Lecture: Psycholinguistics Professor Dr. Neal R. Norrick

Download Report

Transcript Lecture: Psycholinguistics Professor Dr. Neal R. Norrick

Lecture: Psycholinguistics
Professor Dr. Neal R. Norrick
_____________________________________
Psycholinguistics
Universität des Saarlandes
Dept. 4.3: English Linguistics
SS 2009
Organizational matters
attendance:
don‘t miss more than 2 lectures!
(make sure you sign into the
Anwesenheitsliste every week)
tutorial:
mandatory for magister,
Erasmus (to get full credit),
LAG alt (if you‘re taking the exam
or wish to do 3 SWS)
 recommended for everyone as
preparation for the final exam
there will be no lectures in July (7th, 14th, 21th)!!!
 last lecture:
6-30-09
final exam:
will take place on 07-28-09 in the
lecture hall (at the regular time)
signing up for the exam (Neue Studiengänge):
 via sign-up list – will be passed around
in the last two weeks of the lecture in June
(exact procedure to be announced later!)
updated script and bibliography
are now online!!!
website:
http://www.uni-saarland.de/fak4/norrick/index.html
3.2 Sounds and phonemes
phonemes as psychologically real entities
abstract phoneme /p/
versus positionally variant allophones:
• aspirated [ph] word-initial, as in pill
• preglottalized [p] word-final, as in lip
• unaspirated [p-] after initial s, as in spill
these allophones are predictable variants
they don't distinguish meanings
ability to distinguish meanings defines
phonemes
hence: minimal pair test
pill - bill
but experiments show:
words are recognized faster than phonemes
we recognize the letter b and the sound /b/
faster in the word bat than in isolation
words are more salient than phonemes
suprasegmental features are also
psychologically salient
intonation distinguishes statements
and questions
Sally's here. versus Sally's here?
stress focuses on any constituent in questions
Sally gave the new car to Judy today?
• can question whether it was Sally (not Suzy),
• whether she gave (not loaned) the car,
• whether it was the new (not the old) car etc
other salient suprasegmentals are volume
and speed,
they signal speaker attitudes
and emotional states.
3.3 Sentences and propositions
sentences as grammatical representations
of underlying meaning in the form of (logical)
propositions

propositions in language of thought
clarify (logical) relations between words
and sentences, represent entailments,
inferences etc
versus

