language - bilingual processing Flashcards
define bilingualism (de Groot, 2015)
an individual’s ability (and actual practice) of communicating in two languages and the linguistic knowledge base that enables this ability
3 types of bilingual
simultaneous bilingual = more than 1 language learnt from birth
early sequential bilingual
= learning a 2nd language after a 1st language early in life
late sequential bilingual = learning a 2nd language after a 1st language later in life
types of bilingual - learning of L2 (second langage)
second language (L2) learning in a natural environment e.g., moving to a new country
second language (L2) learning at school e.g., only using the L2 at school in class
balanced vs unbalanced bilingualism
balanced/unbalanced bilingual = the way the language is acquired and used will affect how that language is represented in the mind
= balanced bilingual uses both languages equally
features common to all bilinguals
languages affect each other e.g. speaking a second language can affect your first one
- effects of different languages on perception of colour
- mental representation of time
- expressions
- theory of mind
- executive function
3 theories of lexicons and bilingualism
separate lexicons:
- co-ordinate systems
- separate lexicons with separate semantic stores
one lexicon:
- compound system = all representations link to the same semantic store
- subordinative system = first language (L1) links to semantic store - second language (L2) links to L1
co-ordinate systems of lexicon
separate lexicons with separate referents
L1 orthographic and phonological representations <–> L1 semantic representations
L2 orthographic and phonological representations <–> L2 semantic representations
compound system of lexicons
all representations link to the same referent
L1 orthographic and phonological representations + L2 orthographic and phonological representations <–> semantic representations
subordinative system of lexicons
L2 orthographic & phonological <–> L1 orthographic and phonological <–> semantic representations
L2 links to L1 which links to semantic referent
evidence of shared and separate semantic stores - fused vs separate
Lambert et al (1958)
method:
French-English Bilinguals classified into:
- 1 group who learnt their languages in ‘separate’ contexts (e.g. learning french whilst in france and english in england)
- 1 group who learnt their languages in a ‘fused’ context (e.g. learnt french and english in france)
participants rated house, drink, poor, me and their French equivalents along semantic dimensions e.g. fast–slow, large–small
results:
- ‘fused’ group showed less difference in their ratings than the ‘separate’ group
conclusion:
- ‘fused’ group had a shared semantic store
- ‘separate’ group had a semantic store linked to each language
lexicosemantic representation
representation of meaning in a lexicon
bilingual lexicosemantic representation
pure subordinative, compound, or coordinate bilingualism is highly unlikely
representation differs depending on:
- monolingual or bilingual
- context in which languages acquired
- level of L2 proficiency
- L2 learning strategy
- word type
- delay between current and previous use
language switch costs study
Kolers (1966)
participants say name of image in English with a red background or in French when it has a blue background
slower to name images in mixed lists than when they are all to be named in one language
what does language switch cost show about number of lexicons
languages can be switched on or off
effort needed to switch between languages indicated by a delay in production
separate lexicons (language independent)
opposing evidence for language switch costs
Preston and Lambert (1969)
if languages can be switched on or off interference should not be found between language only within languages
evidence of between language interference
one unified lexicon (language interdependent)
language independent view
2 lexicons
no competition between languages (L1 and L2)
competition within a language (within L1)
language interdependent view
1 lexicon
competition between L1 and L2
competition within a language too
bilingual stroop task
Preston and Lambert (1969)
english-french bilinguals
congruent, incongruent, and neutral (asterisks) conditions with both languages used
results:
slower responses to colour words compared to asterisks regardless of the language of the word or the response language
conclusions:
evidence that for a bilingual trying to name the colour of a word produces interference regardless of the language the word is written in or the language of the response
bilinguals do not switch off one of their languages
interference is experienced between languages and within languages
one unified lexicon (Language interdependent)
revised hierarchical model (RHM) of bilingual processing
