stages in lexical access or: the lemma dilemma. tiger (x) tigre noun fem. countable /tigre/ ti g...

Post on 05-Jan-2016

223 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Stages in lexical access

Or: the lemma dilemma

TIGER (X)

Tigre

NounFem.

Countable

/tigre/

t i g

Lexical concept

lemma

lexeme

phonemes

Has Stripes Is Dangerous

Levelt, Roelofs, & Meyer, 1999

Step 1: Selection of semantically and syntactically specified representation (lemma)

Step 2: Selection of the form of the word (lexeme)

Evidence:

Speech Errors Picture/Word interference studies Tip-of-the tongue states Aphasia Cognitive Neuroscience (Evoked Potentials)

Speech errors: word substitution

(1a) I want whipped cream on my mushrooms [intended: strawberries]

(1b) I’ve got some whipped cream on my mushroom [intended: mustache]

Speech errors: exchanges

(2a) you ordered up ending some fish dish (2b) lilting willy [intended: wilting lilly] (2c) that’s why they sell the cheaps drink

(2a) lemma-level (2b) lexeme or phoneme level (2c) lexeme level

Picture/word interference

Ss see a picture (e.g., of a TIGER) and have to name it a.s.a.p.

Concurrently, earlier, or later, they see or hear a distractor word (e.g., llama)

SOA is time between picture and distractor: Negative means distractor first; Positive means distractor last.

Picture/word interference

Schriefers et al., 1990: Semantic interference at early SOAs (-150 ms) Phonological facilitation at late SOA’s (0, 150 ms) No overlap

If one presents the distractor early, it affects the first step, lemma retrieval. Therefore, semantic relation matters.

If one presents the distractor late, it affects the second step, lexeme selection. Therefore, effect of phonological relation.

Picture/word interference: gender

Many languages, e.g., Spanish, Italian, French, German, Dutch, Russian, Arabic have nouns with grammatical gender.

For a majority of nouns, there is no conceptual correlate with gender.

Dependent on lng., gender is marked on the form or not

Gender is a property that drives agreement processes: The red table in Spanish and Dutch:

La mesa roja [the-fem table red-fem] De rode tafel [the-common red-common table]

Picture/word interference: gender

Schriefers (1993): Picture/Word interference in Dutch. Distractors had either the same or a different gender

from the target noun (say tafel-comm): broek-comm /hemd-neuter

Target nouns were either named with a phrase or as a bare noun: de rode tafel or tafel

Gender congruency effect, but only if the entire phrase was named.

Interpretation: distractor boosts one of the two gender nodes. If selection necessary, possible effect.

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of your tongue..

Brown & McNeill (1966):

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of your tongue..

Brown & McNeill (1966):

‘Instrument to determine your position when sailing’

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of your tongue..

Brown & McNeill (1966):

‘Instrument to determine your position when sailing’

‘The capital of Estonia’

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of your tongue..

Brown & McNeill (1966):

‘Instrument to determine your position when sailing’

‘The capital of Estonia’ ‘Electronic device, crucial in early radios’

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of your tongue..

Brown & McNeill (1966):

‘Instrument to determine your position when sailing’

‘The capital of Estonia’ ‘Electronic device, crucial in early radios’ ‘Thing the ancient Egyptians kept a mummy in’

Tot-states

You know a word, and you know you know it, but you can’t say it.. It is on the tip of the tongue..

Often, very partial phonological information available (initial sound, number of syllables)

Claim: preservation of syntactic information (grammatical gender, count noun or mass noun)

Interpretation: lemma retrieval succeeds, but lexeme retrieval fails.

Aphasia

Certain patients, when confronted with a picture of say a pair of scissors:

Ah, I know what it is, it is for ..you know (gestures cutting movement) if there is something in the paper you want to keep you use one to cut it out..

This syndrome is called Anomia

Aphasia

Henaff Gonon et al., 1989 French patient who could not name pictures of objects;

but when asked for the gender of these nouns, he was remarkably accurate.

Miozzo & Caramazza, 1997 similar Italian patients: showed preservation of the

gender of nouns, and of the auxiliary for verbs for pictures he could not name.

Interpretation: correct access to lemma; problems in retrieving lexeme.

Evoked potentials

Electrodes attached to surface of the scalp, registering tiny fluctuations in voltage (EEG)

Time-locking to certain events (I.e., presentation of a certain type of stimulus) and averaging over many trials.

Resulting waveforms vary with stimulus conditions: N400 : semantic anomalies in comprehension P600: syntactic violations in production ERN: error responses in forced-choice paradigms LRP: preparation of motor output, specific for

each side of body (I.e., for each hemisphere)

Evoked potentials

Van Turennout et al., 1997; 1998: Picture of a TIGER Phonological decision AND Gender decision One decision: respond or not respond Another decision: Left hand or Right hand

LRPs: is their preparation in ‘not respond’ trials?

Yes, if gender determines which hand, and phoneme determines whether to respond or not

No, if gender determines whether to respond or not. Thus, after gender selection you can halt and decide

not to prepare.

Summary

There is quite a lot of evidence for the assumption for two sequential stages in lexical access:

(1) Determining a semantically and syntactically specified representation (irrespective of phonology)

(2) Determining a phonologically specified representation (irrespective of semantics/syntax)

But is this compelling evidence for the LRM model??

A. Do we need a lemma? Caramazza & Miozzo, 1997. No! The bulk of the evidence is

also compatible with a model without a lemma. Starreveld & LaHeij, 1995; Starreveld, 2000: Picture/ Word

interference: 1. Interaction between semantic/phonological relatedness 2. Early phonological effects

B. Do we need a lexical concept? LRM: Otherwise, we run into the ‘hyperonym’ problem.

Anything that should activate CAT should also activate ANIMAL

However: there are connectionist solutions to that problem.

top related