You are on page 1of 13

How Liberals and Conservatives Think:

Evidence on Lakoffs Theory


in the Rhetoric of Political Advertising

Glenn W. Richardson Jr.


Department of Political Science
Kutztown University of Pennsylvania
P.O. Box 730
Kutztown, PA 19530
(610) 683-4450
richards@kutztown.edu

Paper prepared for the 64th Annual National Meeting of the


Midwest Political Science Association
Palmer House Hilton, Chicago, Illinois
April 20-23, 2006.

The author wishes to thank the Department of Political Science, the College of Liberal Arts and
Sciences, and the Professional Development Committee at Kutztown University of Pennsylvania
for their support of this research.

Contemporary American politics is characterized by a partisan divide


unequalled in recent history. While popular discussion of states red and blue
oversimplifies the nations geopartisan cleavages, one doesnt have to look far to find
evidence that Democrats and Republicans possess sharply divergent views of the
political world.
Consider a report in the New York Times on the results of recent polling by the
Pew Research Center. According to the polls, conducted in early March of 2006, fully 69
percent of Republicans said the U.S. military effort in Iraq was going very well or
fairly well. The Democratic response to the same question stands the Republican
result on its head: only 30 percent of Democrats said the military effort in Iraq was
going very well or fairly well (Kirkpatrick and Nagourney 2006:A12).
While partisan disagreement has been an enduring feature of the republic, a
cursory examination of the partisan disparity in presidential job approval indicates that
the current divide exceeds historical precedent. Presidential scholars George Edwards
and Stephen Wayne note that the average difference in presidential support between
Democrats and Republicans over the past four decades has been nearly 40 percentage
points. That divide expanded to more than 50 percent during the Reagan and Clinton
presidencies (Edwards and Wayne 2003:114). An April, 2006 CBS News poll, however,
found a 65 percentage point gap between Democratic and Republican identifiers on
President Bushs job approval (www.pollingreport.com/BushJob1.htm; accessed April
17, 2006).
Cognitive linguist George Lakoff (2002) argues that liberals and conservatives
not only view the world differently, but in fact possess fundamentally different

2
worldviews. For Lakoff, Americans worldviews flow from the conceptual metaphors
they use to make sense of politics. The inherent complexity of the political system and
the limited knowledge available to individual citizens makes reasoning about politics
directly too difficult. Instead, people rely on metaphors, which allow them to reason
about what they dont know through analogy with what they do know. Specifically,
Lakoff argues that both liberals and conservatives share a common metaphor for the
political system: the country is a family, politicians are the parents, and citizens are the
children. What distinguishes liberals from conservatives is the particular model of the
family that each embraces.
Conservatives adhere to a strict father model of the family. Liberals, by
contrast, hold a nuturant parent model of the family. For conservatives, the world is a
dangerous place. People, left to their own devices, seek to satisfy their own desires, but
will respond to reward and punishment (Lakoff 2002:65-67). Indeed, Lakoff describes
the morality of reward and punishment as the most basic assumption of the strict father
model (2002:67). In the liberal model of the family, parental nurturance contributes to
self-nurturance among children, and eventually, the ability to nurture others (2002:110).
Lakoff is most compelling when arguing that only by reference to these
metaphorical models of the family can one fully explain (1) why certain stands on issues
go together, (2) why puzzles for liberals are not so for conservatives (and vice versa),
and (3) why liberals and conservatives make characteristic choices of topics, words, and
forms of reasoning (2002:33). Liberals, Lakoff notes, are often bewildered by
conservative claims to be pro-life (on abortion) while simultaneously supporting the
death penalty. Strict father morality allows one to square this circle. The unborn are
innocent and deserve protection, while those sentenced to death have earned their