Kroll & Stewart (1994)
L2 words stored in L2 lexicon that initially links to L1 lexicon
as proficiency increases L2 lexicon develops direct links with semantics
predictions from RHM (3)
assumes links from L2 to L1 are stronger than links from L1 to L2
- translating from L2 to L1 is faster than L1 to L2
- translating from L2 to L1 should be faster than picture naming for beginners
- translating from L2 to L1 should be no faster than picture naming for proficient L2 speakers
test of translation between L1 and L2 speeds
used lists with random organisation
L2–> L1 = faster than L1 –> L2
in line with RHM
test of L1 lexicon linked to semantics, not L2
use of lists organised by semantics for translation
only L1 –> L2 translation affected by semantics
suggests L1 is linked to semantics but not L2
activation of semantics in the L1–>L2 condition interferes with translation
test of translation and picture naming in beginners
picture naming requires semantic representations to be activated
picture naming in L2 for beginners:
L2 o+p –> L1 o+p <–> semantics
L2 –> L1 link is weak, L1 <–> semantics = strong
picture naming in L2 for prolific speakers:
L2 o+p <–> L1 o+p <–> semantics <–> L2 o+p
all 3 components all interlinked strongly
Chen and Leung 1989; Kroll and Curley 1988
results:
picture naming in L2 was slower than translation from L1 to L2 for beginner speakers but not for proficient speakers
conclusions:
links between the semantic store and L2 lexicons develop as speakers become more proficient
test of priming effects from L2 –> L1 and L1 –> L2 according to RHM
Priming effects
L2 –> L1 links are stronger than L1 –> L2
will reaction times for L1 primes then L2 targets be faster than L2 primes then L1 targets
translation priming:
activate L1 prime
- activates semantics (strong link)
- activates L2 lexical (this is a weak link)
L1 –> L2
activate L2 prime
- activates L1 lexical - strong link
- semantic link doesn’t need to be activated for translation but probably are (this is from L1 not from L2)
challenges to the RHM - priming effects
Schoonbaert et al (2009)
L2 –> L1 should be stronger than L1 –> L2
BUT found L1 –> L2 links and smaller effects from L2 –> L1
semantic priming and the RHM - testing this with priming task
L2 –> L1 links are stronger than L1 –> L2
semantics are linked to L1 not L2
need to see if spreading activation works in other languages for semantic priming
L1 prime (“girl”)
- activates semantic representation (+ spreading activations - e.g. to “boy”)
- links between semantics for “boy” and the L” words are weak and unlikely to activate the L2 word for “boy”
- therefore no semantic priming would be found
L2 prime (“jongen”)
- activates L1 word “boy”
- this then activates semantic for “boy”
- this can then spread activation to “girl”
- therefore when L1 target “girl” is presented, response to it will be quick
semantic priming in the RHM summary
- semantic priming effects should be ‘asymmetrical’
- semantic priming should be found from L2 -> L1 but not from L1 -> L2
effects have been found that support the RHM:
- de Groot and Nas (1991) failed to find cross language semantic priming effects from L1 to L2 in Dutch–English bilinguals
effects have been found that do not support the RHM:
- Perea et al (2008) found cross language semantic priming effects for both directions in balanced Basque–Spanish and Spanish– Basque bilinguals
challenges to the RHM - semantic priming effects
Schoonbaert et al (2009)
Semantic priming effects found from both L1 -> L2 and L2 -> L1
numerically larger from L1 -> L2
what does the RHM account for or not
DOES:
- effects seen in the development of proficiency - beginners vs experienced
NOT:
- effects from priming studies don’t fit RHM assumptions
bilingual interactive activation model (BIA+)
Dijkstra & Van Heuven (2002)
- computational model
- 1 lexicon
- node = lexical items, belonging to one or the other language –> language node is connected to all words in that language
language
word
letter
feature
- activation = bottom up –> from features to words ( feature - letter - word - language)
- feedback = top down (language - word - letter - feature)
recognition of a word inhibits activation of other words
activation of letters is not language selective –> all words that match the input are activated regardless of language
once a word in one language is activated all the words in the other language are inhibited
BIA+ - semantic activation
at the word level semantic representations linked to words are activated
selection of an English word inhibits Dutch words –> but a semantically related Dutch target may be recognised quickly due to the activation of related semantic representations
BIA+ - resting levels
words have ‘resting levels’ that are adjusted based on proficiency, frequency etc.