sentences following the rules of some
natural language
grammar rules transform underlying
meanings into grammatical sentences of
natural language
so a single underlying logical proposition
has multiple possible representations in any
given natural language, e.g.
the cat is on the mat, the cat is on
top of the mat
the mat is under the cat, the mat is
beneath the cat etc
But where would such a logical language
of propositions come from if not from
communication in a natural language?
But if our language of thought is some acquired
natural language, then the specific
characteristics of that language determine our
patterns of thinking - and this leads to the
Sapir Whorf Hypothesis.
3.4 Sapir-Whorf Hypothesis
Sapir-Whorf Hypothesis sees language and
human cognition as related in non-arbitrary
ways
Sapir 1921, 1929, 1949, Whorf 1950, 1956
proposed a relationship between language,
meaning, culture, and personality, generally
called the Sapir-Whorf Hypothesis
The strong version of the hypothesis says
our language determines our perception. We
see the things and processes our language
has names for and ignore or cannot see
what our language doesn't name.
The weak version of the hypothesis says our
language influences our perception. We attend
to the things and processes our language has
names for and tend to ignore or find it difficult to
attend to what our language doesn't name, e.g.
English speakers with only a single word wall
find it difficult to understand and make the
distinctions necessary for choosing Wand
versus Mauer in German.
German and English speakers group together
all kinds of spherical objects with the single
word ball, they would not normally distinguish
the objects categorized in French as ball from
those called ballon.
In French, speakers must attend to
differences in size and determine whether
an object is inflated or not to categorize it as
ball versus ballon.
Also, the grammar of the language we're
speaking at any given time (be it our native
language or not) forces us to think in certain
ways.
Slobin's ‘thinking for speaking’ notes that any
language system enforces certain choices in
grammar and lexis, no matter how our
underlying thought patterns work,
e.g. because of the tense/aspect system of
English, all the following questions are relevant
in talking about an event:
• When did the action take place?
present versus past tense
• Is it completed?
perfective versus simple aspect
• Was it an ongoing process or a momentary
activity?
progressive versus simple aspect
• Did it only happen once or does it always
happen?
progressive versus simple aspect
But in various languages, the questions below
are important for determining grammatical
forms (word order, cases):
• Did I (as speaker) see the event or just hear about
it?
• Is this statement a fact or just my opinion?
• What kinds of words are typically subjects?
And what kinds objects?
Compare:
I like it,
mi piace,
mich friert,
isch hann kalt,
mir gefällt es,
I'm cold,
mir ist kalt,
j'ai frois
If we must always attend to certain distinctions
and ignore others, in speaking and thinking,
shouldn't that influence the way we think?
Nevertheless, we manage to translate
between languages and to learn other
languages, so apparently our thought
patterns can extend and adapt.
We can grasp and learn to use words from
other languages, even if they have no
counterpart in our native language, e.g.
Schadenfreude
blind date
4. Words in the Mental Lexicon
Mental Lexicon versus dictionary
• words accessible via sound, meaning,
related words
Mental Lexicon versus encyclopedia
• Encyclopedia contains all kinds of knowledge,
usually unnecessary for normal word use,
e.g. for dog:
• perceptual: four-legged, furry, barking sound etc
• functions: used as pet, for hunting, guarding etc
• behaviors: chases cats, chews bones,
is territorial etc
• origins: animal, mammal, bears litters of
puppies etc
• history: domesticated early, developed into
pet etc
• facts: Suzy has a puppy, Bill's dog chases
cars etc
Encyclopedia contains entries for concepts not
represented by individual words in lexicon
e.g. for "aquatic mammals" and "famous scientists" or
"favorite bars in town"
assume that lexical entries specify only:
–
–
–
–
–
–
sound
morphological irregularities
syntactic properties
core meaning for identification
relations to other words
perhaps spelling
4.1 Word Association Tests (WATs)
Experiments show:
we recognize concrete words like table
faster than abstract words like trouble
table  chair
trouble  bad
faster, more consistent
lower, less consistent
we also recognize familiar words and short words
faster than unfamiliar and long words
compare:
A traveling salesman arrived in town
An itinerate salesman arrived in town
WATs also show paradigmatic versus
syntagmatic relations:
• paradigmatic
• syntagmatic
apple, pear, banana, plum
apple, red, juicy, eat
in WATs:
• adults respond paradigmatically:
pillow  bed
• children respond syntagmatically:
pillow  soft
semantic features (or components) are also
psychologically real for speakers
Woman
Man
 human, adult, female . . .
 human, adult, male . . .
again kids don't analyze, responding syntagmatically:
