Quiz #2 Flashcards
Garret’s model of speech production
Message–>functional (semantic and syntactic relationship of content words)–>positional (function words inserted) –>sound (phonological forms)–>articulatory instructions
Lexicalization: 2-stage process
- Lemma level
2. Lexeme level
What is the lemma level?
Meaning-based: semantics, “pre-phonological form” (concept)
What is the lexeme level?
Sound-based: phonology
Repetition-primed
Given the definition (“It rises in the east and sets in the west”)
Homophone priming
Word form is generated in response to a different meaning (“Like father like ______”)
What happens in the stroop test?
Interference of input perception on language production
What does TOT tell us?
- Conceptualization and semantic info come first
- Phonological form later
- Info can be partially accessible
- Occurs when target word is a low-frequency word
Interloper
Phonological neighbors
Partial activation
Weak connection between semantic and phonological form of intended word
Blocking hypothesis
Target word suppressed by another strong competitor, TOT arises due to competition
What is planned in the processing stage of speech?
- Sentence structure: word order, sentence type
- Concrete items placed early in sentence
- Grammatical role: subject, verb, object
- Use pauses
Evidence for syntactic planning
Visual world paradigm: looking at objects while producing sentences
Coarticulation
The way the brain organizes sequences of vowels and consonants, interweaving into a smooth whole
Acoustic cues used in perception (3)
- Duration of sound
- Spectral properties: pitch, resonant frequencies (formants), amplitude
- Spectral transitions from one sound to another
Context cues used in perception (3)
- Phonetic
- Syntactic
- Pragmatic/semantic
Audio-visual integration cues used in perception (2)
- Mouth movement
2. Spellings
Categorical perception
We tend to perceive “categories”
- Members with subtle differences within the class will most likely still be perceived as same category
- Perceptual boundaries can be shifted
Voice onset time
Length of time between release of a stop consonant and the onset of voicing
Phoneme restoration effect
Even when sounds are missing, speech signal can be restored by brain and may appear to be heard, uses top-down processing
Sound splicing
Create a nonword from 2 different “base” forms (ex: smob from smog, smob from smod-lexical decision task)
Analysis by synthesis
When attempting to synthesize certain speech sounds, one can find out whether previous knowledge about these speech sounds was correct
Problems with analysis by synthesis
- Unclear or opaque mapping from what we mimic in our mind onto actual input
- How to account for clear articulation of words in improbable context
Motor theory of speech perception
- Perceive articulatory (vocal tract) gestures
- We mentally compute what gestures are needed for the sounds perceived
- Evidence from neuroimaging studies
McGurk Effect
Mismatch between auditory and visual information, visual wins
Auditory lexical decision task (French words)
Words that contain rhymes with multiple possible spellings yield slower responses than those with only one possible spelling
Orthographic effect on speech perception
- 2 transcription systems for Chinese
- Phoneme monitoring task-decide if syllable contains [i]
Implications of orthographic effect
- Literacy training and phoneme awareness
- Orthographic effect can occur pre-lexically (after low-level processing but before meaning is retrieved)
Difference between perception and word recognition
From low-level processing (sensory input) to retrieval of meanings (lexical access)
AX discrimination task
Determine if 2 sounds/syllables/nonwords are the same
ABX categorization task
Determine if the 3rd stimulus (X) is more similar to the first (A) or second (B)
Experimental paradigms involving word recognition (3)
- Lexical decision task
- Word spotting task
- Gating task
Gating task
- Paradigm used to study spoken word recognition
- Words are presented segment by segment incrementally
- Isolation point of recognition
- Support for cohort model
Isolation point of recognition
Point at which a word (during gating task) is recognized
Cohort model
When a person hears a speech segment, each segment activates every word in the lexicon beginning with that segment and rules out words as it keeps going
TRACE model
- Connectionist and interactive model
- Simulations are predictions about how a human brain processes speech sounds and words as they are heard in real time
- Emphasis on top-down influence
- 3 levels: 1. features 2. phonemes 3. words
Bottom-up uses:
Perceptual cues, features, transitional cues
Top-down uses:
Semantic, syntactic, pragmatic contexts
Inter-speaker variability
- Pronunciation
- Lexical choices
- Syntactic choices
- Social choices (politeness)
- Assumptions about common ground
Phonetic adaptation
Adjust your categories to fit a new speaker’s productions
How do we study speech perception (3)?
- Categorization
- Discrimination-measures accuracy
- Transcription-measures accuracy and reaction time (RT)
Is adaption just a low-level process applied regardless of context?
No
- Uses world knowledge and causal inference
- Uses prior knowledge
“Explain away”
Comes up with a reason as to why word sounded weird (ex: pen in mouth), less/no adaptation
Can listeners recover from the wrong interference?
Yes, listeners’ perceptual behavior is actually quite sophisticated, use general purpose world knowledge and casual reasoning
fMRI uses:
- Measure individual voxels (3D pixels)
- Get time series of BOLD signal over time
Categorical and gradient
- Representations in categorical areas (IFG) can change during perceptual adaptation
- This change lags behind behavior