- high frequency = high resting level = quick activation
- if dominant L1 –> L1 words have a higher resting activation than L2 words, reach threshold for selection with less activation
translation priming effects in BIA+
translation priming effects found:
- from L1 -> L2 Smaller effects from L2 -> L1 (Schoonbaert et al., 2009)
- from L1 -> L2 only (De Groot and Nas, 1991; Gollan et al., 1997)
BIA+:
- account for the asymmetrical translation priming effects via the slower activation of L2 words due to the lower resting activity
- takes longer for the L2 ‘jongen’ to reach the threshold for activation than the L1 ‘boy’ resulting in smaller priming effects
BIA+ –> switch costs study
Grainger & Beauvillain (1987); Von Studnitz & Green (1997)
method:
Cross Language Lexical Descision Task – press a button when you see a word you recognise
results:
slower to recognise words in a mixed list (both languages) because one language is inhibited
semantic priming effects and the BIA+
Schoonbaert et al (2009)
BIA+ model predicts that symmetrical effects would be seen for L1 -> L2 and L2 -> L1 semantic priming due to the activation of semantic representations for words of both languages
3 things BIA+ can account for
asymmetrical translation priming effects
switch costs
semantic priming effects
the bilingual lexicon two models comparison
evidence suggests that both languages are activated when a bilingual is processing language, regardless of the target language
“language non selective hypothesis”
Models of bilingual processing are divided on the question of whether there are 2 separate lexicons or 1:
- models with 2 lexicons predict competition within languages (RHM - Kroll & Stewart, 1994)
- models with 1 lexicon predict competition between and within languages (BIA+ - Dijkstra and Van Heuven, 2002)
cognitive advantages of bilingualism
neurocognitive adaptation to bilingualism mechanisms are a result of the way the brain handles and uses the languages at one’s disposal
languages are constantly and jointly active, which in turn creates a state of competition for both language comprehension and production
consequences of having more than one language - the BIA+ model
inhibitory feature of the ‘language node’ in the BIA+ model provides an account of how competition between languages might provide bilinguals with stronger inhibitory control compared to monolinguals
language node is assumed to be domain general – not specific to language – suggesting that advantages from inhibiting a language may transfer to non verbal cognitive control
consequences of having more than one language - competition and inhibition - methodology
Blumenfield and Marian (2013)
tested language competition between languages and inhibitory control
- competition between languages was tested using a visual world paradigm
- inhibitory control was tested using a Simon Task
Visual World Paradigm:
- Spanish-English bilinguals listened to words whilst viewing a grid with images
- Bilingual Spanish-English participants should be distracted by the image of a thumb when hearing pool because the Spanish word for thumb is pulgar
- Participants who look at the thumb are assumed to be experiencing cross language competition from the phonological competitor
Simon task:
- participants have to press a key on a keyboard in response to the direction of arrows
- when the direction of the arrow matches the side of the screen it is congruent
- when the direction of the arrow does not match the side of the screen it is incongruent
- for incongruent responses the participant has to inhibit the ‘inappropriate’ response that would match the position of the arrow
- typically congruent responses are faster than incongruent responses
- a person with a small difference between congruent and incongruent responses has strong inhibitory processing
consequences of having more than one language - competition and inhibition - study results
Visual World Paradigm:
proficient = more distracted by image in other language (cross language competitor)
domain specific effect
Simon task
proficient = smaller Simon effect = better inhibitory control
domain general effect
conclusions:
negative correlation between Simon effect ad looking at cross language competitor
means bilinguals are distracted by second language and have better inhibitory processing
shows link between cross language competition and inhibitory processing
cognitive advantages and disadvantages of bilingualism - issues with research into this
lots of evidence for cognitive advantages, but also disadvantages
Judith Kroll argues that:
- much of the previous work has been correlational
- research looking at the causal mechanisms is sparse
- whether advantages are seen or not may depend on the proficiency of the bilingual, how they learnt their language and how we measure it
reasons for failures to find cognitive advantages for bilinguals (6)
- differences in tasks
- differences in Level of proficiency
- differences in the age at which the second language was learnt
- Social Economic Status
- age of the person when tested
- differences between using and knowing a second language