Lecture 3 Flashcards
describe the first stage of reading, visual word recognition
getting letters to the meaning of a word
a small set of symbols in combination makes up an infinite set of words
define mental lexicon
systematic organisation of words in your brain (+_ 60,000 - 70,000)
how does frequency affect word recognition speed?
high frequency words are recognised faster than low frequency words
general information-retrieval mechanism
less frequent words also tend to be less familiar, meaningful, and more nonword-like; if more like non words, more difficult to say “yes, it is a word” -> post-lexical factors
define the age of acquisition effect
words that you learned at a younger age are recognised faster
overlap with frequency, but not completely, e.g. kite
what did RAYNER say about transposed and substituted letters?
substituted letters more disruption than transposed letters
BUT transposed letters still costly compared to identical controls
t and s letter do slow down understanding, especially if the change is at the beginning or, to a lesser extent, the end of a word
what did REY say about transposed and substituted letters?
letter detection: is there an “A” in BOARD / BRASH?
A = grapheme, OA = grapheme
easier to indicate “yes there is an A” in BRASH than BOARD
-> Graphemes are processed as perceptual reading units. To get to the “A” in “BOARD”, you have to break apart the “OA”
define graphemes
letters and letter groups that correspond to one sound (phoneme)
what are morphemes?
the smallest meaningful unit of a language
can be a word itself (e.g., deck) or part of a word (de-brief)
explain roots and affixes in a morpheme
(prefixes and suffixes)
“unreal” has two morphemes “un-“ (prefix) and “real” (root)
“farmhouse” has two morphemes “farm” and “house”
what is compound decomposition?
tests whether compounds are decomposed into their constituents
immediate effects of lexeme (constituent) frequency
later effects of whole-compound frequency
-> immediate decomposition, but also whole word representation
What are pseudo-affixes?
CORNER (pseudo-suffix: CORN + ER)
BROTHEL (no pseudo-suffix, but contains existing word broth)
what occurs during word recognition with pseudo-affixes?
if pseudo-affixes are not distinguished as a unit, then priming comparable for both conditions
if pseudo-affixes are used to divide up words, then greater priming for CORN+ER -> CORN than for BROTHEL -> BROTH
suffixes (and possible suffixes - pseudo-suffixes) are extracted as units early during word recognition
are semantically rich words recognised faster or slower?
faster,
more semantic features; more semantic neighbours higher imageability, more concrete; easier to bodily interact with; degree of emotional valence (positive, negative, neutral: lucky - angry - plain); degree of arousal (high, low: snake -sleep)
how can semantics (supposed to come in only after accessed) affect recognition?
feedback from semantic to orthographic and phonological layer (interactive)?
marker or “flag” associated with orthography?
what is parafoveal processing?
lots of processing is done even before we look at a word
evidence that frequency of next word is extracted within ~100ms of looking at the word before
what is the first model of word recognition?
word entries are searched one at a time (in a series or serially)
word entries are searched all at once (in parallel)
what is the second model of word recognition?
information flows in strictly one way: letters -> words
interactive: letters <-> words
considering information flow, are there feedback connections between letters & words?
letters in words detected better than letters in nonwords
feedback from words to letters
letters in words detected better than letters on their own -> word superiority effect
describe Forster’s search model: a simple serial search
- recognise letters “COW”
- get some initial unit e.g. first letter
this choses a bin (e.g. the letter “C” bin for the word “COW”) - once in a bin, entries are searched one by one (bins are ordered by frequency)
- use master file to get to meaning
describe the parallel model: the Logogen model
send info about letters to all word detectors (logogens) at once
info only feeds forward from letters to words
each logogen has a threshold level based on e.g. frequency
when activation passes threshold -> fire -> word recognised
summarise the Interaction Activation Competition (IAC) model
parallel activation of letters and words
feedback from words to letters
high frequency words have higher resting levels of activation (not lower thresholds - note: difference from logogen mechanism)
all words have the same threshold
describe ASIDE: Nettalk
connectionist neural network from the 1980s
1 input layer, 1 hidden layer, 1 output layer
about 300 “neurons”, interconnected -> 18,000 “synapses”
first random sound generation, with backwards propagation, starts to learn what is good and not (adjusting strength of connections)
does not contain any spelling-to-sound rules
after 1/2 day - can learn 1,000 words
what is the dual-route model?
not just word recognition, but also from letters on a page/screen to speech
reading predictable words out loud
- pronunciation is predictable based on how the word is spelled
e.g. “SAT” “SPEAK” (called regular words)
reading unpredictable words out loud
- pronunciation is not predictable based on how the word is spelled
e.g. “YACHT” “STEAK” (called irregular words)
reading new words
- in experiments we use “nonwords” (nonsense words)
nonwords are letter strings which aren’t real words
two types:
pronounceable (e.g. chotel)
unpronounceable (e.g. rpteeof)
what is neuropsychology?
neuropsychology is the study of patients who have suffered brain damage