my dog.' (Lyons 1999, 127.) The two syntactic variants: the type ni-ciye , and the type mi-ta-ka k. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type mi-ta-ka k, the type ni-ciye ) / in Lakhota I.e. with respect to coding properties, the type mi-ta-ka k is more natural than the

89

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

type ni-ciye , in Lakhota. 1.2. >sem (-alienable, +alienable) / possession I.e. with respect to semantic complexity, inalienable possession is more natural than alienable possession. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type ni-ciye and the type mi-ta-ka k, such that one indicates alienable possession, and the other inalienable possession, it is the type mi-ta-ka k that tends to express alienable possession. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type ni-ciye and the type mi-ta-ka k, such that one indicates alienable possession, and the other inalienable possession, it is the type ni-ciye that tends to express inalienable possession. Q.E.D. 98. Lakhota (a dialect of Dakota, one of the Siouan languages). (a) Within the first person singular, the prefix for kinship terms is mi-; the prefix for body parts is ma-. (Lyons 1999, 129.) (b) There is no such duality in the second person singular.The present deduction deals with case (a). Case (b) is considered in deduction 99. The two syntactic variants: the prefix mi-, and the prefix ma-, both for inalienable possession in the first person singular. 1. The assumptions of Naturalness Theory: 1.1. >sem (body part, kinship relation) / inalienable possession I.e. with respect to semantic complexity, body parts are more natural than kinship relations, as instances of inalienable possession.This is based on the circumstance that some languages, e.g. Dyirbal (Lyons 1999, 129), treat kinship relations as alienable. 1.2. >sem (ma-, mi-) / inalienable possession in the first person singular, in Lakhota I.e. with respect to semantic complexity, ma- is more natural than mi- as an expression of inalienable possession in the first person singular, in Lakhota.This is based on the circumstance that ma- is specialized for inalienable possession. The prefix mi- is used for alienable possession as well, e.g. in 'my dog' (Lyons 1999, 127). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the prefixes ma- and mi- for inalienable possession in the first person singular, such that one prefix is used for body parts, and the other for kinship terms, it is the prefix ma- that tends to be used for body parts. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the prefixes ma- and mi- for inalienable possession in the first person singular, such that one prefix is used for body parts, and the other for kinship terms, it is the prefix mi- that tends to be used for kinship terms. Q.E.D. 99. Lakhota (a dialect of Dakota, one of the Siouan languages). (a) Within the first person singular, the prefix for kinship terms is mi-; the prefix for body parts is ma-. (Lyons 1999, 129.) (b) There is no such duality in the second person singular.The present 90

EXAMPLES

deduction deals with case (b). Case (a) is considered in deduction 98. The two syntactic variants: two prefixes v. one prefix, within expressions of inalienable possession in the non-third person. 1. The assumptions of Naturalness Theory: 1.1. >sem (two prefixes, one prefix) / inalienable possession in the non-third person singular of Lakhota I.e. with respect to semantic complexity, two prefixes is more natural than one prefix, as an expression of inalienable possession in the non-third person singular of Lakhota. The scale has the format >sem (A + B, A). 1.2. >sem (first, second) / grammatical person I.e. with respect to semantic complexity, the first grammatical person is more natural than the second. (Greenberg 1966, 44-5.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within expressions of inalienable possession in the nonthird person, between using two prefixes and using one prefix, such that one option is used in the first person, and the other option is used in the second person, it is two prefixes that tend to be used in the first person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within expressions of inalienable possession in the nonthird person, between using two prefixes and using one prefix, such that one option is used in the first person, and the other option is used in the second person, it is one prefix that tends to be used in the second person. Q.E.D. 100. Lakhota (a dialect of Dakota, one of the Siouan languages). In third-person singular expressions of inalienable possession, body parts enter the following structure: the full noun phrase possessor is placed before the head noun, and the latter is followed by the definite article, e.g. John poge k 'John's nose.' Kinship terms enter the following structure: the full noun phrase possessor is placed before the head noun, the latter has the prefix h-, and is followed by the definite article, e.g. John h-ku k 'John's mother.' (Lyons 1999, 129.) The two syntactic variants: the type John poge k, and the type John h-ku k. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type John h-ku k, the type John poge k ) / the third-person singular inalienable possession in Lakhota I.e. with respect to coding properties, the type John h-ku k is more natural than the type John poge k , in the third person singular expressions of inalienable possession, in Lakhota. 1.2. >sem (body part, kinship relation) / inalienable possession I.e. with respect to semantic complexity, body parts are more natural than kinship relations, as instances of inalienable possession.This is based on the circumstance that

91

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

some languages, e.g. Dyirbal (Lyons 1999, 129), treat kinship relations as alienable. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type John poge k and the type John h-ku k, such that one type expresses inalienable possession of body parts, and the other type expresses inalienable possession of kinship relations, it is the type John h-ku k that tends to express inalienable possession of kinship relations. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type John poge k and the type John h-ku k, such that one type expresses inalienable possession of body parts, and the other type expresses inalienable possession of kinship relations, it is the type John poge k that tends to express inalienable possession of body parts. Q.E.D. 101. Lakhota (a dialect of Dakota, one of the Siouan languages). Inalienable possession in the third person singular involving kinship terms: the full noun phrase possessor + possessum with the suffix -ku + the definite article k, e.g. John h-ku k 'John's mother;' the third person singular pronominal possessor is simply not expressed, e.g. h-ku k 'his/her mother.' (Lyons 1999, 129.) The two syntactic variants: the type John h-ku k, and the type h-ku k. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (the type John h-ku k, the type h-ku k) / in Lakhota I.e. with respect to coding properties, the type John h-ku k is more natural than the type h-ku k, in Lakhota. 1.1.2. >sym (-pronominal noun phrase, +pronominal noun phrase) I.e. with respect to coding properties, a non-pronominal noun phrase is more natural than a pronominal noun phrase. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the type John h-ku k and the type h-ku k, such that one type expresses non-pronominal possessors, and the other type expresses pronominal possessors, it is the type John h-ku k that tends to express non-pronominal possessors. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the type John h-ku k and the type h-ku k, such that one type expresses non-pronominal possessors, and the other type expresses pronominal possessors, it is the type h-ku k that tends to express pronominal possessors.

92

EXAMPLES

Q.E.D. 102. Lakhota (a dialect of Dakota, one of the Siouan languages). There is a +specific indefinite article w. Non-specific indefiniteness is expressed with the numeral 'one,' wi. (Lyons 1999, 99.) The two syntactic variants: the +specific indefinite w and the non-specific indefinite wi. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / unit I.e. with respect to coding properties, a unit of greater transparency is more natural than a corresponding unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (wi, w) / in Lakhota I.e. with respect to coding properties, wi is more natural than w.The former word has more sound body than the latter word. 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+specific, -specific) / indefinite I.e. with respect to semantic complexity, +specific is more natural than -specific, in indefinites.Everything +specific is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between w and wi as expressions of indefiniteness, such that one is +specific and the other -specific, it is wi that tends to express nonspecificness. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between w and wi as expressions of indefiniteness, such that one is +specific and the other -specific, it is w that tends to express specificness. Q.E.D. 103. Lappish, Southern. A definite direct object is in the accusative case, while an indefinite direct object is in the nominative case. (Schachter 1985, 41, referring to Wickman.) My interpretation: A definite direct object is in the accusative which formally does not equal the nominative, while an indefinite direct object is in the accusative case which formally equals the nominative. The two syntactic variants: definite direct object, and indefinite direct object. 1. The assumptions of Naturalness Theory: 1.1. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, the indefinite direct object is more natural than the definite direct object.In the spirit of Mayerthaler 1981, 14 and 1987, 42, concerning the markedness relations of the object as being the opposite of those typical of the subject.

93

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.2. >sem (conflated, separated) / accusative and nominative I.e. with respect to semantic complexity, accusative and nominative conflated is more natural than accusative and nominative separated.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the definite and indefinite direct objects, such that one takes the accusative case which equals the nominative case, and the other takes the accusative case which does not equal the nominative case, it is the indefinite direct object that takes the accusative case which equals the nominative case in form. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the definite and indefinite direct objects, such that one takes the accusative case which equals the nominative case, and the other takes the accusative case which does not equal the nominative case, it is the definite direct object that takes the accusative case which does not equal the nominative case in form. Q.E.D. 104. Latin. IUBERE 'command' takes the accusative with infinitive, the synonymous IMPERARE mostly an ut-clause. (Kopriva 1989, 216-7, 288.) The two syntactic variants: IUBERE + accusative with infinitive, and IMPERARE + ut-clause. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ut-clause, accusative with infinitive) / in Latin I.e. with respect to coding properties, an ut-clause is more natural than the accusative with infinitive, in Latin. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem (IUBERE, IMPERARE ) / in Latin I.e. with respect to semantic complexity, IUBERE is more natural than IMPERARE , in Latin. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between IUBERE and IMPERARE , such that one of them takes the accusative with infinitive, and the other an ut-clause, it is IMPERARE that tends to take an ut-clause. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between IUBERE and IMPERARE , such that one of 94

EXAMPLES

them takes the accusative with infinitive, and the other an ut-clause, it is IUBERE that tends to take the accusative with infinitive. Q.E.D. 105. Latin. The non-reflexive possessive pronouns of the non-third persons are adjectives, e.g. meus 'my,' tuus 'your.' Those of the third persons are genitives, e.g. eius 'his/her.' (Lyons 1999, 125.) The two syntactic variants: the possessive pronouns of the third person, and the possessive pronouns of the first and second persons (all non-reflexive). 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type meus, the type eius) / non-reflexive possessive in Latin I.e. with respect to coding properties, the type meus is more natural than the type eius, as non-reflexive possessive of Latin.The type meus, being inflected, is more transparent than the type eius. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the non-reflexive possessive pronouns of the third and non-third persons, such that one set consists of adjectives, and the other of genitives, it is the pronouns of the non-third person that tend to be adjectives. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the non-reflexive possessive pronouns of the third and non-third persons, such that one set consists of adjectives, and the other of genitives, it is the pronouns of the third person that tend to be genitives. Q.E.D. 4. Note. There is no such duality in the Latin reflexive series meus, tuus, suus. But cf. deduction 106. 106. Latin. Within the third person of pronominal possessives, the reflexive pronoun is an adjective, the non-reflexive a genitive. (Kopriva 1989, 50-60.) The two syntactic variants: the reflexive suus, and the non-reflexive type eius, within the third person of pronominal possessive. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (suus, the type eius) / pronominal possessives in Latin

95

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to coding properties, suus is more natural than the type eius, within the third-person pronominal possessives of Latin.The reflexive suus is inflected, therefore more transparent than eius. 1.2. >sem (-reflexive, +reflexive) / pronominal possessive I.e. with respect to semantic complexity, the non-reflexive pronominal possessive is more natural than the reflexive one.Some languages lack reflexive pronominal possessives. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the reflexive suus and the non-reflexive type eius within the third grammatical person, such that one set is inflected, and the other not, it is the reflexive suus that tends to be inflected. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the reflexive suus and the non-reflexive type eius within the third grammatical person, such that one set is inflected, and the other not, it is the non-reflexive eius that tends not to be inflected. Q.E.D. 107. Latin. In clauses of comparison, the standard of comparison is expressed (a) with an ablative case, e.g. melle dulcior 'sweeter than honey,' (b) with quam + noun phrase, e.g. dulcior quam mel 'sweeter than honey.' Case (a) is prevailingly used in instances bordering on the formulaic, as in the example above. (Szantyr [1965] 1972, 108.) The two syntactic variants: standard of comparison expressed with noun phrase in the ablative, and with quam + noun phrase. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (the type dulcior quam mel , the type melle dulcior) / in Latin I.e. with respect to coding properties, the type dulcior quam mel is more natural than the type melle dulcior, in Latin. 1.1.2. >sym (-formula, +formula) I.e. with respect to coding properties, a non-formula is more natural than a formula. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2, and 2.1 it can be deduced: 3.1. If there is any difference between the types dulcior quam mel and melle dulcior, such that one type is used in formulas, and the other type not, it is the type dulcior quam mel that does not tend to be used in formulas. Q.E.D. From 1.1.1-2, and 2.2 it can be deduced: 3.2. If there is any difference between the types dulcior quam mel and melle dulcior,

96

EXAMPLES

such that one type is used in formulas, and the other type not, it is the type melle dulcior that tends to be used in formulas. Q.E.D. 108. Latin. The ablative introduces the spatial source of the movement denoted by the verb, e.g. Roma profectus sum 'I set out from Rome.' If a common noun is substituted for the place name, the source is expressed by preposition + ablative, e.g. ab urbe profectus sum 'I set out from the city.' (Anderson 1999, 58-9.) The two syntactic variants: the type Roma 'from Rome,' and the type ab urbe 'from the city.' 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type ab urbe, the type Roma) / in Latin I.e. with respect to coding properties, the type ab urbe is more natural than the type Roma, in Latin. 1.2. >sem (proper noun, common noun) I.e. with respect to semantic complexity, a proper noun is more natural than a common noun. (Croft 1990, 112, with references.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between proper nouns and common nouns, such that one kind of nouns is used in expressions of source with the aid of a preposition, and the other kind of nouns is used in expressions of source without the aid of a preposition, it is the common noun that tends to employ the preposition. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between proper nouns and common nouns, such that one kind of nouns is used in expressions of source with the aid of a preposition, and the other kind of nouns is used in expressions of source without the aid of a preposition, it is the proper noun that tends not to employ the preposition. Q.E.D. 109. Macedonian. When the direct object is definite, clitic doubling occurs, e.g. ja vidov maka-ta 'I saw the cat.' (Lyons 1999, 140.) Cf. vidov maka 'I saw a cat.' The two syntactic variants: the type vidov maka , and the type ja vidov maka-ta. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (+clitic doubling, -clitic doubling) / in Macedonian

97

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to coding properties, clitic doubling is more natural than its absence, in Macedonian. 1.2. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, an indefinite direct object is more natural than a definite direct object. (In the spirit of Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type ja vidov maka-ta and the type vidov maka , such that one type shows clitic doubling, and the other lacks clitic doubling, it is the type containing a definite direct object that tends to show clitic doubling. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type ja vidov maka-ta and the type vidov maka , such that one type shows clitic doubling, and the other lacks clitic doubling, it is the type containing an indefinite direct object that tends to lack clitic doubling. Q.E.D. 110. Macedonian. While the l-tenses of the copular 'be' are acceptable in the standard language, the perfect tenses of the copular 'be' are not. (Hendriks 1976, 226 [end of page], Friedman 1976, 98.) The two syntactic variants: most verbal lexemes v. most verbal lexemes except the copular 'be'. 1. The assumptions of Naturalness Theory: 1.1. >sem (the l-tenses, the perfect tenses) / in Macedonian I.e. with respect to semantic complexity, the l-tenses are more natural than the perfect tenses, in Macedonian.This is based on the circumstance that the perfect tenses (auxiliary 'have') are unquestionably of younger origin than the l-tenses (auxiliary 'be'), Lunt 1952, 100 fn. 1.2. >sem (the copular verb 'be' & most other verbal lexemes, most other verbal lexemes) / in Macedonian I.e. with respect to semantic complexity, contracting almost any verbal lexeme is more natural than contracting most verbal lexemes except 'be,' in Macedonian.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between contracting most verbal lexemes and contracting most verbal lexemes expect the copular 'be', such that one option is used in the l-tenses, and the other option is used in the perfect tenses, it is most verbal lexemes that tend to be used in the l-tenses. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between contracting most verbal lexemes and contracting most verbal lexemes expect the copular 'be', such that one option is used in the l-tenses, and the other option is used in the perfect tenses, it is most verbal lexemes except the copular 'be' that tend to be used in the perfect tenses. Q.E.D.

98

EXAMPLES

111. Miriam (a language of Papua-New Guinea). The number marker appears only on the verb. The verb encodes the number of its subject (a four-way singular/dual/trial/ plural contrast), and its object (singular and plural only). Thus, for example, irmile means 'one follows one,' irmirdare means 'three follow one,' dirmiriei means 'two follow many,' and so on. (Cruse 1999, 268.) The two syntactic variants: the set singular, plural, dual & trial, and the set singular & plural. 1. The assumptions of Naturalness Theory: 1.1. >sem (subject, object) I.e. with respect to semantic complexity, the subject is more natural than the object. (Mayerthaler 1981, 14.) 1.2. >sem (singular, plural, dual, and trial; singular and plural) I.e. with respect to semantic complexity, having singular, plural, dual, and trial is more natural than having just singular and plural.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the set singular, plural, dual & trial and the set singular & plural, such that one set encodes the number of the subject, and the other set encodes the number of the object, it is the set singular, plural, dual & trial that tends to encode the number of the subject. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the set singular, plural, dual & trial and the set singular & plural, such that one set encodes the number of the subject, and the other set encodes the number of the object, it is the set singular & plural that tends to encode the number of the object. Q.E.D. 112. Norwegian, Bokml. The definite article is a suffix on the noun, e.g. stol-en 'the chair.' If the noun is preceded by an adjective, the chief rule is that the definite article in addition assumes the form of a word, e.g. den lille stol-en 'the little chair.' (Faarlund et al. 1997, 173 ff., 289, 296 ff.) Similarly in Swedish. The two syntactic variants: the definite article as a suffix only, and as a word + suffix. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (two-part, single-part) / definite article in Bokml I.e. with respect to coding properties, a two-part definite article is more natural than a single-part definite article, in Bokml. 1.1.2. >sym (adjective + head noun, head noun) / in Bokml I.e. with respect to coding properties, the adjective + head noun is more natural than

99

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

the bare head noun, in Bokml. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the definite article in the form of a suffix only, and the definite article consisting of a word + suffix, such that one is used in the construction adjective + head noun, and the other with the bare head noun, it is the definite article consisting of word + suffix that tends to be used in the construction adjective + head noun. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the definite article in the form of a suffix only, and the definite article consisting of a word + suffix, such that one is used in the construction adjective + head noun, and the other with the bare head noun, it is the definite article in the form of a suffix that tends to be used with the bare head noun. Q.E.D. 113. Paiute, Southern (Paiute is a Uto-Aztecan language spoken in small areas of the USA). Demonstrative and third-person personal pronouns are not distinguished, e.g. aa 'that one' and 'he;' ia 'this one' and 'he;' ar 'that one' and 'it.' (Schachter 1985, 30, referring to Sapir.) Since the forms enumerated indicate distant vs. near reference, they are primarily demonstrative pronouns, in addition serving as personal pronouns. Similarly in Latin. The two syntactic variants: third-person personal and demonstrative pronoun, and non-third-person personal pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sem (conflated, separated) / personal and demonstrative pronoun I.e. with respect to semantic complexity, a personal and a demonstrative pronoun conflated is more natural than a personal and a demonstrative pronoun separated.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between being a personal OR a demonstrative pronoun and being a personal AS WELL AS demonstrative pronoun, such that one option is used in the third person, and the other option is used in the non-third person, the option of being a personal AND a demonstrative pronoun tends to be used in the third person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between being a personal OR a demonstrative pronoun and being a personal AS WELL AS demonstrative pronoun, such that one option is used

100

EXAMPLES

in the third person, and the other option is used in the non-third person, the option of being a personal OR a demonstrative pronoun tends to be used in the non-third person. Q.E.D. 114. Persian. In compound verbal tenses, the negative prefix na- is added to the finite verb. Only the verb 'be' used as finite auxiliary cannot be thus negated. (The negative prefix is in that case added to the infinite verb.) (Boyle 1966, 39.) The two morphological variants: a finite verb that can be negated, and a finite verb that cannot be negated. 1. The assumptions of Naturalness Theory: 1.1. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.1: 1.1.1. >sem (the auxiliary 'be,' most other verbal lexemes) / in Persian I.e. with respect to semantic complexity, the auxiliary 'be' is more natural than most other verbal lexemes, in Persian. 1.2. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+[na- + finite verb], -[na- + finite verb]) / in Persian I.e. with respect to coding properties, a finite verb that can be negated is more natural than a finite verb that cannot be negated, in Persian. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between finite verbs that can be negated and finite verbs that cannot be negated, such that the difference is between the auxiliary 'be' and the remaining verbs, it is the finite auxiliary 'be' that tends not to be negated. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between finite verbs that can be negated and finite verbs that cannot be negated, such that the difference is between the auxiliary 'be' and the remaining verbs, it is the remaining verbs that tend to be negated. Q.E.D. 115. Quechua (an Andean language spoken in South America). The equivalent of subject and object personal pronouns is expressed by affixes on the verbs, e.g. maqa-manki hit-me-you 'you hit me.' Such pronominal affixes may co-occur with corresponding non-affixed pronouns when a pronominal subject or object is being emphasized, e.g. qam noqata maqa-ma-nki you me hit-me-you 'YOU hit ME.' (Schachter 1985, 26.) The two syntactic variants: subject and object personal pronouns expressed by affixes on the verb, and expressed as words. 1. The assumptions of Naturalness Theory:

101

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (word, morpheme) I.e. with respect to coding properties, a word is more natural than a morpheme. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between a subject or object pronoun expressed as a word and a subject or object pronoun expressed by an affix, such that one kind is emphatic, and the other not, it is the subject or object pronoun expressed as a word that tends to be emphatic. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between a subject or object pronoun expressed as a word and a subject or object pronoun expressed by an affix, such that one kind is emphatic, and the other not, it is the subject or object pronoun expressed by an affix that tends to be nonemphatic. Q.E.D. 116. Quich (a Mayan language of Mexico). (a) Only animate nouns have plural verb agreement, and (b) only human nouns allow plural marking. Inanimate nouns do not admit plural verb agreement or plural marking. (Croft 1990, 117.) This deduction deals with case (a); case (b) is treated in deduction 117. The two syntactic variants: animate and inanimate nouns. 1. The assumptions of Naturalness Theory: 1.1. >sem (+animate, -animate) I.e. with respect to semantic complexity, +animate is more natural than -animate. (Mayerthaler 1981, 14.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+plural verb agreement, -plural verb agreement) I.e. with respect to coding properties, plural verb agreement is more natural than the lack of plural verb agreement. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences:

102

EXAMPLES

From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between animate and inanimate nouns, such that one kind of nouns has plural verb agreement, and the other kind of nouns lacks plural verb agreement, it is the animate nouns that tend to have plural verb agreement. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.1. If there is any difference between animate and inanimate nouns, such that one kind of nouns has plural verb agreement, and the other kind of nouns lacks plural verb agreement, it is the inanimate nouns that tend to lack plural verb agreement. Q.E.D. 4. Note. Cf. note 4 of deduction 117. 117. Quich (a Mayan language of Mexico). (a) Only animate nouns have plural verb agreement, and (b) only human nouns allow plural marking. Inanimate nouns do not admit plural verb agreement or plural marking. (Croft 1990, 117.) This deduction deals with case (b); case (a) is treated in deduction 116. The two syntactic variants: noun showing two numbers v. noun showing only one number. 1. The assumptions of Naturalness Theory: 1.1. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 1.2. >sem (two numbers, one number) / noun I.e. with respect to semantic complexity, a noun showing two numbers is more natural than a noun showing only one number.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a noun showing two numbers and a noun showing only one number, such that one noun is +human, and the other noun is -human, it is the noun showing two numbers that tends to be +human. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a noun showing two numbers and a noun showing only one number, such that one noun is +human, and the other noun is -human, it is the noun showing only one number that tends to be -human. Q.E.D. 4. Note. It can be seen from the Quich cases (a) and (b) that verb agreement is more widespread than plural marking of nouns. This can be explained with the scale >sem (verb agreement, plural marking of nouns), according to which verb agreement is more natural than plural marking of nouns. The scale is supported by the circumstance that, crosslinguistically, verb agreement is often found where plural marking of nouns is absent, e.g. the sheep are grazing, whereas any plural marking of nouns that does not trigger the corresponding verb agreement is not a valid case of plural marking, e.g. the data is surprising. 118. Quileute (one of the Chimakuan languages, spoken in Washington). Proper nouns are accompanied in the oblique case by what seems to be the indefinite article. (Lyons 1999, 123-4, referring to Andrade.) The two syntactic variants: casus rectus of proper nouns, and casus obliquus of proper

103

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

nouns. 1. The assumptions of Naturalness Theory: 1.1. >sem (rectus, obliquus) / case I.e. with respect to semantic complexity, casus rectus is more natural than casus obliquus. (Mayerthaler 1981, 14.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (indefinite article + proper noun, bare proper noun) / in Quileute I.e. with respect to coding properties, the indefinite article + proper noun is more natural than the bare proper noun, in Quileute. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between proper nouns in casus rectus and casus obliquus, such that one casus is accompanied by the indefinite article, and the other casus not, it is casus rectus that tends not to be accompanied by the indefinite article. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between proper nouns in casus rectus and casus obliquus, such that one casus is accompanied by the indefinite article, and the other casus not, it is casus obliquus that tends to be accompanied by the indefinite article. Q.E.D. 119. Romance languages. While the verb agrees with any subject, it agrees only with pro objects. (Lyons 1999, 140.) E.g. French je l'ai mise dans le tiroir 'I have put it [= la lettre 'letter (fem.)] into the drawer.' The two syntactic variants: the set of noun phrases and the set comprising only pro. 1. The assumptions of Naturalness Theory: 1.1. >sem (subject agreement, object agreement) I.e. with respect to semantic complexity, subject agreement is more natural than object agreement.Subject agreement is commoner cross-linguistically than object agreement. 1.2. >sem (pro & other noun phrase, pro) I.e. with respect to semantic complexity, involving any noun phrase is more natural than involving just pro.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the set of noun phrases and the set comprising only pro, such that one set contracts subject agreement, and the other set contracts object agreement, it is the set of noun phrases that tends to contract subject agreement. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the set of noun phrases and the set comprising

104

EXAMPLES

only pro, such that one set contracts subject agreement, and the other set contracts object agreement, it is the set comprising only pro that tends to contract object agreement. Q.E.D. 120. Romance languages. In dates, the cardinal number is used, e.g. Italian il cinque ottobre 'the fifth of October,' except that ordinalia are used in low numbers, such as 'first, second.' (Mayerthaler 1981, 168.) The two syntactic variants: the type il primo ottobre, and the type il cinque ottobre. 1. The assumptions of Naturalness Theory: 1.1. >sem (ordinal number, cardinal number) / dates I.e. with respect to semantic complexity, an ordinal number is more natural than a cardinal number, in dates.This is based on the assumption that, cross-linguistically, it is the ordinal numbers that prevail in dates. 1.2. >sem (+low number, -low number) I.e. with respect to semantic complexity, low numbers are more natural than non-low numbers. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between cardinal and ordinal numbers in dates, such that the difference involves low and non-low numbers, it is the low numbers that tend to be expressed with ordinalia. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between cardinal and ordinal numbers in dates, such that the difference involves low and non-low numbers, it is the non-low numbers that tend to be expressed with cardinalia. Q.E.D. 4. Note. I am not certain that the matter discussed in this deduction is connatural in the sense of Mayerthaler et al. 1998, 185, referring to Bailey. 121. Romanian. The definite article is systematically omitted after a preposition, e.g. sub o mas 'under a table,' but sub mas 'under the table.' (Lyons 1999, 51, referring to Murrell & tefnescu-Drgneti.) The two syntactic variants: the construction preposition + definite head noun, and the construction preposition + indefinite head noun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (preposition + article + head noun, preposition + head noun) / in Romanian I.e. with respect to coding properties, the construction preposition + article + head noun is more natural than the construction preposition + head noun, in Romanian. 1.2. >sem (+presupposed, -presupposed)

105

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to semantic complexity, +presupposed is more natural than presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the preposition + definite head noun and the preposition + indefinite head noun, such that the article is expressed in one of them, and not in the other, it is the preposition + indefinite head noun that tends to have the article expressed. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the preposition + definite head noun and the preposition + indefinite head noun, such that the article is expressed in one of them, and not in the other, it is the preposition + definite head noun that tends to lack the article. Q.E.D. 4. Note. There is a complication in place names. See deduction 122. 122. Romanian. The definite article is systematically omitted after a preposition, e.g. sub o mas 'under a table,' but sub mas 'under the table.' (Lyons 1999, 51, referring to Murrell & tefnescu-Drgneti.) This matter is dealt with in deduction 121. There is a complication in place names. Masculine place names follow the above rule, and omit the suffixed definite article after a preposition, e.g. din Bucureti 'from Bucharest.' But most feminine place names keep the suffixed definite article - a, e.g. din Constana 'from Constana.' (Lyons 1999, 123 fn.) This matter is considered in the present deduction. The two syntactic variants: preposition + place name, and preposition + place name + suffixed definite article. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (preposition + place name + suffixed definite article, preposition + place name) / in Romanian I.e. with respect to coding properties, the construction preposition + place name + suffixed definite article is more natural than the construction preposition + place name, in Romanian. 1.2. >sem (masculine, feminine) I.e. with respect to semantic complexity, the masculine is more natural than the feminine. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 106

EXAMPLES

2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the construction preposition + masculine place name and the construction preposition + feminine place name, such that one takes the suffixed definite article, and the other does not, it is the preposition + feminine place name that tends to take the article. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the construction preposition + masculine place name and the construction preposition + feminine place name, such that one takes the suffixed definite article, and the other does not, it is the preposition + masculine place name that tends to lack the article. Q.E.D. 123. Romanian. In the construction attribute + head noun, definiteness is mostly expressed with the suffixed definite article (in the Wackernagel position), e.g. buna carte 'the good book.' An exception: if the attribute is a cardinal numeral, definiteness is expressed with a free-form determiner at the beginning of the construction, e.g. cei patru prieteni 'the four friends.' (Lyons 1999, 75.) The two syntactic variants: attribute + head noun, and determiner + attribute + head noun, both constructions containing a marker of definiteness. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (determiner + attribute + head noun, attribute + head noun) / +definite noun phrase, in Romanian I.e. with respect to coding properties, the construction determiner + attribute + head noun is more natural than the construction attribute + head noun, as +definite noun phrase, in Romanian. 1.2. >sem (adjective, cardinal number) / attribute I.e. with respect to semantic complexity, an adjective is more natural than a cardinal number, as attribute.The adjective is the prototypical attribute. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the construction determiner + attribute + head noun and the construction attribute + head noun, such that the attribute is either an adjective or a cardinal number, it is the construction determiner + attribute + head noun that tends to contain a cardinal number as attribute. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the construction determiner + attribute + head noun and the construction attribute + head noun, such that the attribute is either an adjective or a cardinal number, it is the construction attribute + head noun that tends to contain an adjective as attribute. Q.E.D.

107

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

124. Romanian. Personal names occurring as subject and direct object do not require an article, but in "oblique" functions (essentially indirect object and possessive) they are preceded by the form lui. E.g. lui Ana 'to/for/of Ana.' (Lyons 1999, 123 fn., referring to Murrell & tefnescu-Drgneti.) The two syntactic variants: the type Ana, and the type lui Ana. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type lui Ana, the type Ana) / Romanian personal names I.e. with respect to coding properties, the type lui Ana is more natural than the type Ana, in Romanian personal names. 1.2. >sem (nominative/accusative, other cases) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative/accusative is more natural than other cases, in nominative-accusative languages. (Mayerthaler 1987, 41; Mayerthaler et al. 1998, 167.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between definite article + personal name and bare personal name, concerning whether the definite article is used in the subject and direct-object functions of the noun, or in the oblique functions of the noun, it is the type definite article + personal name that tends to be used in the oblique functions of the noun. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between definite article + personal name and barepersonal name, concerning whether the definite article is used in the subject and direct-object functions of the noun, or in the oblique functions of the noun, it is the type bare personal name that tends not to be used in the oblique functions of the noun. Q.E.D. 125. Romanian. In feminine names ending in a (which happens to be identical to the feminine definite article suffix), that a is optionally changed to ei (one of the oblique forms of the feminine definite article suffix) instead of being preceded by the more general definite article lui of oblique forms, see deduction 124. Thus there is (oblique) lui Ana and Anei . (Lyons 1999, 123 fn., referring to Murrell & tefnescu-Drgneti.) The two syntactic variants: the definite articles - ei & lui v. only the definite article lui, within the oblique functions of feminine personal names. 1. The assumptions of Naturalness Theory: 1.1. >sem (-ei and lui, only lui) / definite article in oblique functions of Romanian feminine personal names I.e. with respect to semantic complexity, feminine personal names which admit the oblique definite articles lui and -ei are more natural than feminine personal names which admit only the oblique definite article lui, in Romanian.The scale has the format >sem (A + B, A).

108

EXAMPLES

1.2. >sem (feminine in -a, other feminine) / personal names of Romanian I.e. with respect to semantic complexity, feminine names in - a are more natural than other feminine names, in Romanian.Feminine names in - a are the normal option of feminine personal names in Romanian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the oblique functions of feminine personal names, between contracting the definite articles - ei & lui, and contracting only the definite article lui, such that one option occurs with feminine names in - a, and the other option occurs with "other" feminine names, it is the definite articles - ei and lui that tend to be used with feminine names in -a. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference, within the oblique functions of feminine personal names, between contracting the definite articles - ei & lui, and contracting only the definite article lui, such that one option occurs with feminine names in - a, and the other option occurs with "other" feminine names, it is the definite article lui that tends to be used with "other" feminine names. Q.E.D. 126. Romanian. In compound verbal tenses, the clitic personal pronoun mostly leans proclitically on the auxiliary verb, e.g. mi-a plcut 'I came to like (it).' This applies also to the clitic personal pronoun of the third person singular masculine, e.g. mi l-a artat 'he showed him (it) to me.' The clitic personal pronoun of the third person singular feminine leans enclitically on the (main-verb) participle, e.g. mi a spus-o ieri 'he told it to me yesterday;' thus also in the present tense of the conditional mood. (T. Repina 1968, 61-2. My attention was drawn to this matter by Mitja Skubic viva voce, 2000.) The two syntactic variants: the type mi l-a arrat, and the type mi a spus-o ieri . 1. The assumptions of Naturalness Theory: 1.1. >sem (masculine, feminine) I.e. with respect to semantic complexity, the masculine is more natural than the feminine. (Mayerthaler 1981, 15.) 1.2. >sym (+adjacency, -adjacency) / of units that belong together I.e. with respect to coding properties, adjacency of units that belong together is more natural than their non-adjacency.Behaghel's Law (Behaghel 1932, 4). A special case of 1.2: 1.2.1. >sym (main verb + object clitic, object clitic main verb) / in Romanian I.e. with respect to coding properties, the main verb + its object clitic is more natural than the main verb separated from its object clitic, in Romanian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type mi l-a arrat and the type mi a spus-o ieri, such that in one type the object clitic leans on the main verb, and in the other type the object clitic does not lean on the main verb (whereby one clitic is masculine and the other feminine), it is the type mi l-a arrat that tends to separate the (masculine) object clitic from the main verb. Q.E.D. 109

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type mi l-a arrat and the type mi a spus-o ieri, such that in one type the object clitic leans on the main verb, and in the other type the object clitic does not lean on the main verb (whereby one clitic is masculine and the other feminine), it is the type mi a spus-o ieri that tends to link the (feminine) object clitic to the main verb. Q.E.D. 127. Russian, Old, Northern. Non-animate direct objects of the infinitive stand in the nominative case, e.g. zeml'a paxati land (nom. sg.) plough (inf.) 'to plough the land.' The construction can only be observed in the feminine singular, for morphological reasons. (Tschernych 1957, 265-7; M. Sokolova 1962, 260-2; Ivanov 1964, 429-39; Borkovskij & Kuznecov 1965, 455-9.) The two syntactic variants: zeml'u paxaju 'I plough the land,' and zeml'a paxati 'to plough the land.' 1. The assumptions of Naturalness Theory: 1.1. >sem (+finite, -finite) / verbal form I.e. with respect to semantic complexity, a finite verbal form is more natural than a non-finite verbal form. (Mayerthaler et al. 1993, 145.) A special case of 1.1: 1.1.1. >sem (finite verbal form, infinitive) / in Old Northern Russian I.e. with respect to semantic complexity, a finite verbal form is more natural than an infinitive, in Old Northern Russian. 1.2. >sem (accusative, nominative) / as the case of the direct object I.e. with respect to semantic complexity, the accusative is more natural than the nominative, as the case of the direct object. (In the spirit of Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (accusative, nominative) / as the case of the inanimate direct object in Old Northern Russian I.e. with respect to semantic complexity, the accusative is more natural than the nominative, as the case of the inanimate direct object in Old Northern Russian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the inanimate direct object of a finite verb, and the inanimate direct object of an infinitive (such that the choice is between the direct object in the accusative and in the nominative), it is the inanimate direct object of the finite verb that tends to be in the accusative case. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the inanimate direct object of a finite verb, and the inanimate direct object of an infinitive (such that the choice is between the direct object in the accusative and in the nominative), it is the inanimate direct object of the infinitive that tends to be in the nominative case. Q.E.D. 128. Russian. Gender agreement between the attributive adjective and the noun is limited to the singular. E.g. bol'-oj avtomobil' (masculine) 'large car,' bol'-aja maina (feminine) 'large car,' bol'-oe taksi (neuter) 'large taxi,' as against bol'-ie (plural of all three genders) 'large.' (Corbett 1999, 12-3.) The two syntactic variants: visible and non-visible gender of adjectives in the construction adjective + noun. 110

EXAMPLES

1. The assumptions of Naturalness Theory: 1.1. >sem (+visible, -visible) / gender of adjective in the construction adjective + noun I.e. with respect to semantic complexity, visible gender is more natural than nonvisible gender, in adjectives within the construction adjective + noun.In languages that do have gender distinctions in attributive adjectives, it is most sem-natural that that gender is visible. 1.2. >sem (singular, plural) I.e. with respect to semantic complexity, singular is more natural than plural. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between visible and non-visible gender of adjectives in the construction adjective + noun, such that one kind of gender occurs in the singular, and the other kind of gender occurs in the plural, it is visible gender that tends to occur in the singular. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between visible and non-visible gender of adjectives in the construction adjective + noun, such that one kind of gender occurs in the singular, and the other kind of gender occurs in the plural, it is non-visible gender that tends to occur in the plural. Q.E.D. 129. Russian. Russian favors SVO in the formal written language but SOV in the colloquial spoken language. (A. Siewierska 1999, 413. Partly confirmed in E. Zemskaja 1979, 149-50.) The two syntactic variants: the basic element order SVO, and the basic element order SOV. 1. The assumptions of Naturalness Theory: 1.1. >sem (conversation, the other registers) I.e. with respect to semantic complexity, conversation is more natural than other registers. (Dotter 1990, 228.) 1.2. >sem (SOV, SVO) / basic element order I.e. with respect to semantic complexity, the basic element order SOV is more natural than the basic element order SVO. (Mayerthaler et al. 1998, 309. Cf. the discussion in A. Siewierska 1999, 412-3.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between conversation and the other registers concerning their basic element order, and the choice is restricted to the basic element orders SVO and SOV, it is conversation that tends to favour SOV. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between conversation and the other registers concerning their basic element order, and the choice is restricted to the basic element orders SVO and SOV, it is the other registers that tend to favour SVO. Q.E.D.

