Скачать презентацию Reading Model of word recognition Two Скачать презентацию Reading Model of word recognition Two

6abdb4d8ecc2d7760eca13dc745e06e8.ppt

  • Количество слайдов: 20

Reading • Model of word recognition • Two theories – indirect access – direct Reading • Model of word recognition • Two theories – indirect access – direct access • Individual differences • Comprehension – two measures – effects of prior knowledge – organization of the text • global coherence • local coherence – Kintsch’s model of comprehension and “readablity” Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 1

Word Recognition • The relationship between speech and reading – – silent letters introspective Word Recognition • The relationship between speech and reading – – silent letters introspective connection acoustic recoding what is the role of acoustic recoding in reading? Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 2

Model of Reading sentence Visual encoding Speech recoding Semantic Lexical access LTM Working memory Model of Reading sentence Visual encoding Speech recoding Semantic Lexical access LTM Working memory LTM Tentative propositional structure Store propositional structure Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 3

Two theories • Indirect access – must go through speech (acoustic recoding) to get Two theories • Indirect access – must go through speech (acoustic recoding) to get to meaning • Direct access – skilled readers get immediate visual access to meaning – faster, no recoding Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 4

Evidence For Indirect Access • EMG recordings of vocal apparatus while people read – Evidence For Indirect Access • EMG recordings of vocal apparatus while people read – people moving their lips while reading • Lexical decision task (is this a word) – phonemic similarity increases lexical RT – e. g. it takes longer to say “NO” to “brane” than to “melp” because “brane” is acoustically equal to “brain” – e. g. faster RT to “set - wet” (rhyming pair) than to “few -sew” (graphemically similar but pronounced differently • Errors during letter search and proof reading are influenced by phonemic factors – the silent versus pronounced “e”s – in proof reading “borst” for “burst” is easy, but “hurd” for “heard” is hard Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 5

Evidence For Indirect Access • People often sound-out words when material is difficult – Evidence For Indirect Access • People often sound-out words when material is difficult – when people are prevented from making lip movements they may have more trouble reading difficult material • Mistakes on categorizing homonyms – “rows” as a flower – “which” as a character in a Halloween story • Phonological readiness for reading – e. g. Kyle’s difficulty with phonological understanding Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 6

Evidence for Direct Access • Profoundly deaf can read – but have trouble learning Evidence for Direct Access • Profoundly deaf can read – but have trouble learning and never get as good as hearing readers • Profoundly deaf – not make same letter search errors • equal number of errors on silent and pronounced letters – not have same proofreading errors • Acquired phonemic dyslexia – can repeat words they have heard but cannot read out loud – can not sound-out phonemes – can get direct access to meaning but may be slightly “off”, e. g. read “dream” as “sleep” – also not have same types of proofreading errors as normal readers Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 7

Evidence for Direct Access • Semantic decision task – decisions about individual words not Evidence for Direct Access • Semantic decision task – decisions about individual words not affected by phonemic factors • e. g. vary number of syllables - “clerk” versus “secretary”, same RT when asked if an accupation • Use of homonyms – same sounds that lead to different meanings • In reading aloud “mown- down” and “horse - worse” are problems because of pronunciation, but not a problem in silent reading Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 8

Conclusion • Dual encoding – may speech recode – speech recoding may be necessary Conclusion • Dual encoding – may speech recode – speech recoding may be necessary for learning – skilled readers have direct access • like the horse race model – maybe direct visual access only for overlearned or automatized words • Why speech recode? – Facilitate working memory (ease storage) – especially for: • difficult passages • when memory will be tested • under stress Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 9

Individual Differences • What differentiates good, average, and poor readers? • Many different theories Individual Differences • What differentiates good, average, and poor readers? • Many different theories • Three processing differences – use of phonemic code – capacity of working memory – speed of letter encoding Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 10

