Word Recognition And Sentence Processing (Chapter 4 Part 2)

Front 1

Differentiate between localist and distributed models (give an ex that illustrates distributed)

Back 1

localist: words concept mediation have one to one mapping - earlier models had this localist view
distributed: representations overlap- things that fit in categories deviate from the prototype in different ways and will over lap in conceptual memory. Some words may overlap but have different representations in each language
-ex all these representations are hard for L2 learners to grasp for example black in english has many many meanings and used but the L2 learner may only have one sense- like the colour

Front 2

What is the the embodied approach?

Back 2

-semantics are grounded in experiences that co-occur with words during language use- concrete vs abstract semantics are grounded in sensorimotor vs emotional experiences, respectively.

Front 3

Describe the Titone study on L2 embodied word processing using eye movement reading measures. Why was this interesting to study? what did they do ? what did they find?

Back 3

they compared L2 and L1 readers on words with varying emotionality, concreteness and frequency
this is interesting because words we see in L2 usually co-occur less frequently with experiences that ground semantics- we don't learn sad words as much
They found that emotional words are slower when they are abstract than concrete, but ONLY if they are low frequency

Front 4

what was the point of the study with Russians and English and the continuum of when a cup becomes a mug?

Back 4

basically that Language affects how we categorize objects . L2-L1 influence can occur for very frequent nouns

Front 5

what is the word frequency effect?

Back 5

- an important signature of lexical access= word frequency effect- high frequency words (home) are processed faster and more accurately than low-frequency words

Front 6

what are cognates? Interlexical homographs? inter lexical homophones?

Back 6

cognates (orthography+semantics are similar) ex. jungle
inter lexical or interlingual homographs (orthography is similar but semantics not) ex cat
interlexical homophones: words sound same but are orthographically and semantically different

Front 7

what is cross- language orthographic neighbourhood density?

Back 7

-how many words in that language are similar in orthography or phonology
-this factor can cause facilitation or interference depending on the nature of the task

Front 8

what are early stage vs late stage measures of online eye movement techniques?

Back 8

- early-stage measures (reflects initial word recognition): first fixation duration (time spent reading word the very first time), gaze duration (sum of all fixations on a word during one pass)
- late-stage measures : total gaze duration (sum of all fixations, including second passes)

Front 9

Homographs should produce____ and ___.
Cognates should produce ____.

Back 9

form-level facilitation (the word is same across two languages, so because the words form is familiar to you it facilitates), and Meaning-level conflict (but the more you have to process it for meaning because it isn’t what it is in the other language, it causes conflict) ;
form-level facilitation

Front 10

What did Beauvillian find in his study of interlingual homographs and ambiguity?

Back 10

- cross-language lexical decision task, primed with either language unique or homographs
- would presenting a semantically related prime word in the other language (coin= penny, corner) facilitate the other meaning in the other language?
ex. CORNER. Sandra fell over and dropped her coin
MONAIE. Luc était fâché et il s'est mis dans le coin
- yes! you do get priming across languages, but they only found that you got priming when the interval between the prime and the target is short

Front 11

you are doing a go-no go task. there are homographs and language unique words (task is french-english). The word MANGER appears and you must go if it is an english word. Will you be fast or slow compared to controls?

Back 11

you will be slower because the word is low frequency english but high frequency french
- this is non selective access
**in general, homographs were slower than control words

Front 12

Cross language lexical decision task always results in non-selective access.

Back 12

False-
under some circumstances, language-selective access can occur. we cant conclude this!!...

- had bilinguals naming visually presented english words, some eyer iLH some control
- sometimes they were preceded by english words paragraph, sometimes by french
- L2 bilinguals named ILH more slowly no matter what
- L1 bilinguals names the english ILH more slowly only when preceded by french words
or

- Promoted a global language context- german-english bilinguals were presented with either a german or english silent films (no talking but still context)
- then presented IHL immediately followed by an L2 (english) target word, they had to decide it word was in english
- targets that were preceded by homographic primes did have faster response times
- the N400 in related targets was less negative, meaning earlier semantic integration
- this is clear evidence of non-selective access but only when the german film played first

Front 13

Why is BIA so influential? (3)

Back 13

-BIA is so influential because it uses numbers and is able to form new predictions that are true when tested in the real world
-there is a lot of evidence that says when we store languages we store it in an integrative way
-it is very flexible and allows us to account for the different constraints that provide selective and non selective access

Front 14

describe the Jared and Kroll study on spelling-to-sound conversion with french-english naming task

Back 14

- had them pronounce words that have only-english bodies- like -osh which are pronounced consistently the same
- then words that were english bodies but inconsistently produced like - ead, eak (head, lead; steak, leak)
- then words with different bodies in english vs french like -ait (bait vs était)
- the put a block of french filler words
-They found interference effects of french phonological enemies only when english naming followed french filler block- only when put in french mode!