111

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

130. Sanskrit. In the early history of Sanskrit, univerbation of the adverbial particle and the verb takes place, e.g. pr jaje 'was born forth' becomes prajaje , whereas the univerbation of the adverbial particle + noun does not come about. (Hock 1991, 337-8.) The two syntactic variants: adverbial particle + verb, and adverbial particle + noun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (particle + [(X +) head noun], particle + verb) I.e. with respect to coding properties, particle + (X +) noun is more natural than particle + verb.X represents the material that can intervene between the head noun and the beginning of its noun phrase. 1.1.2. >sym (two words, corresponding one word) I.e. with respect to coding properties, two words are more natural than their univer bation. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between adverbial particle + verb and adverbial particle + noun, such that one of them is subjected to univerbation, and the other not, it is the adverbial particle + noun that tends not to undergo univerbation. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between adverbial particle + verb and adverbial particle + noun, such that one of them is subjected to univerbation, and the other not, it is the adverbial particle + verb that tends to undergo univerbation. Q.E.D. 131. Sanskrit. In clauses expressing past time by aid of the intransitive passive parti ciple, e.g. na gatah 'he did not go,' any pronominal subject is expressed, except if it is of the third person; in the latter case it is usually omitted. (Coulson 1976, 60.) The two syntactic variants: pronominal subject + intransitive passive participle, and the bare intransitive passive participle, in expressions of past time. 1. The assumptions of Naturalness Theory: 1.1. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (pronominal subject + verb, verb with covert pronominal subject)

112

EXAMPLES

I.e. with respect to coding properties, pronominal subject + verb is more natural than the bare verb. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference among the third and the non-third grammatical persons concerning the expression of the pronominal subject, it is the third grammatical person that tends not to be accompanied by a pronominal subject. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference among the third and the non-third grammatical persons concerning the expression of the pronominal subject, it is the non-third grammatical person that tends to be accompanied by a pronominal subject. Q.E.D. 132. Sanskrit. The finite verb of main clauses is unaccented (except when clauseinitial), the finite verb of dependent clauses is accented. (Delbrck 1888, 35 ff.) The two syntactic variants: the finite verb in a main clause, and the finite verb in a dependent clause. 1. The assumptions of Naturalness Theory: 1.1. >sem (main, dependent) / clause I.e. with respect to semantic complexity, a main clause is more natural than a depen dent clause.Main clauses are phylogenetically earlier than dependent clauses. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+accented, -accented) / word I.e. with respect to coding properties, an accented word is more natural than an unaccented word. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between finite verbs in main clauses and finite verbs in dependent clauses, such that one kind of finite verbs is accented, and the other kind is not accented, it is the finite verb in the main clause that tends to be unaccented. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between finite verbs in main clauses and finite verbs in dependent clauses, such that one kind of finite verbs is accented, and the other kind is not accented, it is the finite verb in the dependent clause that tends to be accented. Q.E.D. 4. Note. Cf. deduction 133. 133. Sanskrit. In main clauses, the finite verb is accented when clause-initial: otherwise it is unaccented (and then mostly clause-final). (Delbrck 1888, 35 ff.) Sanskrit

113

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

is an SOV-language. The two syntactic variants: the accented finite verb, and the unaccented finite verb, in main clauses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Ma yerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (clause-initial, clause-final) / position I.e. with respect to coding properties, clause-initial position is more natural than clause-final position. 1.1.2. >sym (+accented, -accented) I.e. with respect to coding properties, being accented is more natural than being unaccented. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the clause-initial and clause-final finite verb, such that one kind is accented, and the other kind is not accented, it is the clause-initial finite verb that tends to be accented. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the clause-initial and clause-final finite verb, such that one kind is accented, and the other kind is not accented, it is the clause-final finite verb that tends to be unaccented. Q.E.D. 4. Note. Cf. deduction 132. 134. Slovenian. If the clause contains the characterizing copular 'be' and a subject complement, the controller of person, number and gender agreement is the subject, e.g. Breice so bile mesto 'Breice was a town,' where the subject Breice is in the third person plural feminine, therefore the target of agreement, namely the complex form of the copular verb so bile , is in the third person plural feminine as well. However, if the copular 'be' is identifying, the controller of person, number and gender agreement is the subject complement (the so-called back agreement, Corbett 1999, 14), e.g. to mesto so bile Breice 'this town was Breice,' where the subject complement Breice is in the third person plural feminine, therefore the target of agreement, namely the complex form of the copular verb so bile , is in the third person plural feminine as well. (Some of this data from Toporii 1976, 478; 2000, 609.) The two syntactic variants: subject and subject complement as controllers of person, number and gender agreement in clauses containing the copular 'be.' 1. The assumptions of Naturalness Theory: 1.1. >sem (subject, subject complement) / controller of agreement I.e. with respect to semantic complexity, the subject as controller of agreement is more natural than the subject complement as controller of agreement.The subject is the normal controller of agreement.

114

EXAMPLES

1.2. >sem (characterizing, identifying) / copula I.e. with respect to semantic complexity, a characterizing copula is more natural than an identifying copula. (Mayerthaler et al. 1998, 197.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between subject and subject complement as controllers of person, number and gender agreement in clauses containing the copular 'be,' such that one controller is used when the copula is characterizing, and the other controller is used when the copula is identifying, it is the subject that tends to be used as controller when the copula is characterizing. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between subject and subject complement as controllers of person, number and gender agreement in clauses containing the copular 'be,' such that one controller is used when the copula is characterizing, and the other controller is used when the copula is identifying, it is the subject complement that tends to be used as controller when the copula is identifying. Q.E.D. 4. Note. A possibly related matter is the back agreement between the subject complement and the pronominal subject in Latin (regularly, sed haec mea culpa est 'but this is my fault'), in Ancient Greek (less regularly), in Italian ( questa la mia casa 'this is my house'), in English (these are vain wishes). (Brandenstein 1966, 141.) 135. Slovenian. The affirmative imperative can be formed from verbal lexemes of per fective and imperfective aspects, e.g. ustreli (pf.), streljaj (ipf.) both in different readings of 'shoot (imperative).' In the negative imperative, it is mostly the imperfective aspect that is used, e.g. ne streljaj 'do not shoot.' (Toporii 1976, 433; 2000, 350-1, 397; Herrity 2000, 188.) The two syntactic variants: the affirmative imperative, and the negative imperative. 1. The assumptions of Naturalness Theory: 1.1. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15.) 1.2. >sem (perfective & imperfective, imperfective) / aspect of verbal forms in Slovenian I.e. with respect to semantic complexity, a verbal form admitting both perfective and imperfective aspect is more natural than a verbal form admitting only the imperfective aspect.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a verbal form admitting both perfective & imperfective aspect and a verbal form admitting only the imperfective aspect, such that one verbal form is affirmative, and the other verbal form is negative, it is the verbal form admitting both perfective & imperfective aspect that tends to be affirmative. Q.E.D.

115

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a verbal form admitting both perfective & imperfective aspect and a verbal form admitting only the imperfective aspect, such that one verbal form is affirmative, and the other verbal form is negative, it is the verbal form admitting only the imperfective aspect that tends to be negative. Q.E.D. 136. Slovenian. The masculine singular nominative of the cardinal numeral 'one' is en when accompanied by a mention of the unit counted, and eden when not accompanied by a mention of the unit counted, e.g. samo en lovek je priel 'only one man came,' samo eden je priel 'only one came.' (Cf. Toporii 2000, 330.) The two syntactic variants: eden, en, the masculine singular nominative of 'one.' 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater transparency is more natural than a corresponding syntactic unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (eden, en) / masculine nominative singular of 'one' in Slovenian I.e. with respect to coding properties, the masculine nominative singular eden is more natural than the masculine nominative singular en, in Slovenian. Eden is more transparent than en, because eden has more sound body and internal structure. 1.2. >sem (+mention, -mention) / of the unit counted with the cardinal numeral I.e. with respect to semantic complexity, a cardinal numeral + the unit counted is more natural than a cardinal numeral not accompanied by the unit counted.The prevalent typological situation is that the cardinal numeral is accompanied by a mention of the unit counted. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the forms of masculine singular eden and en, such that one form is accompanied by a mention of the unit counted, and the other form is not accompanied by a mention of the unit counted, it is the form eden that tends not to be accompanied by a mention of the unit counted. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the forms of masculine singular eden and en, such that one form is accompanied by a mention of the unit counted, and the other form is not accompanied by a mention of the unit counted, it is the form en that tends to be accompanied by a mention of the unit counted. Q.E.D. 4. Note. Cf. deduction 137. 137. Slovenian. The masculine nominative singular cardinal numeral 'one' is en in adjective function, and eden when it is the head of a noun phrase, e.g. samo eden je priel 'only one came.' (Cf. Toporii 2000, 330.) In the remaining forms of 'one,' only en- is used. The two syntactic variants: eden and en- as cardinal numeral 'one.' 1. The assumptions of Naturalness Theory:

116

EXAMPLES

1.1. >sem (nominative masculine singular, other case forms) / of nouns and adjectives, in Slovenian I.e. with respect to semantic complexity, the nominative masculine singular of nouns and adjectives is more natural than other case forms of nouns and adjectives, in Slovenian.The nominative is more sem-natural than other cases. The singular is more sem-natural than other grammatical numbers. The masculine is more sem-natural than other genders. (Mayerthaler 1981, 14-5.) 1.2. >sem (eden & en-, only en-) / cardinal number 'one' in Slovenian I.e. with respect to semantic complexity, 'one' expressed by eden and en- is more natural than 'one' expressed by en- only.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the cardinal number 'one,' between taking the forms eden & en- and taking only the form en-, such that one option occurs in the nominative masculine singular, and the other option occurs in the remaining forms, it is the option taking both eden and en- that tends to occur in the nominative singular masculine. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.1. If there is any difference, within the cardinal number 'one,' between taking the forms eden & en- and taking only the form en-, such that one option occurs in the nominative masculine singular, and the other option occurs in the remaining forms, it is the option taking only en- that tends to occur outside the nominative singular masculine. Q.E.D. 4. Note. Cf. deduction 136. 138-9. Slovenian. The personal pronouns of the first and second person singular do not distinguish genders, whereas the personal pronouns for the remaining persons do distinguish genders, at least in the nominative. (Toporii 1976, 240; 2000, 305-6; Herrity 2000, 88-9.) The two syntactic variants: the personal pronouns of the first and second person singular, and the remaining personal pronouns, in the nominative. 1. The assumptions of Naturalness Theory: 1.1. >sem (conflated, separated) / genders of nominative of personal pronouns in Slovenian I.e. with respect to semantic complexity, conflated genders are more natural than separated genders, in the nominative of personal pronouns in Slovenian.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 1.2. >sem (first and second person singular, the remaining persons) / nominative of personal pronouns in Slovenian I.e. with respect to semantic complexity, the first and second person singular is more natural than the remaining persons, in the nominative of personal pronouns in Slovenian. The "remaining" persons (of all three numbers) do not constitute a "natural class," therefore they possess low sem-naturalness. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem

117

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the first and second person singular personal pronouns on the one side, and the remaining personal pronouns on the other side (within the nominative only), such that one kind distinguishes genders, and the other kind does not, it is the personal pronouns of the first and second person singular that tend not to distinguish genders. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the first and second person singular personal pronouns on the one side, and the remaining personal pronouns on the other side (within the nominative only), such that one kind distinguishes genders, and the other kind does not, it is the "remaining" personal pronouns that tend to distinguish genders. Q.E.D. 4. Note. Cf. deduction 210. 140. Slovenian and elsewhere. The construction of the comparative and the superlative is more complex than the construction of the positive. E.g. Micka je pridneja od Janeza 'Micka is more diligent than Janez,' Micka je najpridneja izmed vseh 'Micka is the most diligent of all.' The two syntactic variants: the construction of the positive, and the construction of the comparative and superlative. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit. 1.2. >sem (positive; comparative, superlative) I.e. with respect to semantic complexity, the positive is more natural than the comparative or the superlative. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the construction of the positive and the con struction of the comparative & superlative, such that one construction is syntactically ex pandable, and the other not, it is the construction of the positive that tends not to be expandable. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the construction of the positive and the con struction of the comparative & superlative, such that one construction is syntactically expandable, and the other not, it is the construction of the comparative & superlative that tends to be expandable. Q.E.D. 141. Slovenian. If the adjective modifier which normally stands to the left of its head is displaced to the right of the head, the adjective modifier expresses emphasis (a

118

EXAMPLES

derogatory or praising meaning), e.g. (derogatory) krava stara zoprna, hudi frdamani, enska pritegnjena, ava avasta, trola zmedena , (praising) srek moj dragi, ljubek moj dragi, ljubica ljubljena, pela pelasta . (The examples supplied by Varja Cvetko-Orenik viva voce, 1996; they are too idiomatic to be translatable.) Cf. Toporii 2000, 562. The two syntactic variants: adjective + head, and head + adjective. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1 >sym (postmodification, premodification) / of noun-phrase head, in Slovenian I.e. with respect to coding properties, the postmodification of the noun-phrase head is more natural than the premodification of the noun-phrase head, in Slovenian.Postmodification is structurally richer than premodification in Slovenian, speaking of averages. 1.2. >sem (-derogatory/praising, +derogatory/praising) I.e. with respect to semantic complexity, non-derogatory/non-praising is more natural than derogatory/praising. (In the spirit of Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the premodification and the postmodification of the noun-phrase head, such that one construction has derogatory or praising meaning, and the other construction not, it is the postmodification that tends to have derogatory or praising meaning. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the premodification and the postmodification of the noun-phrase head, such that one construction has derogatory or praising meaning, and the other construction not, it is the premodification that tends not to have derogatory or praising meaning. Q.E.D. 142. Slovenian. The normal construction of 'ask somebody for something' is prositi nekoga za nekaj , i.e. verb + accusative of person + prepositional case of thing. However, if the thing is a noun phrase of high frequency and pro-clausal, the accusative can be used instead of the prepositional case, e.g. to, nekaj, drugo, kaj te prosim 'I ask you for this, for something, for something else, for what;' kar te prosim 'what I ask you.' (The construction with the double accusative is not registered in the SSKJ s.v. PROSITI. Notice that ROTITI 'entreat,' which allows rotim te, da pomagaj 'I entreat you to help,' does not allow rotim te to 'I entreat you this,' parallel to prosim te, da pomagaj 'I ask you to help,' and to prosim te to 'I ask you this.') The two syntactic variants: prositi nekoga za nekaj , and prositi nekoga nekaj . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency.

119

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

(Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (prositi nekoga za nekaj, prositi nekoga nekaj ) / in Slovenian I.e. with respect to coding properties, prositi nekoga za nekaj is more natural than prositi nekoga nekaj , in Slovenian. 1.2. >sem (few, many) / units I.e. with respect to semantic complexity, a set of only few units is more natural than a set of many units.It can be observed time and again that small (closed) classes are more >sem-natural than large (open) classes. A special case of 1.2: 1.2.1. >sem (to, nekaj, drugo, kaj ; most noun phrases) / clause element in Slovenian I.e. with respect to semantic complexity, a clause element which can be realized only as to, nekaj, drugo, kaj is more natural than the same clause element which can be realized as almost any noun phrase, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between prositi nekoga za nekaj and prositi nekoga nekaj , such that one construction is used with very few noun phrases, and the other construction is used with almost any noun phrase, it is prositi nekoga za nekaj that tends to be used with almost any noun phrase. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between prositi nekoga za nekaj and prositi nekoga nekaj , such that one construction is used with very few noun phrases, and the other construction is used with almost any noun phrase, it is prositi nekoga nekaj that tends to be used with very few noun phrases. Q.E.D. 4. Note. Similarly rogare 'ask' in Latin, where the construction is also used in some formulas, e.g. populum magistratus rogare 'ask the people about the offices' = 'lead the election.' Formulas have low sym-value = high sem-value. 143. Slovenian. The question 'in whose honour did she put on her black dress' can be rendered in two ways as far as 'in whose honour' is concerned: v/na ast komu in honour to whom (left branching), or komu v/na ast to whom in honour (right branching). In the former case 'in whose honour' is emphatic. The two syntactic variants: a syntactic unit admitting both emphasis and nonemphasis v. a syntactic unit admitting only emphasis. 1. The assumptions of Naturalness Theory: 1.1. >sem (right, left) / branching I.e. with respect to semantic complexity, right branching of elements is more natural than left branching of elements.This is supported by the circumstance that in German main clauses right branching prevails, and in German dependent clauses left branching prevails (Mayerthaler et al. 1998, 312, 315-6). 1.2. >sem (+/-emphasis, +emphasis) / syntactic unit I.e. with respect to semantic complexity, a syntactic unit admitting both emphasis and non-emphasis is more natural than a syntactic unit admitting only emphasis.The scale has the format >sem (A + B, A).

120

EXAMPLES

2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a syntactic unit admitting both emphasis & nonemphasis and a syntactic unit admitting only emphasis, such that one syntactic unit shows right branching, and the other syntactic unit shows left branching, it is the syntactic unit admitting both emphasis and non-emphasis that tends to show right branching. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a syntactic unit admitting both emphasis & nonemphasis and a syntactic unit admitting only emphasis, such that one syntactic unit shows right branching, and the other syntactic unit shows left branching, it is the syntactic unit admitting only emphasis that tends to show left branching. Q.E.D. 144. Slovenian. An expressive variant of the imperative is various kinds of nonimperative sentences, for instance da pride tono that you-come accurately 'come in time,' or a ne bo miren not will-you-be quiet 'be quiet.' The two syntactic variants: a syntactic unit admitting both emphasis and nonemphasis v. a syntactic unit admitting only emphasis. 1. The assumptions of Naturalness Theory: 1.1. >sem (+imperative sentence, -imperative sentence) / expressing command I.e. with respect to semantic complexity, an imperative sentence is more natural than a non-imperative sentence, as expression of command.Imperative sentences are the prototypical means of expressing command. 1.2. >sem (+/-emphasis, +emphasis) / syntactic unit I.e. with respect to semantic complexity, a syntactic unit admitting both emphasis and non-emphasis is more natural than a syntactic unit admitting only emphasis.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a syntactic unit admitting both emphasis & nonemphasis and a syntactic unit admitting only emphasis, such that one syntactic unit is an imperative sentence, and the other syntactic unit is a non-imperative sentence (whereby both the imperative and the non-imperative sentences express command), it is the imperative sentence that tends to admit both emphasis and non-emphasis. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a syntactic unit admitting both emphasis & nonemphasis and a syntactic unit admitting only emphasis, such that one syntactic unit is an imperative sentence, and the other syntactic unit is a non-imperative sentence (whereby both the imperative and the non-imperative sentences express command), it is the nonimperative sentence that tends to admit only emphasis. Q.E.D. 4. Note. Similarly in German: wirst du nicht ruhig sein 'be quiet,' da du mir pnktlich kommst 'come on time.' Also in Ancient Greek and in Latin. (Brandenstein 1966, 134-5.)

121

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

145. Slovenian. An imperative clause can express the conditional protasis, e.g. reci bedaku, da je pameten, pa ti bo verjel 'tell an idiot that he is intelligent, and he will believe you.' (Toporii 1976, 433; 2000, 444.) Such a conditional protasis can only express a real condition. (I thank my student Sao ivanovi for help with this example, in 2001.) The two syntactic variants: a conditional protasis which has several meanings, and a conditional protasis which has only one of those meanings. 1. The assumptions of Naturalness Theory: 1.1. >sem ('if' protasis, protasis containing an imperative) / conditional I.e. with respect to semantic complexity, an 'if' protasis is more natural than a protasis containing an imperative.If a language has a conditional protasis introduced by 'if' at its disposal, the 'if' protasis is the normal expression of conditionals in that language. 1.2. >sem (several meanings, one of these meanings) / protasis I.e. with respect to semantic complexity, a protasis capable of expressing several meanings is more natural than a protasis capable of expressing only one of those meanings.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a conditional protasis which has several meanings and a conditional protasis which has only one of those meanings, such that one conditional protasis is an 'if' protasis, and the other conditional protasis is an imperative, it is the conditional protasis which has several meanings that tends to be an 'if' protasis. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a conditional protasis which has several meanings and a conditional protasis which has only one of those meanings, such that one conditional protasis is an 'if' protasis, and the other conditional protasis is an imperative, it is the conditional protasis which has one of "those" meanings that tends to be an imperative. Q.E.D. 4. Note. Similarly in German: sage mir, mit wem du umgehst, und ich will dir sagen, wer du bist 'tell me with whom you associate, and I will tell you who you are.' Also in Ancient Greek and Latin. (Brandenstein 1966, 134.) 146. Slovenian. The affirmative imperative supported with suitable particles can be used ironically instead of the negated imperative, e.g. potem pa e komu zaupaj then but still anybody trust 'trust no-one.' (Toporii 1976, 433.) The two syntactic variants: the negated imperative, and the ironical affirmative imperative, both expressing negative command. 1. The assumptions of Naturalness Theory: 1.1. >sem (negated imperative, affirmative imperative supported with particles) / expressing negative command I.e. with respect to semantic complexity, a negative imperative is more natural than an affirmative imperative supported with particles, as expression of negative command. Negated imperatives are the prototypical means of expressing negative command.

122

EXAMPLES

1.2. >sem (stylistically unmarked, stylistically marked) I.e. with respect to semantic complexity, stylistically unmarked is more natural than stylistically marked.The stylistically marked opposite number may be lacking. A special case of 1.2: 1.2.1. >sem (-ironical, +ironical) I.e. with respect to semantic complexity, lack of irony is more natural than presence of irony. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the affirmative imperative and the negated imperative, such that both express a negative command, and such that one is used ironi cally, and the other not, it is the affirmative imperative that tends to be used ironically. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the affirmative imperative and the negated imperative, such that both express a negative command, and such that one is used ironi cally, and the other not, it is the negated imperative that tends not to be used ironically. Q.E.D. 147. Slovenian. In a variety of non-standard Slovenian, the dual of nouns is preserved only in the nominative/accusative dual of masculine nouns preceded by a quantifying expression meaning 'two,' e.g. dva/oba lonca 'two/both pots.' In the remaining cases of the dual preceded by a quantifying expression, plural forms are used, e.g. gen. dveh/obeh loncov, dat. dvem/obem loncom, etc. (I thank Helena Majcenovi for help with this example, 2001.) The two syntactic variants: the type dva/oba lonca (= nominative/accusative dual), and the other cases of the dual, masculine nouns only. 1. The assumptions of Naturalness Theory: 1.1. >sem (nominative/accusative, the other cases) I.e. with respect to semantic complexity, the nominative/accusative is more natural than the other cases. (Mayerthaler et al. 1998, 167.) 1.2. >sem (dual, plural) / as expression of duality I.e. with respect to semantic complexity, the dual is more natural than the plural as expression of duality. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the dual (of masculine nouns) preceded by a quantifying expression, between the nominative/accusative and the other cases, such that one kind has special forms for the dual, and the other kind uses only plural forms, it is the nominative/accusative that tends to have special forms for the dual. Q.E.D.

123

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within the dual (of masculine nouns) preceded by a quantifying expression, between the nominative/accusative and the other cases, such that one kind has special forms for the dual, and the other kind uses only plural forms, it is the other cases that tend to use plural forms. Q.E.D. 148. Slovenian. One way of stating the full hours of the day is to use a cardinal number in 'one o'clock,' and ordinal numbers elsewhere, e.g. ob eni uri 'at one o'clock' as against ob drugi uri 'at two o'clock.' (Cf. Toporii 1976, 269; 2000, 332.) The two syntactic variants: the type ob eni uri, and the type ob drugi uri. 1. The assumptions of Naturalness Theory: 1.1. >sem (cardinal, ordinal) / numeral I.e. with respect to semantic complexity, a cardinal number is more natural than an ordinal number. (Mayerthaler 1981, 160.) 1.2. >sem (n, n + 1) I.e. with respect to semantic complexity, any number is more natural than the next higher number. (Mayerthaler 1981, 15.) A special case of 1.2: 1.2.1. >sem ('one,' higher than 'one') / in Slovenian I.e. with respect to semantic complexity, the number 'one' is more natural than any number higher than 'one,' in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type ob eni uri and the type ob drugi uri, such that one type uses cardinal numbers, and the other type uses ordinal numbers, it is the type ob eni uri that tends to use cardinal numbers. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type ob eni uri and the type ob drugi uri, such that one type uses cardinal numbers, and the other type uses ordinal numbers, it is the type ob drugi uri that tends to use ordinal numbers. Q.E.D. 4. Note. I cannot account for the placement of the cut-off point between the type ob eni uri and the type ob drugi uri. 149-50. Slovenian. In the sentence kateri kruh hoete, beli ali rni 'which bread do you want, the white or the black one' both accusatives equal the corresponding nominative. Alternatively, the latter accusative can equal the corresponding genitive: kateri kruh hoete, belega ali rnega same meaning. Likewise in kaken kruh hoete, bel ali rn (belega ali rnega) 'what bread do you want, a white or a black one.' I interpret the versions containing the accusative = nominative after the comma as partially copying the modifier kateri/kaken 'which/what'. This approach accounts for the definite form of beli ali rni 'the white or the black one' (the definite form being required by the preceding kateri 'which') and for the indefinite form of bel ali rn 'a white or a black one' (the indefinite form being required by the preceding kaken 'what kind'). I interpret the versions containing the accusative = genitive after the comma as partially

124

EXAMPLES

copying the full noun phrases kateri kruh 'which bread' and kaken kruh 'what bread.' This approach accounts for the identity of the alternative case form, namely accusative = genitive, in both example-sentences. The two syntactic variants: full NP v. its modifier (both meant as partial copies = repetitions). 1. The assumptions of Naturalness Theory: 1.1. >sem (accusative = nominative, accusative = genitive) / in nom.-acc. languages I.e. with respect to semantic complexity, the option accusative = nominative is more natural than the option accusative = genitive, in nominative-accusative languages.The affinity of the accusative with the nominative is greater than the affinity of the accusative with the genitive, in nominative-accusative languages. (Mayerthaler et al. 1998, 167.) 1.2. >sem (modifier, full NP) / repetition I.e. with respect to semantic complexity, the modifier within an NP (as repetition) is more natural than the full NP (as repetition).Repetitions tend to be short. Full NPs are longer than the modifiers that they contain. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a full NP and its modifier (both meant as partial repetitions), such that one of them is accusative = nominative, and the other of them is accusative = genitive, it is the modifier that tends to be accusative = nominative. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a full NP and its modifier (both meant as partial repetitions), such that one of them is accusative = nominative, and the other of them is accusative = genitive, it is the full NP that tends to be accusative = genitive. Q.E.D. 4. Note. With singular neuter adjectivals, the accusative takes the form of the corresponding nominative if the adjectival is accompanied by a noun head, e.g. belo vino 'white wine (acc. sg. neuter).' The accusative optionally takes the form of the corresponding genitive, if the noun head of the adjectival is ellipted, e.g. hoem belo/belega 'I want the white one (scil. the white wine).' (Toporii 2000, 317-8.) I cannot account for the circumstance that the tendency of singular inanimate nouns for accusative = genitive (under ellipsis) is considerably stronger in the masculine than in the neuter. 151. Slovenian. The adverb menda expresses (a) an assumption, e.g. nocoj se menda e ne vrnejo 'presumably they are not returning tonight,' (b) obviousness, e.g. v takem deju pa menda ne boste li od doma 'obviously you won't leave home in such rain,' (c) agreement without reservation, e.g. Se ti zdi reitev pametna? Menda . 'Do you find the solution sensible? Of course.' (Data from SSKJ s.v. menda.) In meaning (c) the word menda is often either pronounced differently from menda in the meanings (a-b), e.g. mnda instead of mend, or replaced by the loanword valjda. Both modifications pertain to the non-standard language. The two syntactic variants: mend in meanings (a-b), and mnda/valjda in meaning (c). 1. The assumptions of Naturalness Theory: 1.1. >sem (mend, mnda/valjda ) I.e. with respect to semantic complexity, mend is more natural than mnda/valjda . Mend is older than mnda/valjda , and native speakers feel this. 1.2. >sym (more transparent, less transparent) / syntactic unit 125

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (extra-clausal, independent use; intra-clausal use) / adverb menda in Slovenian I.e. with respect to coding properties, extra-clausal, independent use of the adverb menda is more natural than its intra-clausal use, in Slovenian.The extra-clausal use is more conspicuous than the intra-clausal use. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the adverbs mend and mnda/valjda , such that one kind is intra-clausal (and expresses assumption, obviousness), and the other kind is extra-clausal (and expresses agreement without reservation), it is mend that tends to be intra-clausal (and to express assumption, obviousness). Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the adverbs mend and mnda/valjda , such that one kind is intra-clausal (and expresses assumption, obviousness), and the other kind is extra-clausal (and expresses agreement without reservation), it is mnda/valjda that tends to be extra-clausal (and to express agreement without reservation). Q.E.D. 152. Slovenian. Sentence negation is expressed with ne, e.g. danes ne deuje 'it is not raining today.' The answer to a yes/no question can be ne as well, e.g. ne, danes ne deuje 'no, it is not raining today;' in that case ne is an extra-clausal adverb. Both negations can be emphatic or not. However, the probability that extra-clausal negation will be emphatic is significantly greater than the probability that sentence negation will be emphatic. The two syntactic variants: the sentence negation ne, and the extra-clausal ne. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (extra-clausal negation, sentence negation) / in Slovenian I.e. with respect to coding properties, extra-clausal negation is more natural than sen tence negation, in Slovenian.Extra-clausal negation is more conspicuous than intraclausal negation. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15.) A special case of 1.2: 1.2.1. >sem (less often emphatic, more often emphatic) I.e. with respect to semantic complexity, less often emphatic is more natural than

126

EXAMPLES

more often emphatic. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between extra-clausal negation and sentence negation, such that one is more often emphatic than the other, it is extra-clausal negation that tends to be more often emphatic than sentence negation. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between extra-clausal negation and sentence negation, such that one is more often emphatic than the other, it is sentence negation that tends to be less often emphatic than extra-clausal negation. Q.E.D. 153. Slovenian. Nothing can intervene between the sentence negator and the finite verb, e.g. danes ne deuje 'it is not raining today.' If the negated verb is infinite, the negator precedes it, is accented, and can be separated from the verb, e.g. ne ga e obiskati not him yet to-visit 'don't visit him yet.' (M. Milojevi-Sheppard & M. Golden, 2000.) Also, ne prav posebno pogosto bivajo v Ljubljani 'not exactly especially often sojourning in Ljubljana.' The two syntactic variants: negated finite and non-finite verbs. 1. The assumptions of Naturalness Theory: 1.1. >sem (+finite, -finite) / verb I.e. with respect to semantic complexity, a finite verb is more natural than an infinite verb. (Mayerthaler et al. 1993, 144.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (ne (X) verb, ne verb) / in Slovenian I.e. with respect to coding properties, ne + verb optionally separated by other material is more natural than ne + verb obligatorily adjacent, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between negated finite and non-finite verbs, such that one kind can be separated from the negator ne by other material, and the other kind cannot be thus separated, it is the finite verb that tends not to be separated from the negator ne by other material. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between negated finite and non-finite verbs, such that one kind can be separated from the negator ne by other material, and the other kind cannot be thus separated, it is the non-finite verb that tends to admit separation from the negator ne by other material. Q.E.D.

127

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

154. Slovenian. The subject complement of middle-voice reflexive verbs can assume the accusative case if the subject is animate, e.g. Micka/medvedka se uti premagano 'Micka/the she-bear feels defeated.' The nominative case is also possible, e.g. Micka/medvedka se uti premagana same meaning. If the subject is not animate, the subject complement must be in the nominative, e.g. blazina se uti mehka 'the pillow feels soft.' The two syntactic variants: the subject complement taking the nominative or the accusative, and the subject complement taking only the nominative. 1. The assumptions of Naturalness Theory: 1.1. >sem (the nominative and the accusative, the nominative only) / the case of the subject complement with reflexive verbs in Slovenian I.e. with respect to semantic complexity, the subject complement taking the nominative or the accusative is more natural than the subject complement taking only the nominative, with reflexive verbs in Slovenian.The scale has the format >sem (A + B, A). 1.2. >sem (+animate, -animate) / the subject I.e. with respect to semantic complexity, an animate subject is more natural than an inanimate subject. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the subject complement taking the nominative or the accusative and the subject complement taking only the nominative, such that one kind of subject complement is used with animate subjects, and the other kind of subject complement is used with inanimate subjects, it is the subject complement taking the nominative or the accusative that tends to be used with animate subjects. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the subject complement taking the nominative or the accusative and the subject complement taking only the nominative, such that one kind of subject complement is used with animate subjects, and the other kind of subject complement is used with inanimate subjects, it is the subject complement taking only the nominative that tends to be used with inanimate subjects. Q.E.D. 155. Slovenian. When an adjective is used as part of a vocative, it takes the definite form, e.g. Dragi Janez! 'dear Janez.' The two syntactic variants: an adjective having the definite and the indefinite form v. an adjective having only the definite form. 1. The assumptions of Naturalness Theory: 1.1. >sem (the definite and the indefinite form, the definite form) / of the adjective in Slovenian I.e. with respect to semantic complexity, an adjective having the definite and the indefinite form is more natural than an adjective having only the definite form, in Slovenian. The scale has the format >sem (A + B, A). 1.2. >sem (-vocative, +vocative) / nominative case in Slovenian I.e. with respect to semantic complexity, the non-vocative use of the nominative is more natural than the vocative use of the nominative, in Slovenian.The prototypical nominative is the case of the subject, not the case of the vocative.

