Layla - Words Sentecnes And Meaning Flashcards
Speech segmentation
Words rarely produced in isolation, speech is transient and sounds tend to run together, multiple segmentations possible e.g. jimsawthecargo could be car go or cargo. Harder w fast speaker. Variability in voice, accents, rate, called noise. Also how carefully words are pronounced e.g. probably becomes probly. Variability in how phonemesa ctually sound, differ based on surrounding one’s called assimilation e.g. don’t be silly becomes dom be silly.
Things that help speech segmentation
Pauses to establish word boundaries, stress patterns- English words usually start with stressed syllables. Phonotactics: some clusters are illegal within a word e.g. tr can be at the start but not the end. Prosodic cues: syllables onset are longer than medial syllables. Length of initial syllable influenced by word length e.g. cap vs captain. Cues differ depending on speaker and people use diff cues
Stages of spoken word recognition
Have activation then selection then integration. Start with spoken input, id the phonemes, activation of lexical candidates, selection where you choose one activated candidate and integration where you take stored knowledge about a word and integrate w sentence context. Selection leads to lexical acess- acess to stored knowledge. Don’t know if Erika acess (wait till hear the whole word, find match and acess meaning) or parallel/incremental (guess word as quick as possible), sounds then partial matches, shortlist as more input comes in . Serial accurate but slow, when do you stop. Parallel is fast but may need revision
Gating studies- grosjean 1980
Hear diff segments of the same word, guess what the word is and rate how confident. More possibilities the less of the word you hear, can immediately acess words in lexicon, multiple options for selection. Isolation point is when listener chooses the target w little confidence, recognition point is when get target w confidence. Uniqueness point is theoretical point where target is the only possibility- w: requires listeners to think about what words they hear. Evidence for parallel
Context in word recognition
Knowledge of words can provide context cues egg, when hear section running together we know spee isn’t a word but speech is. Use more when single has mor noise. Don’t know is context use is autonomous/ where context can t have an effect until word recognised but interactive says context can influence word recoigion
Warren 1970 - evidence for interaction
Ps asked to say where the noise was and whether the sound replaced the phoneme, ps said all phonemes heard and misplaced the noise, ppl hear the full world- phonemes restoration effect. Warra and warran 1970: ps hear diff phonemes depdenit on context e.g. peel on orange but heel on shoe. Gaining effect, mcgurk effect for non lexical input-mouth. Focusing on words can make you hear those words- green needle . Grosjean 1980: camel vs kids rode the canoe vs rode the canoe at the zoo (no/short/long context) and 1/2;3 syllables. Mean time for isolation point was highest for no context and 3, then short then long- context helps us hear words faster
Bilingualism
Increasingly becoming the norm 54% of Europeans, diff in the amount of words possible from none syllable. Lang selective acess: only words in target lang are candidates for activation of language for no selective acess: words in both langs are candidates for activation. Evidence from eye tracking- spivey and Marian- look up?
Reading
Psemantic priming: when read/hear a word, you activate words related in meaning/lexical decision task where ps read string of letters w real or fake words, faster to identify real after word w related meaning. Eye tracking- see when told to click on one, look at related-facilitation. Mediated semantic priming is primed word speeds target id because of 3rd variable e.g. lion and stripes.Inhibition: if prime w stiff, slower to id still. Neighbourhood density effects- harder to get word from memory is similar to many of words