Front 15

describe how you set up a a study on temporary syntactic and lexical ambiguity (with a screen and eye movement tracking) what have these studies shown?

Back 15

ex if told- pick up the CANDY and put it above the fork, when a visual display contains both a candy and a CANdle, saccade latency to target CANDY increases— initial overlapping phonology (activation of phonologically similar representations in memory)
- you can do this within a language, and you can also do it with two languages- pick up the MARKER , when marka in Russian is stamp and there are pictures of a stamp and a marker

they found both within and between language effects which varied based on the persons language "mode"

Front 16

Can orthography and Phonology interact?

Back 16

yes, in cross-language studies where the prime resembled the target in sound or look, people were facilitated- also happens cross-alphabets

Front 17

what is the cognate effect?

Back 17

differences in response times to cognates vs control words
- normally facilitory in nature
- but can be rare inhibitory effects which arise when a cognate pair are phonologically dissimilar, increasing naming latencies- like Jungle in french and english

Front 18

what happens to the cognate effect when you manipulate semantic constraint- Titone!

Back 18

- with the homographs, we see interference, and with the cognates we see facilitation compared to controls— when looking at first fixation time
- in terms of total reading time , they saw a interference of low constraint but not high (for homographs) and they saw a facilitation for low constraint not high for cognate)s
- her data basically points towards the idea that the cross language effect is minimal and the effects are coming from frequency effects

Front 19

what are 3 accounts of the cognate effect

Back 19

Dominant View: cognate processing= driven by co-activation of memory nodes representing elements of non-target language during target processing

Alternate View: cognate processing= driven by qualitatively different representations at the meaning/morphology level for cognates vs non-cognate pairs

Similar but alternate, Titones view: cognate effects arise from lexical-level form frequency difference between cognates and language-unique words

Front 20

how does executive control and proficiency relate to cross-language conflict during reading homographs and cognates? (4)

Back 20

- for homographs: they found that as executive control gets worse, the gaze duration (and total duration) increases meaning they are harder to read - this suggests that individual differences that aren’t linguistic related are penetrating the blue BIA box and influencing language processing
- they also found that for cognates gaze duration more for the control words as a function of L2 proficiency
- thus the homographs are correlating with conflict resolution and the cognates are correlating with the L2 proficiency!!
- the homograph finding threatens BIA+ because because the executive control is effecting the language processing- you need feedback form the top task/ decision system

Front 21

3 take home messages from chapter 4?

Back 21

- cross-language knowledge of words stored in an integrated fashion- when bilinguals encounter written r spoken words they activate in memory all of their language knowledge in a graded (continuous) fashion
-different kinds of constraints can push memory activation around- sentence context, grapheme-to-phoneme conversion, language “mode”, L2 ability and general non-linguistic cognitive capacity
- knowledge of language membership is also encoded in memory in some fashion- either explicitly through language nodes, implicitly through differential similarity of linguistic units or both

Front 22

describe syntax-first models (2)

Back 22

- parser initially ignores syntax-external sources (context, semantics, lexical inch, frequency)
- automatically applies universal parsing strategy assigning simplest structure possible based on syntactic info only, then reanalyze later.- only cares if it is nouns, verbs ect not what it means

Front 23

the syntax models and ambiguity...

Back 23

- temporary syntactic ambiguities- part of the sentence is ambiguous- Garden paths- we will have to reanalyze

Ex. The student graded…
- graded would be interpreted as the main verb, and you expect a direct object like “the paper”
- but if you get ..”by the professor received and A” - sentence compliment
Ex. the ticked agent admitted the mistake might not have been caught
- you want to add a “that the” before mistake

Front 24

problem with the Syntax first model

Back 24

- assumes initial syntactic analysis is impenetrable by other sources of info, but monolingual studies show that parsing cam by influences by semantics like “the paper graded by the professor” is less ambiguous than “the student graded by the professor” because of animacy

Front 25

Can L1 influence L2 in parsing? L2 L1? what does this change as a function of?

Back 25

yes, yes; proficiency - the more proficient means the effects are reduced