128

EXAMPLES

2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the construction adjective + nominative, between an adjective having the definite & the indefinite form and an adjective having only the definite form, such that the accompanying nominative has or does not have the vocative use, it is the adjective having the definite and the indefinite form that tends to combine with a nominative which does not have the vocative use. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within the construction adjective + nominative, between an adjective having the definite & the indefinite form and an adjective having only the definite form, such that the accompanying nominative has or does not have the vocative use, it is the adjective having only the definite form that tends to combine with a nominative which has the vocative use. Q.E.D. 156. Slovenian. (a) In the plural and dual, full noun phrase subjects can be used even with verbs in the non-third verbal person, e.g. tudentje garate 'you students work hard,' oba tudenta garava 'both of us students work hard.' The verbal person is marked only in the desinence of the verb. (b) If the subject contains a cardinal numeral higher than '4' (see note 4), it must be followed by the head noun in the genitive, and agreement with the verb becomes impossible; the verb assumes the default value of the third person singular. The grammatical person of the subject must be expressed with a pronoun clitic copy of the subject's head noun (indicating the case and number of the head noun, as well as the grammatical person of the whole subject): pet tudentov nas/vas gara 'we/you five students work hard.' (All the English glosses are misleading in the crucial syntactic respect.)The present deduction deals with case (a). Case (b) is treated in deduction 157. My interest in case (a) has been instigated by Lyons 1999, 310 ff. The two syntactic variants: the type tudentje garajo 'the students work hard' (occurring in the third person of the non-singular; transparent agreement of the verb with the subject), and the type tudentje garamo 'we students work hard' (occurring in the nonthird person of the non-singular; opaque agreement of the verb with the subject). 1. The assumptions of Naturalness Theory: 1.1. >sem (transparent, opaque) / agreement with full-NP subject I.e. with respect to semantic complexity, transparent agreement with a full-NP subject is more natural than opaque agreement with such a subject. (In the spirit of Mayerthaler et al. 1998, 127.) 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type tudentje garajo and the type tudentje garamo, such that one type is used in the third person, and the other in the non-third

129

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

person, it is the transparent type tudentje garajo that tends to be used in the third person. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type tudentje garajo and the type tudentje garamo, such that one type is used in the third person, and the other in the non-third person, it is the opaque type tudentje garamo that tends to be used in the non-third person. Q.E.D. 4. Note. The above observation about cardinal numerals higher than '4' must be understood as referring to cardinal numbers which, when pronounced, end in 5-99, or are hundred or thousand bases. (Toporii 1976, 268; 2000, 333; Herrity 2000, 132 f.) 157. Slovenian. (a) In the plural and dual, full noun phrase subjects can be used even with verbs in the non-third verbal person, e.g. tudentje garate 'you students work hard,' oba tudenta garava 'both of us students work hard.' The verbal person is marked only in the desinence of the verb. (b) If the subject contains a cardinal numeral higher than '4' (see note 4.2), it must be followed by the head noun in the genitive, and agreement with the verb becomes impossible; the verb assumes the default value of the third person singular. The grammatical person of the subject must be expressed with a pronoun clitic copy of the subject's head noun (indicating the case and number of the head noun, as well as the grammatical person of the whole subject): pet tudentov nas/vas gara 'we/you five students work hard.' (All the English glosses are misleading in the crucial syntactic respect.)The present deduction deals with case (b). Case (a) is treated in deduction 156. The two syntactic variants: transparent and opaque subject agreement. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (clitic pronoun, desinence) / expression of grammatical person I.e. with respect to coding properties, a clitic pronoun is more natural as an expression of grammatical person than a desinence. 1.2. >sem (transparent, opaque) / agreement with full-NP subject I.e. with respect to semantic complexity, transparent agreement with a full-NP subject is more natural than opaque agreement with such a subject. (In the spirit of Mayerthaler et al. 1998, 127.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between transparent and opaque subject agreement, such that one kind of subject agreement is expressed with a desinence, and the other kind of subject agreement is expressed with a clitic pronoun, it is the transparent subject agreement that tends to be expressed with a desinence. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between transparent and opaque subject agreement, such

130

EXAMPLES

that one kind of subject agreement is expressed with a desinence, and the other kind of subject agreement is expressed with a clitic pronoun, it is the opaque subject agreement that tends to be expressed with a clitic pronoun. Q.E.D. 4. Notes. 4.1. If the verb is transitive in the type pet tudentov nas gara, say pet tudentov vas gleda, the sentence is ambiguous between 'you five students are watching' and 'five students are watching you' (seeing that vas can be the genitive of the subject-issued clitic or the accusative of the clitic direct object). Disambiguation is possible using the alternative type mi tudentje vas gledamo (containing the determiner mi 'we' in the subject), which becomes nas pet tudentov vas gleda 'we five students are watching you' when a suitable numeral is present in the subject. Just as mi is at least accented (if not emphatic) in mi tudentje vas gledamo , the genitive nas (corresponding to mi in mi tudentje vas gledamo ) is at least accented (let alone clitic) in nas pet tudentov vas gleda , and of course does not belong to the Wackernagel position. 4.2. The above observation about cardinal numerals higher than '4' must be understood as referring to cardinal numbers which, when pronounced, end in 5-99, or are hundred or thousand bases. (Toporii 1976, 268; 2000, 333; Herrity 2000, 132 f.) 158. Slovenian. In comparison, the adjective DOBER 'good' shows strong suppletion, the adjectives MAJHEN 'little,' VELIK 'big,' LAHEK 'light,' and DOLG 'long' show weak suppletion. (My interpretation. Data from Toporii 1976, 262; 2000, 326. Cf. Herrity 2000, 82.) The forms are: dober, bolji, najbolji; majhen, manji, najmanji; velik, veji, najveji; lahek, laji, najlaji; dolg, dalji, najdalji . The two syntactic variants: the adjectives 'good' and 'little/big/light/long.' 1. The assumptions of Naturalness Theory: 1.1. >sym (weak suppletion, strong suppletion) I.e. with respect to coding properties, weak suppletion is more natural than strong suppletion. (Mayerthaler 1981, 38.) 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem ('good,' 'little/big/light/long') / in Slovenian I.e. with respect to semantic complexity, the adjective lexeme 'good' is more natural than the adjective lexemes 'little/big/light/long,' in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the comparison of the adjective 'good' and of the adjectives 'little/big/light/long,' such that one kind shows strong suppletion, and the other kind weak suppletion, it is the adjectives 'little/big/light/long' that tend to show weak suppletion. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the comparison of the adjective 'good' and of the adjectives 'little/big/light/long,' such that one kind shows strong suppletion, and the other kind weak suppletion, it is the adjective 'good' that tends to show strong suppletion.

131

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

Q.E.D. 159-60. Slovenian. Pronominal possession, alienable and inalienable. (a) Alienable possession with a pronominal possessor is expressed with the possessive adjective of the possessor, e.g. moja hia 'my house.' Inalienable possession is expressed with the bare possessum optionally preceded by the possessive adjective of the possessor, e.g. (moj) brat 'my brother.' (b) In inalienable possession, the possessive adjectives are used obligatorily when disambiguation is necessary, and are then emphatic, e.g. MOJ brat. The present deduction deals with case (a). Case (b) is considered in deduction 161. The two syntactic variants: the type moja hia, and the type (moj) brat. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type moja hia, the type (moj) brat) / in Slovenian I.e. with respect to coding properties, the type moja hia is more natural than the type (moj) brat, in Slovenian. 1.2. >sem (-alienable, +alienable) / possession I.e. with respect to semantic complexity, inalienable possession is more natural than alienable possession. (Mayerthaler 1981, 152; Mayerthaler et al. 1998, 275.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type moja hia and the type (moj) brat, such that one expresses alienable possession, and the other expresses inalienable possession, it is the type moja hia that expresses alienable possession. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type moja hia and the type (moj) brat, such that one expresses alienable possession, and the other expresses inalienable possession, it is the type (moj) brat that expresses inalienable possession. Q.E.D. 161. Slovenian. Pronominal possession, alienable and inalienable. (a) Alienable possession with a pronominal possessor is expressed with the possessive adjective of the possessor, e.g. moja hia 'my house.' Inalienable possession is expressed with the bare possessum optionally preceded by the possessive adjective of the possessor, e.g. (moj) brat 'my brother.' (b) In inalienable possession, the possessive adjectives are used obligatorily when disambiguation is necessary, and are then emphatic, e.g. MOJ brat. The present deduction deals with case (b). Case (a) is considered in deduction 159-60. The two syntactic variants: the type (moj) brat, and the type moj brat. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency.

132

EXAMPLES

(Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type moj brat, the type (moj) brat) / in Slovenian I.e. with respect to coding properties, the type moj brat is more natural than the type (moj) brat, in Slovenian. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type (moj) brat and the type moj brat, such that one type is emphatic, and the other not, it is the type moj brat that tends to be emphatic. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type (moj) brat and the type moj brat, such that one type is emphatic, and the other not, it is the type (moj) brat that tends to be nonemphatic. Q.E.D. 162. Slovenian. Within pronominal possessives, the reflexive svoj distinguishes gender, case and number, whereas the non-reflexive moj 'my' etc. distinguish gender, case, number, and person. (Toporii 1976, 278-9; 2000, 342; Herrity 2000, 94-6.) The two syntactic variants: possessive pronominal distinguishing gender, case, number & person, and possessive pronominal distinguishing only gender, case & number. 1. The assumptions of Naturalness Theory: 1.1. >sem (+gender & +case & +number & +person, +gender & +case & +number) / possessive pronominal I.e. with respect to semantic complexity, a possessive pronominal distinguishing gender, case, number and person is more natural than a possessive pronominal distin guishing just gender, case and number.The scale has the format >sem (A + B, A). 1.2. >sem (-reflexive, +reflexive) / pronominal possessive I.e. with respect to semantic complexity, the non-reflexive pronominal possessive is more natural than the reflexive one.Some languages lack reflexive pronominal possessives. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a possessive pronominal distinguishing gender, case, number & person and a possessive pronominal distinguishing only gender, case & number, such that one kind of possessive pronominal is reflexive, and the other kind of possessive pronominal is non-reflexive, it is the possessive pronominal distinguishing gender, case, number and person that tends to be non-reflexive. Q.E.D. From 1.1-2 and 2.2 it can be deduced:

133

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.2. If there is any difference between a possessive pronominal distinguishing gender, case, number & person and a possessive pronominal distinguishing only gender, case & number, such that one kind of possessive pronominal is reflexive, and the other kind of possessive pronominal is non-reflexive, it is the possessive pronominal distinguishing only gender, case and number that tends to be reflexive. Q.E.D. 163. Slovenian. Within pronominal possessives, the reflexive svoj is used within its clause only, whereas the non-reflexive moj 'my' etc. are not thus limited. The two syntactic variants: pronominal possessive whose referent must be intraclausal, and pronominal possessive which is not thus limited. 1. The assumptions of Naturalness Theory: 1.1. >sem (intra- or extraclausal, intraclausal only) / referent accessible to a pronominal possessive I.e. with respect to semantic complexity, a pronominal possessive whose referent must be intraclausal is less natural than a pronominal possessive which is not thus limited.The scale has the format >sem (A + B, A). 1.2. >sem (-reflexive, +reflexive) / pronominal possessive I.e. with respect to semantic complexity, the non-reflexive pronominal possessive is more natural than the reflexive one.Some languages lack reflexive pronominal possessives. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between a pronominal possessive whose referent must be intraclausal and a pronominal possessive which is not thus limited, such that one kind of pronominal possessive is reflexive, and the other kind of pronominal possessive is not reflexive, it is the pronominal possessive whose referent can be intra- or extraclausal that tends to be non-reflexive. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between a pronominal possessive whose referent must be intraclausal and a pronominal possessive which is not thus limited, such that one kind of pronominal possessive is reflexive, and the other kind of pronominal possessive is not reflexive, it is the pronominal possessive whose referent must be intraclausal that tends to be reflexive. Q.E.D. 164. Slovenian. The main arguments of the verb are the nominative as the case of the subject and the accusative as the case of the direct object. The main argument of the noun is the genitive. (My attention was drawn to the putative unpredictability of the relationship between the main arguments of the verb and of the noun by Martina Kriaj-Ortar viva voce, in the early 1980s.) The two syntactic variants: nominative/accusative of the verb, and genitive of the noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (verb, noun) I.e. with respect to semantic complexity, the verb is more natural than the noun.The morphology of the verb is mostly better developed than the morphology of the noun. Such

134

EXAMPLES

a state of affairs is expected with the less marked (the verb) and the corresponding more marked (the noun) parts of speech. 1.2. >sem (nominative/accusative, genitive) / as argument in nom.-acc. languages I.e. with respect to semantic complexity, the nominative/accusative is more natural than the genitive, as argument in nominative-accusative languages. (Mayerthaler et al. 1998, 167.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes nominatives and accusatives, and the other part of speech takes genitives, it is the verb that tends to take the nominative and the accusative. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes nominatives and accusatives, and the other part of speech takes genitives, it is the noun that tends to take the genitive. Q.E.D. 4. Note. Cf. deductions 7 and 231. 165. Slovenian. (In what follows only the standard usage of those speakers whose pronunciation lacks tonemes is discussed.) Slovenian distinguishes +definite and -definite adjective forms. (a) The formal difference between +definite and -definite adjectives obtains only in the nominative singular masculine where the +definite form ends in - i, and the -definite form lacks any ending, e.g. +definite lep-i, -definite lep, both meaning 'beautiful.' (In the event that the accusative singular masculine equals in form the nominative singular masculine, the formal difference between the +definite and the -definite adjective obtains in that case form too.) (b) With the adjectives 'little' and 'big,' the formal difference between +/-definite obtains throughout the declension: majhen is -definite, mal-i is +definite, both meaning 'little;' vlik is -definite, vlik-i is +definite, both meaning 'big.' (Toporii 1976, 256; 2000, 320; Herrity 2000, 73-4.) The present deduction deals with case (a). For case (b) see deduction 166. The two syntactic variants: the nominative singular masculine, and the remaining case forms of the adjectival declension. 1. The assumptions of Naturalness Theory: 1.1. >sem (+[distinction +/-definite], -[distinction +/-definite]) / case in the adjectival declension of Slovenian I.e. with respect to semantic complexity, a case showing the distinction +/-definite is more natural than a case not showing such a distinction, in the adjectival declension of Slovenian.If a language has the distinction +/-definite, it is most sem-natural that the distinction is realized. 1.2. >sem (nominative singular masculine, the remaining case forms) / the adjectival declension of Slovenian I.e. with respect to semantic complexity, the nominative singular masculine is more natural than the remaining case forms, in the adjectival declension of Slovenian.The nominative is more sem-natural than the remaining cases. The singular is more semnatural than the remaining grammatical numbers. The masculine is more sem-natural

135

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

than the remaining genders. (Mayerthaler 1981, 14-5.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the adjectival declension, between the nominative singular masculine and the remaining case forms, such that the formal difference between +/-definite is or is not expressed, it is the nominative singular masculine that tends to show the formal distinction between +/-definite. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within the adjectival declension, between the nominative singular masculine and the remaining case forms, such that the formal difference between +/-definite is or is not expressed, it is the remaining case forms that tend to lack the formal distinction between +/-definite. Q.E.D. 166. Slovenian. (In what follows only the standard usage of those speakers whose pronunciation lacks tonemes is discussed.) Slovenian distinguishes +definite and -definite adjective forms. (a) The formal difference between +definite and -definite adjectives obtains only in the nominative singular masculine where the +definite form ends in - i, and the -definite form lacks any ending, e.g. +definite lep-i, -definite lep, both meaning 'beautiful.' (In the event that the accusative singular masculine equals in form the nominative singular masculine, the formal difference between the +definite and the -definite adjective obtains in that case form too.) (b) With the adjectives 'little' and 'big,' the formal difference between +/-definite obtains throughout the declension: majhen is -definite, mal-i is +definite, both meaning 'little;' vlik is -definite, vlik-i is +definite, both meaning 'big.' (Toporii 1976, 256; 2000, 320; Herrity 2000, 73-4.) The present deduction deals with case (b). For case (a) see deduction 165. The two syntactic variants: the lexical items 'big' and 'little,' and the remaining adjectival lexical items. 1. The assumptions of Naturalness Theory: 1.1. >sem (+[distinction +/-definite], -[distinction +/-definite]) / case in the adjectival declension of Slovenian I.e. with respect to semantic complexity, a case showing the distinction +/-definite is more natural than a case not showing such a distinction, in the adjectival declension of Slovenian.If a language has the distinction +/-definite, it is most sem-natural that the distinction is realized. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem ('big,' 'little;' most other adjectival lexical items) / in Slovenian I.e. with respect to semantic complexity, the lexical items 'big' and 'little' are more natural than most other adjectival lexical items, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences:

136

EXAMPLES

From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference, within the adjectival declension, between the lexical items 'big' and 'little' on the one side, and the remaining adjectival lexical items on the other, such that the formal difference between +/-definite is or is not expressed, it is the lexical items 'big' and 'little' that tend to show the formal distinction between +/-definite. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference, within the adjectival declension, between the lexical items 'big' and 'little' on the one side, and the remaining adjectival lexical items on the other, such that the formal difference between +/-definite is or is not expressed, it is the remaining adjectival lexical items that tend to lack the formal distinction between +/definite. Q.E.D. 4. Note to item 3.2. The "remaining" adjectival lexical items do show the formal distinction +/-definite in the nominative singular masculine. See deduction 165. 167. Slovenian. There is a type of phrasal verb consisting of a one-unit verb + a semantically empty accusative of the clitic third person singular masculine or feminine personal pronoun, e.g. lomiti ga 'do or say silly things' ( lomiti 'break'), mahniti jo 'set out, go' (mahniti 'wave'). (Toporii 1976, 206; 2000, 270; Herrity 2000, 202-3.) The verb of such phrasal verbs tends to pertain to relatively frequent lexical items. The two syntactic variants: verbal lexemes in general, and frequent verbal lexemes. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (phrasal verb, single-unit verb) / in Slovenian I.e. with respect to coding properties, a phrasal verb is more natural than a single-unit verb, in Slovenian. 1.2. >sem (+frequent & -frequent, +frequent) / verbal lexeme I.e. with respect to semantic complexity, involving any verbal lexeme is more natural than involving a frequent verbal lexeme only.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between verbal lexemes in general and frequent verbal lexemes, such that one kind of verbal lexemes are phrasal verbs, and the other kind of verbal lexemes are single-unit verbs, it is the verbal lexemes in general that tend to be single-unit verbs. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between verbal lexemes in general and frequent verbal lexemes, such that one kind of verbal lexemes are phrasal verbs, and the other kind of verbal lexemes are single-unit verbs, it is the frequent verbal lexemes that tend to be phrasal verbs. Q.E.D.

137

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

4. Note. The same deduction would account for the phrasal verbs of English, for the Slovenian type delati greh to-make sin 'commit a sin/sins,' delati teavo to-make difficulty 'cause a difficulty' (Toporii 1976, 292; 2000, 355), etc. 168. Slovenian, non-standard. The definite article takes the form of the prefix ta- affixed to any initial adjectival in the noun phrase, e.g. ta-zelena obleka 'the green dress/suit.' Bare head nouns cannot take this prefix, and do not express the definiteness at all, e.g. obleka 'dress/suit (+/-def.).' (Cf. Herrity 2000, 76-7.) The two syntactic variants: the type ta-zelena obleka (article-adjective + head noun, +definite) and the type obleka (bare head noun, +definite). 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (prefix, ) / the definite article in non-standard Slovenian I.e. with respect to coding properties, a prefix is more natural than , as the definite article of non-standard Slovenian. 1.1.2. >sym (the type ta-zelena obleka , the type obleka ) / in non-standard Slovenian I.e. with respect to coding properties, the type ta-zelena obleka is more natural than the type obleka , in non-standard Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between adjective + noun (def.) and the bare noun (def.), such that definiteness is expressed with the prefix ta- in one case, and not expressed in the other, it is adjective + noun (def.) that tends to take the prefix ta-. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between adjective + noun (def.) and the bare noun (def.), such that definiteness is expressed with the prefix ta- in one case, and not expressed in the other, it is the bare noun (def.) that tends not to express definiteness. Q.E.D. 169. Slovenian. In the construction type Plenikova hia mojega oeta '[architect] Plenik's house of my father's,' the possessive adjective Plenikov 'of [architect] Plenik' is characterizing, whereas the genitive mojega oeta 'of my father' is identifying. The two syntactic variants: the possessive adjective and the genitive in the construction type Plenikova hia mojega oeta . 1. The assumptions of Naturalness Theory: 1.1. >sem (characterizing, identifying) / attribution I.e. with respect to semantic complexity, characterizing attribution is more natural than identifying attribution. (In the spirit of Mayerthaler et al. 1998, 197.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency.

138

EXAMPLES

(Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.2.1. >sym (the genitive, the possessive adjective) / in Slovenian I.e. with respect to coding properties, the genitive is more natural than the possessive adjective, in Slovenian.The syntactic potential of the genitive is significantly greater than the syntactic potential of the possessive adjective. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the possessive adjective and the genitive in the construction type Plenikova hia mojega oeta , such that one of these modifying elements characterizes, and the other identifies, then it is the genitive that tends to have the identifying function. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the possessive adjective and the genitive in the construction type Plenikova hia mojega oeta , such that one of these modifying elements characterizes, and the other identifies, then it is the possessive adjective that tends to have the characterizing function. Q.E.D. 170. Slovenian. Genitivus subiectivus tends to be expressed with a possessive adjective, morphology permitting, e.g. profesorjevo obudovanje 'the professor's admiration' (scil. 'the professor admires'). Genitivus obiectivus tends to be expressed with the genitive case, e.g. obudovanje profesorja 'the admiration of the professor' (scil. 'one admires the professor'). The two syntactic variants: genitivus subiectivus, and genitivus obiectivus. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (genitive, possessive adjective) / in Slovenian I.e. with respect to coding properties, the genitive is more natural than the possessive adjective, in Slovenian.The syntactic potential of the genitive is significantly greater than the syntactic potential of the possessive adjective. 1.2. >sem (subject, object) / in nom.-acc. languages I.e. with respect to semantic complexity, the subject is more natural than the object, in nominative-accusative languages.This follows from the basic properties of nominativeaccusative languages. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced:

139

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.1. If there is any difference between the genitivus subiectivus and the genitivus obiectivus, such that one of them is expressed with the possessive adjective (morphology permitting), and the other is expressed with the genitive, it is the genitivus obiectivus that tends to be expressed with the genitive. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the genitivus subiectivus and the genitivus obiectivus, such that one of them is expressed with the possessive adjective, and the other is expressed with the genitive, it is the genitivus subiectivus that tends to be expressed with the possessive adjective, morphology permitting. Q.E.D. 171. Slovenian. The direct object mostly takes the accusative, e.g. ela je penico 'she was cutting the wheat.' In obsolete use, the direct object takes the genitive in infinite clauses of purpose (containing the so-called supine), e.g. la je penice et 'she went to cut the wheat.' (Toporii 1976, 204, 293, 338; 2000, 269, 402.) The two syntactic variants: direct object of a finite verb and of an infinite verb. 1. The assumptions of Naturalness Theory: 1.1. >sem (+finite, -finite) verb I.e with respect to semantic complexity, a finite verb is more natural than an infinite verb. (Mayerthaler et al. 1993, 145.) 1.2. >sem (accusative, genitive) / in Slovenian I.e. with respect to semantic complexity, the accusative is more natural than the geni tive, in Slovenian (which is a nominative-accusative language). (Mayerthaler et al. 1998, 167.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the direct object of a finite verb and of an infi nite verb, such that one kind of verb takes the direct object in the accusative, and the other kind of verb takes the direct object in the genitive, it is the finite verb that tends to take the direct object in the accusative. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the direct object of a finite verb and of an infi nite verb, such that one kind of verb takes the direct object in the accusative, and the other kind of verb takes the direct object in the genitive, it is the infinite verb that tends to take the direct object in the genitive. Q.E.D. 172. Slovenian. The construction type 'be' + action noun, e.g. avto je v okvari 'the car is in disrepair,' knjiga je v nastajanju 'the book is in the making' is used in parallel with other constructions which mean approximately the same, e.g. avto je pokvarjen 'the car is out of order' and knjiga nastaja 'the book is in the making.' The construction 'be' + action noun is invariably stative. It differs from the corresponding constructions in one or both of the following respects: (a) 'be' + action noun is more formal in register than any corresponding construction; (b) 'be' + action noun expresses a temporary state, while any corresponding construction is neutral with respect to temporariness. The two syntactic variants: (Case a) all speaker's registers and speaker's formal register. (Case b) +/-permanent and -permanent state/process.

140

EXAMPLES

1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type avto je v okvari, the type avto je pokvarjen ) I.e. with respect to coding properties, the type avto je v okvari is more natural than the type avto je pokvarjen . Case (a) 1.2. >sem (all speaker's registers, speaker's formal register) I.e. with respect to semantic complexity, the speaker's use of several registers is more natural than the speaker's limitation to the formal register.The scale has the format >sem (A + B, A). Case (b) 1.3. >sem (+/-permanent, -permanent) / state, process I.e. with respect to semantic complexity, indifference to the permanency of a state/ process is more natural than a non-permanent state/process.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: Case (a). From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between all speaker's registers and speaker's formal register, such that in one kind of register the type avto je v okvari is used, and in the other type of register the type avto je pokvarjen is used, it is all speaker's registers that tend to use the type avto je pokvarjen . Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between all speaker's registers and speaker's formal register, such that in one kind of register the type avto je v okvari is used, and in the other type of register the type avto je pokvarjen is used, it is a speaker's formal register that tends to use the type avto je v okvari. Q.E.D. Case (b). From 1.1.1, 1.3 and 2.1 it can be deduced: 3.3. If there is any difference between +/-permanent and -permanent state/process, such that one kind of state/process characterizes the type avto je v okvari , and the other type of state/process characterizes the type avto je pokvarjen , it is the +/-permanent state/process that tends to characterize the type avto je pokvarjen . Q.E.D. From 1.1.1, 1.3 and 2.2 it can be deduced: 3.4. If there is any difference between +/-permanent and -permanent state/process, such that one kind of state/process characterizes the type avto je v okvari , and the other type of state/process characterizes the type avto je pokvarjen , it is the -permanent

141

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

state/process that tends to characterize the type avto je v okvari. Q.E.D. 173. Slovenian. Men's surnames are mostly inflected, whereas women's surnames are mostly uninflected. (Toporii 1976, 219, 231; 2000, 296, 560; Herrity 2000, 33.) E.g. Joe Toporii , genitive Joeta Toporiia ; Breda Pogorelec, genitive Brede Pogorelec . The two syntactic variants: man's surname, and woman's surname. 1. The assumptions of Naturalness Theory: 1.1. >sem (+inflected, -inflected) / noun in Slovenian I.e with respect to semantic complexity, an inflected noun is more natural than an uninflected noun, in Slovenian.If a language has inflection of nouns, the implementation of the inflection is the most sem-natural option. 1.2. >sem (masculine, feminine) I.e. with respect to semantic complexity, masculine is more natural than feminine. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between men's surnames and women's surnames, such that one kind is inflected and the other not, it is men's surnames that tend to be inflected. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between men's surnames and women's surnames, such that one kind is inflected and the other not, it is women's surnames that tend not to be inflected. Q.E.D. 174. Slovenian. The Slovenian geographical regions often have names ending in - sko and -ska, e.g. Gorenjsko, Gorenjska 'Upper Carniola.' The name in - sko tends to be used with spatial reading, e.g. na Gorenjskem 'in Upper Carniola,' as against Gorenjska je lepa 'Upper Carniola is beautiful.' (Toporii 1976, 236; 2000, 301; Herrity 2000, 62.) The two syntactic variants: the type Gorenjsko, and the type Gorenjska. 1. The assumptions of Naturalness Theory: 1.1. >sem (the type Gorenjsko, the type Gorenjska) I.e. with respect to semantic complexity, the type Gorenjsko is more natural than the type Gorenjska.The type Gorenjsko is the older of the two. The speakers are aware of this on the basis of the fact that the nominative Gorenjsko is obsolescent or obsolete. 1.2. >sem (spatial reading, other reading) / of place names I.e. with respect to semantic complexity, the spatial reading of a place name is more natural than any other reading of that place name. (Hock 1991, 233.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the region names in - sko and in -ska, such that one tends to be used with spatial reading, and the other not, it is the type Gorenjsko that tends to be used with spatial reading. Q.E.D.

142

EXAMPLES

From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the region names in - sko and in -ska, such that one tends to be used with spatial reading, and the other not, it is the type Gorenjska that tends not to be used with spatial reading. Q.E.D. 175. Slovenian. In the animate nouns of the singular masculine, the accusative equals the genitive in form. In the inanimate nouns of the singular masculine, the accusative equals the nominative in form. The accusative singular masculine of the third person singular personal pronoun invariably (i.e. even when inanimate) takes the form of the corresponding genitive, viz. njega, clitic ga 'him, it (masc.),' cf. nom. sg. masc. on 'he, it.' The two morphological variants: nominative and accusative masculine singular of the third person personal pronoun, of animate and inanimate nouns. 1. The assumptions of Naturalness Theory: 1.1. >sem (inanimate nouns; animate nouns & third person personal pronouns) / masculine singular I.e. with respect to semantic complexity, inanimate masculine singular nouns are more natural than animate masculine singular nouns and third person masculine singular personal pronouns.Animate masculine singular nouns and third person masculine singular personal pronouns (animate and inanimate referents) do not constitute a "natural class," therefore their sem-naturalness is low. 1.2. >sem (accusative = nominative, accusative = genitive) / in nom.-acc. languages I.e. with respect to semantic complexity, the option accusative = nominative is more natural than the option accusative = genitive, in nominative-accusative languages.The affinity of the accusative with the nominative is greater than the affinity of the accusative with the genitive, in nominative-accusative languages. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference, in the nominative and accusative masculine singular, between the third person personal pronoun and animate nouns on the one side, and inani mate nouns on the other side, such that in one set the accusative equals the nominative, and in the other set the accusative equals the genitive, it is the inanimate nouns that tend to have the accusative equal to the nominative. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference, in the nominative and accusative masculine singular, between the third person personal pronoun and animate nouns on the one side, and inani mate nouns on the other side, such that in one set the accusative equals the nominative, and in the other set the accusative equals the genitive, it is the animate nouns and the third person personal pronoun that tend to have the accusative equal to the genitive. Q.E.D. 176. Slovenian. Outside the singular, personal pronouns have forms whose gender is visible in the nominative, and they have forms whose gender is not visible outside the nominative, e.g. midva 'we two (masc.), midve/medve 'we two (fem./neuter),' both nominative forms, as against naju 'of us two (masc./fem./neuter),' etc. (Toporii 1976, 240-1; 2000, 305-6; Herrity 2000, 88-90.)

143

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

The two morphological variants: personal pronouns in the nominative non-singular, and personal pronouns in the non-nominative non-singular. 1. The assumptions of Naturalness Theory: 1.1. >sem (+visible, -visible) / gender in languages which show gender distinctions I.e. with respect to semantic complexity, visible gender is more natural than nonvisible gender, in languages that show gender distinctions.If a language has gender distinctions, the most natural option is for them to be visible. 1.2. >sem (+nominative, -nominative) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than the nonnominative, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between forms whose gender is visible and forms whose gender is non-visible, such that one kind occurs in the nominative, and the other kind occurs in the non-nominative, it is the forms whose gender is visible that tend to be used in the nominative. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between forms whose gender is visible and forms whose gender is non-visible, such that one kind occurs in the nominative, and the other kind occurs in the non-nominative, it is the forms whose gender is not visible that tend to be used in the non-nominative. Q.E.D. 4. Notes. 4.1. Cf. deduction 177. 4.2. "Visible" above means "determinable from the form itself." E.g. naju ni doma 'we two are not at home:' naju, genitive non-singular, lacks visible gender; midva sva doma 'we two are at home:' midva, nominative dual, has visible gender, namely masculine. 177. Slovenian. In the nominative singular, personal pronouns have forms whose gender is not visible in the non-third person, and they have forms whose gender is visible in the third person, e.g. jaz 'I (masc., fem., neuter),' on 'he, it (masc.),' ona 'she, it (fem.),' ono 'he, she, it (neuter),' etc. (Toporii 1976, 240-1; 2000, 305-6; Herrity 2000, 88-90.) The two morphological variants: personal pronouns in the third person singular, and personal pronouns in the non-third person singular, all nominative. 1. The assumptions of Naturalness Theory: 1.1. >sem (+visible, -visible) / gender in languages which show gender distinctions I.e. with respect to semantic complexity, visible gender is more natural than nonvisible gender, in languages that show gender distinctions.If a language has gender distinctions, the most natural option is for them to be visible. 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two morphological

144

EXAMPLES

variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between forms whose gender is visible and forms whose gender is non-visible, such that one kind of forms is used in the third person, and the other kind of forms is used in the non-third person, it is the forms whose gender is visible that tend to be used in the third person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between forms whose gender is visible and forms whose gender is non-visible, such that one kind of forms is used in the third person, and the other kind of forms is used in the non-third person, it is the forms whose gender is not visible that tend to be used in the non-third person. Q.E.D. 4. Notes. 4.1. Cf. deduction 176. 4.2. "Visible" above means "determinable from the form itself." E.g. naju ni doma 'we two are not at home:' naju, genitive non-singular, lacks visible gender; midva sva doma 'we two are at home:' midva, nominative dual, has visible gender, namely masculine. 178. Slovenian. The verb has no desinence in the third person singular of the present tense (= the only synthetic tense of the language). (Toporii 1976, 284, 306; 2000, 369; Herrity 2000, 165.) The two morphological variants: the verb + desinence, the verb lacking desinence, in the present tense. 1. The assumptions of Naturalness Theory: 1.1. >sem (the third person singular, the remaining persons) / verb I.e. with respect to semantic complexity, the third person singular is more natural than the remaining persons.The third person is more sem-natural than the non-third person (according to the linguistic tradition beginning with Jakobson 1932). The singular is more sem-natural than the non-singular (Greenberg 1966, 31-7). 1.2. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (form containing a desinence, form without a desinence) I.e. with respect to coding properties, a form containing a desinence is more natural than a form lacking a desinence. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the third person singular and the remaining persons, such that one kind is without desinence, and the other kind has desinences, it is

145

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

the third person singular that tends to be without desinence. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the third person singular and the remaining persons, such that one kind is without desinence, and the other kind has desinences, it is the remaining persons that tend to use desinences. Q.E.D. 179. Slovenian. In the default verb conjugation (= the type DELATI 'work') the stem ends in a vowel. In the present tense (= the only synthetic tense) the desinences of the singular consist of just one consonant (except that the third person singular lacks any desinence), e.g. dela-m 'I work,' whereas the desinences of the non-singular assume the structure CV, e.g. dela-va 'we two work.' (My attention was drawn to this data by Tatjana Marvin viva voce, 2000.) The two morphological variants: the desinence consisting of C, and the desinence consisting of CV. 1. The assumptions of Naturalness Theory: 1.1. >sem (+singular, -singular) / grammatical number I.e. with respect to semantic complexity, the singular is more natural than the nonsingular. (Greenberg 1966, 31-7.) 1.2. >sym (CV, C) / desinence in environment V + ___ I.e. with respect to coding properties, the vowel-final morpheme + desinence of the structure CV is more natural than the vowel-final morpheme + desinence of the structure C.In contradistinction to the vowel-final morpheme + desinence C, the vowel-final morpheme + desinence CV makes the morpheme and the syllable boundaries coincide. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If, in a morphological paradigm, the pre-desinential morpheme ends in a vowel, and there is a choice between endings of internal structure C and CV, such that one type of endings is used in the singular, and the other type of endings is used outside the singular, it is the singular that tends to assume desinences of the structure C. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If, in a morphological paradigm, the pre-desinential morpheme ends in a vowel, and there is a choice between endings of internal structure C and CV, such that one type of endings is used in the singular, and the other type of endings is used outside the singular, it is the non-singular that tends to assume desinences of the structure CV. Q.E.D. 180. Slovenian. The so-called ethical dative is mostly used in clauses that express the speaker's negative attitude, e.g. on pa ti pozabi 'but he forgets it on you.' Similarly in English, compare the English gloss (containing the prepositional phrase on you). The two syntactic variants: clause capable of expressing positive or negative attitude, and clause capable of expressing only negative attitude. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 146

EXAMPLES

1.1.1. >sym (clause containing the ethical dative, the corresponding clause lacking the ethical dative) / in Slovenian I.e. with respect to coding properties, the clause containing the ethical dative is more natural than the corresponding clause lacking the ethical dative, in Slovenian. 1.2. >sem (+/-positive attitude, -positive attitude) / clause I.e. with respect to semantic complexity, a clause capable of expressing positive or negative attitude is more natural than a clause capable of expressing only negative attitude.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between a clause capable of expressing positive or negative attitude and a clause capable of expressing only negative attitude, such that one kind of clause contains the ethical dative, and the other kind of clause lacks the ethical dative, it is the clause capable of expressing positive or negative attitude that tends to lack the ethical dative. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between a clause capable of expressing positive or negative attitude and a clause capable of expressing only negative attitude, such that one kind of clause contains the ethical dative, and the other kind of clause lacks the ethical dative, it is the clause capable of expressing only negative attitude that tends to contain the ethical dative. Q.E.D. 181. Slovenian. The subject of the clause is usually in the nominative case. A systematic exception: the subject can stand in the genitive case if the clause is negative and its verb is a form of non-auxiliary 'be.' E.g. Janez ni doma Janez (nom.) not-is at-home 'Janez is not at home;' Janeza ni doma Janez (gen.) not-is at-home 'Janez is not at home.' (Toporii 1976, 426; 2000, 269; Herrity 2000, 63.) The two syntactic variants: any verbal lexeme v. only the non-auxiliary verb 'be,' in negative clauses. 1. The assumptions of Naturalness Theory: 1.1. >sem (nominative, other case) / subject in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than other cases, as subject in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. A special case of 1.1: 1.1.1. >sem (nominative, genitive) / subject in Slovenian I.e. with respect to semantic complexity, the nominative is more natural than the genitive, as subject in Slovenian. 1.2. >sem (the non-auxiliary verb 'be' & most other verbal lexemes, the non-auxiliary verb 'be') I.e. with respect to semantic complexity, contracting any verbal lexeme is more natural than contracting just the non-auxiliary verb 'be.'The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 147

