Classifying Speech Errors

Production Exchange Errors

Feature
Phoneme
Clusters of phonemes
Affixes
Syllable
Stems
Lexical Word
Constituent Phrase

Error Bank:

Glear plue sky; I disregard this as precise; blushing crow; I randomed some samply; you hissed all my mystery lectures; I got into this guy with a discussion; pig and vat; serp is souved; the Lord is a shoving leopard; Guess whose mind came to name?; Seymour sliced a knife with the salami; Hey, joke, have you heard the Mike about…?; stick neff; tasted the whole worm; I hate raining on a hitchy day; There are many churches in our minister; heft lemisphere; I’d hear one if I knew it; burst of beaden; Children interfere with your nife lite

Production Blending Errors

Phoneme
Cluster of phonemes
Syllables

Error Bank:

Frowl (frown/scowl); The competition is a little stougher (stiffer/tougher); the choung (children/young) of today; It’s difficult to valify (validate/verify); A swip chick (swinging/hip); splisters (splinters/blisters); It’s a lot of brothel (bother/trouble)

Anticipation/Perseverance Errors

Anticipation
Perseverance

Error Bank:

God rest re (ye) merry gentlemen; The mirst (first) of May; The firing (hiring) of the minority faculty; Bake (take) my bike; He pulled a pantrum (tantrum); cats and cogs (dogs)

Connectionist approaches to modeling speech errors

Connections: Excitatory – send signals to keep firing

Inhibitory – send signals to stop firing

What motivates these signals?

Excitatory: (between levels)

Semantic meaning  to cat and other related animals.

cat receives the strongest activation

cat sends signals to the phonemes that are in that word

rat will receive less activation, and will send signals to the phonemes in the words

dog also receives some activation

Inhibitory: (within levels)

The most activated node in each level sends a signal to everyone else to “turn off”!

The result –the only signals left firing are between the “message” for cat, the word cat, and the phonemes /k/, /ae/, and /t/. (If everything goes right)

What happens if something goes wrong?

How can you get this system to produce errors?

What if certain connections were weaker (between semantic/word level and the word/phoneme level?)

Would you expect to see different patterns of errors in aphasics (suffering from brain damage) who lost more of a tie to sounds/meanings?

Where does this diagram fit in the 3-stage model of speech production? (conceptualization, formulation: functional/positional level, articulation)