Loading presentation...

Present Remotely

Send the link below via email or IM

Copy

Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.

DeleteCancel

Make your likes visible on Facebook?

Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.

No, thanks

LEXICAL ACCESS

No description
by

Pavlina Peskova

on 22 February 2016

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of LEXICAL ACCESS

Lexical Access
Tuesday, February 24, 2015 0930
Lexical Access
LEXICAL ACCESS
PRIMARY METHODS
MODELS OF LEXICAL ACCESS
CAMP 1
SECOND-GENERATION MODELS
TRACE



LEXICAL AMBIGUITY RESOLUTION
How do we process words with more than one meaning?

exhaustive access: all meanings are activated even though only one of the meanings is appropriate

EVIDENCE:
Priming experiments:

The spy swept the room looking for concealed bugs.
The cook picked up a bag of flour in the kitchen and saw the bugs.

target word: listen
control word:
unrelated

Result: Hearing the word bug made people respond faster to target words related to either of its meanings, no matter which meaning was appropriate in the context.


CAMP 2
THIRD-GENERATION MODELS
Base Camp:
First-Generation Models
LOGOGEN MODEL
FOBS MODEL
LOGOGEN
Frequency ordered bin search (FOBS) model
MORPHOLOGY
- smallest units of language that can be assigned meaning
Morphemes:
Derivational morphemes
Inflectional morphemes
Main idea:
Lexical representations are organized into
bins

Important Concepts:
High-frequency x low-frequency words
Self-terminating search
Bin organization (shared roots)
Lexical access in FOBS Model:
1. Morphological DECOMPOSITION
- stimulus is broken down into individual morphemes
E.g. dog-s

Evidence:
FREQUENCY
- surface frequency
- root frequency
PRIMING:
- shared-morpheme priming
- real vs. pseudo-affixes (e.g.
grow-er
vs.
sist-er
)
Conclusion:

Morphological representations and processes play a unique role in lexical access, separate from semantic, phonological and orthographic effects.
COHORT



Main Concepts:
- interactive (bottom-up and top-down activation)
- cascaded activation
- lateral inhibition
- word superiority effect

Advantages:
- can deal with degraded input
TRACE
(McClelland and Rumelhart 1981)
COHORT (Marslen-Wilson 1987)
(Shadowing method)
Main concepts:
- bottom-up model
- activation, selection, and integration
- incremental process of lexical access
- recognition point
EVIDENCE
1) Non-word detection:
trenkitude vs. cathedruke
2) Cross-modal priming
DISTRIBUTED FEATURE MODELS



SIMPLE RECURRENT NETWORK (SRC)
- a multi-leyered network
- distributed representations (subclasses)
- word recognition = pattern of activation
- every instantiation of a word is different
- abstract relationships between constituents
DISTRIBUTED COHORT MODEL (DCM)
- simultaneous activation of phonological and semantic units (phoneme perception and word identification are two parallel processes raher than two sequential stages)
- activation in proportion to conditional probability

Evidence
:
- decoding coarticulation
("the semantic side 'wins' the competition and ends up being selected)
- processing of words wiht multiple meanings
Today's objectives:
1)
Methods
used to investigate lexical access

2) Major
models
of lexical access

The process of identifying words and recovering word-related information from long-term memory.

Most models of lexical access deal with activating word
form
information.

SHADOWING
Shadowing
Gating
Lexical Decision Tasks
Priming
Word Monitoring
Eye-tracking
- participant listens to speech through headphones
- participant repeats speech as quickly as possible
- recorded response is evaluated for accuracy
Observation:

Lexical Access is fast!
Speech: ~ 5 syllables/second
fast shadowers lag ~ 250 ms ~ 1 syllable
Errors are not random: they fit the semantic and syntactic context

Short Demonstration

GATING
LEXICAL DECISION
INTEGRATION


SELECTION


ACTIVATION


INPUT
Stab
Sip
Sing
Speed
Sting
Sink
Say
Slip
Stick
...
S
Stab
Stink
Stick
Stack
Stagger
Stand
Store
...
ST
Stab
Stack
Stagger
Stand
Store
STA
Stab
Stack
Stagger
Stand
STA
Stab
Speak
Sink
Stand
Store
Swim
Steal
...
S
Stand
STAN
word uniqueness point
Task
:
Participants must identify a word after being presented with a fragment of the word.
The duration of the fragment gradually increases in duration (e.g. by 50 msec increments) until the participant correctly guesses the word (i.e. the uniqueness points is reached).
INTERESTING FINDINGS
Interesting Findings
Words are recognized before they are complete
(incremental lexical access)
RECOGNITION POINT
= earliest "gate" at which the participant recognizes the word

Gating shows the effect of context on spoken word recognition:

identifying a word
in isolation
~ 1/3 of a second
identifying a word
in context
~ 1/5 of a second
EXAMPLE
Is it easier to recognize a word if it is preceded by a related stimulus?
TASK:

How quickly do participants recognize stimuli as words and nonwords.