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference, in negative clauses, between contracting any verbal lexeme and contracting only the non-auxiliary verb 'be,' such that one option occurs with the subject in the nominative, and the other option occurs with the subject in the genitive, it is in clauses using any verbal lexeme that the subject tends to be in the nominative. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference, in negative clauses, between contracting any verbal lexeme and contracting only the non-auxiliary verb 'be,' such that one option occurs with the subject in the nominative, and the other option occurs with the subject in the genitive, it is in clauses using only the non-auxiliary 'be' that the subject tends to be in the genitive. Q.E.D. 4. Note. Tjaa Mikli viva voce (in 2000) drew my attention to the following. In the non-standard language, even some other verbs than copular 'be' can be used in the construction under discussion, in which case the subject is invariably the genitive of NI 'nothing,' e.g. niesar se ni zgodilo nothing (gen.) - reflexive pronoun as free morpheme (pertaining to the verb 'happen') - not-is - happen (ptcp.) 'nothing has happened.' (Toporii 1976, 427; 2000, 269: niesar ni res 'nothing is true,' containing non-auxiliary 'be.') Tjaa Mikli suggested that the limitation of this extended usage to the subject NI 'nothing' can best be explained as hypercorrect use of the genitive niesar. In negative clauses in which NI is the direct object, the non-standard language tends to use the accusative ni instead of the prescribed genitive niesar. The pressure of the prescriptive grammar enforces the wrong use of the genitive niesar even in its role as subject. 182. Slovenian. Most verbal lexemes are negated by aid of the negative word ne 'not,' whereas three very frequent verbs ( BITI 'be,' IMETI 'have' and HOTETI 'want') display special negative verbs: nisem 'I am not,' nimam 'I do not have' and noem 'I do not want.' The two syntactic variants: negation of the verb with the negative word ne 'not,' and verb-internal negation. 1. The assumptions of Naturalness Theory: 1.1. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.1: 1.1.1. >sem ('be,' 'have,' 'want;' most other verbs) / in Slovenian I.e. with respect to semantic complexity, the verbs 'be,' 'have' and 'want' are more natural than most other Slovenian verbs. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (ne + verb, internally negated verb) / in Slovenian. I.e. with respect to coding properties, a negative word + verb is more natural than an internally negated verb, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants:

148

EXAMPLES

2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between frequent and non-frequent verbal lexemes concerning the type of negation, such that the choice is between negative word + verb and verb-internal negation, it is the frequent verbal lexemes that tend to be internally negated. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between frequent and non-frequent verbal lexemes concerning the type of negation, such that the choice is between negative word + verb and verb-internal negation, it is the non-frequent verbal lexemes that tend to show ne + verbal form. Q.E.D. 4. Notes. 4.1. Cf. deduction 183. 4.2. I cannot account for the placement of the cut-off point between internally negated verbs and other verbs. 183. Slovenian. The verbs BITI 'be,' IMETI 'have,' and HOTETI 'want' have special negative finite forms nisem 'I am not,' nimam 'I do not have' and noem 'I do not want.' In the corresponding infinite forms, the negative word is used, e.g. ne biti 'not to be,' ne imeti 'not to have,' ne hoteti 'not to want.' The two syntactic variants: the type nisem 'I am not,' and the type ne biti 'not to be.' 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ne + verb, internally negated verb) / in Slovenian. I.e. with respect to coding properties, negative word + verb is more natural than internally negated verb, in Slovenian. 1.2. >sem (+finite, -finite) / verbal form I.e. with respect to semantic complexity, a finite verbal form is more natural than an infinite verbal form. (Mayerthaler et al. 1993, 145.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the finite and infinite forms of the verbs BITI, IMETI, HOTETI, such that one set of forms is negated with a negative word, and the other set of forms has special (synthetic) negative variants, it is the finite forms that tend to have special negative variants. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the finite and infinite forms of the verbs BITI, IMETI, HOTETI, such that one set of forms is negated with a negative word, and the other

149

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

set of forms has special (synthetic) negative variants, it is the infinite forms that tend to be negated with a negative word. Q.E.D. 4. Note. Cf. deduction 182. 184. Slovenian. (a) If the main declarative clause begins with a sentence element other than the subject, the normal element order is VS, e.g. danes bere Janez knjigo today reads Janez book 'today, Janez is reading a/the book.' (b) However, if the main declarative clause is negative, the normal element order is SV, e.g. danes Janez ne bere knjige 'today, Janez is not reading a/the book.' (My attention was drawn to this matter by my students Alja Ferme and Sao ivanovi viva voce , 2000.) This deduction deals with case (a). Case (b) is considered in deduction 189. The two syntactic variants: the type Janez bere danes knjigo , and the type danes bere Janez knjigo. 1. The assumptions of Naturalness Theory: 1.1. >sem (SV, VS) / in SVO-languages I.e. with respect to semantic complexity, the element order subject - verb is more natural than the element order verb - subject, in SVO-languages.This follows from the basicness of SVO in SVO-languages (Mayerthaler et al. 1998, 349). Slovenian displays convincing SVO-properties. 1.2. >sem (+[subject=topic], -[subject=topic) / nom.-acc. languages I.e. with respect to semantic complexity, a subject which is also the topic is more natural than a subject which is not the topic, in nominative-accusative languages.The prototypical subject is also the topic in nominative-accusative languages (Mayerthaler et al. 1998, 344). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the types Janez bere danes knjigo (+subject=topic) and danes bere Janez knjigo (-subject=topic), such that in one type the element order is SV, and in the other type the element order is VS, it is in the type Janez bere danes knjigo (+subject=topic) that the element order is SV. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the types Janez bere danes knjigo (+subject=topic) and danes bere Janez knjigo (-subject=topic), such that in one type the element order is SV, and in the other type the element order is VS, it is in the type danes bere Janez knjigo (-subject=topic) that the element order is VS. Q.E.D. 185. Slovenian. The anaphoric expression for two coordinated genders G1 + G1 also contains gender G1. E.g. the anaphora of the coordination Micka in Ana 'Micka and Ana' (both are of feminine gender) is of feminine gender: onidve 'they two (fem.).' The anaphoric expression for two coordinated genders G1 + G2 contains masculine gender (= the most sem-natural gender). E.g. the anaphora of the coordination Micka in Janez 'Micka (fem.) and Janez (masc.)' contains masculine gender: onadva 'they two (masc.).' Cf. Toporii 1976, 479; 2000, 609; Herrity 2000, 76.) The two syntactic variants: the anaphoric expression of G1 + G1, and the anaphoric expression of G1 + G2, where G1 and G2 are genders.

150

EXAMPLES

1. The assumptions of Naturalness Theory: 1.1. >sem (G1 + G1, G1 + G2) / coordination of two genders I.e. with respect to semantic complexity, the coordination of two instances of the same gender is more natural than the coordination of two different genders.The coordination of two instances of the same gender is cognitively simpler than the coordination of two different genders. 1.2. >sem (expected gender, default gender) / anaphora of two coordinated noun phrases I.e. with respect to semantic complexity, the expected gender is more natural than the default gender, of the anaphora of two coordinated noun phrases.The expected gender is the one used in G1 + G1, thus any gender. The default gender is masculine in Slovenian. The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the anaphoric expression for the coordination of two genders, between the expected gender and the default gender, such that one option is G1 + G1, and the other option is G1 + G2, it is the expected gender that tends to be the anaphora for G1. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.1. If there is any difference, within the anaphoric expression for the coordination of two genders, between the expected gender and the default gender, such that one option is G1 + G1, and the other option is G1 + G2, it is the default gender that tends to be the anaphora for G1 + G2. Q.E.D. 4. Notes. 4.1. This deduction does not address the question as to why the masculine gender is the default gender in Slovenian. 4.2. Cf. deduction 83. 186. Slovenian. In negative clauses, the direct object assumes the genitive case instead of the expected accusative case, e.g. nimam asa I-not-have time-GENITIVE 'I don't have any time.' (Toporii 1976, 426-7; 2000, 579; Herrity 2000, 63.) The two syntactic variants: the affirmative clause containing a direct object in the accusative case, and the negative clause containing a direct object in the genitive case. 1. The assumptions of Naturalness Theory: 1.1. >sem (accusative, genitive) / case of direct object, in nom.-acc. languages I.e. with respect to semantic complexity, the accusative case is more natural than the genitive case, as the case of the direct object, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. 1.2. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences:

151

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the affirmative transitive clause and the negative transitive clause, such that in one type of clauses the direct object takes the accusative case, and in the other type of clauses the direct object takes the genitive case, it is the affirmative clause that tends to take the accusative case. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the affirmative transitive clause and the negative transitive clause, such that in one type of clauses the direct object takes the accusative case, and in the other type of clauses the direct object takes the genitive case, it is the negative clause that tends to take the genitive case. Q.E.D. 4. Note. Cf. deductions 187-8. 187. Slovenian. In negative clauses, the direct object assumes the genitive case instead of the expected accusative case, e.g. nimam asa I-not-have time-GENITIVE 'I don't have any time.' [This subject-matter is dealt with in deduction 186.] In negative clauses of the non-standard language the accusative is sometimes used instead of the genitive if (a) the direct object pertains to a frequent lexical item, such as 'this, he, she, it, nothing,' and/or (b) the direct object precedes the negative word + verb, e.g. to nimamo thisACCUSATIVE we-not-have 'we don't have it.' (Toporii 1976, 427; N. Jovanoski 1997; A. Boi et al. 1998; T. Jakop 1998.) This deduction deals with case (a). The two syntactic variants: negative clause containing the direct object only in the genitive, and negative clause containing the direct object in the genitive or the accusative. 1. The assumptions of Naturalness Theory: 1.1. >sem (accusative & genitive, only genitive) / case of direct object I.e. with respect to semantic complexity, the direct object which can take the accusative or the genitive is more natural than the direct object which can take only the genitive.The scale has the format >sem (A + B, A). 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.2: 1.2.1. >sem ('this' / 'that' / 'he' / 'she' / 'it' / 'nothing,' most other lexical items) I.e. with respect to semantic complexity, lexical items such as 'this,' 'that,' 'he,' 'she,' 'it,' 'nothing' are more natural than most other lexical items. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between negative transitive clauses that take the direct object in the accusative or in the genitive, and negative transitive clauses that take the direct object only in the genitive (such that one case attracts frequent lexical items, and the other non-frequent lexical items), it is the direct object containing a frequent lexical item that tends to take the genitive or the accusative case. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between negative transitive clauses that take the direct object in the accusative or in the genitive, and negative transitive clauses that take the direct object only in the genitive (such that one case attracts frequent lexical items, and the

152

EXAMPLES

other non-frequent lexical items), it is the direct object containing non-frequent lexical items that tends to take only the genitive case. Q.E.D. 4. Note. Cf. deductions 186 and 188. 188. Slovenian. In negative clauses, the direct object assumes the genitive case instead of the expected accusative case, e.g. nimam asa I-not-have time-GENITIVE 'I don't have any time.' [This subject-matter is dealt with in deduction 186.] In the non-standard language the accusative is sometimes used instead of the genitive if (a) the direct object pertains to a frequent lexical item, such as 'this, he, she, it, nothing,' and/or (b) the direct object precedes the negative word + verb, e.g. to nimamo this-ACCUSATIVE we-not-have 'we don't have it.' (Toporii 1976, 427; N. Jovanoski 1997; A. Boi et al. 1998; T. Jakop 1998.) This deduction deals with case (b). The two syntactic variants: negative transitive clauses that take the direct object in the accusative or in the genitive, and negative transitive clauses that take the direct object only in the genitive. 1. The assumptions of Naturalness Theory: 1.1. >sem (accusative & genitive, only genitive) / case of direct object I.e. with respect to semantic complexity, the direct object which can take the accusative or the genitive is more natural than the direct object which can take only the genitive.The scale has the format >sem (A + B, A). 1.2. >sym (direct object following verb, direct object preceding verb) / in Slovenian I.e. with respect to coding properties, the direct object following the verb is more natural than the direct object preceding the verb, in Slovenian.The direct object following the verb is easier for the addressee to process than the direct object preceding the verb. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between negative transitive clauses that take the direct object in the accusative or in the genitive, and negative transitive clauses that take the direct object only in the genitive (such that one case precedes the verb, and the other case follows it), it is the direct object preceding the verb that tends to take the accusative or the genitive. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between negative transitive clauses that take the direct object in the accusative or in the genitive, and negative transitive clauses that take the direct object only in the genitive (such that one case precedes the verb, and the other case follows it), it is the direct object following the verb that tends to take only the genitive case. Q.E.D. 4. Note. Cf. deductions 186-7. 189. Slovenian. (a) If the main declarative clause begins with a sentence element other than the subject, the normal element order is VS, e.g. danes bere Janez knjigo today reads Janez book 'today, Janez is reading a/the book.' (b) However, if the main declarative clause is negative, the normal element order is SV, e.g. danes Janez ne bere knjige 'today, Janez is not reading a/the book.' (My attention was drawn to this matter by my students Alja Ferme and Sao ivanovi viva voce , 2000.) This deduction deals with case (b). Case (a)

153

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

is considered in deduction 184. The two syntactic variants: the type danes bere Janez knjigo , and the type danes Janez ne bere knjige . 1. The assumptions of Naturalness Theory: 1.1. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 1.2. >sem (VS, SV) / element order in main declarative clause whose subject is not clause-initial I.e. with respect to semantic complexity, the element order VS is more natural than the element order SV in main declarative clauses whose subject is not clause-initial.This follows from deduction 184, item 3.2. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the types danes bere Janez knjigo (affirmative) and danes Janez ne bere knjige (negative), such that one type is SV, and the other type is VS, it is the type danes bere Janez knjigo (affirmative) that tends to be VS. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the types danes bere Janez knjigo (affirmative) and danes Janez ne bere knjige (negative), such that one type is SV, and the other type is VS, it is the type danes Janez ne bere knjige (negative) that tends to be SV. Q.E.D. 190. Slovenian. During hair wash the hairdresser asks the customer, "Bo voda dobra?" will-be water good 'is the water good enough (polite version).' A neutral version of the same question would be, "Je voda dobra?" is water good 'is the water good enough.' The two syntactic variants: question 'is the water good enough' in normal and polite versions. 1. The assumptions of Naturalness Theory: 1.1. >sem (present tense, other tense) / verbal tense I.e. with respect to semantic complexity, the present tense is more natural than other verbal tenses. (Mayerthaler 1981, 14.) A special case of 1.1: 1.1.1. >sem (present tense, future tense) / verbal tense I.e. with respect to semantic complexity, the present tense is more natural than the future tense. 1.2. >sem (speaker's normal register, speaker's other registers) I.e. with respect to semantic complexity, a speaker's normal register is more natural than a speaker's other registers. (Dotter 1990, 228.) A special case of 1.2: 1.2.1. >sem (normal register, polite register) I.e. with respect to semantic complexity, a speaker's normal register is more natural than a speaker's polite register. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem

154

EXAMPLES

3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the normal and the polite registers, such that the present tense is used in one register, and the future tense is used in the other register, it is in the normal register that the present tense tends to be used. Q.E.D. From 1.1.1, 1.2.1 and 2.2. it can be deduced: 3.2. If there is any difference between the normal and the polite registers, such that the present tense is used in one register, and the future tense is used in the other register, it is in the polite register that the future tense tends to be used. Q.E.D. 191. Slovenian. The non-honorific personal pronoun of address uses the singular: ti 'you-SINGULAR.' The corresponding honorific pronoun uses the plural: Vi 'youPLURAL.' Similarly in a number of European and other languages. The two syntactic variants: honorific 'you,' and non-honorific 'you.' 1. The assumptions of Naturalness Theory: 1.1. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 1.2. >sem (-honorific, +honorific) / personal pronoun of address I.e. with respect to semantic complexity, the non-honorific personal pronoun of address is more natural than the corresponding honorific pronoun.Primitive and early civilizations employ non-honorific pronouns of address only. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the honorific and non-honorific personal pronouns of address, such that one kind uses the singular, and the other kind uses the plural, it is the non-honorific pronoun that tends to use the singular. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the honorific and non-honorific personal pronouns of address, such that one kind uses the singular, and the other kind uses the plural, it is the honorific pronoun that tends to use the plural. Q.E.D. 4. Note. The deduction treats the addressing of ONE person only. 192. Slovenian, a part of the Gorenjsko dialect. Female persons show masculine agreement with the verb and with the adjective, e.g. jaz sem el I=woman am goPTCP.MASCULINE 'I went.' Non-persons of feminine gender show the expected feminine agreement, e.g. krava je la cow-FEMININE is go-PTCP.FEMININE 'the cow went.' Both examples are here rendered in standard Slovenian. (Cf. M. Beter 1998.) The two syntactic variants: +human nouns and personal pronouns of feminine gender, and -human nouns and personal pronouns of feminine gender. 1. The assumptions of Naturalness Theory: 1.1. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 1.2. >sem (masculine, feminine) I.e. with respect to semantic complexity, masculine is more natural than feminine. (Mayerthaler 1981, 15.) 155

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between persons and non-persons of feminine gender concerning the choice of masculine or feminine gender in agreement, it is the persons of feminine gender that tend to use the masculine gender in agreement. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between persons and non-persons of feminine gender concerning the choice of masculine or feminine gender in agreement, it is the non-persons of feminine gender that tend to use the feminine gender in agreement. Q.E.D. 4. Note. The facts may be misrepresented above: some speakers do not apply the masculinization of females in the third person. This alternative is considered in deduction 193. 193. Slovenian, a part of the Gorenjsko dialect. Subjects that denote female persons show masculine agreement with the verb and with the adjective in the non-third person, e.g. jaz sem el I=woman am go-PTCP.MASCULINE 'I went,' and only optionally in the third person, e.g. la/el je go-PTCP.FEMININE/MASCULINE 'she went.' The example is here rendered in standard Slovenian. (Cf. M. Beter 1998.) The two syntactic variants: masculine or feminine agreement and only masculine agreement, within female persons. 1. The assumptions of Naturalness Theory: 1.1. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third person is more natural than the non-third person. (According to the linguistic tradition beginning with Jakobson 1932.) 1.2. >sem (masculine and feminine, masculine only) / in non-standard Slovenian I.e. with respect to semantic complexity, using masculine and feminine is more natural than using masculine only, in non-standard Slovenian.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within female persons, between using masculine or feminine agreement and using only masculine agreement, such that one option occurs in the third person, and the other option occurs in the non-third person, it is the the use of masculine or feminine agreement that tends to occur in the third person. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within female persons, between using masculine or feminine agreement and using only masculine agreement, such that one option occurs in the third person, and the other option occurs in the non-third person, it is the the exclusive use of masculine agreement that tends to occur in the non-third person. Q.E.D. 4. Note. For a refinement, see deduction 194. 194. Slovenian, a part of the Gorenjsko dialect. Subjects that denote female persons show masculine agreement with the verb and with the adjective in the singular, or at least

156

EXAMPLES

in the non-third person singular, e.g. jaz sem el I=woman am go-PTCP.MASCULINE 'I went.' (The example is here rendered in standard Slovenian.) This situation obtains in familiar usage only. The two syntactic variants: the +intimate use, and the -intimate use, in the singular, or at least in the non-third person singular. 1. The assumptions of Naturalness Theory: 1.1. >sem (masculine, feminine) I.e. with respect to semantic complexity, masculine is more natural than feminine. (Mayerthaler 1981, 15.) 1.2. >sem (+intimate, -intimate) I.e. with respect to semantic complexity, +intimate is more natural than -intimate. This is based on the circumstance that in early and primitive civilizations intimate relations prevail. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the intimate use and the non-intimate use of female-person subjects, such that one use has masculine agreement with the verb and the adjective, and the other use has feminine agreement with the verb and the adjective, it is the +intimate use that tends to have masculine agreement. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the intimate use and the non-intimate use of female-person subjects, such that one use has masculine agreement with the verb and the adjective, and the other use has feminine agreement with the verb and the adjective, it is the -intimate use that tends to have feminine agreement. Q.E.D. 195. Slovenian. In nouns, singular, plural and dual are distinguished. The expression of 'two entities' is distributed between the dual and the plural: the plural is used for two entities that come in pairs, e.g. (moje) roke 'my hands.' The dual is employed for the remaining instances of two entities, e.g. knjigi 'two books.' (Toporii 1976, 207; 2000, 271; Herrity 2000, 37-9.) The two syntactic variants: the dual and the plural of nouns as an expression of 'two entities.' 1. The assumptions of Naturalness Theory: 1.1. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 1.2. >sem (+pairwise, -pairwise) / two entities I.e. with respect to semantic complexity, two entities that come in pairs are more natural than two entities that do not come in pairs. (Mayerthaler et al. 1998, 93.) Some languages, for instance Tocharian, have a grammatical number ("paralis") exclusively for two entities that come in pairs (Vyacheslav V. Ivanov viva voce, 1994). Tocharian has a dual as well, used for two entities that do not come in pairs (Krause & Thomas 1960, 767). Most languages that have a dual use it to express two entities that come in pairs, only some languages also use it to express two entities of any kind. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants:

157

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the dual and plural (as an expression of 'two entities'), such that one number is used with nouns that mean +pairwise, and the other number is used with nouns that mean -pairwise, it is the plural that tends to be used with two entities that come in pairs. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the dual and plural (as an expression of 'two entities'), such that one number is used with nouns that mean +pairwise, and the other number is used with nouns that mean -pairwise, it is the dual that tends to be used with two entities that do not come in pairs. Q.E.D. 4. Note. The consequence 3.2 is not altogether true. This is remedied in deduction 196. 196. Slovenian. In nouns, singular, plural and dual are distinguished. The expression of 'two entities' is distributed between the dual and the plural. The plural is used for two entities that come in pairs and are not accompanied by a quantifier meaning 'two,' e.g. (moje) roke 'my hands.' The dual is employed in the remaining instances, e.g. obe moji roki 'both of my hands,' (dve) knjigi '(two) books.' (Toporii 1976, 207; Herrity 2000, 379.) The two syntactic variants: the dual and the plural of nouns as an expression of 'two entities that come in pairs' and are (not) accompanied by a quantifier meaning 'two.' 1. The assumptions of Naturalness Theory: 1.1. >sem (plural, dual) I.e. with respect to semantic complexity, the plural is more natural than the dual. (Greenberg 1966, 31-7.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+[quantifier 'two'], -[quantifier 'two']) / expressions that mean 'two entities' and come in pairs, in Slovenian I.e. with respect to coding properties, the presence of a quantifier meaning 'two' in expressions that mean 'two entities' and come in pairs is more natural than the absence of such a quantifier, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sem tends to associate with >sym 2.2. >sem tends to associate with <sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the dual and the plural of nouns (as an expression of 'two entities that come in pairs'), such that one number is accompanied by a quantifier meaning 'two,' and the other number is not accompanied by a quantifier meaning 'two,' it is the unit containing the quantifier that tends to use the dual. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced:

158

EXAMPLES

3.2. If there is any difference between the dual and the plural of nouns (as an expression of 'two entities that come in pairs') such that one number is accompanied by a quantifier meaning 'two,' and the other number is not accompanied by a quantifier meaning 'two,' it is the unit lacking the quantifier that tends to use the plural. Q.E.D. 4. Note. This deduction continues deduction 195. 197. Slovenian. The partial involvement of the patient is coded with richer syntactic structure than the total involvement of the patient. E.g. Janez je pojedel malo mesa 'Janez ate up a little meat' v. Janez je pojedel meso 'Janez ate up the meat.' The two syntactic variants: the type Janez je pojedel malo mesa , and the type Janez je pojedel meso. 1. The assumptions of Naturalness Theory: 1.1. >sem (total affectedness, partial affectedness) / of the patient I.e. with respect to semantic complexity, the total affectedness of the patient in the event is more natural than the partial affectedness of the patient. (Hopper & S. Thompson 1980, 287 and passim.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (malo mesa 'a little meat,' meso 'meat') / in Slovenian I.e. with respect to coding properties, malo mesa is more natural than meso, in Slovenian. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the types Janez je pojedel malo mesa and Janez je pojedel meso, such that in one type partial affectedness of the patient is encoded, and in the other type total affectedness of the patient is encoded, it is the type indicating partial affectedness of the patient that tends to be coded more richly than the type indicating total affectedness of the patient. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the types Janez je pojedel malo mesa and Janez je pojedel meso, such that in one type partial affectedness of the patient is encoded, and in the other type total affectedness of the patient is encoded, it is the type indicating total affectedness of the patient that tends to be coded more frugally than the type indicating partial affectedness of the patient. Q.E.D. 4. Note. Additional examples (from Toporii 1976, 293-4; 2000, 356): dobil je ugled 'he acquired esteem' (i.e. initially he had none) = total change of esteem, dobil je na ugledu 'he acquired more esteem' (i.e. initially he had some) = partial change of esteem; izgubiti ast 'lose one's honour' (i.e. have no honour left) = total change of honour, izgubiti na asti 'lose some honour' (i.e. have some honour left) = partial change of honour.

159

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

198. Slovenian. If the subject of the clause stands in the nominative case, reflexivity is expressed with a reflexive pronominal (personal or possessive, as the case may be), e.g. Janez sanja o svojem avtu 'Janez i is dreaming about his i car:' the subject Janez is in the nominative case, svojem is a form of the possessive reflexive pronominal. If the subject of the clause is not in the nominative case, reflexivity is expressed with a non-reflexive pronominal (personal or possessive, as the case may be), e.g. Janezu se sanja o njegovem avtu 'Janez i is dreaming about his i car:' the subject Janezu is in the dative case, njegovem is a form of the possessive non-reflexive pronominal. The two syntactic variants: the type Janez sanja o svojem avtu , and the type Janezu se sanja o njegovem avtu . 1. The assumptions of Naturalness Theory: 1.1. >sem (+nominative, -nominative) / subject in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than the nonnominative, as the subject in nominative-accusative languages. (Cf. Mayerthaler 1981, 14.) 1.2. >sem (+reflexive, -reflexive) / pronominal as expression of reflexivity I.e. with respect to semantic complexity, a reflexive pronominal is more natural than a non-reflexive pronominal, as an expression of reflexivity. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the types Janez sanja o svojem avtu and Janezu se sanja o njegovem avtu , such that in one type a reflexive pronominal is used to express reflexivity, and in the other type a non-reflexive pronominal is used to express reflexivity, it is the type Janez sanja o svojem avtu that tends to employ the reflexive pronominal. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the types Janez sanja o svojem avtu and Janezu se sanja o njegovem avtu such that in one type a reflexive pronominal is used to express reflexivity, and in the other type a non-reflexive pronominal is used to express reflexivity, it is the type Janezu se sanja o njegovem avtu that tends to employ the non-reflexive pronominal. Q.E.D. 199. Slovenian. Reflexivity. If the participant referred to in the subject is represented as totally involved in a reflexive act, the reflexive pronoun used is se (= accusative), e.g. Janez se umiva 'Janez is washing (himself).' If the involvement is only partial, the reflexive pronoun used is si (= dative), e.g. Janez si umiva roke 'Janez is washing his hands.' (Cf. Lichtenberk 1999, 315 on Czech.) The two syntactic variants: the type Janez se umiva, and the type Janez si umiva roke . 1. The assumptions of Naturalness Theory: 1.1. >sem (total affectedness, partial affectedness) / of the patient I.e. with respect to semantic complexity, the total affectedness of the patient in the event is more natural than the partial affectedness of the patient. (Hopper & S. Thompson 1980, 287 and passim.) 1.2. >sem (se, si) / clitic reflexive pronoun of Slovenian I.e. with respect to semantic complexity, the clitic reflexive pronoun se is more natural

160

EXAMPLES

than the clitic reflexive pronoun si, in Slovenian.This is based on the circumstance that se is an accusative, and si is a dative. The accusative is more natural than the dative, in nominative-accusative languages (Mayerthaler et al. 1998, 167). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the types Janez se umiva (total affectedness of the subject) and Janez si umiva roke (partial affectedness of the subject), such that one type uses the clitic reflexive pronoun se, and the other type uses the clitic reflexive pronoun si, it is the type showing total affectedness of the subject that tends to use the clitic reflexive pronoun se. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the types Janez se umiva (total affectedness of the subject) and Janez si umiva roke (partial affectedness of the subject), such that one type uses the clitic reflexive pronoun se, and the other type uses the clitic reflexive pronoun si, it is the type showing partial affectedness of the subject that tends to use the clitic reflexive pronoun si. Q.E.D. 4. Note. The pronoun si is <sem (according to 1.2), therefore tends to associate with >sym, i.e. tends to attract relatively rich syntactic structure. It is indeed the case that the type Janez si umiva roke is richer in structure than the type Janez se umiva . Cf. deduction 197. 200. Slovenian. In the masculine gender, to some extent also in the neuter gender, the accusative singular takes the form of the corresponding nominative if the noun is inani mate, e.g. klobuk 'hat (masc.)' = nominative and accusative singular; polje 'field (neuter)' = nominative and accusative singular. The accusative singular takes the form of the corresponding genitive, if the noun is animate, e.g. otrok 'child (masc.) = nominative singular, otroka = accusative and genitive singular; tele 'calf (neuter)' = nominative and accusative singular, teleta = genitive and accusative singular ( [ne] vidim teleta 'I can[not] see the calf'). The accusative of inanimate masculine singular nouns also takes the form of the corresponding genitive in some (not so few) cases, e.g. jur 'a thousand (banknote),' jurja = accusative and genitive singular. (Toporii 1976, 218 concerning the masculine gender; the neuter type acc. = gen. teleta is not mentioned; Herrity 2000, 34.) The deduction is continued in deduction 201. The two morphological variants: the accusative singular masculine which equals the corresponding nominative, and the accusative singular masculine which does not equal the corresponding nominative. 1. The assumptions of Naturalness Theory: 1.1. >sem (accusative nominative, accusative = nominative) / accusative singular of masculine nouns in Slovenian I.e. with respect to semantic complexity, an accusative which does not equal the corresponding nominative is more natural than an accusative which does equal the corresponding nominative, in Slovenian masculine nouns.If the language has a distinction (here: between the nominative and the corresponding accusative), the realization of that distinction has a high sem-value. 1.2. >sem (+animate, -animate)

161

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to semantic complexity, animate is more natural than inanimate. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within singular masculine nouns, between an accusative which does not equal the corresponding nominative and an accusative which does equal the corresponding nominative, such that one kind of accusatives are animate, and the other kind of accusatives are inanimate, it is the accusative which does not equal the corresponding nominative that tends to be animate. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.21. If there is any difference, within singular masculine nouns, between an accusative which does not equal the corresponding nominative and an accusative which does equal the corresponding nominative, such that one kind of accusatives are animate, and the other kind of accusatives are inanimate, it is the accusative which equals the corresponding nominative that tends to be inanimate. Q.E.D. 4. Note. The type jur 'a thousand (banknote)' must be marked as +animate in the lexicon. 201. Slovenian. In the masculine gender, to some extent also in the neuter gender, the accusative singular takes the form of the corresponding nominative if the noun is inani mate, e.g. klobuk 'hat (masc.)' = nominative and accusative singular; polje 'field (neuter)' = nominative and accusative singular. The accusative singular takes the form of the corresponding genitive, if the noun is animate, e.g. otrok 'child (masc.) = nominative singular, otroka = accusative and genitive singular; tele 'calf (neuter)' = nominative and accusative singular, teleta = genitive and accusative singular ( [ne] vidim teleta 'I can[not] see the calf'). The accusative of inanimate masculine singular nouns also takes the form of the corresponding genitive in some (not so few) cases, e.g. jur 'a thousand (banknote),' jurja = accusative and genitive singular. (Toporii 1976, 218 concerning the masculine gender; the neuter type acc. = gen. teleta is not mentioned; Herrity 2000, 34.) The deduction continues deduction 200. The two morphological variants: accusative = nominative or accusative nominative, and accusative = nominative, within neuter singular nouns. 1. The assumptions of Naturalness Theory: 1.1. >sem (+animate, -animate) I.e. with respect to semantic complexity, animate is more natural than inanimate. (Mayerthaler 1981, 14.) 1.2. >sem (accusative = nominative & accusative nominative, only accusative = nominative) / neuter singular nouns in Slovenian I.e. with respect to semantic complexity, a neuter singular noun whose accusative = nominative as well as accusative nominative is more natural than a neuter singular noun whose accusative = nominative.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem

162

EXAMPLES

2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within neuter singular nouns, between the accusative = nominative or accusative nominative and the accusative = nominative, such that one kind of accusatives is animate, and the other kind of accusatives is inanimate, it is the accusative which does or does not equal the nominative that tends to be animate. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within neuter singular nouns, between the accusative = nominative or accusative nominative and the accusative = nominative, such that one kind of accusatives is animate, and the other kind of accusatives is inanimate, it is the accusative which invariably equals the nominative that tends to be inanimate. Q.E.D. 202. Slovenian. In the genitive, dative and accusative, personal pronouns have full and clitic forms, e.g. meni (= full form) and mi (= corresponding clitic form) 'to me.' The full forms are used when the pronoun is emphatic, e.g. in the saying danes meni, jutri tebi 'today [it happens] to me, tomorrow [it will happen] to you.' (Toporii 1976, 240, 242; 2000, 307; Herrity 2000, 89-90.) The two syntactic variants: the full and the clitic forms of the personal pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (full form, clitic form) / of personal pronoun in Slovenian I.e. with respect to coding properties, the full form of the personal pronoun is more natural than the corresponding clitic form, in Slovenian. 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the full and the corresponding clitic forms of personal pronouns, such that one kind of personal pronouns is emphatic, and the other kind of personal pronouns is not emphatic, it is the full forms that tend to be used emphatically. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the full and the corresponding clitic forms of personal pronouns, such that one kind of personal pronouns is emphatic, and the other kind of personal pronouns is not emphatic, it is the clitic forms that tend to be used nonemphatically. Q.E.D. 203. Slovenian. Some quantifiers, e.g. nekaj 'some' and ve 'several,' including the

163

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

cardinal numbers 5-99 and the round cardinal numbers, behave as nouns in the nominative and the accusative, i.e. take the genitive case of the unit counted, e.g. pet deklet 'five young ladies' (deklet = gen. pl.). Outside the two cases those quantifiers behave as adjectives, e.g. s pet(imi) dekleti 'with five young ladies' ( dekleti = instr. pl. required by the preposition s 'with'). (Toporii 1976, 268; 2000, 331-2; Herrity 2000, 132-3.) The two syntactic variants: the type pet deklet , and the type s pet(imi) dekleti . 1. The assumptions of Naturalness Theory: 1.1. >sem (noun, adjective) I.e. with respect to semantic complexity, nouns are more natural than adjectives. Adjectives are not a universal category. (Mayerthaler et al. 1998, 24.) A special case of 1.1: 1.1.1. >sem (noun, adjective) / quantifier in Slovenian I.e. with respect to semantic complexity, nouns are more natural than adjectives, when they are quantifiers in Slovenian. 1.2. >sem (nominative/accusative, other cases) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative and the accusative are more natural than other cases, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. Cf. Mayerthaler 1987, 41; Mayerthaler et al. 1998, 167. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the quantifiers in the nominative & accusative and the quantifiers in the other cases, such that the quantifiers in one kind of cases behave as nouns, and the quantifiers in the other kind of cases behave as adjectives, it is the quantifiers in the nominative & the accusative that tend to behave as nouns. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the quantifiers in the nominative & accusative and the quantifiers in the other cases, such that the quantifiers in one kind of cases behave as nouns, and the quantifiers in the other kind of cases behave as adjectives, it is the quantifiers in the cases other than the nominative & the accusative that tend to behave as adjectives. Q.E.D. 204. Slovenian. The so-called l-participles (whose main function is to form several verbal tenses) can be used as adjectives if derived from ergative verbs. E.g. PASTI 'fall', its l-participle padel ; the latter can be used as adjective, e.g. vojak, padel v vojni 'a/the soldier fallen in the war', padli vojaki '(the) fallen soldiers'. The l-participles of other than ergative verbs cannot be used as adjectives. (My attention was drawn to this data by Tatjana Marvin viva voce , 2001.) The two syntactic variants: an l-participle which can also be used as an adjective, and an l-participle which can only be used as a participle. 1. The assumptions of Naturalness Theory: 1.1. >sem (participle & adjective, only participle) / l-participle in Slovenian I.e. with respect to semantic complexity, a verb whose l-participle can also be used as adjective is more natural than a verb whose l-participle cannot be used as adjective, in Slovenian.The scale has the format >sem (A + B, A).

