Speech Flashcards
what is the structure of language?
- phonetics
- phonology
- morphology
- syntax
- semantics
- pragmatics
what is phonetics
-describing and classifying speech sounds
-these aren’t language specific
what is phonology
- sound system
- speech sounds of a particular language
-even if you don’t know the language, you know how to pronounce it
what is morphology
- word formation
-e.g. combines stem and suffix
What is syntax
- sentence structure
what is semantics
meaning
what is pragmatics
-language in context
-how language is used in context
-e.g. using language to convey blame
what do phonetics consist of?
- articulatory (how speech sounds are produced)
- auditory (how speech sounds are perceived)
- acoustic (the physical properties of sounds)
distinction between letter and sound
-Each letter stands for a single sound but only sometimes
-Sometimes each sound is not represented by the same letter or letter combination
- e.g. The ‘e’ is spelled differently in ‘he’ and ‘seas’
what is phonology concerned with?
-the way speech sounds form systems within a language
- so the phonology of English is different to the phonology of Japanese, only phonetics would stay the same
what are phones?
-the inventory of phonetic segments and distinct [sounds] in a language
- regardless of whether the exact sound is critical to the meanings of words
- the physical characteristics of a phoneme
what are phonemes?
-the /smallest/ sounds in language that distinguish between words
-if swapped with another phoneme, could change one word to another
what is used to determine between the phonemes of a language?
-minimal pairs
- e.g sip and zip
how can languages differ in regard to phonemes?
-in which phones they choose to use as phonemes
- when two sounds are allophones (variations of the same phoneme) it is difficult to distinguish between them
-e.g. In Thai, the use of an excavated puh sound changes the meaning of the word as they are two different phonemes. English doesn’t making them allophones
process of language production, from thought to speech:
- conceptualisation/ preverbal
- formulation
- articulation
- self-monitoring
what does conceptualisation consist of?
-prelinguistic message planning of what to express
-language neutral (Pinkers mentalese)
Whats another word for word selection
lemmas
Whats another word for sound processing
lexemes
what does formulation consist of?
determining how to express information by using lemmas and lexemes
- lexicalisation (picking out the right words )
- syntactic planning (putting into syntactic frame)
- phonological encoding (activates sounds of word)
- phonetic planning (preparing to say word)
what is articulation?
expressing/pronouncing information
what component did weaver introduce, and what does this involve?
self-monitoring
- internal monitoring of what you’re going to say
- external monitoring that exists during speech
What is evidence for speech monitoring
- speech errors
-tip of the tongue - picture naming
- picture-word interference
how many speech sounds are made per second?
around 15
why do large speech errors occur?
-is it a fast automatic process
-less attention is paid to speech production rather than comprehension so we pick up errors later
what do speech errors tell us
- a persons capacity for using language and its components
where do speech errors occur
all levels
- phoneme, morpheme and words
type of speech error: shift
- in case she DECIDE to HITS it
- The s shifted from one word to another word
type of speech error: exchange
-fancy getting your MODEL RENOSED
-two words shifted (remodeled )
type of speech error: anticipation
-BAKE my bike
-Where you anticipate an upcoming sound, bake and bike are similar causing this error
type of speech error: perseveration
-he pulled a PANTRUM
- The same sound that you already produced is used for a bit longer
type of speech error: addition
i didn’t explain this CLAREFULLY
- another sound is added
type of speech error: deletion
mutter INTELLIGIBLY
- part of word/sound is deleted (unintelligibly)
type of speech error: substitution
it’s too LIGHT
- word is substitutted for a different word (heavy)
type of speech error: blend
john is quite CABLE
- two words are activated and used together (calm/stable)
what are common properties of speech errors
- phonemes exchange in similar positions
- consonant wont be replaced for a vowel (vice versa)
- when a word that doesn’t exist is introduced, it still follows the phonological rules of language
what do common speech errors suggest?
there are two different processes:
1. word retrieval
2. syntactic frame being constructed (tenses, plurals, and grammar)
components of word errors
-not restricted by distance
always happen of the same type
-happens early in speech production
components of sound errors
-made closer together
can cross word types
- happens later in speech production
what does garrett’s model of speech production break formulation stage into?
