Language and (Social) Thought

Issue: Are thoughts and thought processes influenced, constrained, or determined by the language one uses?

Note: Implications for modularity; if language is a separate module that developed relatively late (e.g., Chomsky, Pinker), then language should have relatively little effect on thought

Whorf-Sapir Hypothesis

Whorf - amateur anthropologist and linguist (Native American languages). Contended that our experience and representation of the world is a function of our language:

"We cut nature up, organize it into concepts, and ascribe significance as we do, largely because we are parties to an agreement to organize it in this way - an agreement that holds throughout our speech community and is codified in the patterns of our language" (1956)

Example: Hopi have no imaginary plurals (e.g., 10 days); time not objectified as in English

200 words for snow? (Some shoddy research)

Linguistic determinism (language determines thought) or linguistic relativity (language and thought are related)

Early Tests

Carrol & Casagrande (1958): matching task; Navajo-dominant (language emphasizes structure) more likely to categorize based on form than English dominant. But English speakers from Boston most likely to categorize based on form (cited by both supporters and detractors)

Color perception and memory

Within-language effect: color codability (judgment consensus) affects memory (Brown & Lennnenberg, 1954)


Rosch (1973); Dani have 2 color words but shown enhanced memory for primary colors (color spectrum impacts memory independent of language).

Berlin & Kay (1969): underlying, universal, logic to assignment of color names (black,white, red). Cross-linguistic agreement on prototypical examples of color categories.


Primary colors are more distinctive and hence memorable (Lucy & Sheweder, 1979)

Kay & Kempton (1984) Categorical perception by English speakers (lexical items for green and blue) but not Tarahumara speakers (no blue - green lexical items)(Linguistic relativity)

1 2 3 4|5 6 7 8

Overall: Conflicting results; color perception/memory may not be good candidate because of physiological basis of color perception


Language differences in number lexicalization; English vs. Chinese differences and relationship to children's number learning

Counterfactual reasoning (Bloom; 1981)

English: lexical devices for expressing counterfactual statements;

e.g., past tense and subjunctive:

"If it hadn't rained we would have played ball."

Bloom argues no parallel constructions in Chinese (translate: "It rained

yesterday so we did not play ball")

Example 1:

English and Chinese speaking students read:

"If Brier could have read Chinese, he would have discovered the relevance of Chinese philosophy for his own work"...............

____ He led Western philosophy one step closer to Chinese philosophy

____ None are appropriate (correct response; N. Americans 96- 98%; Chinese: 6-63%)

Example 2:

If all circles were large, and this small triangle was a circle, would it be large?

____Yes (85% N. Americans; 25% Taiwanese)

____ No

Bloom criticized by Au (83) and others for using less idiomatic Chinese expressions; However, Au's work used English-Chinese bilinguals

Counterfactual reasoning possible in Chinese (e.g., regret, hope, etc. are basic) but possibly more difficult, relative to other languages

Person Perception - differences as function of available lexical items (good candidate for linguistic effects due to ambiguity of person perception)

Hoffman et al (1986): Chinese-English bilinguals read Chinese/English descriptions consistent with one-word label in Chinese (shi gu) or English (artistic).

Ss impressions and memory were schema congruent (e.g., false recognition) if read in language with descriptor

Fundamental Attribution Error - No research, but effect is:

culture specific (Miller, 1984), and

possibly related to available lexicon (more descriptors in English)

Language Use

Examine effects of language use (rather than language as an abstract system) on cognitions.

Major comparison is effects of using vs. not using language (rather than cross-linguistic comparison) on cognitive processes

Carmichael et al - memory task: reconstruction of ambiguous stimuli affected by label (glass or dumbbell). But: language effect (encoding) or retrieval effect (label recalled and used to reconstruct)

Verbal overshadowing (Schooler); Ss who verbalized descriptions of stimuli (e.g., color chips, faces) had poorer memory for stimuli (verbal representation interferes with visual representation)

Note: describing can enhance verbal memory (Bahrick)

Loftus (1977) and memory:

- 'blue' car affected memory

- affects memory trace or retrieval bias?

Schematic effects - e.g., Bransford, Cohen, Schank & Abelson, etc. .

These schematic effects represent chunks of knowledge (re. librarians, eating in a restaurant, washing clothes, etc.)activated by lexical items (may be pushing it here)

Story-telling - Narrative representation; The stories we tell (vs. Not verbalizing) influences later memory and representation of information.