164

EXAMPLES

1.2. >sem (ergative, other) / verb I.e. with respect to semantic complexity, an ergative verb is more natural than other verbs.Other verbs do not constitute a "natural class," therefore their sem-value is low. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between an l-participle which can also be used as an adjective and an l-participle which can only be used as a participle, such that one kind of l-participles belongs to ergative verbs, and the other kind of l-participles belongs to other verbs, it is the l-participle which can also be used as an adjective that tends to belong to ergative verbs. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between an l-participle which can also be used as an adjective and an l-participle which can only be used as a participle, such that one kind of l-participles belongs to ergative verbs, and the other kind of l-participles belongs to other verbs, it is the l-participle which can only be used as a participle that tends to belong to other verbs. Q.E.D. 205. Slovenian. When the relative clause is introduced by the relative particle ki, the latter is accompanied by a corresponding form of the personal pronoun, usually clitic, elsewhere in the clause, e.g. lovek, ki ga vidi man ki him you-see 'the man whom you see.' The personal pronoun is normally omitted in the nominative case, e.g. lovek, ki stoji tam man ki stands there 'the man who is standing there.' (Slovenian is a pro-drop language.) The two syntactic variants: the relative clause introduced by the particle ki and containing a form of the personal pronoun, and the relative clause introduced by the particle ki and lacking the personal pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ki + personal pronoun, the bare ki) / in Slovenian I.e. with respect to coding properties, the particle ki followed by a corresponding personal pronoun is more natural than the bare ki, in Slovenian. 1.2. >sem (nominative, other cases) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than other cases, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced:

165

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.1. If there is any difference between the presence and absence of the personal pronoun in the relative clause introduced by ki, such that one situation obtains in the nominative, and the other situation obtains outside the nominative, it is outside the nomi native case that the personal pronoun tends to be used. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the presence and absence of the personal pronoun in the relative clause introduced by ki, such that one situation obtains in the nominative, and the other situation obtains outside the nominative, it is in the nominative case that the personal pronoun tends not to be used. Q.E.D. 206. Slovenian. Relative clauses are introduced either by the relative particle ki followed somewhere in the clause by a personal pronoun (usually a clitic, and normally omitted in the nominative case), or by the relative pronoun kater-i inflected for number, case and gender. Ki prevails in non-prepositional cases, especially in the nominative. Kateri is normal in prepositional cases, and to some extent in the non-prepositional cases outside the nominative. The two syntactic variants: relative clause introduced by kater-i, and the relative clause introduced by ki. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (kater-i, ki ) / in Slovenian I.e. with respect to coding properties, kater-i is more natural than ki, in Slovenian. 1.1.2. >sym (+prepositional case, -prepositional case) I.e. with respect to coding properties, a prepositional case is more natural than a nonprepositional case. 1.2. >sem (nominative, other cases) / in nom.-acc. languages I.e. with respect to semantic complexity, the nominative is more natural than other cases, in nominative-accusative languages. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 2.3. >sym tends to associate with <sem 2.4. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.1.2, 1.2, 2.1 and 2.3 it can be deduced: 3.1. If there is any difference between the relative clauses introduced by ki or kater-i, such that one kind of relative clauses is associated with prepositional cases, and the other kind of relative clauses is associated with non-prepositional cases, and such that one kind of relative clauses is associated with the nominative, and the other kind of relative clauses is used generally outside the nominative, it is kater-i that tends to be associated with prepositional cases and generally with cases outside the nominative. Q.E.D. From 1.1.1, 1.1.2, 1.2, 2.2 and 2.4 it can be deduced: 3.2. If there is any difference between the relative clauses introduced by ki or kater-i,

166

EXAMPLES

such that one kind of relative clauses is associated with prepositional cases, and the other kind of relative clauses is associated with non-prepositional cases, and such that one kind of relative clauses is associated with the nominative, and the other kind of relative clauses is used generally outside the nominative, it is ki that tends to associate with nonprepositional cases, and especially with the nominative. Q.E.D. 4. Note. Our theory can account for the competition between the subordinators ki and kater-i only if it is arbitrarily decided (as has in fact been done here) that the personal pronoun which normally accompanies ki within the same clause, be ignored in the naturalness scale 1.1.1. 207. Slovenian. Slovenian used to have two demonstrative pronominals meaning 'such:' tak and taken . Now tak survives in neutral usage, whereas taken is slightly expressive. Slovenian also used to have two interrogative pronominals meaning 'what like:' kak and kaken . Now kak is obsolescent, whereas kaken has survived in neutral usage. (My attention was drawn to this matter by Joe Toporii viva voce , 2000.) The two morphological variants: the type ending in ken, and the type ending in k. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type ending in ken, the type ending in k) / demonstrative and interrogative pronominals in Slovenian I.e. with respect to coding properties, the type ending in ken is more natural than the type ending in k, within demonstrative and interrogative pronominals of Slovenian.The type ending in ken has more sound body than the type ending in k. 1.2. >sem (demonstrative, interrogative) / pronominal I.e. with respect to semantic complexity, a demonstrative pronominal is more natural than an interrogative pronominal.Phylogenetically demonstratives are earlier than interrogatives. Suppletion is more likely in demonstratives than in interrogatives (Varja Cvetko-Orenik viva voce , 2001). 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference, within interrogative and demonstrative pronominals, between the type ending in ken and the type ending in k, such that one type pertains to demonstratives, and the other type pertains to interrogatives, it is the type ending in ken that tends to assert itself in interrogatives. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference, within interrogative and demonstrative pronominals, between the type ending in ken and the type ending in k, such that one type pertains to demonstratives, and the other type pertains to interrogatives, it is the type ending in k that tends to assert itself in demonstratives. Q.E.D. 4. Note. For those speakers who feel that even taken is in neutral usage, see 167

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

deduction 208. 208. Slovenian. Slovenian used to have two demonstrative pronominals meaning 'such:' tak and taken. Both are still in neutral use, for some speakers. Slovenian also used to have two interrogative pronominals meaning 'what like:' kak and kaken. The variant kak is obsolescent or obsolete. (My interpretation. My attention was drawn to this matter by Joe Toporii viva voce , 2000.) The two morphological variants: the use both of the type ending in ken and of the type ending in k, and the exclusive use of the type ending in ken, within interrogative and demonstrative pronominals. 1. The assumptions of Naturalness Theory: 1.1. >sem (the type ending in ken & the type ending in k, only the type ending in ken) / demonstrative and interrogative pronominals in Slovenian I.e. with respect to semantic complexity, demonstrative and interrogative pronominals admitting both the type ending in ken and the type ending in k are more natural than demonstrative and interrogative pronominals admitting only the type ending in ken, in Slovenian.The scale has the format >sem (A + B, A). 1.2. >sem (demonstrative, interrogative) / pronominal I.e. with respect to semantic complexity, a demonstrative pronominal is more natural than an interrogative pronominal.Phylogenetically demonstratives are earlier than inter rogatives. Suppletion is more likely in demonstratives than in interrogatives (Varja Cvetko-Orenik viva voce , 2001). 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within interrogative and demonstrative pronominals, between the use both of the type ending in ken and of the type ending in k and the exclusive use of the type ending in ken, such that one type occurs in demonstrative pronominals, and the other type occurs in interrogative pronominals, it is the use both of the type ending in ken and of the type ending in k that tends to occur in demonstrative pronominals. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within interrogative and demonstrative pronominals, between the use both of the type ending in ken and of the type ending in k and the exclusive use of the type ending in ken, such that one type occurs in demonstrative pronominals, and the other type occurs in interrogative pronominals, it is the exclusive use of the type ending in ken that tends to occur in interrogative pronominals. Q.E.D. 4. Note. For those speakers who feel that taken is not in neutral usage, see deduction 207. 209. Slovenian and elsewhere. Vivid narrative utilizes the present tense. (My attention was drawn to this matter by Milan Dolgan viva voce, 1997.) The two syntactic variants: narrative in the past tense, and narrative in the present tense. 1. The assumptions of Naturalness Theory: 1.1. >sem (past tense, present tense) / narrative I.e. with respect to semantic complexity, the past tense is more natural than the

168

EXAMPLES

present tense, in narrative.This is based on the circumstance that the past tenses are the prototypical tenses in narrative. 1.2. >sem (stylistically unmarked, stylistically marked) I.e. with respect to semantic complexity, a stylistically unmarked expression is more natural than a stylistically marked expression.This is based on the circumstance that the stylistically marked counterpart may simply be lacking. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between normal and vivid narrative, such that one kind of narrative uses the present tense, and the other kind of narrative uses the past tense, it is the past tense that tends to be used in normal narrative. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between normal and vivid narrative, such that one kind of narrative uses the present tense, and the other kind of narrative uses the past tense, it is the present tense that tends to be used in vivid narrative. Q.E.D. 210. Spanish. Personal pronouns. Gender is indicated (a) in third person singular only (English, German), (b) in third person only (French), (c) in third person and plural generally (Spanish), (d) in all persons, singular and plural, except first (Arabic). (Lyons 1999, 137.) Case (c) is treated in the present deduction. Case (a) is dealt with in deduction 57, case (b) in deduction 43 and case (d) is considered in deduction 6. The two syntactic variants: the personal pronouns of the non-third person singular, and the personal pronouns of the third person singular and of the plural. 1. The assumptions of Naturalness Theory: 1.1. >sem (non-third person singular; plural & third person singular) / grammatical person I.e. with respect to semantic complexity, the non-third person singular is more natural than the persons of the plural & the third person singular.The non-third person singular does form a "natural class," whereas the third person singular in conjunction with the persons of the plural does not form a "natural class," therefore their sem-naturalness is low. 1.2. >sem (genders conflated, genders separated) I.e. with respect to semantic complexity, conflated genders are more natural than separated genders.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the personal pronouns of the non-third person singular and the personal pronouns of the plural & of the third person singular, such that one kind has separated genders, and the other kind not, it is the personal pronouns of the non-third person singular that tend to have conflated genders. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the personal pronouns of the non-third person

169

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

singular and the personal pronouns of the plural & of the third person singular, such that one kind has separated genders, and the other kind not, it is the personal pronouns of the plural & of the third person singular that tend to have separated genders. Q.E.D. 4. Note. Cf. deduction 138. 211. Spanish. The personal pronouns of the third person keep gender distinctions in the corresponding clitic forms lo, la, los, las, whereas the personal pronouns of the nonthird person plural lose gender distinctions in the corresponding clitic forms nos, os. (Lyons 1999, 138.) The two syntactic variants: the personal pronouns of the non-third person plural: full and clitic forms. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (full form, clitic form) / of the non-third person plural pronouns in Spanish I.e. with respect to coding properties, the full forms of the non-third person plural personal pronouns are more natural than the corresponding clitic forms, in Spanish. 1.2. >sem (genders conflated, genders separated) / the personal pronouns of the nonthird person plural in Spanish I.e. with respect to semantic complexity, conflated genders are more natural than separated genders, in the personal pronouns of the non-third person plural in Spanish.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference, within the personal pronouns of the non-third person plural, between their full and clitic forms, such that one kind distinguishes genders, and the other kind does not, it is the full forms that tend to distinguish genders. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference, within the personal pronouns of the non-third person plural, between their full and clitic forms, such that one kind distinguishes genders, and the other kind does not, it is the clitic forms that tend not to distinguish genders. Q.E.D. 212. Spanish. The indefinite article can indicate a non-literal reading, e.g. son unos payasos 'they are (like) clowns,' as opposed to son payasos 'they are clowns' (literal reading). (Lyons 1999, 104, partly referring to Butt & C. Benjamin.) The two syntactic variants: the type son payasos, and the type son unos payasos. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see 170

EXAMPLES

Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type son unos payasos, the type son payasos) / in Spanish I.e. with respect to coding properties, the type son unos payasos is more natural than the type son payasos, in Spanish. 1.2. >sem (+literal, -literal) / reading I.e. with respect to coding properties, the literal reading is more natural than any corresponding non-literal reading.The non-literal reading may simply be lacking in some cases. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type son payasos and the type son unos payasos, such that one type has the literal reading, and the other type the non-literal reading, it is the type son unos payasos that tends to have the non-literal meaning. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type son payasos and the type son unos payasos, such that one type has the literal reading, and the other type the non-literal reading, it is the type son payasos that tends to have the literal meaning. Q.E.D. 213. Spanish. (a) The indefinite article tends not to be used in predicative noun phrases, e.g. Mara se hizo dentista 'Mara became a dentist.' In two cases the indefinite article tends to be used even in predicative noun phrases: (b) if the head noun of the noun phrase is modified, e.g. son unos conservadores arrepentidos 'they are repentant conservaives;' (c) if the head noun of the noun phrase is -human, in which case the use of the indefinite article is either obligatory (e.g. es un problema 'it is a problem') or optional (e.g. es (una) cuestin de dinero 'it is a question of money'). (My interpretation of Lyons 1999, 104-5, who partly refers to Butt & C. Benjamin.) The present deduction considers case (a). For cases (b-c) see deductions 214-5. The two syntactic variants: indefinite noun phrases used predicatively and non-predi catively. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type un dentista, the type dentista) / in Spanish I.e. with respect to coding properties, the type un dentista is more natural than the type dentista, in Spanish. 1.2. >sem (predicative noun phrase, other uses of noun phrase) I.e. with respect to semantic complexity, the predicative noun phrase is more natural than other uses of the noun phrase.This is based on the circumstance that the other uses of the noun phrase do not constitute a "natural class." 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 171

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type dentista and the type un dentista, such that one type is used in predicative noun phrases, and the other type elsewhere, it is the type un dentista that tends to be used elsewhere. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type dentista and the type un dentista, such that one type is used in predicative noun phrases, and the other type elsewhere, it is the type dentista that tends to be used in predicative noun phrases. Q.E.D. 214. Spanish. (a) The indefinite article tends not to be used in predicative noun phrases, e.g. Mara se hizo dentista 'Mara became a dentist.' In two cases the indefinite article tends to be used even in predicative noun phrases: (b) if the head noun of the noun phrase is modified, e.g. son unos conservadores arrepentidos 'they are repentant conservatives;' (c) if the head noun of the noun phrase is -human, in which case the use of the indefinite article is either obligatory (e.g. es un problema 'it is a problem') or optional (e.g. es (una) cuestin de dinero 'it is a question of money'). (My interpretation of Lyons 1999, 104-5, who partly refers to Butt & C. Benjamin.) The present deduction considers case (b). For case (a) see deduction 213; for case (c) see deduction 215. The two syntactic variants: the type conservadores, and the type unos conservadores arrepentidos, both types used as predicative noun phrases. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (+modified, -modified) / predicative noun phrase, in Spanish I.e. with respect to coding properties, a modified predicative noun phrase is more natural than a non-modified predicative noun phrase, in Spanish. 1.1.2. >sym (+indefinite article, -indefinite article) / predicative noun phrase in Spanish I.e. with respect to coding properties, the presence of the indefinite article in the predicative noun phrase is more natural than the absence of the indefinite article from the predicative noun phrase, in Spanish. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the modified and the non-modified predicative noun phrases, such that one type of predicative noun phrase is used with the indefinite article, and the other type not, it is the modified type that tends to be used with the indefinite article. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the modified and the non-modified predicative noun phrases, such that one type of predicative noun phrase is used with the indefinite article, and the other type not, it is the non-modified type that tends not to be used with the

172

EXAMPLES

indefinite article. Q.E.D. 215. Spanish. (a) The indefinite article tends not to be used in predicative noun phrases, e.g. Mara se hizo dentista 'Mara became a dentist.' In two cases the indefinite article tends to be used even in predicative noun phrases: (b) if the head noun of the noun phrase is modified, e.g. son unos conservadores arrepentidos 'they are repentant conservatives;' (c) if the head noun of the noun phrase is -human, in which case the use of the indefinite article is either obligatory (e.g. es un problema 'it is a problem') or optional (e.g. es (una) cuestin de dinero 'it is a question of money'). (My interpretation of Lyons 1999, 104-5, who partly refers to Butt & C. Benjamin.) The present deduction considers case (c). For case (a) see deduction 213; for case (b) see deduction 214. The two syntactic variants: the type (se hizo) dentista , and the type (es) un problema , in predicative noun phrases. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type un problema, and the type dentista) / predicative noun phrase of Spanish I.e. with respect to coding properties, the type un problema is more natural than the type dentista, in predicative noun phrases of Spanish. 1.2. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type dentista (= +human) and the type problema (= -human), within predicative noun phrases, such that one type is used with the indefinite article, and the other type not, it is the type problema that tends to be used with the indefinite article. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type dentista (= +human) and the type problema (= -human), within predicative noun phrases, such that one type is used with the indefinite article, and the other type not, it is the type dentista that tends not to be used with the indefinite article. Q.E.D. 4. Notes. 4.1. I have no explanation for the contrast between es un problema 'it is a problem,' where the indefinite article is normally present, and es (una) cuestin de dinero 'it is a question of money,' where the indefinite article is optionally omitted. 4.2. For no es problema 'it is not a problem,' with the indefinite article normally omitted in the negative context, although the article is usually present in the affirmative es un problema 'it is a problem,' see item 4 of deduction 46.

173

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

216. Spanish. The expected indefinite article is omitted in the combination verb + its object and in the combination preposition + its object if the expression is near-formulaic, e.g. tener coche 'have a car,' pedir hora 'ask for an appointment,' casa con jardin 'house with a garden.' (If the head noun is modified, the expression of course forfeits its nearformulaic character, and the omission of the indefinite article does not obtain.) (My interpretation of Lyons 1999, 105, who partly refers to Butt & C. Benjamin.) The two syntactic variants: near-formulaic verb/preposition + its object, and nonformulaic verb/preposition + its object. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (-formula, +formula) I.e. with respect to coding properties, a non-formula is more natural than a formula. 1.1.2. >sym (indefinite article + head noun, bare head noun) / in Spanish I.e. with respect to coding properties, the indefinite article + head noun is more natural than the bare head noun, in Spanish. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between near-formulaic verb/preposition + its object and non-formulaic verb/preposition + its object, such that one expression is used with the indefinite article, and the other not, it is the non-formulaic expression that tends to use the indefinite article. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between near-formulaic verb/preposition + its object and non-formulaic verb/preposition + its object, such that one expression is used with the indefinite article, and the other not, it is the near-formulaic expression that tends not to use the indefinite article. Q.E.D. 217. Spanish. 'This country' is este pas and el pas este . (Lyons 1999, 120.) The two syntactic variants: the type este pas, and the type el pas este . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type el pas este, the type este pas) / in Spanish I.e. with respect to coding properties, the type el pas este is more natural than the type este pas, in Spanish.

174

EXAMPLES

1.2. >sym (+postposed, -postposed) / modifier in Romance languages I.e. with respect to coding properties, a postposed modifier is more natural than a preposed modifier, in Romance languages. (Mayerthaler et al. 1998, 21.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the (less transparent) type este pas and the (more transparent) type el pas este , such that in one of them the demonstrative precedes the head noun, and in the other the demonstrative follows the head noun, it is in the more transparent type el pas este that the demonstrative tends to follow the head noun. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the (less transparent) type este pas and the (more transparent) type el pas este , such that in one of them the demonstrative precedes the head noun, and in the other the demonstrative follows the head noun, it is in the less transparent type este pas that the demonstrative tends to precede the head noun. Q.E.D. 218. Spanish. In the construction modal or auxiliary verb + main verb, any object clitic can lean on either constituent of the construction, e.g. est haciendo-lo and lo-est haciendo 'he is doing it.' Only if the auxiliary is a form of haber 'have,' any object clitic MUST lean on it, e.g. lo-ha hecho 'he has done it.' (Givn 1995, 196-7.) The two syntactic variants: the type est haciendo-lo (with the clitic optionally attached to the main verb), and the type lo-ha hecho (with the clitic obligatorily attached to the auxiliary verb). 1. The assumptions of Naturalness Theory: 1.1. >sem (auxiliary haber, other auxiliaries as well as modal verbs) / in Spanish I.e. with respect to semantic complexity, the auxiliary haber is more natural than other auxiliaries or any modal verbs, in Spanish.In Spanish, the auxiliary haber is semantically the emptiest verb. 1.2. >sym (-detachment, +detachment) / of object clitic from main verb, in Spanish I.e. with respect to coding properties, non-detachment of the object clitic from the main verb is more natural than detachment of the object clitic from the main verb.The object clitic pertains to the main verb, therefore Behaghel's Law applies (Behaghel 1932, 4). A special case of 1.2: 1.2.1. >sym (optional detachment, obligatory detachment) / of object clitic from main verb, in Spanish I.e. with respect to coding properties, optional detachment of the object clitic from the main verb is more natural than obligatory detachment of the object clitic from the main verb. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the type est haciendo-lo and the type lo-ha hecho, such that the clitic is attached to the main verb or to the auxiliary in the one type,

175

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

and obligatorily separated from the main verb in the other type, it is in the type lo-ha hecho that the clitic tends to be obligatorily separated from the main verb. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the type est haciendo-lo and the type lo-ha hecho, such that the clitic is attached to the main verb or to the auxiliary in the one type, and obligatorily separated from the main verb in the other type, it is in the type est haciendo-lo that the clitic tends to be attached to the main verb or to the auxiliary. Q.E.D. 219. Spanish. In the reflexive passive, any patient expressed with the prepositional accusative is human, e.g. se mataba a los cristianos 'the Christians were being killed.' Otherwise the patient is in the nominative or in the non-prepositional accusative, e.g. en el jardin se ve(n) rosas 'in the garden roses are seen.' (My attention was drawn to this matter by Milena Milojevi-Sheppard viva voce, 1999.) The two syntactic variants are: the reflexive passive containing the patient in the accusative, and the reflexive passive containing the patient in the nominative or in the accusative. 1. The assumptions of Naturalness Theory: 1.1. >sem (-human, +human) / patient I.e. with respect to semantic complexity, -human patient is more natural than +human patient. (In the spirit of Mayerthaler 1981, 14.) 1.2. >sem (nominative & accusative, accusative only) / the case of the patient in Spanish I.e. with respect to semantic complexity, the patient in the nominative or accusative is more natural than the patient in the accusative only, in Spanish.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. Consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the expression of the patient ( a + accusative v. bare nominative or bare accusative) in the reflexive passive, such that one expression is used with +human patients, and the other with -human patients, it is the -human patient that tends to take the bare nominative or the bare accusative. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference in the expression of the patient ( a + accusative v. bare nominative or bare accusative) in the reflexive passive, such that one expression is used with +human patients, and the other with -human patients, it is the +human patient that tends to take a + accusative. Q.E.D. 220. Spanish. Certain verbs can express the inchoative aktionsart by adding the reflexive pronoun se: la paja ardi 'the straw burned,' as against la paja se ardi 'the straw caught fire.' (Hopper & S. Thompson 1980, 266, where it is maintained that the latter example-sentence expresses punctuality.) The two syntactic variants: the type la paja ardi, and the type la paja se ardi . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency

176

EXAMPLES

is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (la paja se ardi, la paja ardi ) / in Spanish I.e. with respect to coding properties, the type la paja se ardi is more natural than the type la paja ardi. 1.1.2. >sym (+aktionsart, -aktionsart) I.e. with respect to coding properties, the expression of an aktionsart is more natural than the non-expression of that aktionsart. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the type la paja ardi and the type la paja se ardi, such that one type expresses some aktionsart, and the other type does not express it, it is the type la paja se ardi that tends to express some aktionsart. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the type la paja ardi and the type la paja se ardi, such that one type expresses some aktionsart, and the other type does not express it, it is the type la paja ardi that tends not to express any aktionsart. Q.E.D. 221. Spanish. The +specific human direct object is introduced by the preposition a, the remaining direct objects are not introduced by any preposition. E.g. busco a mi madre 'I am looking for my mother,' as against busco cocinero 'I am looking for a (non-specific) cook.' (Krger 1924, 15.) The syntactic variants are: prepositionless direct object, and preposition + direct object. 1. The assumptions of Naturalness Theory: 1.1. >sem (-human, +human) / direct object I.e. with respect to semantic complexity, -human is more natural than +human, as direct object. (Mayerthaler 1981, 14.) A special case of 1.1: 1.1.1. >sem (-human, +specific +human) / direct object I.e. with respect to semantic complexity, -human is more natural than +specific +human, as direct object. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (preposition + case, prepositionless case) I.e. with respect to coding properties, preposition + case is more natural than a prepositionless case. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym

177

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between +specific +human direct object and other objects, such that one object is introduced by a preposition, and the other not, it "other" object that tends to lack the preposition. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between +specific +human direct object and other objects, such that one object is introduced by a preposition, and the other not, it +specific +human object that tends to be introduced by a preposition. Q.E.D.

direct is the direct is the

222. Swedish, Old. In the conjugation of the present indicative of most verbs, the grammatical persons were distinguished in the plural only (Ferguson 1996, 243). E.g. sg. lsk-ar, pl. lsk-um, lsk-in, lsk-a 'love' (Wessn 1956, 117). My attention was drawn to this matter by John C. Smith viva voce, 2000. The two syntactic variants: the singular and the plural of the present indicative. 1. The assumptions of Naturalness Theory: 1.1. >sem (verbal persons conflated, verbal persons separated) / per number in the present indicative of most Old Swedish verbs I.e. with respect to semantic complexity, verbal persons conflated is more natural than verbal persons separated, in the present indicative of most Old Swedish verbs.A form with several meanings has a higher sem-value than a corresponding form with fewer meanings. 1.2. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the singular and the plural of the present indicative, such that one number has the verbal persons separated, and the other number not, it is the singular that tends not to have the verbal persons separated. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the singular and the plural of the present indicative, such that one number has the verbal persons separated, and the other number not, it is the plural that tends to have the verbal persons separated. Q.E.D. 4. Note. Cf. deduction 29, in which a similar situation is treated. 223. Swedish. Adjectives show two declensions, 'strong' = -definite, and 'weak' = +definite. The weak declension displays only one form, e.g. klok-a, the strong declension has several forms: klok, klok-t, klok-a (of KLOK 'wise, shrewd'). (Lyons 1999, 85.) The two morphological variants: the strong and the weak adjectival declensions. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 178

EXAMPLES

1.1.1. >sym (strong, weak) / adjectival declension in Swedish I.e. with respect to coding properties, the strong adjectival declension is more natural than the weak one, in Swedish. 1.2. >sem (+presupposed, -presupposed) I.e. with respect to semantic complexity, +presupposed is more natural than -presupposed. (Mayerthaler 1981, 14.) A special case of 1.2: 1.2.1. >sem (+definite, -definite) I.e. with respect to semantic complexity, +definite is more natural than -definite. Everything definite is presupposed. 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sym tends to associate with >sem 2.2. <sym tends to associate with <sem 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the strong and weak adjectival declensions, such that one declension expresses +definite, and the other declension expresses -definite, it is the strong declension that tends to express -definiteness. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the strong and weak adjectival declensions, such that one declension expresses +definite, and the other declension expresses -definite, it is the weak declension that tends to express +definiteness. Q.E.D. 224. Swedish. There are two definite articles, a suffix and a free form. The two cooccur in the presence of a pre-nominal modifier, e.g. den lnga resan 'the long journey.' However, the free-form article is omitted if the noun phrase is a common one, a fixed expression (e.g. katolska kyrkan 'the Catholic Church'), or contains one of the very frequent adjectives, such as 'old,' 'young,' 'big,' 'little,' e.g. stora flickan 'the big girl.' (Lyons 1999, 78 + fn.) The two syntactic variants: the type den lnga resan, and the type stora flickan . 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (the type den lnga resan, the type stora flickan ) / in Swedish I.e. with respect to coding properties, the type den lnga resan is more natural than the type stora flickan , in Swedish. 1.1.2. >sym (-formula, +formula) I.e. with respect to coding properties, a non-formula is more natural than a formula. 1.3. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of G. Fenk-Oczlon 1991.) A special case of 1.3: 1.3.1. >sem ('old,' 'young,' 'big,' 'little;' most other adjectives) / in Swedish

179

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

I.e. with respect to semantic complexity, the adjectives 'old,' 'young,' 'big,' 'little' are more natural than most other adjectives, in Swedish. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with another >sym 2.2. <sym tends to associate with another <sym 2.3. >sym tends to associate with <sem 2.4. <sym tends to associate with >sem 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the type den lnga resan and the type stora flickan , such that one is used in formulaic expressions and the other not, it is the type den lnga resan that tends not to be used in formulaic expressions. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the type den lnga resan and the type stora flickan , such that one is used in formulaic expressions and the other not, it is the type stora flickan that tends to be used in formulaic expressions (such as katolska kyrkan ). Q.E.D. From 1.1.1, 1.3.1 and 2.3 it can be deduced: 3.3. If there is any difference between the type den lnga resan and the type stora flickan , such that one is used with very frequent adjectives, and the other not, it is the type den lnga resan that tends not to be used with very frequent adjectives. Q.E.D. From 1.1.1, 1.3.1 and 2.4 it can be deduced: 3.4. If there is any difference between the type den lnga resan and the type stora flickan , such that one is used with very frequent adjectives, and the other not, it is the type stora flickan that tends to be used with very frequent adjectives. Q.E.D. 225. Swedish. Whenever the indirect object can be paraphrased with a prepositional phrase containing the prepositions till/t/fr, the paraphrase is preferred in the case that the indirect object is less prominent (e.g. is indefinite) than the direct object, e.g. Arne gav boken till en granne 'Arne gave the book to a neighbour' is better than Arne gav en granne boken 'Arne gave a neighbour the book.' If the indirect object is more prominent (e.g. is definite) than the direct object, both constructions are equally good: Arne gav sin granne en bok 'Arne gave his neighbour a book' and Arne gav en bok till sin granne 'Arne gave a book to his neighbour.' (Teleman et al. 1999, vol. 3, 299.) The syntactic variants: the type Arne gav boken till en granne , the type Arne gav sin granne boken , and the type Arne gav boken till sin granne . 1. The assumptions of Naturalness Theory: 1.1. >sem (+definite, -definite) / indirect object I.e. with respect to semantic complexity, a definite indirect object is more natural than an indefinite direct object.It is reported from many languages that the indirect object tends to be definite (as well as animate, or even human). A special case of 1.1: 1.1.1. >sem (indirect object more prominent than direct object, indirect object less prominent than direct object) I.e. with respect to semantic complexity, an indirect object that is more prominent than a direct object is more natural than an indirect object that is less prominent than a direct object.Mostly, the indirect object's prominence depends on its definiteness. If the indirect object is not +definite, it can usually not be more prominent than the direct object. 1.2. >sem (the construction verb + direct object + prepositional indirect object & the 180

EXAMPLES

construction verb + indirect object + direct object, the construction verb + direct object + prepositional indirect object) / in Swedish I.e. with respect to semantic complexity, contracting either mutual order of the direct and indirect object is more natural than contracting just the order direct object - indirect object, in Swedish.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between contracting the construction verb + indirect object + direct object & the construction verb + direct object + prepositional indirect object and contracting only the construction verb + direct object + prepositional indirect object, such that the prominence of the direct object prevails in one case, and the prominence of the indirect object prevails in the other case, it is both constructions that tend to be used if the indirect object is more prominent than the direct object. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between contracting the construction verb + indirect object + direct object & the construction verb + direct object + prepositional indirect object and contracting only the construction verb + direct object + prepositional indirect object, such that the prominence of the direct object prevails in one case, and the prominence of the indirect object prevails in the other case, it is the construction verb + direct object + prepositional indirect object that tends to be used if the indirect object is less prominent than the direct object. Q.E.D. 226. Tadzhik (an Iranian language). The definite (or referentially prominent) object is marked with the postposition - ro, e.g. xona-i surx-ro 'the red house.' (Lyons 1999, 86, referring to Comrie.) The two syntactic variants: the definite and the indefinite direct objects. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (noun phrase + -ro, bare noun phrase) / the direct object of Tadzhik I.e. with respect to coding properties, a noun phrase marked with the postposition - ro is more natural than a bare noun phrase, in the direct object of Tadzhik. 1.2. >sem (-definite, +definite) / direct object I.e. with respect to semantic complexity, a -definite direct object is more natural than a +definite direct object. (In the spirit of Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the definite and indefinite direct objects, such that one of them is marked with a postposition, and the other not, it is the definite direct object that tends to be marked with a postposition. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 181

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3.2. If there is any difference between the definite and indefinite direct objects, such that one of them is marked with a postposition, and the other not, it is the indefinite direct object that tends not to be marked with a postposition. Q.E.D. 4. Note. Similarly in Hebrew (Lyons 1999, 86). 227. Tagalog (an Austronesian language spoken mostly on Luzn). The definite article is ang, e.g. ang gur 'the teacher.' A special article is used with proper names, e.g. si Marya 'Marya.' (Lyons 1999, 123, referring to Krmsk.) The two syntactic variants: the type ang gur, and the type si Marya. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / morphological unit I.e. with respect to coding properties, a morphological unit of greater transparency is more natural than a corresponding morphological unit of lesser transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (ang, si) / definite article of Tagalog I.e. with respect to coding properties, the article ang is more natural than the article si, in Tagalog.Ang has more sound body than si. 1.2. >sem (proper noun, common noun) I.e. with respect to semantic complexity, a proper noun is more natural than a common noun.This follows from the animacy hierarchy (Lyons 1999, 214, with references). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the definite articles ang and si, such that one is used with common nouns and the other with proper nouns, it is ang that tends to be used with common nouns. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the definite articles ang and si, such that one is used with common nouns and the other with proper nouns, it is si that tends to be used with proper nouns. Q.E.D. 228. Tagalog (an Austronesian language spoken mostly on Luzn). In answer to yesno questions the words meaning 'yes' and 'no' are used. Exception: if the yes-no question is existential (e.g. 'Is there any food?'), the answer repeats the positive or negative existential marker of the question. (Schachter 1985, 32-3.) The two syntactic variants: answer to non-existential yes-no question, and answer to existential yes-no question. 1. The assumptions of Naturalness Theory: 1.1. >sem (incomplete repetition of question, other means) / used in answering a yesno question I.e. with respect to semantic complexity, an incomplete repetition of the material of the question is more natural than other means, in answers to yes-no questions.This is based on the circumstance that answering a yes-no question by repeating some of the material of the question incompletely is widespread as a means of answering a yes-no question. Even languages that use 'yes' and 'no' as answers employ the said repetition as an alternative or enlarging means of answering, e.g. Are you hungry? (Yes,) I am.

