The Internal Lexicon

1
The Internal Lexicon
2
Word Knowledge

• phonological
• syntactic
• morphological
• (e.g., pronunciate, orientate, typewrite,
  baby-sit, edit, ush, pea, matchmake)
• semantic
• sense vs. reference
• possible world semantics / mental models
• truth value

3
TOT demonstration

• What is the word for
• a childrens doctor?
• the counting machine that uses sliding counters
  along rods or grooves?

4
TOT demonstration

• Pronounce each word and rate pronunciation
  difficulty on a scale of 1 (very easy) to 5 (very
  difficult)
• indigent
• rejoicing
• serious
• tappet
• abstract
• velvet
• truncate
• tradition
• publish
• locate

5

• What word means to formally renounce a throne?

6
abdicate

• How many of you had the TOT state?
• Could you recall part of the word?

7
What causes TOT-state?

• Dense Phonological neighborhoods
• (lots of words that sound alike)
• We induce TOT by priming (activating) those
  neighborhoods so theyll interfere with
  phonological retrieval
• indigent
• rejoicing
• serious
• tappet
• abstract
• velvet
• truncate
• tradition
• publish
• locate

Target word Abdicate Notice that we had you
pronounce every part of the word, by having you
pronounce the list of words before!
8
Semantic Relationships

• synonymy (fear panic)
• coordination (cat dog)
• hypernymy (bird sparrow)
• hyponymy (sparrow bird)
• meronymy (seat chair)
• denotation vs. connotation
• bachelor/spinster

9
In-class experiment

• Write down each word on this list, along with the
  next word that comes to mind
• Doctor
• King
• Lets go over
• Always remember
• The child threw

10
In-class experiment

• Then write down
• the name of a fruit
• the name of a vegetable
• the name of a color

11
In-class experiment

• How many of you gave these answers?
• Doctor ? Nurse
• King ? Queen
• Lets go over ? to/there
• Always remember ? to/that
• The child threw ? up/the ball
• (fruit) ? apple
• (vegetable) ? carrot
• (color) ? red

12
Lexical retrieval

• Interactive or discrete?
• Interactive phonology influences semantic
  choices, and semantics affect phonological
  retrieval
• Discrete Words are selected based on their
  semantics

13
MessageFormulationLexical SelectionMorpholo
gy RulesPhonological RulesVocal Tract
Instructions
A simple, serial (non-interactive) model!
14
An interactive model

• The arrows point both ways. There is feedback
  between levels!
• (Bock Levelt, 1994)

15
How do we know that we retrieve words in two
stages?

• Repetition priming lasts over 100 trials, but
  only if it has the same phonology, syntax, and
  semantics
• Priming homophones
• weight for wait is not as strong. Neither is
  that as demonstrative That is a huge cup of
  coffee vs. that as complementizer I know that
  she will pass all her classes. (so its not just
  phonological)
• No priming across languages, as long as the
  phonological forms differ
• Monsell, Matthews, and Miller (1992) bilingual
  Welsh-English speakers. (so its not just
  semantic)

16
Lexical Access Models

• Autonomous Search Model
• like a dictionary search, using orthography or
  phonetics
• search most frequent words first
• no semantic/syntactic influence on selection
• points to entry in master lexicon
• Logogen Model
• each word has a logogen, which specifies its
  attributes
• logogen activated via sensory input or context
• context lowers the activation threshold
• Cohort Model
• initial activation is strictly bottom-up
• once cohort is activated, frequency and context
  come into play

17
Lexical decision task
18

• gambastya

19

• gambastya

20

• revery

21

• voitle

22

• chard

23

• wefe

24

• cratily

25

• decoy

26

• puldow

27

• raflot

28

• oriole

29

• voluble

30

• boovle

31

• chalt

32

• awry

33

• signet

34

• trave

35

• crock

36

• cryptic

37

• ewe

38

• himpola

39
Lexical decision task, pt. 2

• (frequent words)

40

• mulvow

41

• governor

42

• bless

43

• tuglety

44

• gare

45

• relief

46

• ruftily

47

• history

48

• pindle

49

• develop

50

• gardot

51

• norve

52

• busy

53

• effort

54

• garvola

55

• match

56

• sard

57

• pleasant

58

• coin

59

• maisle

60
How do we know that we retrieve words in two
stages?

• Tip-of-the-Tongue State
• Know the meaning, but cant produce word
• Syntax is retrieved. Speaker can reliably tell
• syntactic category
• gender of word (in languages with grammatical
  gender)
• Incomplete or faulty phonology
• cant reliably tell first letter of word

61
Why is lexical retrieval interactive?

• Neighborhood effects
• Dense semantic neighborhoods faster production
• Dense phonological neighborhoods slower
  production
• Semantic interference with speech errors
• Get one 4x more likely to be pronounced Wet
  gun if the speaker has recently read damp
  rifle
• Mixed errors are more common than youd expect by
  chance
• Bat or Rat instead of Cat

62
Disambiguating Homophones

• All meanings accessed at first, in all parts of
  speech
• (Swinney, 1979)
• Heard Rumor had it that, for years, the
  government building had been plagued with
  problems. The man was not surprised when he found
  several spiders, roaches, and other bugs 1 in
  the 2 corner of the room.
• Seen ant or spy or control sew
• Task Lexical decision
• Facilitated for both meanings at position 1,
  but only for the appropriate meaning at position
  2
• Context disambiguates after lexical retrieval