Fuzzy Logic Model of Perception (FLMP)

Key point: We try to categorize speech sounds

Categorization involves fuzzy boundaries

Categorization process in FLMP

  1. Evaluation

    2. Sound and vocal movements are analyzed

  2. Integration

    3. Features of acoustic signal are compared against prototypes in memory (Feature matching)

  3. Decision

Classify the sound based on overall evidence

 

Dual route model

McClelland and Rumelhart model posits a direct route from orthography to words (lexicon).

Do we also have an assembled route?

Letters (Syllables) ‡ Speech sounds ‡ words

Evidence that we use both routes.

 

 

Evidence for Dual-routes

Frost, Katz & Bentin (1987)

1. Hebrew: Deep orthography (vowels optional)
  1. English Moderately deep orthography (letters don't map perfectly to phonemes)
  2. Serbo-Croatian Shallow orthography (one-to-one correspondence of letters to phonemes)

Task: Naming target words and whether word was primed by a category name:

Bird Robin

Results: Size of semantic priming effects varies with orthographic depth

No effect of priming in Serbo-Croatian, because they used the assembled route and the pronunciation was available before the meaning was activated in the lexicon.

Strong effects of priming in Hebrew, lesser priming effects in English

Conclusion:

Hebrew readers rely on direct route to lexicon

Serbo-Croatian use assembled route

English use both

VanOrden (1987)

Task: read category name and then word and say whether it fits the category

Subjects are slower at food-meet than food-melt

Why? Conflict between visual information and phonological information.

 

 

Short note on Dyslexia

Many dyslexics (about 60%) show problems with using the phonological route.

Difficulty on going from printed word to phonemic representation to meaning of a word.

Because of poor bottom-up processing, they often rely on more top-down processing.