182

EXAMPLES

1.2. >sem (existential clause, most other kinds of clauses) I.e. with respect to semantic complexity, existential clauses are more natural than most other kinds of clauses.This is based on the circumstance that existential clauses are a universal kind of clauses, whose structure often contains simple and idiosyncratic language material. (For some examples see Schachter 1985, 57-8.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between existential yes-no questions and other yes-no questions, concerning the means used in answering such a question (partial repetition of the question v. other means), it is the existential question that tends to have the answer consisting of a partial repetition of the question. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between existential yes-no questions and other yes-no questions, concerning the means used in answering such a question (partial repetition of the question v. other means), it is the non-existential question that tends to have the answer consisting of "other means." Q.E.D. 229. Thai. Classifiers are used with nouns, but are obligatory only when they are accompanied by a low definite number, e.g. maa saam tua dog three classifier 'three dogs.' (Schachter 1985, 39, referring to an unpublished paper by K. Adams & N. Conklin.) The two syntactic variants: noun + cardinal numeral + obligatory classifier, and noun + cardinal numeral + optional classifier. 1. The assumptions of Naturalness Theory: 1.1. >sem (+/-classifier, +classifier) / in languages that have classifiers I.e. with respect to semantic complexity, using an optional classifier is more natural than using an obligatory classifier, in languages which have classifiers.The scale has the format >sem (A + B, A). 1.2. >sem (-low, +low) / cardinal numeral in languages showing e-perspective I.e. with respect to semantic complexity, non-low cardinal numerals are more natural than low cardinal numerals, in languages showing e-perspective = ensemble perspective in nouns.On i-perspective v. e-perspective see Mayerthaler et al. 1998, 138. Thai shows eperspective. The scale is my guess. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between nouns that are accompanied by an obligatory classifier and nouns that are accompanied by an optional classifier, such that one kind of nouns is accompanied by a low cardinal numeral, and the other kind of nouns is accompanied by a non-low cardinal numeral, it is the low cardinal numerals that tend to occur in the pattern noun + cardinal numeral + obligatory classifier. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between nouns that are accompanied by an obligatory classifier and nouns that are accompanied by an optional classifier, such that one kind of nouns is accompanied by a low cardinal numeral, and the other kind of nouns is accom-

183

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

panied by a non-low cardinal numeral, it is the non-low cardinal numerals that tend to occur in the pattern noun + cardinal numeral + optional classifier. Q.E.D. 4. Note. I cannot "generate" the placement of the cut-off point between low and nonlow cardinal numerals. 230. Tiwi (an Australian language). Pronouns and +human nouns have a number di stinction, the remaining common nouns do not. E.g. wuralaka 'young girl,' wawuralakawi 'young girls' v. waliwalini 'ant(s).' (Croft 1990, 111, referring to Osborne.) The two syntactic variants: pronoun & +human noun, and -human noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (+number distinction, -number distinction) / in languages which have a number distinction. I.e. with respect to semantic complexity, a realized number distinction is more natural than a lacking number distinction, in languages which have a number distinction. 1.2. >sem (personal pronoun & +human noun, -human noun) I.e. with respect to semantic complexity, a personal pronoun and a +human noun are more natural than a -human noun. (Croft 1990, 112-3 [with references] on the animacy hierarchy.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between personal pronouns and +human nouns on the one side, and -human nouns on the other, such that one kind expresses the grammatical number, and the other kind does not express it, it is the pronouns and the +human nouns that tend to express the number. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between personal pronouns and +human nouns on the one side, and -human nouns on the other, such that one kind expresses the grammatical number, and the other kind does not express it, it is the -human nouns that tend not to express the number. Q.E.D. 231. Turkish. The pronominal possessors are realized as affixes on the head noun = the possessum. The affixes are phonologically identical to personal inflections of the verb, representing the subject argument of the verb. E.g. ky-m 'my village.' (Lyons 1999, 1256.) Cf. grr-m 'I see.' The two syntactic variants: affix as subject argument of the verb, and affix as pronominal possessor on the noun. 1. The assumptions of Naturalness Theory: 1.1. >sem (verb, noun) I.e. with respect to semantic complexity, the verb is more natural than the noun.The morphology of the verb is mostly better developed than the morphology of the noun. Such a state of affairs is expected with the less marked (the verb) and the corresponding more marked (the noun) entities. 1.2. >sem (nominative, genitive) / affix as argument in nom.-acc. languages I.e. with respect to semantic complexity, the nominative affix is more natural than the genitive affix, as argument in nominative-accusative languages. (In the spirit of Mayerthaler et al. 1998, 167.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 184

EXAMPLES

2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes the nominative affix, and the other part of speech takes the genitive affix, it is the verb that tends to take the nominative affix. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the noun and the verb concerning their main arguments, such that one part of speech takes the nominative affix, and the other part of speech takes the genitive affix, it is the noun that tends to take the genitive affix. Q.E.D. 4. Note. Cf. deductions 7 and 164. 232. Turkish. A function of the plural is to indicate intensification: kap yklyordu 'there was a heavy knocking at the door' (lit. 'the door was being knocked down'), plural kaplar yklyordu 'there was a tremendous knocking at the door' (lit. 'the doors were being knocked down'). (Cruse 1999, 271.) The two syntactic variants: the singular type kap yklyordu 'there was a heavy knocking at the door,' and the plural type kaplar yklyordu 'there was a tremendous knocking at the door.' 1. The assumptions of Naturalness Theory: 1.1. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 1.2. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.) A special case of 1.2: 1.2.1. >sem (first degree, second degree) / of intensity I.e. with respect to semantic complexity, intensity of the first degree is more natural than intensity of the second degree. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the singular type kap yklyordu 'there was a heavy knocking at the door' and the plural type kaplar yklyordu 'there was a tremendous knocking at the door,' such that one type is expressing intensity of the first degree, and the other type is expressing intensity of the second degree, it is the singular type that tends to express intensity of the first degree. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the singular type kap yklyordu 'there was a heavy knocking at the door' and the plural type kaplar yklyordu 'there was a tremendous knocking at the door,' such that one type is expressing intensity of the first degree, and the other type is expressing intensity of the second degree, it is the plural type that tends to express intensity of the second degree. Q.E.D. 233. Turkish. (a) The plural marker is normally excluded after numerals and other expressions of numerosity: bir inek 'a cow,' inek-ler 'cows,' yedi inek 'seven cows.' (b) The 185

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

plural marker appears after a numeral only in references to well-known collectivities like 'The Seven Dwarfs' (Yedi Cceler ) and 'The Forty Thieves' ( Krk Hrszlar). (Cruse 1999, 269.)This deduction treats case (a). Case (b) is dealt with in deduction 234. The two syntactic variants: the type inek-ler 'cows,' and the type yedi inek 'seven cows.' 1. The assumptions of Naturalness Theory: 1.1. >sem (-redundant, +redundant) / cardinal numeral in expressions of exact quantity I.e. with respect to semantic complexity, non-redundant cardinal numeral is more natural than redundant cardinal numeral, in expressions of exact quantity.This is based on the circumstance that cardinal numerals are the prototypical expressions of exact quantity. A special case of 1.1: 1.1.1. >sem (present, absent) / cardinal numeral in expressions of exact quantity I.e. with respect to semantic complexity, a present cardinal numeral is more natural than a lacking cardinal numeral, in expressions of exact quantity. 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (the type inek-ler, the type inek ) / noun plural in Turkish I.e. with respect to coding properties, the plural type inek-ler is more natural than the plural type inek, in Turkish. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the plural noun preceded by a cardinal number and a plural noun not preceded by a cardinal number, such that in one construction the plural ending is present, and in the other construction the plural ending is lacking, it is in the plural noun preceded by the cardinal number that the plural ending tends to be lacking. Q.E.D. From 1.1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the plural noun preceded by a cardinal number and a plural noun not preceded by a cardinal number, such that in one construction the plural ending is present, and in the other construction the plural ending is lacking, it is in the plural noun not preceded by the cardinal number that the plural ending tends to be present. Q.E.D. 234. Turkish. (a) The plural marker is normally excluded after numerals and other expressions of numerosity: bir inek 'a cow,' inekler 'cows,' yedi inek 'seven cows.' (b) The plural marker appears after a numeral only in references to well-known collectivities like 'The Seven Dwarfs' (Yedi Cceler), and 'The Forty Thieves' (Krk Hrszlar). (Cruse 1999, 269.)This deduction treats case (b). Case (a) is dealt with in deduction 233. The two syntactic variants: the type yedi cce '(the) seven dwarfs,' and the type Yedi Cceler 'The Seven Dwarfs.' 1. The assumptions of Naturalness Theory: 186

EXAMPLES

1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the type Yedi Cceler , the type yedi cce ) / in Turkish I.e. with respect to coding properties, the type Yedi Cceler is more natural than the type yedi cce , in Turkish. 1.2. >sem (plurality, collectivity) / expressed by the plural I.e. with respect to semantic complexity, plurality is a more natural expression of the plural than collectivity.Cross-linguistically, the plural marker denotes first of all plurality. Cruse 1999, 267 mentions only few languages expressing the collective plural. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the type yedi cce and the type Yedi Cceler, such that one type expresses plurality, and the other type collectivity, it is the type Yedi Cceler that tends to express collectivity. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the type yedi cce and the type Yedi Cceler, such that one type expresses plurality, and the other type collectivity, it is the type yedi cce that tends to express plurality. Q.E.D. 235. Turkish. A function of the plural is to convey vagueness: burada 'here' (lit. 'in this place'), plural buralarda 'hereabouts' (lit. 'in these places'). (Cruse 1999, 271.) The two syntactic variants: the same expression in the singular and in the plural. 1. The assumptions of Naturalness Theory: 1.1. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 1.2. >sem (+precise, -precise) / meaning I.e. with respect to semantic complexity, a precise meaning (reading) is more natural than the corresponding imprecise (vague) meaning (reading).This is based on the circumstance that the prototypical speaker tries to speak as precisely as possible. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between an expression in the singular and in the plural, such that one expression has precise meaning, and the other expression has vague meaning, it is the expression in the singular that tends to have precise meaning. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between an expression in the singular and in the plural, such that one expression has precise meaning, and the other expression has vague

187

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

meaning, it is the expression in the plural that tends to have imprecise (vague) meaning. Q.E.D. 4. Note. An English plural of vagueness is seen in every student thinks they have passed the exam , where they stands for the singular he/she (example from Lyons 1999, 31). 236. Walbiri (spoken in the Northern Territory, Australia). Head-modifier case agreement is lacking when the head and the modifier are adjacent: the case suffix appears only once at the end of the noun phrase, e.g. tjantu wiri-nki tji yalkunu dog big bit me. When the head and modifier are separated, case agreement is obligatory, wiri-nki tji yalkunu tjantu-nku big bit me dog. (Givn 1995, 220, referring to Hale.) The two syntactic variants: the adjacent head and modifier, and the separated head and modifier. 1. The assumptions of Naturalness Theory: 1.1. >sym (+adjacency, -adjacency) / of units that belong together I.e. with respect to coding properties, adjacency of units that belong together is more natural than their non-adjacency.Behaghel's Law (Behaghel 1932, 4). 1.2. >sem (+agreement, -agreement) / head-modifier in languages that have headmodifier agreement I.e. with respect to semantic complexity, the use of head-modifier agreement is more natural than its lack, in languages that have head-modifier agreement. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between adjacent head & modifier and separated head & modifier, such that in one configuration the ending occurs only once, and in the other con figuration there is agreement between head & modifier, it is the adjacent head & modifier that tend to have the ending expressed only once. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between adjacent head & modifier and separated head & modifier, such that in one configuration the ending occurs only once, and in the other configuration there is agreement between head & modifier, it is the separated head & modifier that tend to show agreement between head and modifier. Q.E.D. 237. Welsh. Subject agreement is limited to pronominal arguments (overt and null), e.g. agoron (hwy) y drws opened-3PL (they) the door 'they opened the door,' as against agorodd y dynion y drws opened the men the door 'the men opened the door.' (Lyons 1999, 212-3.) The two syntactic variants: the type agoron (hwy) y drws, and the type agorodd y dynion y drws. Only the third persons are considered. 1. The assumptions of Naturalness Theory: 1.1. >sem (+subject agreement, -subject agreement) / in languages that have subject agreement I.e. with respect to semantic complexity, the use of subject agreement is more natural than the lack of subject agreement, in languages that have subject agreement. 1.2. >sem (pronominal argument, other argument) / subject

188

EXAMPLES

I.e. with respect to semantic complexity, a pronominal argument is more natural than any other argument, as subject.This in accordance with the animacy hierarchy. (Croft 1990, 112, with references.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between the subject which is a pronominal argument and the subject which is some other argument, such that in one type there is subject agreement, and in the other type there is no subject agreement, it is with the subject which is a pronominal argument that there tends to be subject agreement. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between the subject which is a pronominal argument and the subject which is some other argument, such that in one type there is subject agreement, and in the other type there is no subject agreement, it is with the subject which is some other argument that there tends to be no subject agreement. Q.E.D. 238. Wikchamni dialect of Yokuts (a Penutian language spoken in California). Singular, plural and dual are distinguished. While the singular and plural are in general use with nouns and pronouns, the dual is found only with personal and demonstrative pronouns. (Croft 1990, 100, referring to Gamble.) The two syntactic variants: pronouns distinguishing singular, plural and dual; pronouns and nouns distinguishing only singular and plural. 1. The assumptions of Naturalness Theory: 1.1. >sem (personal & demonstrative pronouns, other pronouns & all nouns) I.e. with respect to semantic complexity, personal and demonstrative pronouns are more natural than other pronouns and all nouns.Personal and demonstrative pronouns are the prototypical pronouns. Other pronouns + nouns do not form a "natural class." 1.2. >sem (singular, plural & dual; singular & plural) I.e. with respect to semantic complexity, having singular, plural and dual is more natural than having just singular and plural.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between nominal units which distinguish singular, plural and dual, and nominal units which distinguish only singular and plural, such that one kind of units comprises personal and demonstrative pronouns, and the other kind of units comprises the remaining pronouns and all nouns, it is the nominal units which distinguish singular, plural and dual that tend to comprise personal and demonstrative pronouns. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between nominal units which distinguish singular, plural and dual, and nominal units which distinguish only singular and plural, such that one kind of units comprises personal and demonstrative pronouns, and the other kind of units comprises the remaining pronouns and all nouns, it is the nominal units which distinguish

189

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

only singular and plural that tend to comprise the remaining pronouns and all nouns. Q.E.D. 239. Cross-linguistically, the comment has more sound body and/or more internal structure than the topic. The two syntactic variants: the topic, and the comment. 1. The assumptions of Naturalness Theory: 1.1. >sem (topic, comment) I.e. with respect to semantic complexity, the topic is more natural than the comment. (Mayerthaler 1981, 15.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerhaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit.The ability to expand involves the tendency to enhance transparency. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the topic and the comment, such that one of them is syntactically expandable, and the other is not syntactically expandable, it is the topic that tends to lack the ability to expand. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the topic and the comment, such that one of them is syntactically expandable, and the other is not syntactically expandable, it is the comment that tends to display the ability to expand. Q.E.D. 240. Equative clauses. Absence of copular 'be' in the third grammatical person only. (Mayerthaler 1981, 157-9, referring to Ferguson.) The two syntactic variants: clause containing copular 'be' in the non-third grammatical persons, and corresponding clause lacking copular 'be' in the third grammatical persons. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (clause containing copular 'be,' corresponding clause without copular 'be') I.e. with respect to coding properties, a clause containing the copular 'be' is more natural than a corresponding clause without copular 'be.' 1.2. >sem (+third, -third) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural

190

EXAMPLES

than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with >sem 2.2. >sym tends to associate with <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the use and non-use of the copular 'be,' such that the difference is between the third and non-third grammatical persons, it is the clause without 'be' that tends to be used in the third grammatical person. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference in the use and non-use of the copular 'be,' such that the difference is between the third and non-third grammatical persons, it is the clause con taining 'be' that tends to be used in the non-third grammatical persons. Q.E.D. 241. Equative clauses. The use of the copular 'be' outside the present tense, or under emphasis, or when negated. (Mayerthaler 1981, 157-9, referring to Ferguson.) The two syntactic variants: (a) clause without the copular 'be' in the present tense, and corresponding clause containing the copular 'be' outside the present tense; (b) clause containing the copular 'be' under emphasis, and corresponding clause without the copular 'be' under lack of emphasis; (c) clause containing the copular 'be' negated, and corresponding clause without copular 'be' and lacking negation. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (clause containing copular 'be,' corresponding clause without copular 'be') I.e. with respect to coding properties, a clause containing the copular 'be' is more natural than a corresponding clause without copular 'be.' 1.2. >sem (present tense, other tenses) I.e. with respect to semantic complexity, the present tense is more natural than other tenses. (Mayerthaler 1981, 14, 158.) 1.3. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.) 1.4. >sem (affirmation, negation) I.e. with respect to semantic complexity, affirmation is more natural than negation. (Mayerthaler 1981, 15, 159.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with >sem 2.2. >sym tends to associate with <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the (non-)use of the copular 'be' in clauses, such that in one clause the present tense is used, and in the other clause non-present tenses are used, it is the clause without the copular 'be' that tends to be used in the present tense. Q.E.D. 191

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1.1, 1.3 and 2.1 it can be deduced: 3.2. If there is any difference in the (non-)use of the copular 'be' in clauses, such that one clause expresses emphasis, and the other clause not, it is the clause without the copular 'be' that tends to be used non-emphatically. Q.E.D. From 1.1.1, 1.4 and 2.1 it can be deduced: 3.3. If there is any difference in the (non-)use of the copular 'be' in clauses, such that one clause expresses affirmation, and the other clause expresses negation, it is the clause without the copular 'be' that tends to be used affirmatively. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.4. If there is any difference in the (non-)use of the copular 'be' in clauses, such that in one clause the present tense is used, and in the other clause non-present tenses are used, it is the clause with the copular 'be' that tends to be used in the non-present tenses. Q.E.D. From 1.1.1, 1.3 and 2.2 it can be deduced: 3.5. If there is any difference in the (non-)use of the copular 'be' in clauses, such that one clause expresses emphasis, and the other clause not, it is the clause with the copular 'be' that tends to be used under emphasis. Q.E.D. From 1.1.1, 1.4 and 2.2 it can be deduced: 3.6. If there is any difference in the (non-)use of the copular 'be' in clauses, such that one clause expresses affirmation, and the other clause expresses negation, it is the clause with the copular 'be' that tends to be used when negated. Q.E.D. 242. Equative clauses. Copular 'be' is used in dependent clauses only. (Mayerthaler 1981, 157-9, referring to Ferguson.) The two syntactic variants: main clause lacking copular 'be,' and dependent clause containing copular 'be.' 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (clause containing copular 'be,' clause without copular 'be') I.e. with respect to coding properties, a clause containing the copular 'be' is more natural than a clause without copular 'be.' 1.2. >sem (main clause, dependent clause) I.e. with respect to semantic complexity, a main clause is more natural than a depen dent clause.Phylogenetically, main clauses are of earlier origin than dependent clauses. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with >sem 2.2. >sym tends to associate with <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the (non-)use of the copular 'be,' such that one configuration is a main clause, and the other configuration is a dependent clause, it is from the main clauses that the copular 'be' tends to be absent. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference in the (non-)use of the copular 'be,' such that one configuration is a main clause, and the other configuration is a dependent clause, it is in the dependent clauses that the copular 'be' tends to be present. Q.E.D. 243. Equative clauses. The absence of the copular 'be' in text types such as telegram, 192

EXAMPLES

proverb, newspaper headline only. (Mayerthaler 1981, 157-9, referring to Ferguson.) The two syntactic variants: clause containing the copular 'be' outside the text types just mentioned, and corresponding clause without the copular 'be' in the text types just mentioned. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (clause containing copular 'be,' corresponding clause without copular 'be') I.e. with respect to coding properties, a clause containing the copular 'be' is more natural than a clause without copular 'be.' 1.1.2. >sym (-[syntactically abbreviated], +[syntactically abbreviated]) / text I.e. with respect to coding properties, syntactically non-abbreviated texts are more natural than syntactically abbreviated texts. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with another <sym 2.2. >sym tends to associate with another >sym 3. The consequences: From 1.1.1, 1.1.2 and 2.1 it can be deduced: 3.1. If there is any difference in the (non-)use of the copular 'be,' such that in one configuration the text type is syntactically abbreviated, and in the other configuration the text type is not syntactically abbreviated, it is the clause without the copular 'be' that tends to be used in text types characterized by syntactic abbreviation. Q.E.D. From 1.1.1, 1.1.2 and 2.2 it can be deduced: 3.2. If there is any difference in the (non-)use of the copular 'be,' such that in one configuration the text type is syntactically abbreviated, and in the other configuration the text type is not syntactically abbreviated, it is the clause with the copular 'be' that tends to be used in text types lacking syntactic abbreviation. Q.E.D. 244. Cross-linguistically, inflection appears in nouns, adjectives and verbs, much less in other parts of speech. The two morphological variants: part of speech displaying inflection, and part of speech lacking inflection. 1. The assumptions of Naturalness Theory: 1.1. >sem (several forms, one form) / of a lexical item I.e. with respect to semantic complexity, a lexical item with several forms is more natural than a lexical item with only one form.The scale has the format >sem (A + B, A). 1.2. >sem (noun/adjective/verb, other parts of speech) I.e. with respect to semantic complexity, the nouns, adjectives and verbs are more natural than the other parts of speech. (In the spirit of the so-called cardinal word classes, Mayerthaler et al. 1998, 389.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem

193

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

3. The consequences: From 1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between lexical items able to inflect and lexical items not able to inflect, such that one kind of lexical items comprise nouns, adjectives and verbs, and the other kind of lexical items comprise "other" parts of speech, it is the lexical items able to inflect that tend to comprise nouns, adjectives and verbs. Q.E.D. From 1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between lexical items able to inflect and lexical items not able to inflect, such that one kind of lexical items comprise nouns, adjectives and verbs, and the other kind of lexical items comprise "other" parts of speech, it is the lexical items not able to inflect that tend to comprise "other" parts of speech. Q.E.D. 245. Cross-linguistically, pronouns display much less syntactic structure than nouns. The two syntactic variants: noun and pronoun. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.1: 1.1.1. >sym (noun, pronoun) I.e. with respect to coding properties, nouns are more natural than pronouns. 1.1.2. >sym (+expandable, -expandable) / syntactic unit I.e. with respect to coding properties, an expandable syntactic unit is more natural than a non-expandable syntactic unit.The ability to expand involves the tendency to enhance transparency. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. <sym tends to associate with another <sym 2.2. >sym tends to associate with another >sym 3. The consequences: From 1.1.1-2 and 2.1 it can be deduced: 3.1. If there is any difference between nouns and pronouns, such that one part of speech shows relatively much structure, and the other part of speech shows relatively little structure, it is the pronouns that tend to show relatively little structure. Q.E.D. From 1.1.1-2 and 2.2 it can be deduced: 3.2. If there is any difference between nouns and pronouns, such that one part of speech shows relatively much structure, and the other part of speech shows relatively little structure, it is the nouns that tend to show relatively much structure. Q.E.D. 4. Note. A similar case could be made for function words in general as opposed to lexical words. 246. In some languages, personal pronouns can be expressed by affixes on the verb, particularly subject and object pronouns, e.g. in Quechua (Schachter 1985, 26), see deduction 115 for examples. The two syntactic variants: object and subject personal pronouns expressed by affixes on the verb, and object and subject personal pronouns expressed as words. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit

194

EXAMPLES

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (word, morpheme) I.e. with respect to coding properties, a word is more natural than a morpheme. 1.2. >sem (subject/object, other clause elements) / in nom.-acc. languages I.e. with respect to semantic complexity, the subject and the object are more natural than other clause elements, in nominative-accusative languages.This follows from the basic properties of nominative-accusative languages. (Mayerthaler 1981, 14.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between a personal pronoun expressed as a word and a personal pronoun expressed by an affix, such that one kind of pronoun is subject or object, and the other kind of pronoun is not subject or object, it is the personal pronoun expressed as a word that tends not to be a subject or object pronoun. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between a personal pronoun expressed as a word and a personal pronoun expressed by an affix, such that one kind of pronoun is subject or object, and the other kind of pronoun is not subject or object, it is the personal pronoun expressed by an affix that tends to be a subject or object pronoun. Q.E.D. 247. In most of the few languages showing trial or paucal forms, those forms are generally found only with pronouns, while nouns have only singular, plural and dual forms. (Croft 1990, 100.) The two syntactic variants: nouns and pronouns. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (noun, pronoun) I.e. with respect to coding properties, nouns are more natural than pronouns. 1.2. >sem (singular, plural, dual and trial/paucal; singular, plural, dual) / nouns and pronouns I.e. with respect to coding properties, nouns and pronouns having singular, plural, dual and trial/paucal are more natural than nouns and pronouns having singular, plural and dual only.The scale has the format >sem (A + B, A). 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences:

195

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between part-of-speech units showing singular, plural, dual and trial/paucal, and part-of-speech units showing only singular, plural and dual, such that one class comprises only pronouns, and the other class comprises only nouns, it is the part-of-speech units showing singular, plural, dual and trial/paucal that tend to comprise pronouns. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between part-of-speech units showing singular, plural, dual and trial/paucal, and part-of-speech units showing only singular, plural and dual, such that one class comprises only pronouns, and the other class comprises only nouns, it is the part-of-speech units showing only singular, plural and dual that tend to comprise nouns. Q.E.D. 248. The imperative, notably its second person, is often used without an overt subject even in non-pro-drop languages, and is relatively short. The two syntactic variants: the second person v. the other persons of the imperative. 1. The assumptions of Naturalness Theory: 1.1. >sem (second person, other persons) / imperative I.e. with respect to semantic complexity, the second person is more natural than the other persons, in the imperative.Orders are prototypical imperatives, and they occur in the second person. (In the spirit of Mayerthaler et al. 1998, 140.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) Two special cases of 1.2: 1.2.1. >sym (long, short) / word form I.e. with respect to coding properties, a long word form is more natural than a short word form.Length enhances transparency. 1.2.2. >sym (+overt, -overt) / subject I.e. with respect to coding properties, an overt subject is more natural than a covert subject. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1-2 and 2.1 it can be deduced: 3.1. If there is any difference, within the imperative, between the second person and the other persons, concerning the length of their forms and the overtness of their subjects, it is the second-person imperative that tends to have short forms and covert subjects. Q.E.D. From 1.1, 1.2.1-2 and 2.2 it can be deduced: 3.2. If there is any difference, within the imperative, between the second person and the other persons, concerning the length of their forms and the overtness of their subjects, it is the other persons that tend to have long forms and overt subjects. Q.E.D. 249. There are languages in which transitive clauses containing two full noun-phrase

196

EXAMPLES

participants are nonexistent (e.g. Puget Salish, spoken in the state of Washington) or un common (e.g. Zapotec, spoken in Mexico) since the subject and/or object must be, or tend to be, expressed solely by pronominal affixes on the verb or by clitics elsewhere in the clause. (A. Siewierska 1999, 413.) The two syntactic variants: subject/object, and other NP-relations. 1. The assumptions of Naturalness Theory: 1.1. >sem (subject, object; other NP-relations) I.e. with respect to semantic complexity, the subject and object are more natural than other NP-relations. (Mayerthaler 1981, 14.) 1.2. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.2: 1.2.1. >sym (noun, pronominal affix & pronominal clitic) I.e. with respect to coding properties, nouns are more natural than pronominal affixes and pronominal clitics. 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1, 1.2.1 and 2.1 it can be deduced: 3.1. If there is any difference between the expression of the subject/object and the expression of other NP-relations in a transitive clause, so that the choice is limited to nouns and pronominal clitics & pronominal affixes, it is the subject/object that tends to be expressed with the pronominal affix or pronominal clitic. Q.E.D. From 1.1, 1.2.1 and 2.2 it can be deduced: 3.2. If there is any difference between the expression of the subject/object and the expression of other NP-relations in a transitive clause, so that the choice is restricted to nouns and pronominal clitics & pronominal affixes, it is the other NP-relations that tend to be expressed with the noun. Q.E.D. 250. The UNEMPHATIC vocative as a separate morphological form different from the nominative (henceforth: unemphatic vocative) is more likely to occur in the singular than in the plural; in the masculine gender than in the feminine gender; in persons than in non-persons; in nouns than in adjectives.On the difference between emphatic and nonemphatic vocatives see Mayerthaler 1981, 31-2. Every EMPHATIC vocative is in some way different from the corresponding non-emphatic nominative. The two morphological variants: unemphatic vocative in the singular, and unemphatic vocative in the plural; unemphatic vocative in the masculine gender, and unemphatic vocative in the non-masculine gender; unemphatic vocative +human, and unemphatic vocative -human; unemphatic vocative in the noun, and unemphatic vocative in the adjective. 1. The assumptions of Naturalness Theory: 1.1. >sem (-emphasis, +emphasis) I.e. with respect to semantic complexity, non-emphasis is more natural than emphasis. (Mayerthaler 1981, 15, 159.)

197

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

A special case of 1.1: 1.1.1. >sem (-emphatic vocative, +emphatic vocative) I.e. with respect to semantic complexity, the unemphatic vocative is more natural than the emphatic vocative. 1.2. >sem (+singular, -singular) I.e. with respect to semantic complexity, the singular is more natural than the nonsingular. (Greenberg 1966, 31-7.) 1.3. >sem (+masculine, -masculine) I.e. with respect to semantic complexity, the masculine gender is more natural than the non-masculine gender. (Greenberg 1966, 39.) 1.4. >sem (+human, -human) I.e. with respect to semantic complexity, +human is more natural than -human. (Mayerthaler 1981, 14.) 1.5. >sem (noun, adjective) I.e. with respect to semantic complexity, a noun is more natural than an adjective. The adjective is not universal. (Mayerthaler et al. 1998, 24.) 2. The assumptions of the Slovenian Theory, concerning any two morphological variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1.1, 1.2 and 2.1-2 it can be deduced: 3.1. If there is a morphological difference between the unemphatic vocative in the singular and the unemphatic vocative in the non-singular, the unemphatic vocative tends to have a special form in the singular, rather than in the non-singular. Q.E.D. From 1.1.1, 1.3 and 2.1-2 it can be deduced: 3.2. If there is a morphological difference between the unemphatic vocative in the masculine gender and the unemphatic vocative in the non-masculine gender, the unemphatic vocative tends to have a special form in the masculine gender, rather than in the non-masculine gender. Q.E.D. From 1.1.1, 1.4 and 2.1-2 it can be deduced: 3.3. If there is a morphological difference between the +human unemphatic vocative and -human unemphatic vocative, the +human unemphatic vocative tends to have a special form, rather than the -human unemphatic vocative. Q.E.D. From 1.1.1, 1.5 and 2.1-2 it can be deduced: 3.4. If there is a morphological difference between the unemphatic vocative noun and the unemphatic vocative adjective, the unemphatic vocative noun tends to have a special form, rather than the unemphatic vocative adjective. Q.E.D. 251. Cross-linguistically, specificity and non-specificity can be overtly distinguished in indefinites only. (Lyons 1999, 178.) The two syntactic variants: definites and indefinites in the languages in which indefinite specific is coded differently from indefinite non-specific. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see

198

EXAMPLES

Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (distinction overt, distinction covert) / between specificity and nonspecificity I.e. with respect to coding properties, the distinction between specificity and nonspecificity is more natural than its absence. 1.2. >sem (definites, indefinites) I.e. with respect to semantic complexity, definites are more natural than indefinites. (Mayerthaler 1981, 14 on presupposed.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with <sym 2.2. <sem tends to associate with >sym 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between definites and indefinites, such that one kind sets apart specificity from non-specificity, and the other kind not, it is the definites that tend not to make the distinction. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between definites and indefinites, such that one kind sets apart specificity from non-specificity, and the other kind not, it is the indefinites that tend to make the distinction. Q.E.D. 252. Cross-linguistically, the anaphora for the personal pronoun of the first person singular + any other NP is the personal pronoun of the first person non-singular. The anaphora for the personal pronoun of the second person singular + any other NP (except the personal pronoun of the first person singular) is the personal pronoun of the second person non-singular. (Mayerthaler et al. 1998, 271; "=p3" should be "=p2.") The two syntactic variants: the anaphora for the personal pronoun of the first person singular + any other NP, and the anaphora for the personal pronoun of the second person singular + any other NP (except the personal pronoun of the first person singular). 1. The assumptions of Naturalness Theory: 1.1. >sem (first, second) / person non-singular of the personal pronoun I.e. with respect to semantic complexity, the first person is more natural than the second person, within the non-singular of the personal pronoun. (According to the Jakobson-Greenberg linguistic tradition.) 1.2. >sem (personal pronoun of the first person singular + other NP, personal pronoun of the second person singular + other NP [except the personal pronoun of the first person singular]) I.e. with respect to semantic complexity, the personal pronoun of the first person singular + other NP is more natural than the personal pronoun of the second person singular + other NP (except the personal pronoun of the first person singular). (According to the Jakobson-Greenberg linguistic tradition.) 2. The assumptions of the Slovenian Theory, concerning any two syntactic variants: 2.1. >sem tends to associate with another >sem 2.2. <sem tends to associate with another <sem 3. The consequences: From 1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the anaphora for the personal pronoun of the

199

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

first person singular + any other NP and the anaphora for the personal pronoun of the second person singular + any other NP (except the personal pronoun of the first person singular), such that one anaphora is the personal pronoun of the first person non-singular, and the other anaphora is the personal pronoun of the second person non-singular, it is the anaphora for the personal pronoun of the first person singular + any other NP that tends to be the personal pronoun of the first person non-singular. Q.E.D. From 1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the anaphora for the personal pronoun of the first person singular + any other NP and the anaphora for the personal pronoun of the second person singular + any other NP (except the personal pronoun of the first person singular), such that one anaphora is the personal pronoun of the first person non-singular, and the other anaphora is the personal pronoun of the second person non-singular, it is the anaphora for the personal pronoun of the second person singular + any other NP (except the personal pronoun of the first person singular) that tends to be the personal pronoun of the second person non-singular. Q.E.D. 4. Note. The present deduction is almost vacuous. It is included to show that the matter considered does not contradict the Jakobson-Greenberg scale >sem (third, first, second) / grammatical person.

200

The preterite and present perfect tenses in German


(An abbreviated and revised version of Orenik 2000) This brief chapter deals with the two chief German verbal tenses expressing past time: the preterite and the present perfect. As is well known, there is an asymmetrical competition going on between them: the present perfect is gaining ground, the preterite is retreating. The process is especially advanced in spoken "Southern" German, including the standard language as used in the area. My data are taken from Hauser-Suida & HoppeBeugel 1972 and from Latzel 1977. (The conditions in the dialects will not be considered.) I propose to treat some aspects of the competition between the present perfect and the preterite in detail. An investigation of the dialogues in a number of German novels has shown that the present perfect is used more in the non-third grammatical persons, and the preterite more in the third persons. See Table 1. preterite first p. sg. 44,81% first p. pl. 40,98% second p. sg. 24,42% second p. pl. 21,21% Sie 'you (hon.)'30,00% third p. sg. 55,74% third p. pl. 54,50% perfect 55,19% 59,02% 75,58% 78,79% 70,00% 42,26% 45,50% N 790 122 303 33 230 1715 189

TABLE 1. The preterite and the present perfect in the dialogues of a number of Ger man novels. The figures stem from Hauser-Suida & Hoppe-Beugel 1972, but are quoted here after the more comfortable source Latzel 1977, 93. My comment: It is indeed the case that the use of the present perfect prevails over the use of the preterite in the non-third grammatical persons. While this circumstance is not accounted for by the two German monographs, the data are given the following synchronic explanation within the Slovenian Theory: 253. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit 202

THE PRETERITE AND PRESENT PERFECT TENSES IN GERMAN

I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (present perfect tense, preterite tense) / in German I.e. with respect to coding properties, the present perfect is more natural than the preterite, in German. 1.2. >sem (third grammatical person, non-third grammatical person) I.e. with respect to semantic complexity, the third grammatical person is more natural than the non-third grammatical person. (According to the linguistic tradition beginning with Jakobson 1932.) 1.3. >sem (singular, plural) I.e. with respect to semantic complexity, the singular is more natural than the plural. (Mayerthaler 1981, 15.) 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the preterite and the present perfect, such that one tense prevails in the third persons, and the other tense prevails in the non-third persons, it is the present perfect that tends to prevail in the non-third grammatical persons. Q.E.D. From 1.1.1, 1.3 and 2.1 it can be deduced: 3.2. If there is any difference between the preterite and the present perfect, such that one tense prevails in the singular, and the other tense prevails in the plural, it is the present perfect that tends to prevail in the plural. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.3. If there is any difference between the preterite and the present perfect, such that one tense prevails in the third persons, and the other tense prevails in the non-third persons, it is the preterite that tends to prevail in the third grammatical persons. Q.E.D. From 1.1.1, 1.3 and 2.2 it can be deduced: 3.2. If there is any difference between the preterite and the present perfect, such that one tense prevails in the singular, and the other tense prevails in the plural, it is the preterite that tends to prevail in the singular. Q.E.D. The same deduction accounts for the conspicuously high percentage of the present perfect tense in the second persons, if the naturalness scale in 253.1.2 is replaced by the more specific one in 253.1.2a: 253. 1.2a. >sem (third, first, second) / grammatical person I.e. with respect to semantic complexity, the third grammatical person is more natural than the first grammatical person; the first grammatical person is more natural than the second grammatical person. (According to the linguistic tradition beginning with Jakobson 1932; cf. especially Greenberg 1966, 44-5.) Given deduction 253, the numerical data of both German monographs are accounted for in synchronic terms. To be sure, the statistics do not allow us to observe the gram 203

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

matical persons separated from the grammatical number; yet the cumulative evidence of Table 1 speaks convincingly in favour of the predictions made by the Slovenian Theory. As already pointed out, the prevalence of the present perfect over the preterite is particularly conspicuous in the second grammatical person (of both singular and plural), especially if the data on the polite Sie 'you (one or several addressees)' are thrown in, a reasonable step in spite of the circumstance that Sie invariably combines with the THIRD person plural of the verb. Hauser-Suida & Hoppe-Beugel 1972, 119 ff. suggest an expla nation of the predominant present perfect in the second persons only. (They leave aside the first persons as well as Sie.) They suggest that the reason for the avoidance of the preterite in the second person singular lies in the purportedly difficult pronunciation of those forms. I disagree with their claim: 1. In so far as the relevant statistical data are extracted from WRITTEN sources, the reference to pronunciation requires caution. (This counter-argument is pointed out by Hauser-Suida & Hoppe-Beugel 1972, 120.) 2. The pronunciation of the verbal forms plays no impeding role in the case of the subject pronoun Sie, yet the statistical data indicate the avoidance of the preterite even with Sie. 3. If I judge correctly, it is easier to pronounce the preteritival form of the second person plural than of the singular ( [ihr] fochtet 'ye fought' as against [du] fochtst 'thou fought'), yet the preterite is avoided more in the plural than in the singular. 4. How can it be that the pronunciation difficulties do not lead to any consequences in the present tense, where the conditions regarding the pronunciation are comparable to those in the preterite, e.g. [du] fichtst, bindst 'thou fight, bind'? 5. Asked about the use of the preterite in the second persons, native speakers of German did not allude to any difficulty of pronunciation, but insisted that the form was anti quated and stylistically questionable. (This is reported in Hauser-Suida & HoppeBeugel 1972, 124.) 6. In the language material from which Hauser-Suida & Hoppe-Beugel have extracted the data about the two tenses, two texts use the present perfect in the second person to the complete exclusion of the preterite, whereas the remaining texts employ the preterite with the following verbal lexemes only: SEIN 'be,' HABEN 'have,' the modal verbs, as well as verbs of saying and thinking. On these frequent lexical items (excepting verbs of saying and thinking, about which the figures are too low to be utilizable in the present context) see Table 2, whose data have been culled from the spoken German texts of the Freiburg corpus. The two German monographs mention that the preterite (regardless of the grammatical person) is favoured by some additional frequent verbal lexemes. To those already enumerated under (6) above, KOMMEN 'come,' GEHEN 'go,' MACHEN 'make,' WISSEN 'know,' BLEIBEN 'remain,' STEHEN 'stand,' and DENKEN 'believe' can be added. This state of affairs is accounted for by the Slovenian Theory in the following way: 254. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.)