Man  work instead of woman  adult or female
based on response times to questions like:
Can a canary sing?
Can a canary fly?
Does a canary have skin?
Collins & Quillan (1989) postulate
memory structures
But other tests show pure frequency of occurrence
in discourse counts for more than response times in
WATs:
A canary sings/is yellow more frequent
versus
A canary flies/eats less frequent
WATs show faster recognition after
associated words:
we recognize roof faster after house
than after some unrelated word like apple
so Lindsay & Norman (1972) postulate
lexical networks
WATs are a questionable method:
• WATs elicit unnatural verbal behavior
• WATs develop quantitative results, but they're
always fuzzy
• WATs are usually limited to nouns, usually
concrete but consider, e.g. colors,fruits, games etc
• WATs are unnecessary, given discourse analysis,
especially now with computers available
4.2 Tip-of-the-Tongue Phenomena
Thinking on Tip-of-the-Tongue (TOT)
phenomena begins with James (1890)
James speaks of “a gap that is intensively
active” in consciousness when we try to recall a
forgotten name.
Meringer and Mayer (1895), Fromkin (1973) kept
personal catalogues of error types to gather natural
data.
Brown and McNeill (1966) collected intuitions
on remembering in diary studies, e.g.
unable to recall the name of the street on
which a relative lives,
one of us thought of Congress and
Corinth and Concord
and then looked up the address and
learned that it was Cornish.
Brown and McNeill also induce TOT states, by
reading definitions of uncommon words to
subjects, who then answer questions about their
search for the missing word, e.g.
subjects asked to identify the target word
sextant based on a dictionary definition
“A navigational instrument used in measuring
angular distances, especially the altitude of sun,
moon and stars at sea.”
Burke et al. (1991) write, “When a TOT occurs, a
lexical node in a semantic system becomes
activated, giving access to semantic information
about the target word, but at least some
phonological information remains inaccessible.”
Subjects in the TOT state often report that a word
related to the target comes repeatedly and
involuntarily to mind, yielding ‘blockers’,‘interlopers’
or ‘persistent alternates’, e.g.
sexton or sextet for sextant
Burke et al. (1991) developed an experimental task,
using prompts like those in a trivia game presented
on a computer, where subjects typed responses,
e.g.
What is the old name of Taiwan?
target: Formosa
foils: Taipei, Canton, Ceylon
The foils often acted as blockers for the target word
They then asked questions like:
“How familiar do you think the word is?”
“How certain are you that you can recall the word?“
“What is the first letter or group of letters in the word?”
Burke et al. (1991) identify a semantic system or
network of nodes connecting concepts
• the concept chastity is connected with “is a virtue,”
“take a vow of” etc
• the concept baker with “bake bread” “get up early”
“sell cakes” “knead dough” etc
Compare scripts of Schank and Abelson (1977),
cognitive models of Lakoff (1987):
Cognitive model for chastity would identify
prototypes for the virtue like saints, and distinguish
characteristics like “is a virtue” from linguistic
constructions in which the word chastity occurs such
as “take a vow of chastity.”
Cognitive model for baker would identify prototypes
for profession like the owner of the bakery at the foot
of the hill.
Burke et al. (1991) say one word may prime,
i.e. facilitate recognition of, another word,
the activation of nurse facilitates activation of
doctor because priming spreads and summates
via these many shared connections.
Cognitive processes recoded in diary studies
and lab experiments differ from TOT searches
in real conversation, e.g.
1 Helen: in Hammond, north Hammond. Junior Toy Company.
2
they used to make toys, little tricycles and scooters and everything.
3 David: and where was it?
4 Helen: I don't remember the street.
5
Hoffman?
6
No.
7
it was a little beyond right here, you know,
8
it wasn't right in north Hammond.
9
t was around that street that turns into Illinois,
10
there when you go to the cemetery.
11
here's a tavern on one side and a VFW on one side.
12
forget the street.
13 David: Gosselin?
14 Helen: Could be, I don't know.
15
I don't know,
16
but that's where Junior Toy was in the low corner there.
• Helen expresses her forgetfulness at line 4 with
“I don’t remember the street.”
• She takes a guess at the name in line 5, but
immediately rejects the guess in line 6.
• She begins an extended description of the area in
terms of landmarks in lines 7-11.
• She concludes, “I forget the street” at line 12,
but David offers a guess of his own, since he’s
familiar with the local neighborhood.
• Helen expects help with name or at least
assurance that David can identify the place.
Storytellers often name landmarks and major
streets, not phonetically similar words.
Note references to cemetery (l. 10),
tavern and VFW (l.. 11)
Description “in the low corner” implies that the
teller can visualize the scene.
So why not search corpora for natural
instances of TOT?