- stimuli are either visual or auditory
- analysis is based on reaction times
- experimenter manipulates characteristics of words to identify factors that influence lexical access (e.g. frequency, duration, etc.)
EXAMPLE:
PRIMING
EXAMPLE
DOUBLE Lexical Decision Task
(Meyer and Schvaneveidt 1971)
Double LDT:
prime and target strings are displayed simultaneously, and participants indicate by a single response whether both are real words.

RESULT:
Responses were 80 ms faster for related pairs such as "NURSE - DOCTOR" than for unrelated pairs such as "BREAD - DOCTOR".
Semantic priming
: participants make semantic judgement
(Eg. hat may prime head)
Associative priming
: participants make an association between two words
(Eg. bread may prime butter)
Cross-modal priming
: participants make an association between text and audio
ASSOCIATIVE PRIMING TEST
PRIMED

Lexical Access
UNPRIMED
Lexical Access
TINDLE
BREAD
BUTTER
Prime
Target
TINDLE
BREAK
BUTTER
Control
Target
400 ms
550 ms
PRIMING GIVES EVIDENCE OF
SPREADING ACTIVATION
Activation of one lexical entry causes activation of associated entries
Node activation
Spreading activation is
automatic:
fast and outside of consicous control

FACTORS:
Word frequency
Recency
Semantic similarity
Form similarity


EXAMPLES OF SPREADING ACTIVATION
WORD MONITORING
EYE-TRACKING
TASK:
Participants listen to:
words, sentences, or discourse

Participants respond to
target words
(e.g. target = CAKE )
EXAMPLE
TARGET WORD =
CAKES
1) experimental sentences
(include grammatical and ungrammatical sentences testing the research question(s)
2) filler sentences
(to make participants unaware of the purpose of the taks and to avoid their familiarization with types and ungrammaticality used)
3) catch trials
(do not include the word being monitored, use of catch trials avoids participants pressing the button mechanically)
GOAL:

Test RTs for lexical retrieval in (un)grammatical sentences
WORD MONITORING:
EXPERIMENTAL SENTENCES
(1)
John's mother is a great baker. Most days she
bakes

cakes
.
EXPERIMENTAL sentence
(2)
John's mother is a great baker. Most days she
bake

cakes
.
EXPERIMENTAL sentence
Blom and Unsworth 2010
MAIN ASSUMPTIONS:
MAIN ASSUMPTIONS:
Most days she
bake

cakes.
We unconsciously slow down whenever we are faced with ungrammaticality
- the word monitored is NOT the word that causes the ungrammaticality but the word AFTER the ungrammaticality.
RTs are longer for the word following ungrammaticality
(Eg. RT for
cakes
in 'she
bake
cakes' LONGER than
cakes
in 'she
bakes
cakes')
TASK:
Participants view
images on screen
Participants listen to commands to click on images, or to move them to new locations
MAIN ASSUMPTIONS:
EXAMPLE:
MAIN ASSUMPTIONS:
People tend to direct their eye gaze to things they are
attending
to in their visual environment.

Language guides
visual attention
- eye movements reflect

highly incremental


linguistic interpretation
- eye movemens reflect a
referentially-driven

interpretation of language
Participants listen to commands to click on images, or to move them to new locations
Target = "candle"
Cohort = "candy"
Rhyme = "sandal"
Cohort + Rhyme = "candy" and "sandal"

RESEARCH Q:
Which words
"COMPETE" with
candle for activation?
RESULTS
Word onset
more influential than rhyme
Lexical access is i
ncremental
Captured bottom-up activation of word forms from sensory input

Accounted for frequency & recency effects by proposing lower activation thresholds

Accounted for semantic effects by proposing higher activation of semantically related logogens
Morton, 1969
VISUAL RECOGNITION POINT
Mediated priming:
WHY does lion prime stripes?
(HINT: think of the semantic category animals)
Perception can be influenced by contextual information:
(Warren and Warren 1970)
It was found that the *eel was on the shoe.
It was found that the *eel was on the orange.
It was found that the *eel was on the table.
Experiment
: the /h,p,m/ sounds from
heel, peel,
and
meal
was replaced with a cough.

Result
: listeners were unable to detect that the phoneme was missing.
(Forster 1989)
Friday, July 19th, 1805

A fine morning. At 9 we came to high parts of the mountains, which had a good deal of pine, spruce and cedar [11] on them, and where there were not so many rocks; but no timber in the bottoms except some small willows. About 1 o'clock we had thunder, lightening and rain, which continued an hour or two, and then the weather became clear. This afternoon we passed parts of the mountains, that were very high, and mostly of solid rock of a light colour. The mountains are so close on the river on both sides that we scarcely could find room to encamp. We went 20 miles and encamped [12] on the south side. After night some rain fell.
THE JOURNALS OF THE LEWIS AND CLARK EXPEDITION
"The Corps of Discovery Expedition" (1804-1806)
Elman 2004
Gaskell and Marslen-Wilson 2002
Full transcript