204

THE PRETERITE AND PRESENT PERFECT TENSES IN GERMAN

preterite SEIN HABEN modal verbs other verbs 178831,14% 295 5,14% 75513,15% 290350,57% 5741100,00%

perfect 153 2,21% 112 1,62% 15 0,22% 663895,95% 6918100,00%

TABLE 2. The preterite and the present perfect in the spoken German texts of the Freiburg corpus. Source: Latzel 1977, 84. A special case of 1.1: 1.1.1. >sym (the present perfect tense, the preterite tense) / in German I.e. with respect to coding properties, the present perfect tense is more natural than the preterite tense, in German. 1.2. >sem (more frequent, less frequent) / unit I.e. with respect to semantic complexity, a more frequent unit is more natural than a less frequent unit. (In the spirit of Fenk-Oczlon 1991.) 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the preterite and present perfect tenses, such that one tense is used in verbal lexemes of high frequency, and the other tense is used in verbal lexemes of non-high frequency, it is the present perfect tense that tends to be used with verbal lexemes of non-high frequency. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the preterite and present perfect tenses, such that one tense is used in verbal lexemes of high frequency, and the other tense is used in verbal lexemes of non-high frequency, it is the preterite tense that tends to be used with verbal lexemes of high frequency. Q.E.D. The passive (auxiliary: WERDEN 'become') and the stative (auxiliary: SEIN 'be') also prefer the preterite to the present perfect. More precisely, the figures culled from the Freiburg corpus (Latzel 1977, 88-9) show that the stative uses the preterite almost exclusively, whereas the situation in the passive is as presented in Table 3. active ("other" verbs) preterite perfect 1475 23,14% 4898 76,86% 6373 100,00% passive ("other" verbs) 318 354 47,32% 52,68%

672 100,00%

TABLE 3. The preterite and the present perfect in the German passive, the Freiburg corpus. Source: Latzel 1977, 88-9.

205

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

As can be seen in the left column of Table 3, the preterite clearly falls behind the present perfect in the Freiburg corpus as a whole. In the passive, however, the incidence of the preterite is doubled. The two German monographs offer no explanation of this state of affairs, whereas the Slovenian Theory appeals to the circumstance that the auxiliary of the passive and of the stative is a very frequent lexical item, i.e. WERDEN, SEIN, so that deduction 254 applies. To account for the numerical difference between the stative and the passive, the following naturalness scale is needed: >sem (stative, passive) / in German I.e. with respect to semantic complexity, the German stative is more natural than the German passive.This scale can be supported with the well-known circumstance that, in German, the stative is the older of the two constructions. What develops earliest is likely to be easy in terms of semantic complexity. (This follows from Mayerthaler's tenet, e.g. 1987, 51, that what is more marked implies the existence of what is less marked.) Compare deduction 255: 255. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the present perfect tense, the preterite tense) / in German I.e. with respect to coding properties, the present perfect tense is more natural than the preterite tense, in German. 1.2. >sem (stative, passive) / German I.e. with respect to semantic complexity, the stative is more natural than the passive, in German. 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 2.2. <sym tends to associate with >sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the preterite and present perfect tenses, such that one tense prevails in the stative, and the other tense prevails in the passive, it is the present perfect that tends to be used in the passive. Q.E.D. From 1.1.1, 1.2 and 2.2 it can be deduced: 3.2. If there is any difference between the preterite and present perfect tenses, such that one tense prevails in the stative, and the other tense prevails in the passive, it is the preterite that tends to be used in the stative. Q.E.D. The numerical difference already mentioned between the use of the preterite in the stative and the use of the preterite in the passive can be accounted for in the following way. (Recall that deduction 254 also applies to the German stative and passive.) The use of the preterite predicted by item 254.3.1 is further FED by item 255.3.2, therefore the 206

THE PRETERITE AND PRESENT PERFECT TENSES IN GERMAN

preterite clearly prevails over the present perfect in the stative. The use of the preterite predicted by item 254.3.1 is BLED by item 255.3.1, therefore the use of the two tenses is more balanced in the passive than in the stative. The prevalence of the preterite also obtains in relative clauses, see Table 4, whose data have been culled from short notices of 28 German daily newspapers. superordinate clause preterite perfect 178 131 57,61% 42,39% relative clause 272 88,03% 37 11,97% 309100,00%

309 100,00%

TABLE 4. The preterite and the present perfect in relative clauses. The figures have been culled from short notices of 28 German daily newspapers. Source: Latzel 1977, 91-2. As can be seen from the left column of Table 4, the present perfect only slightly lags behind the preterite in clauses superordinate to relative clauses, whereas the share of the present perfect in the relative clauses is four times smaller. Latzel has no explanation. The Slovenian Theory fares better; compare the following deduction: 256. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the present perfect tense, the preterite tense) / in German I.e. with respect to coding properties, the present perfect tense is more natural than the preterite tense, in German. 1.2. >sym (non-relative clauses, relative clauses) / dependent clauses I.e. with respect to coding properties, relative clauses are less natural than other kinds of dependent clauses.This guess of mine is based on the circumstance that relative clauses on the average display the simplest syntax of all dependent clauses (Orenik 1992). 1.3. >sem (relative clauses, non-relative clauses) / dependent clauses I.e. with respect to semantic complexity, relative clauses are more natural than other kinds of dependent clauses.This guess of mine is based on historical linguistics, which notes that the relative clauses come into being earliest of all dependent clauses, in the history of languages. Well-known cases in point are old Indo-Iranian languages and Ancient Greek. Practically all subordinators of those languages are relative pronouns or derived from the latter (Schwyzer 1950, 11, 637). 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. <sym (of 1.1.1) tends to associate with >sem 3. The consequences: From 1.1.1, 1.2, 1.3 and 2.1 it can be deduced: 3.1. If there is any difference between the preterite and present perfect tenses, such that one tense is used in relative clauses, and the other tense outside them, it is the 207

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

preterite that tends to be used in relative clauses. Q.E.D. 4. Note. An asymmetrical deduction. In interrogative clauses, the present perfect prevails over the preterite, see Table 5, whose figures have been culled from the dialogues in a number of German novels. interrogative clauses preterite perfect 46 15,08% 259 84,92% 305100,00% other clauses 837 33,84% 163666,16% 2473100,00%

TABLE 5. The preterite and the present perfect in interrogative clauses. The figures, culled from the dialogues in a number of German novels (excluded are SEIN, HABEN , and the modal verbs), stem from Hauser-Suida & Hoppe-Beugel 1972, but are quoted from the more comfortable source Latzel 1977, 95-6. It can be seen in the rightmost column of Table 5 that the proportion of preterite and present perfect is 1 to 2 outside interrogative clauses, whereas in the latter the already small share of the preterite is further reduced by half. The two German monographs do not offer any explanation. Latzel states that, of those 15% of interrogative clauses that do contain the preterite, 11% display the verb in the third verbal persons, which are known to favour the preterite anyway. In addition, the interrogative clauses containing the preterite are mostly of a special kind, namely second-instance solicitation for further information about the narra tive, therefore they stick with the preterite as the verbal tense of the narration. If this peculiarity is eliminated from the statistics, practically only the present perfect is left as the verbal tense of the interrogative clauses expressing past time. This state of affairs can be accounted for by the following deduction: 257. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the present perfect tense, the preterite tense) / in German I.e. with respect to coding properties, the present perfect tense is more natural than the preterite tense, in German. 1.2. >sem (declarative clause, interrogative clause) I.e. in terms of semantic complexity, the declarative clause is more natural than the interrogative clause. (Mayerthaler et al. 1998, 326.) 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 3. The consequences: From 1.1.1, 1.2 and 2.1 it can be deduced:

208

THE PRETERITE AND PRESENT PERFECT TENSES IN GERMAN

3.1. If there is any difference between the preterite and present perfect tenses, such that one tense prevails in declarative clauses, and the other tense prevails in interrogative clauses, it is the present perfect that tends to prevail in interrogative clauses. Q.E.D. 4. Note. An asymmetrical deduction. In the oral narrative of "Northern" German, the preterite of course outweighs the present perfect. However, the conditions are different in clauses that begin with 'and then,' 'and now,' 'in sum:' here the present perfect is more prominent than elsewhere in the narrative. See Table 6, whose figures have been culled from the Bottrop interviews (recorded in the town of Bottrop in the region of Ruhr). normal distribution clauses preterite perfect 1111 63,13% 649 36,87% 1760 100,00% containing 'and then' etc. 113 104 52,07% 47,93%

217 100,00%

TABLE 6. The preterite and the present perfect in clauses containing 'and then' etc., the Bottrop interviews. Source: Latzel 1977, 97. It can be seen in Table 6 that the share of the present perfect rises in clauses containing 'and then' and similar introductory expressions. Latzel 1977, 97 points out that the rise is in reality more dramatic than revealed by the table, seeing that the preterites of such clauses mostly belong to the lexical items 'be,' 'have,' and the modal verbs, which are known to avoid the present perfect anyway. Latzel offers no explanation. According to Liberman 1990, one function of the German present perfect throughout its history has been to foreground the (past-time) message of its clause. Duden Grammatik 1998, 152 mentions the frequent use of the present perfect in the very initial clause or sentence and in the very final clause or sentence of a piece of narrative, even when that narrative otherwise utilizes the preterite. The same foregrounding function of the present perfect can be assumed in the clauses beginning with 'and then,' etc. See the following deduction: 258. The two syntactic variants: the German preterite and present perfect tenses. 1. The assumptions of Naturalness Theory: 1.1. >sym (more transparent, less transparent) / syntactic unit I.e. with respect to coding properties, a syntactic unit of greater syntactic transparency is more natural than a corresponding syntactic unit of lesser syntactic transparency. (Mayerthaler 1981, 35; Mayerthaler et al. 1998, 186. On the notion of transparency see Mayerthaler 1987, 49.) A special case of 1.1: 1.1.1. >sym (the present perfect tense, the preterite tense) / in German I.e. with respect to coding properties, the present perfect tense is more natural than the preterite tense, in German. 1.2. >sem (backgrounding, foregrounding) I.e. in terms of semantic complexity, backgrounding is more natural than foregrounding. (In the spirit of Dotter 1990, 29; Givn 1995, 28.) 2. The assumptions of the Slovenian Theory concerning any two syntactic variants: 2.1. >sym tends to associate with <sem 3. The consequences: 209

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

From 1.1.1, 1.2 and 2.1 it can be deduced: 3.1. If there is any difference between the preterite and present perfect tenses, such that one tense prevails in foregrounding, and the other tense prevails in backgrounding, it is the present perfect tense that tends to be used in foregrounding. Q.E.D. 4. Note. An asymmetrical deduction. I hope to have shown that the Slovenian Theory can pour life into some of the statistical data which have been gathered about the German preterite and present perfect. Without the aid of the Slovenian Theory those data would remain unaccounted for. The two German tenses support the Slovenian Theory well, at least for the following reasons. (1) The case is relatively simple and clear. (2) The data of the case have been gathered, processed and published by linguists not associated with the Slovenian Theory. (3) Most naturalness scales utilized here in the interpretation of the case have likewise been posited and published by linguists not associated with the Slovenian Theory.

210

Slovenski povzetek Slovenian Summary


(Kar sledi, je slovenski razliek uvodnega poglavja v angleini. Da bi besedili lae primerjali, znaki/simboli tu spodaj niso poslovenjeni; ohranjeno je tudi otevilenje an glekega besedila.) Tvarina te knjige je (univerzalna jezikovna) teorija, ki jo je v Sloveniji (pod mojim vodstvom) razvilo nekaj mlajih jezikoslovcev; svoja dognanja preverjamo zveine ob an glekem, nemkem in slovenskem jezikovnem gradivu. Podlaga naega dela je (jezikovna) teorija naravnosti, kakor so jo razvili predvsem na nekaterih avstrijskih in nemkih uni verzah; prim. Mayerthaler 1981, Wurzel 1984, Dressler in dr. 1987, Stolz 1992, Dressler 2000. Teorijo naravnosti so izkoristili tudi v skladnji, zlasti na Univerzi v Celovcu; osnovne objave so Dotter 1990, Mayerthaler & Fliedl 1993, Mayerthaler in dr. 1993, 1995, 1998. K celovki naravni skladnji je slovenska delovna skupina izdelala podaljek, ki se v nadaljevanju imenuje "slovenska teorija". Slovenska teorija preuuje vednje (veinoma skladenjskih in oblikoskladenjskih, v manji meri oblikoslovnih) izrazov, tu imenovanih dvojnice. Kadar se dasta dve dvojnici umestiti v isto lestvico naravnosti, in se torej sme trditi, da je ena dvojnica bolj naravna od druge, more slovenska teorija poblie oznaiti slovnine lastnosti takih dvojnic. Teorija naravnosti se naslanja na dve temeljni povedji: "zaznamovan" in "naraven". Iz zornega kota slovenske teorije ni videti razloga za razlikovanje med njima, zato v nadaljevanju rabim dosledno samo eno povedje, namre "naraven". (Tko stalie je nakazano e v Mayerthaler 1987, 50.) Poleg strokovnih izrazov "naraven", "naravnost" in "lestvica naravnosti" je treba uvesti izraza "sym-vrednost" in "sem-vrednost" (prevzeta iz Mayerthaler 1981, 10 in passim). Sym-vrednost zadeva naravnost izraza glede na nain kodiranja. Sem-vrednost zadeva naravnost izraza glede na zapletenost njegove pomenske strani. Lestvice naravnosti so oblike >sem (a, b) ali oblike >sym (c, d). Oblika >sem (a, b) se ubesedi takole: glede na pomensko zapletenost je a bolj naraven od b. Zaradi lajega raunanja se priredi a-ju visoka sem-vrednost (symbol: >sem "glede na pomensko zapletenost bolj naraven") in b-ju se priredi nizka sem-vrednost (symbol: <sem "glede na pomensko zapletenost manj naraven"). Oblika >sym (c, d) se ubesedi takole: glede na kodiranje je c bolj naraven od d-ja. Zaradi lajega raunanja se priredi c-ju visoka sym-vrednost (symbol: >sym "glede na kodiranje bolj naraven") in d-ju se priredi nizka sym-vrednost (symbol: <sym "glede na kodiranje manj naraven"). Predpostavke slovenske teorije (v mojem najnovejem razliku) so kratko naslednje. V paru dvojnic nastopa (v vsaki dvojnici) ena izmed naslednjih izbirnih monosti:

212

SLOVENSKI POVZETEK SLOVENIAN SUMMARY

1. najmanj ena >sym-vrednost tei po povezavi z najmanj e eno >sym-vrednostjo in/ali z najmanj eno <sem-vrednostjo; 2 najmanj ena <sym-vrednost tei po povezavi z najmanj e eno <sym-vrednostjo in/ali z najmanj eno >sem-vrednostjo; 3. najmanj ena >sem-vrednost tei po povezavi z najmanj e eno >sem-vrednostjo in/ali z najmanj eno <sym-vrednostjo; 4. najmanj ena <sem-vrednost tei po povezavi z najmanj e eno <sem-vrednostjo in/ali z najmanj eno >sym-vrednostjo. Skratka, dovoljene so vse teoretino mone povezave sem- in sym-vrednosti razen >sem v povezavi s >sym ali <sem v povezavi s <sym ali >sem v povezavi s <sem ali >sym v povezavi s <sym. Na primer, ne priakuje se, da bi se zgradba z izpustom (= <sym) rabila izkljuno v netretji osebi (= <sem) ali da bi se opisna zgradba (= >sym) rabila izkljuno v tretji osebi (= >sem) kadar je na voljo resnina izbira med slovninimi osebami. Dalje se ne priakuje, da bi se netretja glagolska oseba (= <sem) rabila samo v ednini (= >sem) ali da bi se opisna zgradba (= >sym) rabila samo v povezavi z izpustom (= <sym). V gornjih tokah 1-4 omenjena "povezava" se dogaja v notranjosti opazovane enote ALI v neposrednem okolju opazovane enote. Slovenska teorija zajema obe monosti. Forschungsgeschichtlich je prednik predpostavk 1-4 znano naelo o konstrukcijskem ikonizmu, kakor ga je postavilo naravno oblikoslovje. Naelo se glasi takole. e in samo e je pomensko bolj zaznamovana kategorija C j kodirana z ve potezami kot manj zaznamovana kategorija Ci, pravimo, da je kodiranje od C j ikonino (Mayerthaler 1987, 48-9). S povedjem "naraven" se naelo izree kratko takole: <sem v povezavi s >sym je ikonino. Slovenska teorija je naelo prenesla e v (obliko)skladnjo in ga nekoliko razirila. Slovenska teorija deluje izkljuno ex post facto. Zgled izpeljave: x. Angleina. Kratnostni prislovi (izraz po Toporii 2000, 408) se tvorijo z glavnimi tevniki in besedo times, npr. four times 'tirikrat', le za najnija tevila so na voljo enobesedni izrazi: once 'enkrat', twice 'dvakrat' in starinsko thrice 'trikrat'. (Collins Cobuild 1990, 270-1.) Skladenjski dvojnici: vzorec once in vzorec four times. x.1. Predpostavke teorije naravnosti: x.1.1. >sym (bolj razvidna, manj razvidna) / skladenjska enota Tj. glede na kodiranje je skladenjska enota veje skladenjske razvidnosti bolj naravna kot soodnosna skladenjska enota manje skladenjske razvidnosti. (Mayerthaler 1981, 35; Mayerthaler in dr. 1998, 186. O pojmu razvidnosti glej Mayerthaler 1987, 49.) Posebni primer od x.1.1: x.1.1.1. >sym (vzorec four times, vzorec once) / v angleini Tj. glede na kodiranje je vzorec four times bolj naraven od vzorca once, v angleini. x.1.2. >sem (nizko tevilo, nenizko tevilo) Tj. glede na pomensko zapletenost je nizko tevilo bolj naravno od nenizkega tevila. (Mayerthaler 1981, 15.) x.2. Predpostavke slovenske teorije o katerem koli paru skladenjskih dvojnic: x.2.1. >sym tei po povezavi s <sem x.2.2. <sym tei po povezavi s >sem x.3. Posledice: 213

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

Iz x.1.1.1, x.1.2 in x.2.1 se da sklepati: x.3.1. e je v kratnostnih prislovih kak razloek med nizkimi in nenizkimi tevili, tako da ena vrsta kratnostnih prislovov rabi vzorec four times in druga vrsta kratnostnih prislovov rabi vzorec once, potem nenizka tevila teijo po rabi vzorca four times. Q.E.D. Iz x.1.1.1, x.1.2 in x.2.2 se da sklepati: x.3.2. e je v kratnostnih prislovih kak razloek med nizkimi in nenizkimi tevili, tako da ena vrsta kratnostnih prislovov rabi vzorec four times in druga vrsta kratnostnih prislovov rabi vzorec once, potem nizka tevila teijo po rabi vzorca once. Q.E.D. Prehajamo na razlago izpeljave x. Uporabili bomo jezikovne lastnosti a-d, ki so bile e omenjene. Iz toke x.1.1.1 se vidi, da ima "vzorec four times" (recimo, da predstavlja lastnost a) veliko sym-vrednost. Dalje se iz toke x.1.1.1 vidi, da ima "vzorec once" (recimo, da predstavlja lastnost b) majhno sym-vrednost. Iz toke x.1.2 se vidi, da ima "nizko tevilo" (recimo, da predstavlja lastnost c) veliko sem-vrednost. Dalje se iz toke x.1.2 vidi, da ima "nenizko tevilo" (recimo, da predstavlja lastnost d) majhno sem-vrednost. V skladu s toko x.2.1 tei velika sym-vrednost (= lastnost a naega zgleda) po povezavi z majhno sem-vrednostjo (= lastnost d naega zgleda). V skladu s toko x.2.2 tei majhna sym-vrednost (= lastnost b naega zgleda) po povezavi z veliko semvrednostjo (= lastnost c naega zgleda). Iz vsega povedanega se da sklepati naslednje. Kjer je podana lastnost a (= "vzorec four times"), se pojavlja tudi lastnost d (= nenizko tevilo). Kjer je podana lastnost b (= "vzorec once"), se pojavlja tudi lastnost c (= nizko tevilo). Podrtana stavka sta napovedi. Slednji obveata, da je naravno, da se "vzorec once" pojavlja v nizkih tevilih in da se "vzorec four times" pojavlja v nenizkih tevilih. (Ni pa nikake napovedi o tem, kje poteka meja med nizkimi in nenizkimi tevili.) Vpraanje jezikoslovcu: "e bi Vam povedali, da loi neki jezik toilnike in rodilnike preme predmete in da rabi ene v trdilnih stavkih, druge pa v nikalnih, v katere stavke bi na podlagi svoje skunje z jeziki dali toilnike in v katere stavke rodilnike preme predmete?" Jezikoslovev odgovor: "Napovedal bi, da se toilniki premi predmeti rabijo v trdilnih stavkih, rodilniki pa v nikalnih." Ravno takih napovedi je zmona slovenska teorija. Samo tisti bralec, ki je opazil in doumel ozkost vpraanja, bo pripoznal tudi vsaj delno koristnost te knjige. e neka tvorbena teorija tvori preme predmete v toilniku in rodilniku in predvidi, da se ena vrsta rabi v trdilnih stavkih, druga pa v nikalnih, se lahko slovenski teoriji prepusti, da obe vrsti premih predmetov razporedi, in sicer toilnike v trdilne stavke, rodilnike v nikalne. Tako slovenska teorija namesto tvorbene teorije dokonuje tvorbo nekaterih zgradb, namre tistih, ki se tu imenujejo dvojnice. (To je kajpak kvejemu liliputanski korak za jezikoslovje, zame pa vsekakor velik!) V osrednjem delu monografije je slovenska teorija ponazorjena z izbranimi zgledi. Vsak je obravnavan v izpeljavi, katere predloga y je takale: y. Ime/imena jezikov. Kratek opis obravnavanega jezikovnega gradiva. Obe dvojnici: V1 (vsebujoa lastnosti a in b) in V2 (vsebujoa lastnosti c in d). y.1. Predpostavke teorije naravnosti: y.1.1. navaja neko lestvico naravnosti, ki se prilega obravnavanemu jezikovnemu gradivu. Lestvica naravnosti omenja lastnosti a in c ter jima prireja visoko oz. nizko sem/sym-vrednost. y.1.2. navaja drugo lestvico naravnosti, ki se prilega obravnavanemu jezikovnemu gradivu. Lestvica omenja lastnosti b in d ter jima prireja visoko oz. nizko sem/symvrednost.

214

SLOVENSKI POVZETEK SLOVENIAN SUMMARY

y.2. Predpostavke slovenske teorije o katerem koli paru dvojnic: y.2.1. navaja neko predpostavko, ki smiselno velja za y.1.1-2. y.2.2. navaja drugo predpostavko, ki smiselno velja za y.1.1-2. (Predpostavke slovenske teorije so izrpno natete v nadaljevanju.) y.3. Posledice: Iz y.1.1, y.1.2 in y.2.1 se da sklepati: y.3.1. e je med lastnostma a in b kak razloek, tako da se ena izmed lastnostjo c in druga izmed njih z lastnostjo d, potem lastnost a tei, da bi lastnostjo c. Q.E.D. Iz y.1.1, y.1.2 in y.2.2 se da sklepati: y.3.2. e je med lastnostma a in b kak razloek, tako da se ena izmed lastnostjo c in druga izmed njih z lastnostjo d, potem lastnost b tei, da bi lastnostjo d. Q.E.D. y.4. Opombe (po potrebi).

njih vee z se vezala z njih vee z se vezala z

Preostalo besedilo je komentar k posaminim tokam predloge. V zgledih sem si prizadeval rabiti jezikovno gradivo, ki bi bilo karseda preprosto, jasno in pestro ter upam nesporno. Najraje sem se odloal za zglede, ki so e bili omenjani in/ali obravnavani v strokovni literaturi. Popolnoma pa sem obel nekaj nujnega, namre dosledno izrabo jezikovnega gradiva v kaki jezikoslovni knjini razpravi ali slovnici, da bi zbral morebitne zglede, ki bi slovenski teoriji nasprotovali. Sodim, da za tko potezo as ni zrel, saj precej lestvic naravnosti, ki bi nam mogle biti v pomo, e ni na voljo. al se za sedaj ne da potrditi, da je kak moen protizgled res pravi. V knjigi je obravnavanih 250-60 zgledov iz ve kakor 50 jezikov in nareij. Bralec se bo udil skoraj popolni odsotnosti anglekih zgledov. V zapiskih jih imam mnogo, a jih v knjigo nisem mogel vkljuiti iz nejezikoslovnih razlogov. Upam, da bo prilonost za objavo kdaj pozneje. tevilo gradivskih virov sem skual karseda kriti, da bi se izognil nevarnemu meanju razlinih sistemov strokovnega izrazja. Iz istega razloga le redko navajam bibliografske podatke, s katerimi se moji viri sklicujejo na e bolj temeljna dela. Kljub tem zavestnim naporom moj metateoretini aparat ni popolnoma usklajen. V knjigi je med kljunimi pojem (oblikoslovne, oblikoskladenjske, skladenjske) dvojnice. Slednje se domala vedno pojavljajo v parih. Na zgodnjih stopnjah naih raziskav smo domnevali, da vlada med lenoma vsakega para (vsaj priblina) sopomenskost. Toda v vrstah jezikoslovja, ki jih poznam, ni na voljo natannega merila (pribline) sopomenskosti katerega koli para dvojnic (z nekaterimi izjemami). Sploh pa so mi noveje raziskave pokazale, da slovenska teorija ohranja veljavo, etudi se pojem dvojnice raziri tako, da vkljuuje (oblikoslovne, oblikoskladenjske, skladenjske) enote, ki gredo pod isto nadrejeno (oblikoslovno, oblikoskladenjsko, skladenjsko) enoto ali kategorijo. Tako nista dvojnici samo oba angleka nedolona lena, a(n) and sm (npr. a book 'knjiga', sm milk 'mleko', oboje z nedolonim lenom), temve tvorita par dvojnic tudi angleki doloni in nedoloni len (ker jima je nadrejen pojem "len"). Ker je ta pojem dvojnice premalo jasen, v vsaki izpeljavi uvodoma posebej natevam dvojnici (redko ve kot dve), ki bosta v obravnavi. (V nadaljevanju se oblikoskladenjske in skladenjske dvojnice kratko imenujejo kar "skladenjske dvojnice".) Knjiga ne uvaja v teorijo naravnosti. Glede nael in drugih naukov slednje bralcu pri poroam temeljna dela, nateta na zaetku povzetka. (Poleg znanja angleine je bistveno 215

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

razumevanje pisne nemine!) V mojem znanstvenem okolju se teorija naravnosti ne da gojiti, ker manjkajo potrebni viri. (To se utegne spremeniti na bolje, e bodo na medmreju zlagoma res dostopni bogati zadevni podatki, glej Plank & E. Filimonova 2000). Kljub temu sem si dovolil predlagati po kako novo lestvico naravnosti. Do mojih tovrstnih ugibanj naj ostane bralec zadran kljub utemeljevanju, s katerim ponekod podpiram svoje poskuse. Bolj pogumno ponujam dve VRSTI lestvic, in sicer takih, ki po moji sodbi prej koreninijo v logiki jezikovnih odnosov kot v jezikovnem gradivu eprav seveda verjamem, da so predlagane lestvice z dejstvi skladne: (a) >sem (x, y), kjer y ne tvori "naravnega razreda". V takem primeru je y manj semnaraven od x-a. Prim. izpeljave 46, 210 in 213. (b) >sem (A + B, A), tj. vsak pojav je bolj sem-naraven kot neki njegov del. Prim. izpeljave 33, 62 in 93. Okolnost, da je neka lestvica oblike >sem (x, y) (kjer y ne tvori "naravnega razreda") ali oblike >sem (A + B, A), se bo v knjigi navajala v podporo tisti lestvici. Na novi vrsti lestvic nisem ravno ponosen, kajti zelo veata mo teorije naravnosti; v tem smislu sta korak v napano smer. A kar precej jezikovnih pojavov bi ostalo zunaj dosega teorije, ko bi takih lestvic ne privzeli. So nujno zlo. Vsaka lestvica predloge >sem (A + B, A) odseva v povezavi z drugo lestvico znano opaanje, da je bolj sem-naravno (manj zaznamovano) tudi bolj pestro. Predloga >sem (A + B, A) ima sistematino slabost, ki se pokae, e se predloga zapie kot >sem (A + B, X), kjer je X bodisi A bodisi B, tako da je izbira med A in B (v X) samovoljna. Zato ne morejo na X-u predloge temeljiti nikake napovedi. Lestvice predloge >sem (A + B, A) se smejo rabiti v toki Posledice katere koli izpeljave samo tako, da so vsebinsko popolnoma predpostavljene; to se pravi, da morajo stati pred prvo omembo besed "tako da". al more lestvica zvrsti >sem (A + B, A) vsebovati tudi nesmisel, npr. >sem (dovrni in nedovrni vid, dovrni vid) / samostalnika zveza. Nasploh je verjetnost, da bi samostalnike zveze razvijale vidske razloke, majhna; pod to predpostavko primer ne bi smel dobiti iga "pravilna tvorba". Vendar to ni zadrega samo moje knjige, marve kolikor vem teava ni odpravljena v nobeni jezikovni teoriji. elim si, da bi vzorec >sem (A + B, A) kmalu nadomestili z bistveno manj monim in bolj obvladljivim >sem (+A in -A, +A ali -A). In v resnici je veina mojih zgledov v skladu s slednjim vzorcem a ne vsi. Teoretino moen primerek vzorca >sem (A + B, A) je naslednji: >sem (ednina & mnoina & dvojina, ednina & mnoina) / samostalnika zveza, tj. samostalnika zveza, ki loi ednino, mnoino in dvojino je bolj sem-naravna od samostalnike zveze, ki loi samo ednino in mnoino. Ni miljeno, da so samostalnike zveze, ki loijo ednino, mnoino in dvojino, po jezikih sveta bolj razirjene od samostalnikih zvez, ki loijo zgolj ednino in mnoino. Lestvica pravi, da so v jeziku, katerega samostalnike zveze deloma loijo ednino, mnoino in dvojino, deloma pa samo ednino in mnoino, samostalnike zveze prve vrste bolj sem-naravne kakor samostalnike zveze druge vrste. Na primer, samostalnike zveze prve vrste bi zajemale osebne zaimke, samostalnike zveze druge vrste pa vse preostale samostalnike zveze tistega jezika. Nota bene: jezikovni pojav, ki je po jezikih sveta pogosten, sme dobiti visoko sem-vrednost. Toda okolnost, da ima neki pojav visoko sem-vrednost, e ne dovoljuje obratnega sklepa, da je tisti pojav po jezikih sveta pogosten. Kakor je znano, izhajajo tevilne lestvice naravnosti od Mayerthalerja, zato je njegovo delo v tej knjigi navajano pogostoma. Celo kadar bi mogel bralca napotiti na drugega pisca, sem se ceteris paribus navadno odloil za Mayerthalerja, da bi se izognil 216

SLOVENSKI POVZETEK SLOVENIAN SUMMARY

morebitnim protislovjem. V asu tega pisanja imam v zapiskih ok. 460 izpeljav. V njih se nekaj lestvic narav nosti ponavlja z dokajnjo vztrajnostjo. Na primer, lestvica razvidnosti se rabi 229-krat, lestvica pogostnosti 74-krat. To najbr kae na realistinost priakovanja, da bo skupna koliina lestvic omejena. e ve, iz okolnosti, da se najve, in z uspehom, rabi ravno lestvica razvidnosti, smemo sklepati, da ima t.i. jezikovna zmonost najraji parametre, ki so karseda preprosti. (Preostale lestvice se pojavljajo vsaka kvejemu ok. 20-krat in so po vsebini precej manj splone od pravkar omenjenih bolj pogostnih lestvic.) Z nekaterimi lestvicami naravnosti so zdruene resnine ali navidezne teave: (a) Slovit primer protislovnih "zakonov" naravnosti so lestvice o slovnini osebi. Loimo najmanj dve izroili. Eno poteka od Jakobsona 1932, e ne e od prej, in je doivelo dopolnitev predvsem v Greenbergovem delu: >sem (tretja, prva, druga) / slovnina oseba. Drugo izroilo se navadno pripisuje Bhlerju (npr. Bhler 1934): >sem (prva, druga, tretja) / slovnina oseba. (Zapis obeh lestvic je tu "posodobljen".) Na podlagi jezikovnega gradiva se vidi, da v veini primerov pomaga Jakobsonova lestvica. Tako igra slovnina oseba kljuno vlogo v nemkih glagolskih asih preteritu in perfektu (glej poglavje o njiju) in oba asa prepriljivo podpirata Jakobson-Greenbergovo izroilo. V opazno manjem deleu zagotavlja dober izid Bhlerjev razliek. Toda kolikor je Bhlerjeva lestvica rabljena v mojih zgledih, nastopa v skreni obliki >sem (prva, neprva) / slovnina oseba. "Neprva slovnina oseba" tega razlika se more razlagati kot primerek "nenaravnega razreda". Jakobson-Greenbergovo izroilo namre daje misliti, da tvorita "naravni razred" prva in druga oseba; e je tako, ne moreta tudi druga in tretja oseba Bhlerjeve lestvice tvoriti "naravnega razreda". Z drugimi besedami: Bhlerjeva lestvica se sprevre v primerek nove vrste lestvic, ki sem jo predlagal zgoraj, in ni resnina protiute Jakobson-Greenbergovemu izroilu. Priredno razmerje dveh slovninih oseb kot 'ti in X' (kar se izide v navezovalnik [anaforo] v drugi osebi) se obvlada z Jakobson-Greenbergom, glej izpeljavo 252. Primerljivo mnenje imam o lestvici >sem (druga, nedruga) / slovnina oseba. Lestvica je uporabna v podsistemu pozivanja, zlasti v zvezi z velelnikom in z zaimki nagovora (npr. nemko du 'ti' nasproti Sie 'Vi'). Nedruga oseba ne tvori "naravnega razreda" in je zato manj sem-naravna od druge osebe. Ta lestvica se rabi v malotevilnih zgledih in je v tem podobna Bhlerjevi ter drugana od Jakobson-Greenbergove. Ni videti potrebe po e kakem etrtem razliku lestvice o slovnini osebi, vendar vabim bralca, naj si ogleda Croft 1990 passim, npr. 111-3, z navedbo literature. (b) Ferguson 1996, 243 je opozoril na paradigmo staroanglekega in starovedskega sedanjika v povednem naklonu. Stara angleina loi slovnine osebe samo v ednini, stara vedina samo v mnoini. Oboje razmere se dajo razloiti s slovensko teorijo, glej izpeljavi 20 in 222. Zgleda kaeta, da so si nekatere lestvice naravnosti v navzkriju, kar je razumeti v lui znane predpostavke, da delujejo mogani z deloma nasprotujoimi si parametri. Vasih se da neko jezikovno gradivo obravnavati v dveh izpeljavah in napovedi so v obeh primerih enake (temeljijo pa na razlinih predpostavkah!). Take razmere so realistine; nimajo namre vsi govorci ponotranjene enake slovnice. Pravila vzporejanja slovenske teorije se verjetno dajo spreprostiti, kakor pokae naslednji premislek. Iz para lestvic >sem (a, b) in >sym (c, d) sledi: a se vee z d, b se vee s c. Iz para lestvic >sem (a, b) in >sem (d, c) sledi isto: a se vee z d, b se vee s c. Iz tega sledi enaba: >sem (a, b) = >sym (b, a). Zaradi te enabe se da vsaka sem-lestvica pretvoriti v enakovredno sym-lestvico in obratno. Torej zadoa imeti samo sem- ali samo

217

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

sym-vrednosti. Sledi, da je razlikovanje med sem- in sym-vrednostmi nepotrebno in da se lahko "vrnemo" k zaznamovanosti. Enabe, ki omogoajo to pretvorbo, so: >sem (a, b) = <m (a, b), tj. a je manj zaznamovan od b; >sym (a, b) = >sem (b, a) = <m (b, a), tj. b je manj zaznamovan od a. Edini pravili vzporejanja, ki ju e potrebujemo, sta >m se vee z najmanj enim >m in <m se vee z najmanj enim <m. To resnino predlaga neodvisno od teorije naravnosti Andersen 2001, ki se pri tem naslanja na Andersen 1972 in na e zgodnejega Mare 1952. Pojav, zajet s praviloma vzporejanja, imenuje Andersen 2001 markedness agreement. Te spreprostitve v svoji knjigi ne uveljavljam, ker je gornje utemeljevanje podprto samo z razmerami pri oblikoslovnih in (obliko)skladenjskih dvojnicah, kar je morda prepila podlaga za tako daljnoseno spremembo. Veina izpeljav se izide v po dve posledici. V manjem tevilu izpeljav pa se da izrei samo po ena posledica. Take izpeljave se tu imenujejo nesimetrine. Razlog za obstoj nesimetrinih izpeljav je v okolnosti, da lestvice naravnosti neke vrste ali nekih vrst po moji vednosti e niso na voljo. Vsekakor je delitev izpeljav na simetrine in nesimetrine brez teoretinega pomena. V nasprotju s tem je kljunega pomena VSEBINA posledic (neglede na njihovo tevilo). Vsaka posledica ima obliko pogojnega podredja. V njem nateva odvisnik pogoje, pod katerimi je glavni stavek veljaven. Glavni stavek napoveduje navzonost ali odsotnost neke jezikovne lastnosti v eni izmed obravnavanih dvojnic. Napovedati X se pravi trditi, da X ne more biti drugaen. V tem smislu je v glavnem stavku vednje tiste jezikovne lastnosti razloeno (v okviru sinhronega jezikoslovja, kajpak). Uspenost vsake napovedi je odvisna od zanesljivosti predpostavk, ki so podlaga napovedi. Zgodovinske razlage niso predmet te knjige. S tem pristopom je zagotovljeno, da so sinhrone razlage podatkov izreene v okviru neke teorije v nasprotju z izvedbo v tevilnih slovnicah; slednje dovoljujejo predteoretine razlage. e bolj se bo zaostrilo v naslednjih letih, ko bodo izhajale slovnice, obogatene z analizo korpusov (dober primerek e v prodaji je Biber in dr. 1999) in v skunjavi, da bi tevilne podatke oivile s poskusi razlag. Seveda morajo razlage jemati v potev mnoge vidike. Naj omenim, da so posebno zahtevne razlage tistih podatkov, ki se pojavljajo samo v nekaterih besedilnih zvrsteh. Zajeti je treba tudi nejezikovne dejavnike. Zato se vsaj pri sedanjem stanju stvari ne pomilja na enotno teorijo sinhronih razlag. Toda nekateri podatki, zlasti neodvisni od besedilnih zvrsti, se dajo takoj izpopolniti s sinhrono razlago v okviru take teorije, kot je naa. Sledea izpeljava z, ki vsebuje zgled iz resninega ivljenja, namenjena pa je bralcu v zabavo ponazarja moje preprianje (ki ga ne bom utemeljeval), da more povezava lestvic naravnosti s postavkami slovenske teorije napovedovati celo vidike nejezikovnega vednja: z. Psikov odziv. Kadar na psiek Zen srea osebo, ki prenaa opazno veliko breme, vanjo zalaja. Dvojnici: oseba, ki nosi (ne nosi) opazno veliko breme. z.1. Predpostavke teorije naravnosti: z.1.1. >sym (bolj razvidna, manj razvidna) / enota Tj. glede na kodiranje je enota veje razvidnosti bolj naravna kot soodnosna enota manje razvidnosti. (Mayerthaler 1981, 35; Mayerthaler in dr. 1998, 186. O pojmu

218

SLOVENSKI POVZETEK SLOVENIAN SUMMARY

razvidnosti glej Mayerthaler 1987, 49.) Posebna primera od z.1.1: z.1.1.1. >sym (oseba, ki nosi veliko breme; oseba, ki ne nosi velikega bremena) Tj. glede na kodiranje je oseba, ki nosi veliko breme, bolj naravna kot oseba, ki ne nosi velikega bremena. z.1.1.2. >sym (psiek laja, psiek ne laja) Tj. glede na kodiranje je psiek, ki laja, bolj naraven kot psiek, ki ne laja. z.2. Predpostavke slovenske teorije glede katerega koli para dvojnic: z.2.1. >sym tei po povezavi s e enim >sym z.2.2. <sym tei po povezavi s e enim <sym z.3. Posledice: Iz z.1.1.1-2 in z.2.1 se da sklepati: z.3.1. e je kak razloek med osebo, ki nosi veliko breme, in osebo, ki ne nosi velikega bremena, tako da v enem primeru na psiek zalaja v tko osebo, v drugem primeru pa ne, potem tei na psiek po tem, da bi zalajal v osebo, ki nosi veliko breme. Q.E.D. Iz z.1.1.1-2 in z.2.2 se da sklepati: z.3.2. e je kak razloek med osebo, ki nosi veliko breme, in osebo, ki ne nosi velikega bremena, tako da v enem primeru na psiek zalaja v tko osebo, v drugem primeru pa ne, potem na psiek ne tei po tem, da bi zalajal v osebo brez velikega bremena. Q.E.D.

