Psy1306 Language and Thought

transcript

Lectures 6Path and Manner

ThemesKey Ideas Discussed:

Language is sketchy/selective

Use-it-or-loose-it/Functional reorganizing

Thinking for SpeakingThinking for Later

Speaking

ExperimentsPopulation:

Prelinguistic Infants vs. Adults

Adults with different language background

Methodology: Categorization via

habituation or preferential looking

Triads Similarity Judgment Recognition Memory Eye-tracking as a window

into thought Labeling vs. No Labeling

Prior to doing the above 3 tasks

(Slides with white background are from A. Papafragou)

Path and Manner Components of Motion

figureSNOOPY BALL HOLE

ground ground

FROM...TO... path

ROLLING

manner

English Predominantly Manner VerbsThe bottle floated out of the cave.

Spanish (or Greek) Predominantly Path VerbsLa botella salió flotando de la cueva.(The bottle exited floating from the cave.)

Crosslinguistic Differences:Path and Manner Verb Preferences

Example from Talmy 1985…children’s attention is heavily channeled in the direction of those semantic distinctions that are grammatically marked in the language. (Berman and Slobin, 1994)

Framing motion events cross-linguistically

Greek Mia petaluda‘a butterfly

petai. is flying’ MANNER V

English A butterfly is flying.MANNER V

Framing motion events cross-linguistically

English A butterflyis flyingMANNER V

to a flower.PATH PP

Greek Mia petaluda‘a butterfly

pai is goingPATH V

s’ena luludi.to a flower’PATH PP

BoundedPathconstraint

Natural Divison

Using gestures, describe an event in which:A cat, having swallowed a bowling ball,

proceeds rapidly down a steep street in a wobbling, rolling manner.

Spanish speaker vs. NSL speaker

Videos available fromScience.

Prelinguistic InfantsPulverman & Golinkoff (2004): 7-months-

olds*Casasola, Hohenstein, & Naigles (2003): 10-

months-oldsPulverman et al. (2007): 14-, 17-months-olds,

Spanish vs. English learners*Havasi & Snedeker (2004a, 2004b): Adults

and children

* Same problem of variance in objects used (as mentioned in last class)

Learning Path-Manner distinctionsChildren converge rapidly on language-specific

syntactic and semantic properties of motion Vs (Bowerman 1996, Choi & Bowerman 1991, Slobin 1996).

Adults are sensitive to the statistical regularities of linguistic packaging of motion eventsin guessing meaning of novel motion Vs, Spanish

speakers make more path conjectures than English speakers (Naigles & Terrazas 1998)

Path vs. Manner salience in motion cognition?

Do speakers of English and Greek become differentially sensitive to Manner & Path of motion?

Do linguistic categories affect non-linguistic categorization? (Papafragou, Massey & Gleitman, Cognition 2002)

Subjects:Monolingual native speakers of English and

Greek.

Two Age Groups: 8-year-olds (14 English speakers and 22 Greek

speakers); Adults (20 English speakers and 21 Greek

speakers).

Sample event: man running up stairs

Same-manner foil: man running down hallway

Same-path foil: man walking up stairs

Linguistic descriptions differ

M a n n e r : 6 9%P a th : 2 5 .5%

E n g lis h

M a n n e r : 2 2 .5%P a th : 6 4 .1%

G re ek

M a in V e rb

…across age groups

Distribution of Verbs: English

Path V Manner V

EngChildren

EngAdults

Distribution of Verbs: Greek

Path V Manner VPr

GrChildren

GrAdults

Categorization does not differ!

0102030405060708090

English Greek

Children vs. adults

0102030405060708090

EnglishChildren

EnglishAdults

GreekChildren

GreekAdults

Do linguistic categories affect memory? (Papafragou et al, Cognition 2002)

Subjects:Monolingual native speakers of English and

Greek. Three Age Groups: 38 5/6-year-olds; 38 11/12-year-olds;21 Adults.

Two Sessions: (1) inspect and describe 6 pictures (2) new set of pictures: ‘Same or different?’

Session 1 boy jumping over log

Session 2: Manner changeboy tripping over log

Session 1 frog jumping into bathroom

Session 2: Path change frog jumping out of bathroom

Memory is not affected!

0102030405060708090

MannerChange

PathChange

NoChange

EnglishGreek

Children vs. adults

0102030405060708090

YoungEnglish

MiddleEnglish

AdultEnglish

YoungGreek

MiddleGreek

AdultGreek

Children vs. adults (2)

0102030405060708090

YoungEnglish

MiddleEnglish

AdultEnglish

YoungGreek

MiddleGreek

AdultGreek

• RecognitionParticipant sees events. Decide later (varied delays) whether they saw the events earlier.Question: Are manner language speakers better at noticing

manner changes and vice versa?• Similarity Judgment (Triad Task) Which one is like the target? Question: Do manner language speakers prefer same manner

and vice versa?

3 crosslinguistic studies on Manner vs. PathKrych (2001). Doctoral dissertation, Stanford (English vs. Spanish)Gennari et al. (2002). Cognition (English vs. Spanish)Papafragou et al. (2002). Cognition (English vs. Greek)

TargetM: carry, P: exit

Same MannerM: carry, P: enter

Same PathM: drag, P: exit

RESULTS: Crosslinguistic Difference?