3
punishment. The centrality of reward and punishment in strict father morality explains
both stances where other potential explanations do not.
In fact, Lakoff identifies three analytical failures of liberalism: the notions that
conservatives merely adhere to the ethos of selfishness (Lerner 1994), that
conservatives just believe in less government, and that conservatism is no more than a
conspiracy of the ultra-rich to protect their privilege (Lakoff 2002:143). If, for example,
conservatives were driven by selfishness, then they should respond to pragmatic liberal
appeals to their economic self interest, which they do not. The sustained growth in
government under the all-Republican regime of the early 21st century puts the lie to the
notion that conservatives simply favor small government. While acknowledging much
merit in the cynical liberal claim that conservatism is a conspiracy of the ultra-rich,
Lakoff notes that there are too many idealistic conservatives for this to be completely
true (2002:147). In short, Lakoff presents a plausible interpretation and analysis of one
of the defining features of contemporary American politics, its partisan and ideological
divisiveness.
Part of Lakoffs claim is that the competing models of the family generate
characteristic patterns of reasoning, topic choice, and use of words. Lakoff offers
examples of conservative and liberal editorials demonstrating such characteristic
patterns of language and reasoning (see for example 2002:5-6). One might logically
expect such patterns to manifest themselves in other political communication as well,
notably political advertising.
Political advertising has become one of the principal vehicles of political
communication in the televisual age. Ads are directed at a mass audience, and attempt
to communicate a lot in a short time. Accordingly, political advertisements would seem

4
to be fertile ground for harvesting evidence of the characteristic worldviews and
language Lakoff argues defines contemporary American politics and how liberals and
conservatives think.
Political advertising has attracted an extremely voluminous scholarly literature.
While the bulk of work has been devoted great effort to analyzing campaign ads in
terms of their tone (positive/negative) and content (issue/image), some researchers
have trained their focus upon partisan differences in advertising appeals. William
Benoit (1999:179) analyzed thousands of presidential campaign ads between 1952 and
1996 and found that Republican ads were more likely to acclaim (tout the virtues of
their candidate) than were Democratic ads (64 percent compared to 57 percent).
Similarly, Democratic spots were more likely to attack than GOP commercials (39
percent compared to 35 percent).
Darrell West (2005:46) analyzed 452 prominent presidential ads between 1952
and 2004. He found that Republican ads included more specific pledges than
Democratic ads (26 percent to 21 percent), were more likely to emphasize international
affairs than Democratic ads (10 percent to 3 percent), but were less likely to emphasize
personal qualities (29 percent compared to 40 percent in Democratic ads).
Roderick P. Hart (2000) analyzed a huge database of political communication
(the Campaign Mapping Project) including ads, speeches, debates, and news coverage
and found that the parties have distinctive political voices (p. 160). Democratic
rhetoric proved to be grounded (high in realism and familiarity), humanistic (high in
commonality and voter references) and institutional (high leader and party references).
Republican rhetoric outscored Democratic rhetoric in the language of inspiration,

5
liberation, patriotism and religion. Hart also found a weakening in partisan style over
time.
DATA
To explore partisan variation in campaign advertising, I collected the scripts of
660 political advertisements run between 1990 and 2000 by searching the Lexis-Nexis
database for news stories including the words ad watch, adwatch or spot check.
Both television and radio advertisements were included in the sample; 95 percent of the
ads were television ads. I included all scripts that appeared to be complete, and omitted
those that appeared to be partial transcripts. The ads were spread across several levels
of elective office, with U.S. Senate and governor accounting for the largest share of spots.
Table 1 breaks the ads down by office (including ads for and against ballot initiatives
and referenda).
Table 1. Ads by Office
Office

Number

Percent

President

84

12.7

U.S. Senate

217

32.9

Governor

218

33.0

Statewide (non-governor)

40

6.1

U.S. House

87

13.2

Local/State Legislature

10

1.5

Ballot Measure

.3

Generic Party Ads

.3

660

100.0

Total

6
The ads are split fairly evenly between Democratic candidates (339 ads, 48.3 percent)
and Republicans (321 ads, 45.7 percent).
METHODS
The computer program DICTION was used to analyze the ads in this study.
DICTION uses a series of dictionaries to search a passage for five semantic features
Activity, Optimism, Certainty, Realism and Commonalityas well as thirty-five
subfeatures (Hart 2000: 245). It also allows users to create up to ten custom
dictionaries. Both DICTIONs word lists and custom dictionaries created specifically for
this research were used.
To test Lakoffs claims about liberal and conservative word choice, two
DICTION custom dictionaries were created. Based on Lakoffs listing of characteristic
liberal language (2002:30), the following words were used to constitute the liberal
language custom dictionary: social, forces, expression, human, rights, equal, concern,
care, help, health, safety, nutrition, dignity, oppression, diversity, deprivation,
alienation, corporations, corporate, welfare, ecology, ecosystem, biodiversity, [and]
pollution. Based on Lakoffs listing of characteristic conservative language (2002:30),
the following words were used to constitute the conservative language custom
dictionary: character, virtue, discipline, tough, strong, self-reliance, self-reliant,
individual, responsibility, backbone, standards, authority, heritage, competition, earn,
hard, work, enterprise, property, reward, freedom, intrusion, interference, meddling,
punishment, traditional, dependency, self-indulgent, elite, quotas, breakdown, corrupt,
decay, rot, degenerate, deviant, [and] lifestyle.
In order to assess whether there are significant differences between the campaign
language of candidates of different parties, a series of ANOVA tests were performed.