219

References
Andersen, Henning, "Diphthongization". Language 48, 11-50. 1972. Andersen, Henning, "Markedness and the theory of linguistic change". In press, 2001. Anderson, John M., "Case." In: Brown & Miller eds. 1999, 58-65. Askedal, John Ole, ""Ersatzinfinitiv/Partizipersatz" und Verwandtes". Zeitschrift fr germanistische Linguistik 19, 1-23. 1991. Askedal, John Ole, "Kontrastive Analysen Deutsch-Norwegisch: eine bersicht". In: Helbig et al. eds. 2001, 351-8. Atmosumarto, Sutanto, Colloquial Indonesian . London, Routledge. 1994. Behaghel, Otto, Deutsche Syntax . Volume 4. Heidelberg, Winter. 1932. Beter, Marja, "Govorjenje na fanta v gorenjskem nareju." In: E. Krinik ed. 1998, 7180. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward Finegan, Longman grammar of spoken and written English . London, Longman. 1999. Bloomfield, Leonard, Language. New York, Holt & Co. 1933. Booij, Geert, Christian Lehmann and Joachim Mugdan (eds.), Morphologie: ein internationales Handbuch zur Flexion und Wortbildung . Volume I. Berlin, de Gruyter. 2000. Borkovskij, Viktor I., and Petr S. Kuznecov, Istorieskaja grammatika russkogo jazyka . Moscow, Nauka. 1965. Bornemann, Eduard, and Ernst Risch, Griechische Grammatik . Second edition. Frankfurt, Diesterweg. 1978. Boyle, John A., Grammar of Modern Persian . Wiesbaden, Harrassowitz. 1966. Boi, Ana, Nataa Komac, Petra Tomain, and Tina Verovnik, "Rodilnik in toilnik v zanikanju." Blapis 2, 2-10. 1998. Brandenstein, Wilhelm, Griechische Sprachwissenschaft. III. Syntax I . Berlin, de Gruyter. 1966. Braune, Wilhelm, Gotische Grammatik . Seventeenth edition, revised by Ernst A. Ebbinghaus. Tbingen, Niemeyer. 1966. Brown, Keith, and Jim Miller (eds.), Concise encyclopedia of grammatical categories . Amsterdam, Elsevier. 1999. Bhler, Karl, Sprachtheorie . Stuttgart, Fischer. 1982. Impression of the 1934 edition. Charisteria Gvilelmo Mathesio qvinqvagenario a discipulis et Circuli lingvistici pragensis sodalibvs oblata . Prague, Prask linguistick krouek. 1932. Collins Cobuild English Grammar. London, HarperCollins. 1990. Conti Jimnez, Luz, "Synchronie und Diachronie in der homerischen Kasussyntax." Paper abstract, 1999. Corbett, Greville G., "Agreement." In: Brown & Miller eds. 1999, 12-8.

220

REFERENCES

Coulmas, Florian (ed.), Direct and indirect speech . Berlin, Mouton de Gruyter. 1986. Coulson, Michael, Sanskrit. London, Teach Yourself Books. 1976. Croft, William, Typology and universals. Cambridge, Cambridge University Press. 1990. Cruse, D. Alan, "Number and number systems." In: Brown & Miller eds. 1999, 267-71. Delbrck, Bertold, Altindische Syntax . Darmstadt, Wissenschaftliche Buchgesellschaft. 1968. Impression of the 1888 edition. Diderichsen, Paul, Element r Dansk Grammatik . Copenhagen, Gyldendal. 1957. Dotter, Franz, Nichtarbitraritt und Ikonizitt in der Syntax . Hamburg, Buske. 1990. Dressler, Wolfgang U., Willi Mayerthaler, Oswald Panagl, and Wolfgang U. Wurzel, Leitmotifs in natural morphology . Amsterdam, John Benjamins. 1987. Dressler, Wolfgang U., "Naturalness." In: Booij et al. eds. 2000, 288-96. Duden. Grammatik der deutschen Gegenwartssprache . Sixth edition. Mannheim, Dudenverlag. 1998. Einarsson, Stefn, Icelandic. Grammar. Texts. Glossary . Baltimore, The Johns Hopkins Press. [1945] 1967. Engel, Ulrich, Deutsche Grammatik . Heidelberg, Julius Groos. 1988. Eroms, Hans-Werner, "Indikativische periphrastische Formen mit DO im Bairischen als Beispiel fr latente und virulente syntaktische Regeln." In: Wiesinger ed. 1984, 12335. Faarlund, Jan Terje, Svein Lie, and Kjell Ivar Vannebo, Norsk referansegrammatikk . Oslo, Universitetsforlaget. 1997. Fenk-Oczlon, Gertraud, "Frequenz und Kognition Frequenz und Markiertheit." Folia Linguistica 25, 361-94. 1991. Ferguson, Charles A., "Variation and drift: Agreement in Germanic." In: Huebner ed. 1996, 241-60. Friedman, Victor A., "Dialectal synchrony and diachronic syntax: The Macedonian Perfect." In: Steever et al. eds. 1976, 96-104. Givn, Talmy, Functionalism and grammar. Amsterdam, John Benjamins. 1995. Grande, Bencion M., Vvedenie v sravnitel'noe izuenie semitskix jazykov . Moscow, Nauka. 1972. Greenberg, Joseph H., Language universals. The Hague, Mouton. 1966. Greenberg, Joseph H. (ed.), Universals of human language . Volume 4: Syntax. Stanford, CA, Stanford University Press. 1978. Grevisse, Maurice, Le bon usage. Thirteenth edition, revised by Andr Goosse. Gembloux, Duculot. 1993. Hauser-Suida, Ulrike, and Gabriele Hoppe-Beugel, Die Vergangenheitstempora in der deutschen geschriebenen Sprache der Gegenwart. Mnchen, Hueber. 1972. Helbig, Gerhard, Lutz Gtze, Gert Henrici and Hans-Jrgen Krumm (eds.), Deutsch als Fremdsprache. Berlin, de Gruyter. 2001. Hendriks, Peter, The Radoda-Vevani dialect of Macedonian . Lisse, Peter de Ridder. 1976. Herrity, Peter, Slovene: A comprehensive grammar. London, Routledge. 2000. Hirt, Hermann, Handbuch des Urgermanischen. II. Stammbildungs- und Flexionslehre . Heidelberg, Winter. 1932. Hirt, Hermann, Handbuch des Urgermanischen. III. Abriss der Syntax . Heidelberg, Winter. 1934. Hock, Hans Henrich, Principles of historical linguistics . Berlin, de Gruyter. 1986. Hopper, Paul J., and Sandra A. Thompson, "Transitivity in grammar and discourse." 221

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

Language 56, 251-99. 1980. Hualde, Jos Ignacio, Catalan. London, Routledge. 1992. Huebner, Thom (ed.), Sociolinguistic perspectives. Papers on language and society, 1959-1994. Charles A. Ferguson. New York, Oxford University Press. 1996. Humbert, Jean, Syntaxe grecque. Paris, Klincksieck. 1972. Ivanov, Valerij V., Istorieskaja grammatika russkogo jazyka . Moscow, Prosveenie. 1964. Jacobs, Joachim, Armin von Stechow, Wolfgang Sternefeld, and Theo Vennemann (eds.), Syntax. Volume I. Berlin, de Gruyter. 1993. Jakobson, Roman, "Zur Struktur des russischen Verbums." In: Charisteria 1932, 74-84. Jakop, Tjaa, "Raba rodilnika pri zanikanju." Seminar paper, manuscript. Ljubljana, Filozofska fakulteta. 1998. Johannessen, Janne Bondi, Coordination. Oxford, Oxford University Press. 1998. Jovanoski, Natalie, "Raba rodilnika ob zanikanem povedku." Seminar paper, manuscript. Ljubljana, Filozofska fakulteta. 1997. Kenesei, Istvn, Robert M. Vago, and Anna Fenyvesi, Hungarian. London, Routledge. 1998. Kopriva, Silvo, Latinska slovnica . Maribor, Obzorja. 1989. Kovai, Irena, Milena Milojevi-Sheppard, Silvana Orel-Kos and Janez Orenik (eds.), Linguistics and language studies: Exploring language from different perspectives . Ljubljana, Filozofska fakulteta. 2000. Kozianka, Maria, "Negation und Modus im Germanischen." Paper abstract, 1999. Krause, Wolfgang, and Werner Thomas, Tocharisches Elementarbuch . Volume I. Heidelberg, Winter. 1960. Krger, Fritz, Einfhrung in das Neuspanische . Leipzig, Teubner. 1924. Krinik, Erika (ed.), XXXIV. seminar slovenskega jezika, literature in kulture . Ljubljana, Filozofska fakulteta. 1998. Latzel, Sigbert, Die deutschen Tempora Perfekt und Prteritum . Mnchen, Hueber. 1977. Li, Charles N., "Direct and indirect speech: A functional study." In: Coulmas ed. 1986, 29-45. Liberman, Anatoly, "The present perfect in Old and Modern Icelandic." In: Raible ed. 1990, 263-79. Lichtenberk, Frantisek, "Reflexives and reciprocals." In: Brown & Miller eds. 1999, 3139. Louis-Jensen, Jonna, and Jhan Hendrik W. Poulsen (eds.), The Nordic languages and modern linguistics 7. Proceedings of the Seventh International Conference of Nordic and General Linguistics in Trshavn, 7-11 August 1989 = Annales Societatis scienti arum froensis Supplementum XVIII. Trshavn, Frskaparsetur. 1992. Lunt, Horace G., A grammar of the Macedonian literary language . Skopje, Dravno knigoizdatelstvo na NR Makedonija. 1952. Lyons, Christopher, Definiteness. Cambridge, Cambridge University Press. 1999. Mare, Frantiek V., Diachronische Phonologie des Ur- und Frhslavischen . Mnchen, Sagner. [[1952]] 1969. Mayerthaler, Willi, Morphologische Natrlichkeit . Wiesbaden, Athenaion. 1981.English version: Mayerthaler 1988. Mayerthaler, Willi, "System-independent morphological naturalness." In: Dressler et al. 1987, 25-58. Mayerthaler, Willi, Morphological naturalness. Ann Arbor, Karoma. 1988. Mayerthaler, Willi, and Gnther Fliedl, "Natrlichkeitstheoretische Syntax." In: Jacobs et

222

REFERENCES

al. eds. 1993, 610-35. Mayerthaler, Willi, Gnther Fliedl, and Christian Winkler, Infinitivprominenz in europischen Sprachen . Teil I: Die Romania (samt Baskisch). Tbingen, Narr. 1993. Mayerthaler, Willi, Gnther Fliedl, and Christian Winkler, Infinitivprominenz in europischen Sprachen . Teil II: Der Alpen-Adria-Raum als Schnittstelle von Germanisch, Romanisch und Slawisch. Tbingen, Narr. 1995. Mayerthaler, Willi, Gnther Fliedl, and Christian Winkler, Lexikon der natrlichkeitstheoretischen Syntax und Morphosyntax . Tbingen, Stauffenburg. 1998. Milojevi-Sheppard, Milena, and Marija Golden, "Imperatives, negation and clitics in Slovene". In: Toporii ed. 2000, 93-109. Orenik, Janez, "Relative clauses are relatively simple." In: J. Louis-Jensen & Poulsen eds. 1992, 75-92. Orenik, Janez, "Naturalness: The preterite and present perfect tenses in German". In: I. Kovai et al. eds. 2000, 21-43. Plank, Frans, and Elena Filimonova, "The universals archive: A brief introduction for pro spective users." Sprachtypologie und Universalienforschung 53, 109-23. 2000. Posten, Jan Peter, "Der Ersatz- oder Scheininfinitiv". Wirkendes Wort 23, 73-85. 1973. Raible, Wolfgang (ed.), Erscheinungsformen kultureller Prozesse . Tbingen, Narr. 1990. Repina, Tamara A., Rumynskij jazyk. Moscow, Izdatel'stvo Moskovskogo universiteta. 1968. Rizzi, Luigi, Issues in Italian syntax . Dordrecht, Foris. 1982. Sandig, Barbara, Syntaktische Typologie der Schlagzeile . Mnchen, Hueber. 1971. Schachter, Paul, "Part-of-speech systems." In: Shopen ed. I 1985, 3-61. Schwyzer, Eduard, Griechische Grammatik . Volume II. Supplemented by Albert Debrunner. Mnchen, Beck. 1950. Shopen, Timothy (ed.), Language typology and syntactic description . Three volumes. Cambridge, Cambridge University Press. 1985. Siewierska, Anna, "Word order and linearization." In: Brown & Miller eds. 1999, 412-8. SSKJ = Slovar slovenskega knjinega jezika , 5 volumes. Ljubljana, Dravna zaloba Slovenije. 1970-91. [The Academy dictionary of Slovenian.] Snoj, Marko, Kratka albanska slovnica . Ljubljana, Filozofska fakulteta. 1991. Sokolova, Marija A., Oerki po istorieskoj grammatike russkogo jazyka . Leningrad, Izdatel'stvo Leningradskogo universiteta. 1962. Steele, Susan, "Word order variation: A typological study." In: Greenberg ed. 1978, 585623. Steever, Sanford B., Carol A. Walker, and Salikoko S. Mufwene (eds.), Papers from the Parasession on diachronic syntax, April 22, 1976 . Chicago, Chicago Linguistic Society. 1976. Stolz, Thomas, Sekundre Flexionsbildung. Eine Polemik zur Zielgerichtetheit im Sprachwandel. Two volumes. Bochum, Brockmeyer. 1992. Szantyr, Anton, Lateinische Syntax und Stilistik . Improved impression. Mnchen, Beck. [1965] 1972. Teleman, Ulf, Staffan Hellberg, and Erik Andersson, Svenska Akademiens grammatik . Four volumes. Stockholm, Svenska Akademien. 1999. Toporii, Joe, Slovenska slovnica . Maribor, Obzorja. 1976. Toporii, Joe, Slovenska slovnica . Fourth (revised and augmented) edition. Maribor, Obzorja. 2000. Toporii, Joe (ed.), Razprave XVII. Ljubljana, Slovenska akademija znanosti in umetnosti. 2000.

223

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

Tschernych, Pawel J., Historische Grammatik der russischen Sprache . Halle, Niemeyer. 1957. Tsunoda, Tasaku, "Transitivity." In: Brown & Miller eds. 1999, 383-91. Vaillant, Andr, Grammaire compare des langues slaves. III. Le verbe . Paris, Klincksieck. 1966. Wessn, Elias, Svensk sprkhistoria III. Grundlinjer till en historisk syntax . Stockholm, Almqvist & Wiksell. 1956. Wiesinger, Peter (ed.), Beitrge zur bairischen und ostfrnkischen Dialektologie . Gppingen, Kmmerle. 1984. Wright, Joseph, and Elizabeth Mary Wright, Old English grammar. Third edition. Oxford, Oxford University Press. 1925. Wurzel, Wolfgang Ullrich, Flexionsmorphologie und Natrlichkeit . Berlin, AkademieVerlag. 1984. Zemskaja, Elena A., Russkaja razgovornaja re': lingvistieskij analiz i problemy obuenija . Moscow, Russkij jazyk. 1979.

224

Index
A. Boi et al. 160 accusative = nominative/genitive 131 f., 150 f., 170-2 adjective as vocative 135 f. adjective modifier displaced 125 adjective v. cardinal numeral as attribute 112 adjective-noun gender agreement 116 f. adverbial particle + verb 117 f. affected object 56-7 affirmation v. negation/interrogation 38 affirmative reply 57 affirmative v. non-affirmative clause 33 Akan 58 Albanian 21-5 alignment rule 18 analytic construction 12 anaphora for two coordinated genders 84 f., 159 anaphora of personal pronoun + NP 210 f. Ancient Greek 75-9, 121, 128 f., 219 Andersen, Henning 18, 229 Anderson, John M. 43, 101 Andersson, Erik Teleman et al. answer to yes-no question 36-8, 193 Arabic 25-30 Arawak 30 'ask somebody for something' 125 f. Askedal, John Ole 9, 69 asymmetrical deduction 18 Atmosumarto, Sutanto 30 f. attributive v. predicative adjective 70-1 auxiliary haber v. other auxiliaries and modals in Spanish 185 auxiliary v. other verb 39 back agreement 120 f. Bahasa Indonesia 30-2 Bambara 32 basic element order 82-3, 117 'be' + action noun 148 f. Behaghel, Otto 48, 69, 115, 185, 199 Beter, Marja 164 Biber et al. 18 Biber, Douglas Biber et al. Bloomfield, Leonard 33 body part v. kinship relation (inalienable possession) 94, 96 Borkovskij & Kuznecov 115 Borkovskij, Viktor I. Borkovskij & Kuznecov Bornemann & Risch 75-6 Bornemann, Eduard Bornemann & Risch Boyle, John A. 106 Boi, Ana A. Boi et al. Brandenstein, Wilhelm 121, 128 f. Braune & Ebbinghaus 73-4 Braune, Wilhelm Braune & Ebbinghaus Breton 32 f. Browne, Wayles 9 Bhler, Karl 17, 228 calender date 110 Cantonese 36 cardinal numeral 27 f. cardinal numeral 'one' 122 f. cardinal numeral +/-unit counted 122 cardinal numeral in expressions of exact quantity 27, 196 cardinal numerals in languages showing eperspective 194

226

REFERENCES

cardinal v. ordinal numerals in dates 110 Carib 33 f. case agreement 198 f. Catalan 35 characterizing and identifying attribution 146 Chinese, Mandarin 35-8 classifier 193 f. clause of comparison 101 cleft sentence 39 clitic doubling 102 f. clitic form of personal pronoun 114 f., 172 closed (small) v. open (large) class 126 Collins Cobuild 12 comparative and superlative constructions 124 f. conflated v. separated genders 25, 123, 179 conflated v. separated meanings 23 conflated v. separated nominative and accusative 98 conflated v. separated personal and demonstrative pronouns 105 conflated v. separated verbal persons 188 conjugation of present indicative 42 f., 85 f., 188 conjunction of two time clauses 72-3 connatural 110 Conrad, Susan Biber et al. consequence-section of deduction 18 constructional iconicity 12 Conti Jimnez, Luz 79 Corbett, Greville G. 116, 120 corpus analysis 18 Coulson, Michael 118 Croatian 38-40 Croft, William 17, 52, 78, 92, 107 f., 194, 199, 206, 228 Cruse, D. Alan 27-30, 103, 195-8 Cvetko-Orenik, Varja 9, 125, 176 f. Cvetko-Rasmussen, Karin 9 Danish 40-2 Davis, Margaret G. 9 de-verbal noun 62-3 deduction 12 ff. definite & indefinite article 26 f. definite article 40 f., 80-2, 91 f., 104 f., 189 f., 192 definite article + adjective 145 f. 227

definite article + feminine name 87, 114 definite article + first name 64-5 definite article + personal name 35, 113 f. definite article + place name 111 f. definite article + surname 65 f., 87 f. definite object 191 f. definite v. indefinite indirect object 190 definiteness of adjective 142-4 definiteness of attribute + head noun 112 f. Delbrck, Bertold 119 demonstrative and personal pronouns 105 demonstrative pronominal 176-8 demonstrative v. interrogative pronominal 176 f. dependent clause with and without subordinator 88-9 desinences in verb conjugation 153 f. Diderichsen, Paul 40 f. direct object 13 f., 93 direct object in negative clauses 159-62 direct object in the genitive 147 f. distinction +/-definite v. lack of that distinction 142 f. distinction between accusative and nominative v. lack of that distinction 170 Dolgan, Milan 178 Dotter, Franz 11, 64, 66, 87, 163, 221, 223 Dressler et al. 11, 223 Dressler, Wolfgang U. 11, 223; Dressler et al. dual 27-30, 33 f., 73-5, 78, 129 f., 166 f. dual v. plural as expression of duality 130 Duden Grammatik 69-71, 221 Dutch 68, 70, 73 Dyirbal 94, 96 easy to access v. difficult to access (referent) 80 Ebbinghaus, Ernst A. Braune & Ebbinghaus Einarsson, Stefn 84 f. element order 158, 162 element order in main and dependent clauses 68 element order in verb chains 69-70 element order when subject is not clauseinitial 162 ellipsis 12 Engel, Ulrich 63

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

English 12 f., 15, 60, 108, 121, 145, 154, 198 equative clause 201-4 ergative v. other verb 174 Eroms, Hans-Werner 67 Estonian 43 ethical dative 154 f. everyday life v. life on stage 75 existential clause v. other clause 193 Faarlund et al. 104 Faarlund, Jan Terje Faarlund et al. female speech in first person plural masculine 75 feminine personal names in -a in Romanian 114 Fenk-Oczlon, Gertraud 31, 46, 48, 50, 70, 99, 106, 138, 144, 156, 160, 190, 216 Fenyvesi, Anna Kenesei et al. Ferguson, Charles A. 17, 42, 188, 228 Ferme, Alja 158, 162 Filimonova, Elena Plank & E. Filimonova Finegan, Edward Biber et al. finite verb + desinence - 153 finite verb of main clause 119 f. Finnish 43-6 first grammatical person 25 first person singular v. other persons 85 first v. second conjunct in coordination 72 Fliedl, Gnther Mayerthaler & Fliedl, Mayerthaler et al. foregrounding 220-1 format of deduction 14 format of naturalness scale 11 f. French 46-62 frequency adverbs 12 f. frequent lexical items 216 Friedman, Victor A. 103 Frisian 68 future tense 38-42 future tense for present tense 162 f. future v. present as expression of future time 39, 41 gender 1 + gender 1 v. gender 1 + gender 2 84, 159 gender of personal pronoun 151-3, 178-80 genitive v. other case dependent on noun 55 228

genitivus explicativus 63 genitivus subiectivus and g. objectivus 146 f. German 17, 60, 62-73, 128 f., 213-22 Germanic languages (old) 73 Givn, Talmy 93, 185, 198, 221 Golden, Marija M. Milojevi-Sheppard & M. Golden Gothic 73-5 grammatical number 16 grammatical person 12, 17, 213-5 Grande, Bencion M. 25, 27 Grdina, Igor 9 Greenberg, Joseph H. 17, 27, 33, 73 f., 78, 84, 95, 153 f., 166 f., 209, 211, 215, 228 Grevisse, Maurice 46-51 Guaran 79-80 Hausa 80-1 Hauser-Suida & Hoppe-Beugel 213, 215, 219 Hauser-Suida, Ulrike Hauser-Suida & Hoppe-Beugel head-modifier agreement 28-30, 199 Hebrew 192 Hellberg, Staffan Teleman et al. Hendriks, Peter 103 Herrity, Peter 60, 121, 123, 138, 140, 1425, 149-53, 155, 159, 166-7, 170-3 Hidatsa 81-2 Hirt, Hermann 74 Hock, Hans Henrich 49-51, 117, 150 honorific v. non-honorific personal pronoun 89, 163 Hoppe-Beugel, Gabriele Hauser-Suida & Hoppe-Beugel Hopper & S. Thompson 56, 83, 168 f., 186 Hopper, Paul J. Hopper & S. Thompson hours of the day 130 f. Hualde, Jos Ignacio 35 human direct object 187 f. Humbert, Jean 75 Hungarian 82-3 'I and he' v. 'I and you' 33 Icelandic 84-6 imperative 89 f., 121 f., 127f., 207 f. imperative + accusative pronoun 60 imperative = conditional protasis 128 f.

REFERENCES

imperative v. lack of imperative in commands 127 'in whose honour' 127 inanimate direct object of infinitive 115 f. inchoative aktionsart 186 f. indefinite article 40 f., 97 f., 180-4 indefinite article + proper name 109 indirect object 190 Indo-Iranian languages (old) 219 infinitive v. that-clause 98 f. inflected v. non-inflected noun 149 inflection of parts of speech 204 f. inflection of surnames 149 f. interrogative clause 30-2, 219 f. interrogative pronominal 176-8 intimate v. non-intimate human relations 165 intransitive past participle 118 f. Italian 86-90, 121 Ivanov, Valerij V. 115 Ivanov, Vyacheslav V. 166 Jakobson, Roman 17, 58-60, 64, 74, 90, 100, 105, 118, 136, 152 f., 165, 201, 211, 214 f., 228 Jakop, Tjaa 160 f. Jakopin, Franc 9 Japanese 90-1 Johannessen, Janne Bondi 72 f. Johansson, Stig Biber et al. Jovanoski, Natalie 160 f. Kavi, Domen 9 Kekchi 91-2 Kenesei et al. 82 Kenesei, Istvn Kenesei et al. Kharia 92 f. KinyaRwanda 93 Komac, Nataa A. Boi et al. Kopriva, Silvo 98, 100 Kozianka, Maria 73 Krause & Thomas 166 Krause, Wolfgang Krause & Thomas Kriaj-Ortar, Martina 141 Krger, Fritz 187 Kuznecov, Petr S. Borkovskij & Kuznecov l-tenses v. perfect tenses in Macedonian 103 Lakhota 93-8

Lappish 98 Latin 98-100, 105, 121, 126, 128 f. Latzel, Sigbert 213, 216-21 Leech, Geoffrey Biber et al. lexical item 'perhaps' 132 f. Li, Charles N. 61 Liberman, Anatoly 221 Lichtenberk, Frantisek 57, 169 Lie, Svein Faarlund et al. Lindeman, Frederik Otto 32 literal v. non-literal reading 180 Lunt, Horace G. 103 Lyons, Christopher more than 70 mentions Macedonian 102-3 main argument of adjective 141 f. main arguments of verb 141 f. main v. dependent clause 39, 68 f., 119, 203 Mare, Frantiek V. 18, 229 marked 11 markedness 18 markedness agreement 18 Marvin, Tatjana 154, 173 Mayerthaler & Fliedl 11, 223 Mayerthaler et al. more than 180 mentions Mayerthaler, Willi more than 570 mentions; Dressler et al., Mayerthaler & Fliedl, Mayerthaler et al. mediopassive 76-7 mend v. mnda in Slovenian 132 Menomini 33 Mikli, Tjaa 156 Milojevi-Sheppard, Milena 9, 58, 186; M. Milojevi-Sheppard & M. Golden M. Milojevi-Sheppard & M. Golden 134 Miriam 103-4 modifier + noun v. noun + modifier 184 modifier v. full NP as repetition 131 morphosyntactic variant 15 Muhlstein, Tom 9 name of geographical region 150 name of street/square 46 f. natural 11 natural class 15 f., 17, 38, 55, 64, 79, 85 f., 124, 179, 181, 200 naturalness scale 15-7 Naturalness Theory 11, 15 and many men-

229

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

tions Neak-Lk, Albina 9 negated imperative as prototypical means of expressing negative command 129 negation 47-52, 133 f. negative context v. other context 55 negative prefix in compound verbal tense 105 f. neutral register v. air of familiarity 66 nominative singular + desinence 61-2, 84 nominative-accusative frame v. dative-nominative frame 91 non-third person pronouns v. other NP 79 Norwegian 73, 104 f. number distinction in pronouns and nouns 36, 194 f. number distinction v. lack of that distinction 92, 194 number in pronouns and animate nouns 92 f. number marking v. lack of number marking 79 number of the subject on the verb 103 f. objective and subjective conjugations 83 obligatory v. optional number marking in nouns 30 Old English 17, 42 f., 68 Old Swedish 17, 188 optative of (plu)perfect 75-7 order of subject and verb in interrogative clauses 31, 51 Orenik, Janez 213, 219, (address of) 9 oui v. si as affirmative reply in French 57 ownership v. non-ownership in possession 45 Paiute 105 Panagl, Oswald Dressler et al. partial v. total involvement of the patient 167 f. partitive case 55 f. passive 216-8 past participle of ergative verb 173 past tense 32 past tense v. present tense in narrative 178 patient v. non-patient (direct object) 93 periphrastic conjugation 67 Persian 105-6 personal pronoun 53, 63, 79 f., 90 f., 123 f.

personal pronoun as affix 206 personal pronoun of address 89, 163 f. phrasal verb 144 f. Plank & E. Filimonova 15, 227 Plank, Frans Plank & E. Filimonova plural 28-30, 195-8 plural v. dual 200 plural verb agreement of animate nouns 107 f. plural verb agreement of inanimate nouns 108 plural, inclusive/exclusive 33 f. plurale tantum 66-7 plurality v. collectivity 197 Poganik, Vladimir 9 possessive construction 21 ff., 34, 43-6, 535, 86 f., 93-7, 138-41, 195 possessive (morphology) 77 f. possessive genitive 47 possessive pronoun 99 f. Posten, Jan Peter 70 praesens pro futuro 41 f. precise v. imprecise meaning 198 predicative noun 43 predicative NP v. other NP 181 prediction made by deduction 13 preposed v. postposed adjective 21 preposition + case 79 preposition + definite article 110 f. present tense in narrative 178 preterite v. present perfect tense 213-21 pronominal possessive 25 f. pronoun v. noun 205 proper noun and definiteness 23-5 Puget Salish 208 Quechua 106-7, 206 question tag 60-1 Quich 107-8 Quileute 109 reciprocal pronoun 58-9 reflexive anaphora 168 f. reflexive passive 186 reflexive pronoun 57-9 reflexive v. non-reflexive pronoun in expressions of reflexivity 57, 169 reflexive v. non-reflexive pronominal possessive 100, 140 f. reflexive v. reciprocal use of pronoun 58

230

REFERENCES

relative clause 174-6, 218 f. repetition of question in answer 37, 193 repetition v. its original 60 Repina, Tamara A. 114 right v. left branching 69, 127 Risch, Ernst Bornemann & Risch Rizzi, Luigi 88 Romance languages 109-10 Romanian 110-5 Russian 115-7 S and V separated v. S and V together 68 Sandig, Barbara 62 Sanskrit 117-20 Schachter, Paul 36 f., 57 f., 60, 69, 98, 105 f., 193, 206 Schwyzer, Eduard 75, 219 second grammatical person 207 second person personal pronoun 89 sem-value 11, 16 Shigemori-Buar, Chikako 9 Siewierska, Anna 82, 117, 208 Skubic, Mitja 9, 115 Slovenian 58, 60, 120-78 Slovenian Academy of Sciences and Arts 9 Slovenian Theory 11 f., 17 f. Slovenian Theory & non-linguistic behaviour 19 Smith, John C. 9, 42, 188 Snoj, Marko 22 f. Sokolova, Marija A. 115 Spanish 56, 178-88 spatial source of movement 101 f. specificity of indefinites 210 SSKJ 132 stative 216-8 stative v. passive 217 Steele, Susan 31 Stolz, Thomas 11, 223 strong v. weak declension of adjective 188 f. stylistically marked v. unmarked 178 subject + unaccusative verb 35 f. subject agreement 32 f., 136-8, 199 f. subject agreement of female persons 164-6 subject agreement v. its lack 199 subject agreement v. object agreement 109 subject and object as pronominal affix 208 subject and object personal pronouns 106 f.

subject complement of middle-voice reflexive verb 134 f. subject of 'be' 155 f. subject pronoun 59 subject v. subject complement as controller of agreement 120 subjunctive of (plu)perfect 75-7 subsystem of appeal 17, 89 suppletion in comparison of adjectives 138 Swedish 104, 188-91 sym-value 11 synchronic explanation 18 syntactic variant 15 syntax of object clitic 185 f. syntax of quantifier 172 f. Szantyr, Anton 101 Tadzhik 191-2 Tagalog 192-3 Teleman et al. 190 Teleman, Ulf Teleman et al. tenses of copular 'be' 103 Thai 193-4 the type Gorenjsko v. the type Gorenjska in Slovenian 150 third grammatical person Jakobson, Roman 'this country' 184 f. Thomas, Werner Krause & Thomas Thompson, Sandra A. Hopper & S. Thompson Tiwi 194-5 Tomain, Petra A. Boi et al. topic v. comment 200 f. Toporii, Joe 60, 120-3, 125, 128-30, 137 f., 140, 142-5, 147, 149 f., 151-3, 155 f., 159-61, 166-8, 170-3, 176 f., 224 total v. partial involvement of participant 169 f. trial and paucal 206 f. Tschernych, Pawel J. 115 Tsunoda, Tasaku 91 Turkish 195-8 two entities pairwise 166 University of Klagenfurt 11 Vago, Robert M. Kenesei et al. Vaillant, Andr 38 f. Vannebo, Kjell Ivar Faarlund et al.

231

A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS

variant 11, 15 verb + direct object v. direct object + verb 161 verb + displaced subject v. subject + verb 88 verb + particle 68-9 verb agreement with object 109 f. verb v. noun 26, 142, 195 verb-internal negation 156-8 verbal aspect 41 f. Verovnik, Tina A. Boi et al. visible v. invisible gender 53, 64, 116, 151 f. vocative 24, 52, 209 f. vocative of person name 24 vocative v. non-vocative function of nominative 135 volitionality 91 Walbiri 198-9 Welsh 199-200 Wessn, Elias 188 Wikchamni 200 Winkler, Christian Mayerthaler et al. Wright & E. Wright 42 Wright, Elizabeth Mary Wright & E. Wright Wright, Joseph Wright & E. Wright Wurzel, Wolfgang U. 11, 223; Dressler et al. Zapotec 208 Zemskaja, Elena A. 117 Zorman, Marina 9 ivanovi, Sao 128, 158, 162

232

ISBN 961-6242-42-3 JANEZ ORENIK A PREDICTABLE ASPECT OF (MORPHO)SYNTACTIC VARIANTS PREDVIDLJIV VIDIK (OBLIKO)SKLADENJSKIH DVOJNIC Izdala Slovenska akademija znanosti in umetnosti v Ljubljani

Jezikovni pregled Varja Cvetko Orenik

Stavek in prelom Alenka Maek Tisk Tiskarna Skuek v Ljubljani

Ljubljana 2001

You might also like