PSY2002 SEMESTER 1 - WEEK 1 Flashcards
define model
simplified/idealised representation of thing
explain difference between statistical and theoretical cognition model
statistical does math relationship between variables held under specific assumptions
theoretical cognition describes relationship between different mental processes that makes assumptions about nature of these processes
give example for a computational model
implemented as functioning computational models
eg; neural network with node representing group of neuron
name 2 strategies to reduce processing demands while planning what next to then say
preformulation, underspecification
define preformulation
producing phrases used before
define underspecification
using simplified expressions in which full meaning is not explicit
what are hesistations for and why can we use them to test speechs production
how long it takes to initiate word (find and plan relevant syntax)
explain tip of tongue state
unable to retrieve necessary word, due to planned sentence and structure, with syntax in place but specific levels of processing is required to produce speech
name 3 levels of tip-of-tonge state
semantic processing
syntactic and morphemic processing
articulation (form processing)
speech error tend to occur within specific process
when do speech errors tend to occur
within following processes, rather than crossing boundaries between them:
1. semantic processing
2. syntactic and morphemic processing
3. articulation (form processing)
name errors at semantic processing level
semantic blend error at conceptual level eg; spork
or semantic substitution- correct word replaced by word of similar meaning eg, bat/raquet
name errors in syntactic and morphemic processing
syntactic and morphemic exchange errors at formulation level eg; slicly thinned
inflection/suffix remains in place but attached to wrong words eg; already suitcased 2 packs (already packed 2 suitcases)
name errors in articulation processing levels
word and phoneme exchange eg; forks of a prong
what is Fromkin 1971 model of utterance generation (5 stages)
serial processing (discrete)= each stage must be completed before next can begin
1. meaning generated (eg; prongs of fork)
2. syntactic structure generated and associated with semantic features (work out layout)
3. position of intonations placed (first article ‘prong’, second ‘fork’)
4. lexicon look up find word, generate phonological segment (finds words prongs and fork)
5. morphophonemic constraints added, phonemes selected for utterance (the, of, a)
give an overview of Dells Spreading Activation Theory (computational)
spreading activation results in activation of related items at all level of processing at same time, cascaded interactive model, semantic and phonemic info processed simultaneously, processes at one level can influence processes at other level
name 4 levels of speech production
- semantic level
- syntactic level
- morphological level
- phonological level
what is the semantic level of speech production
meaning of what is to be said - planning level
what is the syntactic level of speech production
grammatical structure of words in planned utterance
what is the morphological level of speech production
morphemes (basic units of meaning)
what is the phonological level of speech production
phonemes (basic units of sound)
what is a anticipatory speech error?
speech sound made too early in sentence (Tanadian, from Toronto) due to many of words in sentence being activated during speech planning, sometimes later word more activated than correct one
what is a perservatory speech error?
sounds of words spoken later than should be, reflecting planning/monitor failure (eg; red cet, red cat)
name 2 types of speech error predicted by spreading-activatio theory
- anticipatory
- perservatory
what is mixed-error effect
incorrect spoken word semantically and phonemically related to correct word suggesting various levels of processing interacts flexibly and also suggests semantic and phonological factors can influence word selection at same time
what is lexical bias effect
speech errors tend to consist of actual words, not non-words due to it being easier for words to be activated as have representations in lexicon
outline Ferreira and Griffin study into mixed error effect
ppt name target pic after reading sentence priming semantic competitor, semantic and phonological competitor or unrelated word
semantic = nun (woman went to convent to become__)
semantic+phonological= none (thought there’d be cookie, there were__)
unrelated= match (he lit candles with just one__)
for semantic competitor, nun and priest activated but ppt chose nun as primed to think of it
for semantic and phonological, interactive model combined influence of semantic/phonological similarity caussing error. priest spreads activation to nun, reinforce phonology none
demonstrate semantic and phonological similarity can result in speech errors suggesting processing is fluid and interactive
when are speech errors most likely?
when speaker not formed coherent speech plan. few anticipatory error as overall error rate will correlate negatively with anticipatory proportion
name strengths of Dell spreading activation
- convincing account of speech errors, evidence from lab-induced error suggests processing interactive (however some suggests parallel processing in more speech errors)
- mixed error and lexical bias effects indicates processing associated with speech production is high interactive
- accounts for other speech error types
name weaknesses of Dell spreading activation
- de-emphasise processes involved in construction of message and intended meanings
- doesn’t predict time taken to produce correct and incorrect words
- interactive process is less apparent in error-free data than speech-error data
what does WEAVER stand for?
Word-Form Encoding by Activation and VERification
outline Weaver (levelt) model, computational
processing items which are semantically similar will be inhibitive
discrete, feedforward activation-spreading model (one-direction, process must finish before start next) and serial (one at time)
name 3 levels of WEAVER
- conceptual level- nodes representing lexical concepts
- nodes representing lemma, from mental lexicon, activate concept then connect with syntactic info (noun, requires article of “the”)
- nodes representing word forms in terms of morphemes and their phonemes
define lemma
word representation linking with word form, specified syntactically, semantically but not phonologically
in WEAVER model, explain its competitive processes
selection of items at each level is based on competition between items
once has been selected, node sends inhibitory signals to other related concepts to stop activation (serial fashion)
predicts items which are semantically similar inhibits processing eg; DOG lemma being activated inhibits cat lemma processing
outline Wheeldon & Monsell (1994) experiment for WEAVER model
testing speech production so ppts asked to answer question out loud then name picture
answer to question either semantically/unrelated to pic eg; whale living in sea, apple/pear
found when semantically related, longer RT
when do speech errors in levelt model occur
if wrong concept is selected and processed to phonological level, can be avoided by checking mechanism
how do WEAVER and Dell differ regarding speech errors
weaver account for inhibitory effects in speech production, Dell does not
give strengths of WEAVER
- lemma contain syntactic info (TOT, Italian speakers know grammatical words gender)
- word production involves series of stages from lexical selection, morphological encoding, phonological encoding is accurate
- shifted balance of research away from error to precise timing of words production
- simple, elegant, easier to test than Dell
give weaknesses of WEAVER
- narrow focus on processes involved in production of single word, meaning several processes involved in planning + producing entire sentence ignored
- more interactions between differing processing levels than in WEAVER
- evidence from speech errors (eg: mixed-error effect, lexical bias effect, word-exchange errors) suggests parallel processing during speech production occur to greater extents
if processing of spoken words is discrete and feedforward (levelt) then how is phonological processing mean to be occurring
phonological process should only occur for items that have been selected and should not be affected by non target info
if processing of spoken words is interactive and parallel (Dell) then how should phonological processing occur
phonological processing is likely to occur for all items that are related by semantics and/or phonology to target item
outline Meyer & Damians study for phonologically related item
ppts shown 2 items (overlays) and instructed to name green, and ignore red
phonologically related (dog/doll), not related (doll/bat)
if Dell (interactive cascading), should process both in parallel but if Levelt (competitive discrete), should ignore other and only process down the selected one
what did Meyer, Damien conclude
- selection of concepts is competitive
- evidence for lemma
- multiple phonological representations can be active at once
- semantics can influence phonology and vice versa