Three processing differences • Use of phonemic code – good readers have greater reliance Three processing differences • Use of phonemic code – good readers have greater reliance on phonemic code • for second graders good readers how more interference from the rhyming words (set-wet, few-sew) than poor readers • general use of phonemic code first appears around age 5 – “reading readiness” • Capacity of working memory – reading span task • read aloud a series of sentences and then recall the last word of each sentence • measures how many sentences can be held in STM • normal span is 2 -5 sentences • correlated with reading comprehension scores and SATs Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 11

Three processing differences • Capacity continued – some forms of dyslexia are based on Three processing differences • Capacity continued – some forms of dyslexia are based on STM problems – What causes the differences? • More storage • faster processing • efficiency • can this be learned • Speed of letter encoding – Posner letter matching task • not faster at perceptual match • faster at same name and homophone task – not a perceptual processing difference, but a difference in access to phonetics and LTM Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 12

Reading Comprehension • Study how individual differences in prior knowledge and information processing characteristics Reading Comprehension • Study how individual differences in prior knowledge and information processing characteristics interact with the organization of ideas in the text • Two measures of comprehension – subjective - rate ease of understanding – objective - number of ideas recalled Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 13

Prior Knowledge • Context – picture as context, measure number of ideas recalled • Prior Knowledge • Context – picture as context, measure number of ideas recalled • no picture average 3. 6 ideas recalled • picture before text average 8. 0 ideas recalled • picture after text recalled 3. 6 ideas – context must come before the text • not just provide retrieval cues • provide organization – titles also act as context (familiar activities) Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 14

Prior Knowledge • Can lead to false recall – general passage about a man Prior Knowledge • Can lead to false recall – general passage about a man – half subjects are told the man is Gerald Martin and half are told it is about Adolf Hitler • the “Hitler” subjects primarily recalled information consistent with their prior knowledge of Hitler • after one week could not distinguish past knowledge of Hitler from information provided in the text Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 15

Prior Knowledge • Perspective influences recall – Anderson and Pritchert ‘s study of a Prior Knowledge • Perspective influences recall – Anderson and Pritchert ‘s study of a long story about two boys who play hookey and stay at one boy’s wealthy parents’ house. • lots of details about their valuable possessions • also details about the house being old with a leaky roof, damp basement and other problems – Subjects are told to recall ALL information, but • group one is told to take the perspective of a burglar • group two is told to take the perspective of a home buyer – groups recall information relevant to their perspective, then • half are given a new perspective • half are told to try again with same perspective – groups that switch perspective recall more Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 16

Prior Knowledge • Perspective continued – three reasons why a change in perspective may Prior Knowledge • Perspective continued – three reasons why a change in perspective may increase recall • guess ideas consistent with new perspective • not try to recall irrelevant info • perspective provides a plan for searching memory!!!!! Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 17

Organization of the text • Global coherence – main events, characters and goals – Organization of the text • Global coherence – main events, characters and goals – depends upon type of text • story grammars (goals, conflict, and resolution) • cultural differences in narratives (organization and flow) • text books and how to set up global coherence – history – subfields – major problems (questions) • Local coherence – integration of immediate ideas – causal connections – anaphoric reference Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 18

Kintsch’s Model of Comprehension • Simplified model – major point is that incoming information Kintsch’s Model of Comprehension • Simplified model – major point is that incoming information can be understood more easily if it can be integrated with information the reader has already encountered – easiest is when new information fits with information currently in working memory – if not currently in working memory must do a “reinstatement search” • look for related propositions in LTM and transfer them to working memory – hardest is when no relevant information is found during “reinstatement search” • must start a new association network – less well recalled – not well integrated Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 19

Kintsch’s Readability formula • Formula – readability is predicted by the number of recalled Kintsch’s Readability formula • Formula – readability is predicted by the number of recalled propositions divided by the reading time – this correlated best with • word frequency • number of required reinstatement searches • number of inferences required Sensation and Perception - reading. ppt © 2001 Laura Snodgrass, Ph. D. 20