PS2111-Cracking-orthographic-code-2024.pptx

Cracking the Orthographic Code Kevin Paterson [email protected] School of Psychology University of Leicester www2.le.ac.uk/departments/npb/people/kbp3 Introduction to studies of visual word recognition o Aim of the lecture: o To introduce research on how words are recognised in the brain o Outline how t...

Cracking the Orthographic Code Kevin Paterson [email protected] School of Psychology University of Leicester www2.le.ac.uk/departments/npb/people/kbp3 Introduction to studies of visual word recognition o Aim of the lecture: o To introduce research on how words are recognised in the brain o Outline how this has informed the development of computational models o Examine 4 phenomena: 1. Word superiority effect 2. Lexical frequency effect 3. Orthographic neighbour effects 4. Letter transposition effects Visual Word Recognition 1. Word superiority effect – evidence that letters are recognised more easily in words than in isolation. 2. Lexical frequency effect – evidence that words are encoded in terms of frequency of exposure. 3. Orthographic neighbour effects – evidence that similarity between words, in terms of their letter composition, can affect reading. 4. Letter transposition effects – evidence for flexibility in the encoding of letter position information. The Word Superiority Effect o Describes benefit for recognising letters in words. o Standard finding: Letters in words are recognised more accurately in word contexts than either pseudoword (i.e., not-word) contexts or when presented in isolation (Cattell, 1886; Reicher, 1968; Wheeler, 1970) The Reicher-Wheeler Task o Provides clear evidence for the word superiority effect. o Stimulus is displayed briefly then masked, after which the participants makes a forced choice between two options for a specific letter display (2AFC). o Letters are identified more accurately in word than nonword displays or when presented alone. o Evidence that word context facilitates letter recognition. Explaining the Word Superiority Effect o Interactive Activation Model (McClelland & Rumelhart (1981) provides mechanisms for both bottom-up and top-down influences on letter recognition. o Bottom-up input provides one source for letter identification – but remember this input is brief and masked and so degraded. o Top-down evidence from lexical level when the letter appears in a real word provides a further source. o These two sources combined provides stronger evidence for letter identify than one source alone - hence the effect. Visual Word Recognition 1. Word superiority effect – evidence that letters are recognised more easily in words than in isolation. 2. Lexical frequency effect – evidence that words are encoded in terms of frequency of exposure. 3. Orthographic neighbour effects – evidence that similarity between words, in terms of their letter composition, can affect reading. 4. Letter transposition effects – evidence for flexibility in the encoding of letter position information. Word Frequency Effects Readers are quicker to recognise common words than rare words. This indicates that the brain keeps track of word exposure statistics. - e.g., ANCIENT vs. ARCHAIC - Effect is observed in lexical decision task & eye fixation times on words in reading (e.g., Schilling, Rayner, & Chumbley, 1998) Visual Word Recognition 1. Word superiority effect – evidence that letters are recognised more easily in words than in isolation. 2. Lexical frequency effect – evidence that words are encoded in terms of frequency of exposure. 3. Orthographic neighbour effects – evidence that similarity between words, in terms of their letter composition, can affect reading. 4. Letter transposition effects – evidence for flexibility in the encoding of letter position information. Who plays Wordle? Orthographic Neighbours o The IAM model includes the assumption that letter input will active all words that share these letters in the same locations. o That is, input like WAVE will activate WAKE, SAVE, and WIRE, amongst many others. o These words are called orthographic ”neighbours” – defined that differ by only one letter when word length and letter position are preserved (Coltheart et al.,1977) o How does a word’s relationship with its orthographic neighour(s) influence word recognition? Helpful Neighbours (Andrews, 1989) o o o Some words have many neighbours, while others have few. yes no Words with many neighbours include “game”, “hand”, “word”. WORD Words with few neighbours include “folk”, “huge”, and ”size”. o Andrews examined lexical decision (word / not word judgements) and