Mental Lexicon Body of knowledge we hold in

Mental Lexicon Body of knowledge we hold in our minds about words Includes pronunciation, spelling, meaning syntactic roles Recognition of words—whether listening or reading Very Rapid For example, when listening, recognition is completed within about 275 msec even though the average word takes about 370 msec to say How is this accomplished? ? ? One model: Morton’s Direct Access Model

Morton’s Direct Access Model • Feature of words activate detectors for possible units (logogens) until one dominates • Like Treisman's Attenuation Model of Selective Attention or The Bruce-Young Model or The Embedded Processes Model -- It is based on the concept of thresholds • The logogen builds up inputs until its individual threshold level is reached; when it fires, the word is recognized • Lower thresholds relate to high frequency (Foss, 1969; Rayner & Duffy, 1986) recent activation context– semantic priming (Zola, 1984; Meyer & Schvaneveldt, 1971)

Lexical Decision Task: Is it a word? Vote Vate

Lexicalization The process by which the thought that underlies a word is turned into the sound of the word There are 2 stages: 1) A concept activates a lemma, an abstract representation that includes meaning and syntactic information 2) The lexeme is accessed—the phonological form of the word Tip of the Tongue Phenomenon = lemma without lexeme

Studying Sentence Comprehension An analysis of parsing by looking at structural ambiguities Lexical ambiguity example: “They threw stones at the bank. ” Structural ambiguity example: The professor gave a talk on Mars. ” Garden Path Sentences People have a strong tendency to construe (parse) the early portion in a way that the later portion shows to be incorrect e. g. , “I told the woman serving food was too hard for her. ” Immediacy Principle: Make decision about syntactic role of each word as we hear/read them

SENTENCE PRODUCTION Garrett’s Model Message Level Functional Level Positional Level Phonetic Level Articulation Level

Message Level Basic idea is formulated – NOT in verbal form

Functional Level Semantic representations and thematic roles for content words are chosen e. g. , agents, actions, objects

Positional Level Syntactic form of the sentence is specified and Phonological forms for function words and bound morphemes are selected

Phonetic Level Phonological forms (lexemes) of content words are specified and Inserted into the syntactic form already generated

Articulation Level Commands are created and sent to the vocal apparatus to speak

Using Garrett’s Model to explain • Slips of the tongue (including “Freudian” slips) • Pauses • Clinical Cases Jane EST Paul Keith

Butterworth (1975) had speakers give an unplanned monologue He found fluent and hesitant phases with cycles 10 -40 sec, M = 18 sec Pattern: Plan-execute, plan-execute, etc.

What Butterworth found about pauses during speech: • Idea boundaries tended to fall at the end of a fluent phase, beginning of a hesitant phase • words in the hesitant phase were not spoken more slowly; rather, the amount of time spent NOT talking increased • pauses (>250 msec) took up 35 -67% of the time (in interviews 4 -54%) • in general, about 40 -50% of the total “speaking time” was soundless • If pauses are reduced, the repetition of words & phrases doubled • For pauses that occur when looking at the listener, more false starts & repetitions

Sample Monologue with pauses, hesitations, etc. illustrating the Plan-execute cycle