Recognition Similarity

Participant asked to label aloud event No Yesprior to task

Participant not asked to label No No

3 crosslinguistic studies on Manner vs. PathKrych (2001). Doctoral dissertation, Stanford (English vs. Spanish)Gennari et al. (2002). Cognition (English vs. Spanish)Papafragou et al. (2002). Cognition (English vs. Greek)

Different ResultsNo language effect unless labeling occurs beforehand

Language effect w/o labeling

Any thoughts on why?

Language is sketchy

Language is not logically explicit, thought is:John and Mary bought a nice house [TOGETHER].

John and Mary got a good grade [EACH].

Language is selectiveNot everything that is represented in mentalese is

expressed when we speak:

Mary: Let’s go out.John: It’s snowing. [and so we can’t go out].

Because communication takes time and effort, only a fraction of the thought (in mentalese) is encoded in language. Speakers trust hearers to fill in the rest.

When is Manner of motion included?Hypothesis: Manner information is included when

manner of motion not predictable (esp. in Greek).

Manner in Greek!

‘Inferable manner’ scenes: Man walking up stairs.

‘Opaque manner’ scenes: Man running up stairs.

Inferability affects Manner encoding in Greek (Papafragou, Massey & Gleitman, Cognition 2006)

Gr children Gr adults Eng children Eng adults

'Inferable manner' scenes 'Opaque manner' scenes

% of all descriptions which includes Manner

Linguistic descriptions are flexible

Linguistic encoding of motion is selective.

Formulation of event descriptions flexibly adjusts to changing conversational pressures on-linein speech of both young and more experienced

speakers

Gap between linguistic descriptions and rich conceptual representations.

Can language affect motion event perception?

‘(Speakers) code spatial perceptions at the time of experience in whatever output frameworks the speaker’s dominant language offers’.

(Levinson 1996, p. 156)

An online study (Papafragou, Hulbert & Trueswell, 2008, Cognition)

Eye-movements as window onto what sorts of information humans use to build event representations, and when.We compare how English and Greek speakers

interrogate motion scenes while preparing linguistic descriptions vs. encoding information in memory

English vs. Greek speakers…

Linguistic Condition: Ss had to describe what happened. Nonlinguistic Condition: Ss had to remember what they saw. At the end of the experiment they were shown a still image and were asked if it belonged to any of the clips.

A sample trial

Eye movements were recorded throughout.

animation unfolds (3 sec) animation freezes

L group: Ss inspect scenes…

NL group: Ss inspect scenes…Ss study scenes more

Ss describe scenes

Two types of events

BoundedPath

UnboundedPath

Linguistic task: Bounded events

English: 78% manner Vs “A boy is driving his bike to a tent.”

Greek: 36% manner Vs“A boy is going to a tent, on a bike.”

BoundedPath

Linguistic task: Unbounded events

English: 74% manner Vs “A boy is driving his bike.”

Greek: 56% manner Vs“A boy is driving his bike.”

UnboundedPath

0.500 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Greek SpeakersEnglish Speakers

pathmanner

Eye movementdata: Bounded events

Time in 1/30th of Second Units

Animation (3 Seconds) Linguistic Description / Study Phase

clip freezes

… …

0.500 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Animation (3 Seconds) Linguistic Description

* p < 0.05

0.500 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Animation (3 Seconds) Linguistic Description

Bounded events: In Linguistic Task, the 2 populations differ: Ss look for what their language needsPr

Nonlinguistic Task: Event perception the same! But differences in Study Phase: Ss study what their language doesn’t routinely encode

0.500 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Animation (3 Seconds) Study Phase

* p < 0.05

Linguistic Task Nonlinguistic Task

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

ConclusionsHow we inspect a scene depends on task/

Same is true if the task is linguistic: Speakers focus on aspects of scenes which are

routinely encoded by their language Cf. eye-movement production studies within a given

language by Levelt, Bock, Griffin, a.o.

For the first time we show here that these looking patterns differ cross-linguistically: Where languages differ from each other in how they

encode event structure, this difference shows up in how scenes are interrogated during speech planning

Conclusions (cont.)

But when inspecting the world freely, all humans are alike, regardless of the language they speakInterrogation of an unfolding event (cf. our

nonling task) generates nearly identical sequences of shifts in attention

Conclusions (cont.)

Nevertheless, important cross-linguistic differences in how perceptions are encoded in memoryDifferences in nonling task when video

freezes: Ss presumably encode events rapidly in declarative

memoryTruly contra-Whorfian result:

People then proceed to interrogate those aspects of the scene that they couldn't map onto an accessible precompiled linguistic-semantic form (e.g., the lexical semantics of verbs and their argument structures).

Language effects on thought?

Test case: space and motionSalience effects not found

Path/Manner asymmetries in language are not reflected in categorization, memory or apprehension of motion

Linguistic effects emerge when language is implicated in task.

Discussion QsTest case generalizability?Differences between the test case of this

class and last class?

Language affects category salience…learning a language can affect nonlinguistic cognition by selectively maintaining or discouraging sensitivity to (…) distinctions that are, or are not, relevant to that language.

Bowerman & Choi (2003)

Testing the ‘boundary’ and ‘salience’ hypothesesWhat is a non-linguistic task?

Most tasks (memory, categorization) involve the use of language in the instructions.

People may be covertly using linguistic labels to remember or categorize ambiguous stimuli. Does this count as a ‘Whorfian’ effect or as a

‘language on language’ effect?

Testing the ‘boundary’ and ‘salience’ hypothesesWhat counts as a cross-linguistic difference?

Whether a category is grammatical or lexicalWhether a category is obligatory or notWhether a category is frequently or

infrequently used in ordinary communicationHow different ARE languages?

Psy1306 Language and Thought

Documents