Sounds to Words Flashcards
What are some difficulties with recognising speech sounds?
speech, speaker variability (lack of invariance), segmentation problem and impoverished signal
What is categorical perception?
the categories we form about speech sounds affects how we then perceive them - this is why lack of invariance & accents are not a problem, they become categorised
What were the results of a word mispronunciation experiment?
errors were noticed;
more at beginnings of words
more in predictable sequences
more in certain types of sound switches
Briefly explain Forster’s serial search model
a serial model in which a specific word is in one place in the mental lexicon but can be determined from orthographic, phonological or semantic information. Associated words are linked within the lexicon and primed when neighbour is activated
What are some pro’s of Forster’s serial search model?
accounts for some priming and contextual effects
accounts for longer processing of potential non-words
What are some con’s of Forster’s serial search model?
doesn’t account for beginning of word effects
slow
Briefly explain Marsden-Wilson et al.’s Cohort model
sounds and letters activate as many relevant words as possible based on sounds/letters (bottom-up), further information whittles down candidates until only 1 is left
What are the 3 stages of the Cohort model?
access
selection
integration
What happens in the access stage of the cohort model?
auditory word detectors are activated as soon as the beginning of a word is perceived
top down processing used to generate semantically related items
bottom up processing used to generate phonologically related words
What happens in the selection stage of the cohort model?
the cohort gets smaller and smaller with increasing information, once the cohort reaches one possible word the recognition point is reached
What happens in the integration stage of the cohort model?
the syntactic and semantic properties of the word are used to integrate the word into our understanding of the sentence
What are some pro’s of the Cohort model?
allows very quick processing - recognition before they’ve finished speaking
accounts for nonsense words as cohort reaches 0
top down processing helps words be identified quicker
What are some con’s of the Cohort model?
doesn’t account for recognising word with missing beginning
Briefly describe Morton’s Logogen model
a connectionist model in which there is a logogen for every word which can be activated by perceptual and contextual features, the logogen build up information to reach a threshold which then leads to recognition
What are some pro’s of the logogen model?
accounts for frequency effects - threshold is lowered
accounts for priming - frequency effects are just long term priming
What are some con’s of the logogen model?
doesn’t account for deciding a word is not a word
does’t account for ability to read non/unknown words
Briefly describe Trace theory
a connectionist model in which consists of facilitory and inhibitory connections with word, phoneme and feature levels which can interact
What are some pro’s of Trace theory?
accounts for
- frequency effects
- slower response to non-words
- phonological/semantic priming
What are some con’s of Trace theory?
issues of time progression - too inefficient
How are consonants distinguished?
place and manner of articulation
voicing
How are vowels distinguished?
tongue height
tongue backness
lip rounding
What is lack of invariance?
the large variation between speakers pronunciation
differs based on regional/social accents, formality, context, emotions - fixed by categorical perception
What is segmentation?
speech is a continuous stream which must be broken up into distinct words
What is an impoverished signal?
we often don’t hear every acoustic signal, vowels are perceived more accurately than consonants
What is categorical perception?
perception of different things as being the same
What are fricatives?
a type of voiceless consonant produced by an air flow, place of articulation changes the phoneme
What are liquids?
a type of voiced consonant produced by a partial closure in the mouth, place of articulation changes the phoneme
When are mispronunciations noticed the most?
at the beginning of words
more in predictable sequences of words
more in certain types of voicing
What is a mondegreen?
the incorrect perception of the separate words
e.g. excuse my while i.. kiss the sky/kiss this guy
