Psycholinguistics

Tóth Ágoston

Psycholinguistics FAQ

Language Processing

What happens in the human mind when we

read/listen/speak/write?

What are the processes and mechanisms underlying these

complex activities?

How can we discover and document these processes?

Language Acquisition

How do children acquire their mother tongue?

How languages are learnt?

1. Words

II. Sentences

The Word Superiority Effect

J. M. Cattel, 1880s

G. Reicher, 1969

James Ridley Stroop, 1935

Word Superiority Effect – Cattel, 1880s

Experiment: we can recall 4-6 random letters but more when they form words.

a) FON HGT AEW

b) FOG HAT NEW

The same letters are much easier to recall in case b.

In general, letters are easier to read when they form a word than when they do not. BUT: does guessing help? (not all word combinations are words)

Word Superiority Effect – G. Reicher 1969

Name the last letter after short exposure time.

a) AEHR

b) HEAR

"Was the last letter an R or a D?”

Here guessing is ruled out since we both head and hear are posssible.

Do letters continue to be easier to perceive in words in this experiment?

Word Superiority Effect (overview)

Yes; accuracy in reporting the final letter is

significantly better in words than in nonwords.

Accuracy in reporting a letter is significantly

better in words than in isolation, too.

(„Have you just seen a D or an R?”)

Explanation:

We acquire a large store of complex visual units

that are immediately available in reading.

Surface dyslexics do not have/do not access these

visual units, they need to use spelling-to-sound

rules instead. (This is much slower and

problematic in some languages, incl. English)

James Ridley Stroop, 1935

Words are

unitary

immediate and

involuntary

percepts.

RED

BLUE

GREEN

YELLOW

BLACK

GRAY

BROWN

ORANGE

The Stroop effect – PART 1: read these words

RED

BLUE

GREEN

YELLOW

BLACK

GRAY

BROWN

ORANGE

The Stroop effect – PART 2: read these words

RED

BLUE

GREEN

YELLOW

BLACK

GRAY

BROWN

ORANGE

The Stroop effect – PART 3: name the colors

The Word Superiority Effect (summing up)

Words are

unitary

immediate and

involuntary

percepts.

The Familiarity Effect

The familiarity effect

Reading is a learnt skill, improves with repetition

↓

The more frequently a word is seen, the more rapidly it can be recognized.

Which of the following word do you think is the most frequent in Wikipedia articles?

summer

introduce

the

temperature

year

buy

The familiarity effect

The more frequently a word is seen, the more rapidly it can be recognized.

Word frequency top-10 lists:

1. the

2. be

3. and

4. of

5. a

6. in

7. to

8. have

9. it

10. I

1. I

2. and

3. the

4. to

5. that

6. you

7. it

8. of

9. a

10. know

Based on the Corpus

of Contemporary

American English

(450 million words of

written language)

Hartvig Dahl

1M words of

spoken language

The correlation of frequency and similarity

We can measure the frequency of words in large corpora (see previous slide; frequency correlates with familiarity) corpus linguistics

Two ways of measuring familiarity in psycholinguistics:

1. subjects’ overlooking letters in words (e.g. find all the t letters; the t in the is likely to get overlooked because our seeing words as a whole rather than as a sequence of letters)

2. measuring lexical decision times

Lexical decision and

Priming experiments

Lexical decision time

Decide whether the following string is a word in your native language?

LABAKA

Y/N

(reaction time is measured, e.g. 280 msec after stimulus;

reaction time contains the linguistic & motor reaction time)

Lexical decision time

More frequent words: faster response

Words with many meanings: faster response (e.g. chest, crane) – even low freq. words

The effect of morphological processing (data from an experiment on Serbo-Croatian):

frula (nominative): 31% of the cases – fastest lexical decision

fruli (dative/locative): 10% – slower than nominative

frulom (instrumental): 1% – as fast as the dative/locative (!)

Priming

Priming is a modified lexical decision time experiment.

Lexical decision on a target word becomes faster when the

subject is exposed to a related word (prime) before seeing

or hearing the target word.

Phases of a priming experiment

(with semantically related words in this case):

Priming - semantic

Hodgson’s (1991) examined the following types of semantic

relatedness between prime and target words:

synonymy,

antonymy,

superordination and subordination (e.g. chair – furniture) ,

coordinates, words with the same hypernym (e.g. truck – train),

conceptual association (e.g. leash – dog),

phrasal association, words that co-occur in phrases (e.g. private

– property).

Hodgson found semantic priming in all of the above categories.

Priming - morphological

Priming effects arises with morphologically related words,

too (inflection, derivation).

E.g.

manager, management and manages prime manage

cancel does not prime can (morphologically unrelated words)

The mental lexicon may be a huge collection of individual

words but on top of this lengthy list there seems to be a

network of morphological and semantic connections among

words.

1. Words

II. Sentences

Garden-path sentences (O’Grady p.453)

„Since Jay always walks a mile seems a short distance to him.”

Default reading: „… walks a mile …”

Intended meaning: „… a mile seems a short distance to him.”

Psycholinguistic explanations:

Late closure: we prefer to attach new words to the clause currently processed

Minimal attachment: we do not postulate new constituents unless needed

Event-related potentials

The subject sits in front of a computer screen and reads

Electrical activity of his/her brain is measured using EEG

A computer program removes unneeded information

The resulting wave form is an event-related potential

(see the diagram in the upper right corner)

Phase 1: 100-300 msec

„Early left anterior negativity” (ELAN) pattern: a negative-going wave that peaks around 100-300 milliseconds after the onset of a stimulus; it signals problem(s) within phrases, e.g. *The scientist criticized Max’s of proof the theorem.

ELAN has been reported in English, German, Dutch, Chinese and Japanese (may or may not be language specific)

seems to support a "syntax-first model" of sentence processing

ELAN follows the stimulus soon (100-300 msec = 1 - 2 - 3 tenths of a second)

Seems connected to phrase structure building based on syntactic word category information („local syntactic structure”)

Phase 2: 300-500 msec

unusual N400 (stronger negative EEG spike at 400 msec):

argument structure violations, lexical-semantic problems

The pizza was too hot to eat. → normal N400

The pizza was too hot to drink. → stronger N400

unusual LAN (left anterior negativity) EEG pattern:

problems with grammatical relations, e.g. verb inflection

violation

Diagram source: https://ebrary.net/12276/language_literature/event-related_brain_potentials

Examples of abnormal N400 EEG data

A. D. Friederici (2011:1381)

to hire a bodyguard / baby / hen / wire

Phase 3: 500-1000 msec, P600 signal

Integration of all knowledge about the sentence (a more general sense-making activity)

P600 abnormality:

E.g.

agreement problems (subject-verb, tense, gender, case), e.g. *Tom wear a hat.

semantically anomalous sentences e.g. ?Colorless green ideas sleep furiously.

well-formed sentece but large complexity, incl. garden path (The guy persuaded to buy the hat was bald.) or sentences with many thematic roles

Diagram source: https://ebrary.net/12276/language_literature/event-related_brain_potentials

A. D. Friederici (2011:1377)

It’s all in

your head.

For the exam

Chapter_13.pdf

(O’Grady et al. 435-461)