naming (saying a word aloud). o Faster responses for words large compared to small neighbourhoods, especially for lower frequency words. * o Neighbours can conspire to give feeling of knowing? Hostile Neighbours (Perea & Pollatsek, 1998) Target words were either: Words that have a higher frequency neighbour, e.g. ”space” is HFN of “spice” Words that DON’T have a higher frequency neighbour, e.g. ”sauce”. Slower responses for (low frequency) words with a HFN. Evidence for lexical competition in reading: HFN competes for selection with the presented word. Hostile Neighbours (Perea & Pollatsek (1998) Words with and without HFN were embedded in a sentence, which participants read while their eye movements were recorded. ”spice” has “space” as HFN, :sauce” has no HFN. More eye movements back to target words, and longer reading times, for targets with HFN. Do readers sometimes misread words as their HFN? Visual Word Recognition 1. Word superiority effect – evidence that letters are recognised more easily in words than in isolation. 2. Lexical frequency effect – evidence that words are encoded in terms of frequency of exposure. 3. Orthographic neighbour effects – evidence that similarity between words, in terms of their letter composition, can affect reading. 4. Letter transposition effects – evidence for flexibility in the encoding of letter position information. Can you do anagrams? An Email from Cambridge University Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe. Can you read the following words? CANIS O CHOLOCAT E Priming Using Transposed Words & Nonwords o o Evidence for flexible letter position encoding from studies examining priming effects for words and nonwords with transposed letters. yes no CASINO Andrews showed priming of words like ”slat” by transposed letter counterparts like “salt” compared to control condition. o Perea showed priming effects between transposed letter nonwords like “caniso” and words like “casino”. o Also readers prefer to read nonwords like “cholocate” as the correctly spelled word. caniso £$%@ * + &$£@#& caniso CASINO White et al. (2008): Eye Movement Study Sentence Reading Time The teacher gave a difficult problem…. Internal Beginning: The taecher gave a dfificult porblem… Internal End: The teacehr gave a difficlut probelm…. External Beginning: The etacher gave a idfficult rpoblem…. External End: The teachre gave a difficutl problme… Average sentence reading time (ms) Control: 4000 3500 3000 2500 2000 1500 1000 500 0 Letter transposition disrupts normal reading. But this disruption is less for interior than exterior letters. o tr n Co l na r te In e lB g in n n gi a rn e t In nd lE na r te Ex g in n n gi e lB na r te Ex nd E l Cambridge University vs. Hebrew University Velan & Frost (2007) assessed reading of transposed text in English & Hebrew Transposed text read easily in English but not Hebrew. Cambridge University vs. Hebrew University o Semitic languages (Arabic, Hebrew) have a non-concatenative morphology – words formed by interleaving root (usually 3 consonants) with word pattern. o Transpositions of root consonants disrupts word recognition. Flexible Letter Position Encoding o Word recognition system must encode both the identity and position of letters in words. o However, letter position appears to be encoded relatively flexibly, so that misspelled words like ”caniso” can be easily read as the correct word “casino”. o The IAM model assumes letters are encoded in set channels and so does not allow this flexibility. o Modern models based on IAM have introduced flexibility in letter position encoding. Letter Position Dyslexia o Letter position dyslexia provides evidence for difficulty in encoding letter position. o Children misread words by transposing adjacent letters in a word, e.g., bread > beard.Friedmann & Rahamim (2007) o Also evidence that typically developing children show the same reading behaviour, only to a lesser extent perhaps compared to children with letter position dyslexia. Kohnen & Castles (2013) o Children and adults experience competition between words and their Anagrams: Interior: pirates = parties, bread = beard Exterior: being = begin Summary o Provided an introduction to how the brain might recognise words by looking at role of: o The word superiority effect o The lexical frequency effect o Orthographic neighbour effects o Letter position encoding o These are important for understanding typical and atypical reading and reading development and for the development of sophisticated computational models of reading. o Crucial to assess effects across different languages to