1.functional level
-2. positional level
3. sound level
What is the functional level of the formulation stage
lexical/word selection
-location of word errors
- as words haven’t been produced and put into syntactic frame, error can happen anywhere
What is the positional level of the formulation stage
-grammatical encoding
-puts words in correct word order
What is the sound level of the formulation stage (Garrets model of speech production)
-sound form encoding
-location of sound errors
what does lexicalisation consist of?
a two-stage retrieval process
1. meaning (lemma)
2. form (lexeme)
semantic word substitution errors
-glass/cup
- meanings are related
phonological word substitution errors
-historical/hysterical
- sound of two words are related
what do different types of word substitution errors suggest?
semantic and phonological processes are seperate
Neurological evidence for lexicalisation
double disassociation for retrieving lexical meaning and lexical form
what does tip-of-the-tongue involve?
remembering partial information, e.g., initials, sounds, syllables, and stress patterns
what can ToTs result in?
phonologically related words (interlopers) being activated instead
how can lexicalisation explain ToTs?
failure in the second stage of lexicalisation
1. lemma (meaning/syntactic)
2. lexeme (form and sound)
failure in the lexeme stage results in having meaning without sound
the blocking hypothesis believes…
interlopers prevent activation of the correct word
transmission deficit hypothesis believes…
weak links between meaning and word form result in limited activation of the correct word
which hypothesis is favoured by evidence?
the transmission-deficit account
- ToTs do not increase alongside phonological neighbors, actually the opposite happens with phonological neighbor acting as retrieval cues (going against the blocking hypothesis)
Evidence for transmission-deficit account ( bilingualism )
- bilingual speakers have more ToT because they have weaker links between meaning/sound as they dont use the language as much as native speakers
Evidence for transmission-deficit account (syntax properties vs sound)
-the transmission account confirms that both syntax properties and sounds are two separate things
- Gender is non semantic in Italian so we can refer to something masculine using a feminine gender
- So gender doesn’t tell you if the meaning of the word if masc or fem
- When Italians are in a ToT state they can still remember the grammatical gender of the word
Evidence for transmission-deficit account (dyslexic children)
- have more ToT
- have no difference in recalling semantic meaning but more errors at the phonological stage
- this causes a weaker link between both of them
picture naming findings
- long-term facilitation for naming lasts over many trials
- no effect for homophone priming (sound without meaning) neither if we activated meaning without sound
-It is the link between the lemma and lexeme that needs to be activated , we need to have both for successful priming
what does picture word interference show?
- phonologically related words/ distractors result in faster naming
- semantically related words/ distractors lead to slower naming
schriefers et al (1990) in picture word interference
- early semantic priming resulted in inhibition/ takes longer
- later phonological priming resulted in facilitation/ faster
discrete lexicalisation
-semantics (meaning) are activated before phonology
- it is selected based on semantic info first then phonological form is retrieved
interactive lexicalisation: cascading
-activation flows to the form lexicon before a single lemma (word/meaning) has been selected
- multiple word forms are activated
interactive lexicalisation: feedback
activation cascades down and feeds back to the above level
- this results in lemma activation of related word forms
what is mediated priming
- When something has an effect on something else not because of what it is in itself, but mediated by the step in between
- e.g. hearing lion then asking if stripes is a word or not
- lion doesn’t have stripes but ‘lion’ activates tiger as they are closely related
- These have stripes and so it helps us say stripes is a real word
Evidence against cascading/feedback (Levelt et al., 1991)
- picture of sheep shown, mediated priming used
- found early inhibatory priming of ‘goat’ (semantic) at the lemma level
- found late priming of ‘sheet ‘(phonological) at the lexeme level
- found no priming of ‘goal’ (Unrelated word) , supports discrete
- if the cascading model is correct , we expect sheet to be fast but also ‘goal’ as we activated goat
evidence in support of cascading
- near synonyms acted as a successful prime
what does feedback activation suggest?
errors are not random
lexical bias
- sound speech errors result in words more often by chance
similarity effects
- mixed substitution errors (semantically and phonologically related) more often than chance
- e.g. comma –> colon
feedback activation support (lexical bias)
- If it was serially from the meaning to the sound level, there should be no lexical bias
- If you were to make an error, it would just be a phonological error and wrong sound would just be made
HOWEVER - If you make a phonological error it results in real words much more often than non-words
- So there must be a feedback from the sound level back to the meaning level
What do people propose instead of feedback
-an output monitor
- we monitor what were going to say and if we notice something that isnt a word we substitute it for a real word
evidence of the linguistic genius of babies
- critical period for language acquisition (0-7y) decreases with age
- babies have the ability to discriminate between all sounds from all languages due to them taking statistics which they are sensitive to
- exposure to real life mandarin improved percentage of sounds correct (no effect of TV/audio exposure)
- social brain is in control of if and when babies take note of statistics
when is information encoded?
in the first 50-60ms of fixating, and this duration is also dependent on phonology (sound systems)
subvocalisation and what does it show us
- this is our inner speech
- having more stressed syllables can increase word reading time, even when reading silently
- subvocalisation during silent reading can also reflect regional accents
- shows we activate sound when when reading silently
stutterers had ______ ________ during silent reading for words they tend to stutter on? what does this mean
-longer fixations
- means stuttering event already started before the actual output
34% of adults with dyslexia stuttered as a child, what moderated this?
- this was moderated by the severity of their dyslexia
50% of adults with stutters…
fulfilled the criteria of dyslexia, despite never being diagnosed
