PSYC3311 Lecture Notes - Lecture 10: Super Bowl Li, Bigram, Parsing
Week 5 lec 2
Yesterday
- Letters of meta-coding
o Open bigram coding
o Spatial coding
▪ Stamp- expects highest level of activation for letter s and decreases till
p
▪ If u were to transpose T and A – A will now have more activation than
T
▪ TL effect there cos activation is still similar
Spatial Coding. The activation levels for letter representations are graded decreasingly in weight
from left to right, and the word representation is sensitive to the pattern of activation appropriate
to it. The activation levels for s, m, and p are the same for stamp and satmp, while those for t and
a are only slightly different. (e.g., Davis, 2010)
3rd account: Sub-Syllabic Processing (SSP). Consonants need to be assigned to their correct
subsyllabic slot (i.e., onset or coda), and are tried out in all possible slots. When the t of satmp is
tried in the 2nd position of the onset slot, the representation for stamp is activated. (e.g., Taft &
Krebs-Lazendic, 2013)
- Features the onset + vowel + coda
- Vowel and coda is the body
- Onset- ST. and body AMP activate word STAMP.
- Body AMP is activated by A and MP – vowel and coda
- There r sub-syllabic slots corresponding to the onset, vowel and coda – 2 slots in each
here in the e.g. below
- possible to have 3 stree as in street
- letters assigned to relevant slot
- when stamp is presented – vowels r differentiated from consonant and is assigned to
vowel slot
- ST is waiting for s to fill onset slot and t to fill other onset slot. Same for a and mp
- Assignment of letters to slots is imprecise. For stamp to b activated must have s and t in
onset slot and others in their correct slots
- If we do SATMP – the t is tried out In diff slots (2nd onset slot ) – this one will activate the
word stamp
find more resources at oneclass.com
find more resources at oneclass.com
- When u transpose 2 letters u activate base word – TL effect
These three models address the question of how letter information might activate whole-word
information. In addition to examining the nature of TL effects, this issue can be explored by
looking at the impact of embedded words:
For example, what would it mean if a nonword that starts with a word (e.g., furb) takes longer to
classify as a nonword than a control nonword that does not (e.g., lurb), while a nonword that ends
in a word (e.g., clid) does not take longer than a control (e.g., clig)?
- Cos fur is embedded at the start of the word furb but lurb – no word at the start
- Lid embedded on end of clid – slows u down cos confusion, but less of an effect than if
word is embedded in start of the word
Suggests left-to-right bias in parsing the letter-string. → which is y more of an effect when word
is embedded at start of the word.
However, outer embedding also creates interference: e.g., nimp vs fimp, so not pure left-to-right
parsing.
- Nip – outer embedded. Interference here as well.
Taft, xu, & li (2017)
- If the initially embedded word (teaf = neaf)
- Tea – is consonant and vowel. No coda. this is diff to furb which has consonant vowel
and coda
- This shows that internal structure of lil embedded word does matter
- Vowel case- teaf, coda case- furb
Open Bigram can handle interference arising from initial, final, and outer embedding because the
open bigrams that activate the embedded word are found in the nonword stimulus. However, it
needs extra assumptions to explain why a final embedding creates weaker interference than an
initial embedding.
The same is true of the Spatial Coding model, where the relative activation pattern of the letters
of the embedded word remains intact (or close to intact) within the nonword stimulus.
The SSP account can explain the interference from the embedded word in terms of competition
between the body of the embedded word (e.g., the ur of furb) and that of the nonword stimulus
(urb), as well as the coda of the embedded word (the r of fur) and the coda of the nonword
stimulus (rb).
(where the purple link thing refers to an inhibitory link, i.e., competition) the swirl in the middle
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Letters of meta-coding: open bigram coding, spatial coding, stamp- expects highest level of activation for letter s and decreases till p. If u were to transpose t and a a will now have more activation than. T: tl effect there cos activation is still similar. The activation levels for letter representations are graded decreasingly in weight from left to right, and the word representation is sensitive to the pattern of activation appropriate to it. The activation levels for s, m, and p are the same for stamp and satmp, while those for t and a are only slightly different. (e. g. , davis, 2010) Consonants need to be assigned to their correct subsyllabic slot (i. e. , onset or coda), and are tried out in all possible slots. When the t of satmp is tried in the 2nd position of the onset slot, the representation for stamp is activated. (e. g. , taft & Onset- st. and body amp activate word stamp.