7
Because of the generally exploratory nature of the hypotheses and the likelihood that
extraneous factors would be randomly distributed and not systematically related to the
variables of interest, simple bivariate models are appropriate.
FINDINGS
The first test concerns whether the language of Democratic and Republican
candidates differs or is largely similar in terms of their use of liberal words as identified
by Lakoff. Table 2 presents the scores of Democratic and Republican candidates on the
liberal language custom dictionary.
Table 2. Liberal Language Custom Dictionary Scores by Party
Party

Mean

Std. Deviation

Democratic

1.08

339

1.716

Republican

.65

321

1.244

Total

.87

660

1.519

ANOVA Table

Liberal
Language *
Party

Sum of
Squares

df

Mean
Square

Significance

13.38

.000

Between
Groups
(Combined)
Within
Groups

30.281

30.281

1489.772

658

2.264

Total

1520.053

659

Eta = .141 Eta Squared = .020


Democratic candidates clearly used more words from the liberal language
custom dictionary in their advertising than Republican candidates did. DICTION
extrapolates small files (less than 500 words) to a 500-word norm. The average

8
Democratic candidate ad used just over one liberal word, while the average Republican
ad used, on average, one liberal word in every two ads. The ANOVA test indicates that
the difference was statistically significant. The small Eta squared value reflects the low
raw scores resulting in muted variation on this measure.
Table 3 presents the scores of Democratic and Republican candidates on the
conservative language custom dictionary.
Table 3. Conservative Language Custom Dictionary Scores by Party
Party

Mean

Std. Deviation

Democratic

.53

339

1.05

Republican

.56

321

.95

Total

.54

660

1.00

ANOVA Table

Conservative
Language *
Party

Sum of
Squares

df

Mean
Square

Significance

.178

.67

Between
Groups
(Combined)
Within
Groups

.178

.178

659.366

658

1.00

Total

659.544

659

Eta = .016 Eta Squared = .000


Once again, the mean scores for the use of partisan words varies in the expected
direction, with Republican candidates proving more likely than Democratic ones to use
conservative language. The difference here, however, is substantively and statistically
insignificant.

9
DISCUSSION
The data presented here offer mixed evidence of distinctive partisan patterns in
the use of liberal and conservative language. Democrats, in their ads, clearly use more
liberal words than do Republicans in their ads. Consider this ad, Cause, from Hillary
Clintons 2000 New York Senate campaign, which had the highest score for liberal
language of any ad in this study:
She's talking about health care, because it's always been her cause.
When her health care plan was defeated, she kept on fighting step by step and made a
difference. Mammograms covered by Medicare. Longer hospital coverage for new
mothers.
In the Senate, Hillary will fight to provide health care access to every child, for a real
Patients' Bill of Rights and tax credits to help those without coverage on the job. Who'll
be a strong voice for better health care?
Hillary. Always has been, always will be.
The ad could be read as Lakoffs nurturant parent model personified. Longer hospital
coverage for new mothers, health care access to every child, a real Patients
Bill of Rights, help, and, of course, health care.
Not all ads scoring high in liberal language, however, were sponsored by
Democrats. Oliver L. Norths 1994 ad Mean, for example, scored nearly as high in
liberal language as Clintons Cause:
You've paid into Social Security most of your life. You're counting on it being there. It's
your money. But Chuck Robb thinks that the government ought to get some of it before
you do.
That's right. Chuck Robb cast the deciding vote in favor of Bill Clinton's five-year, $ 25
billion tax increase on Social Security benefits for seniors.
But higher Social Security taxes won't be all if Chuck Robb has his way. Robb says he is
willing to cut Social Security benefits. He wants higher taxes and cuts on your Social
Security.
Chuck Robb wants to break the Social Security contract with seniors . . . but he didnt
mind voting to raise his own pay $ 23,000 a year. He's dipping into the Social Security
'trust fund' to pay for new liberal programs.