E.g., McGregor & Holmes (99): Ss pretend to be lawyers and tell story from one side or the other; story told affect later memory

Reason giving (You think too much effect) - Wilson; Ss who were asked to analyze reasons (language use) for decisions (jams, college courses) made relatively less optimal decisions

- reasons verbalized often most easily articulated reasons (which then have undue influence)

Implicit Causality

Brown and Fish (1983) Model:


Subject: Object:

Ann helped Barb. (Ann perceived as cause)

Ann hated Barb. (Barb perceived as cause)

Why? Barb might be needy; Ann might be hateful.

Action verbs (help); Agent - Patient

Agent greater causal weight than patient (regardless of grammatical role; active - passive)

State verbs (hate); Stimulus - Experiencer

Stimulus greater causal weight than experiencer

Stimulus-experiencer (impress)

Experiencer-stimulus (hate)

Method: causal ratings, sentence completion, name memory

Results: effect reliable for state verbs

Action verbs - agent perceived as cause if action initiated by agent, patient if initiated by patient (e.g., Ann congratulated Barb)

Linguistic Category Model (Semin, Fiedler)

- multiple psychological dimensions of verbs

                                            Most         Least             Most

                                        Concrete     Revealing     Verifiable

DAV talk

IAV help

State action amaze (specific)

State hate

Adjective helpful

                                            Least         Most             Least

                                        Concrete     Revealing     Verifiable

Andy hit Bob         DAV             Andy talked to Bob

Andy hurt Bob     IAV                 Andy helped Bob

Andy hated Bob     State             Andy liked Bob

Linguistic intergroup bias (Maas)

Describe negative behaviors of outgroup with adjectives

Describe negative behaviors of ingroup with DAVs

Implicit Causality Explanations

(Language -> thought or thought -> language?)

1. Priming/lexical model

Verbs have derived dispositional terms; terms not distributed symmetrically:

Action verbs - terms reference agent (help - helpful)

State verbs - terms reference stimulus (like - likable)

Hoffman & Tchir (1990)

Larger implicit causality effect for verbs with dispositional terms referencing expected causal locus

Holtgraves & Raymond (1995)

Ss more likely to recall names of agents than patients, etc.

Priming effect is a type of Whorfian effect; perceptions of causality constrained by lexicon (but not by any inherent limitation of language; dispositional terms referencing either role are possible)

2. Causal schema

Basic patterns of causal thinking exist (we tend to think of agents as causing actions) independent of language -> influences language (e.g., distribution of dispositional terms)

Pragmatics and Cognition

Cognitive processes affected by pragmatic principles rather lexical, syntactic effects.

An (under researched) example: Languages differ in the extent to which status is grammaticalized (e.g., Korean vs. English); the former's social perceptions should be more attuned to status than the latter.

Heuristics and Conversation Principles

Research suggests much reasoning is not optimal; underestimate role of situation, fail to use base-rate, info. (Khaneman & Tversky, etc. etc.)

Schwarz, Hilton, and others have reinterpreted these effects in terms of the operation of Grice's conversational principle (maxims)

Grice: speakers expected to abide be four maxims: be clear, relevant, truthful, and appropriately informative

Maxims assumed to operate in all communicative settings, including experimental settings.

Example (from Khaneman & Tversky, 1973):

Ss read the following:

Jack is a 45-year old man. He is married and has four children. He is generally conservative, careful, and ambitious. He shows no interest in political and social issues and spends most of his free time on his many hobbies, which include home carpentry, sailing, and mathematical puzzles.

Then the following:

A panel of psychologists have interviewed and administered personality tests to 30 (70) engineers and 30 (70) lawyers, all successful in their respective fields. On the basis of this information, thumbnail descriptions of the 30 engineers and 70 lawyers have been written. You will find on your forms five descriptions, chosen at random from the 100 available descriptions. For each description, please indicate your probability that the person described is an engineer on a scale from 0 to 100.

Ss more likely to predict Jack was an engineer (the description is 'representative' of engineers) even when the base-rate for lawyers was far greater (tendency to ignore base-rate information).

Participants may have assumed the information provided by Exp. (personality test descriptions - individuating information) was relevant and should be used (why else would it be provided?), and so they use it.

Bias lessened if effects of conversational maxims is reduced; e.g., Ss told descriptions computer generated (rather than result of personality measure)

Another example:

Fundamental attribution error (Jones & Davis)

-infer internal attitudes even though behavior constrained by instructions (effect is removed if randomly generated)

And another example:

Survey question interpretation (Strack) and quantity maxim violation:

Example: How happy are you?

How satisfied are you with your life?

Above items are highly correlated if part of separate surveys; not as highly correlated if part of same survey (items are redundant; Ss assume something different must be meant (quantity maxim)