10

Chuck Robb didn't stand up for seniors, and he won't protect Social Security. Ollie
North will fight to preserve Social Security so it is there when you need it.
Make the change - send Ollie North to the U.S. Senate.
Norths ad co-opts a normally Democratic stronghold, Social Security, while also
launching a more conventional conservative attack on higher taxes and liberal spending.
Overall, this study turned up little difference in the use of conservative language
between Democratic and Republican ads. Perhaps this reflects the sense among
Democrats during the period encompassed by this study that a conservative tide had
swept the nation. Or it may just be that there are now more conservative Democrats
than there are liberal Republicans. Consider this ad from Zell Millers 1994 campaign,
Simpler:
You know, somebody said that as parents, we've been so eager and worked so hard to give
our children what we didn't have that sometimes we failed to give them what we did have
- like values . . . [h]ard work and discipline and respect for other people's property. These
are the values that have guided him as governor. Hard work. Students who work hard
and keep B averages can go to college tuition-free. Discipline. He's brought computers
into the classrooms - and is getting drugs and guns out. Self- reliance. Zell Miller is
moving people from welfare to work. Respect for the law. Two-time violent felons will go
to prison for life. Zell Miller calls it, 'Two strikes and you're out.'
Simpler scored higher on conservative language than any other ad in this study. Of
course, Millers rightward drift earned him the nickname What the Hell, Zell, and his
unhinged assault on Democrats at the 2004 GOP convention (and subsequent challenge
to duel MSNBCs Chris Matthews) qualify him as a Republican in all but name.
CONCLUSION
The evidence presented above provides mixed support for the claim that
Democrats and Republicans use characteristic language (based on their conceptual
metaphors for politics) in their campaign ads. The liberal voice does appear to ring true

11
in Democratic spots but not Republican ones, but the use of conservative words does not
appear to adhere to a partisan framework.
There are many potential explanations for these results. The limitations of
content analysis software (and DICTION in particular) have to be acknowledged. More
significantly, Lakoff argues that conservatives do more than use certain words. They
also use certain modes of reasoning, and this study was not designed to assess modes of
reasoning in ads. It may also be that Democratic ads consciously embrace appeals to
conservative values. That said, it should be noted that even Republican spots used
about half as many conservative words as Democratic ads used liberal words.
Ultimately, however, the fact that liberal language is characteristic of Democratic
ads but not Republicans, while no such partisan pattern obtains in the use of
conservative language is something of a puzzle. Especially since Lakoff argues that
conservatives are much more self-aware of the role of the strict father model than
liberals are of their nurturant parent model. One possible explanation can be drawn
from Ted Braders finding that positive ads are the most polarizing, firing up both ones
base and the opposition (Brader 2005). Interestingly, Braders study was based on
Massachusetts voters, only one in ten of which are Republicans. That is, there may be
partisan differences not only in the language used in ads but also in how partisans
respond to ads. That would explain why, in a world of focus group tested advertising,
only one partys ads embrace characteristic ideological language. It remains for future
research to explore this and other possible resolutions to the paradox of asymmetric use
of ideological language in partisan advertising.

12
REFERENCES

Benoit, William L. 1999. Seeing Spots: A Functional Analysis of Presidential Television


Advertisements, 1952-1996. Westport, CT: Praeger.
Brader, Ted. 2005. Campaigning for Hearts and Minds: How Emotional Appeals in Political
Ads Work. Chicago: University of Chicago Press.
Edwards, George C. III and Stephen J. Wayne. 2003. Presidential Leadership: Politics and
Policy Making (6th ed.). Belmont, CA: Wadsworth/Thompson Learning.
Hart, Roderick P. 2000. Campaign Talk: Why Elections are Good for Us. Princeton:
Princeton University Press.
Kirkpatrick, David D. and Adam Nagourney. 2006. In an Election Year, a Shift in
Public Opinion on the War. New York Times. March 27. A12.
Lakoff, George. 2002. Moral Politics: How Liberals and Conservatives Think. Chicago:
University of Chicago Press.
Lerner, Michael. 1994. Tikkun, November/December pp. 12, 18 (cited in Lakoff)
West, Darrell M. 2005. Air Wars: Television Advertising in Election Campaigns, 1952-2004.
Washington, D.C.: CQ Press.

You might also like