l2 endstate and_dynamic_l2_interlanguage.edited

65
L2 Endstate: L2 Endstate: a non-age based theory a non-age based theory Dynamic interlanguage systems and usage-based learning mechanisms By Nigel P. Daly

Upload: nigel-daly

Post on 13-Jan-2015

511 views

Category:

Education


3 download

DESCRIPTION

 

TRANSCRIPT

Page 1: L2 endstate and_dynamic_l2_interlanguage.edited

L2 Endstate: L2 Endstate: a non-age based theory a non-age based theory

Dynamic interlanguage systems and usage-based learning mechanisms

By Nigel P. Daly

Page 2: L2 endstate and_dynamic_l2_interlanguage.edited

Today’s planToday’s plan

1. Dynamic systems

2. Article Language learning mechanisms L2 language learning mechanisms

limited endstate

3. Discussion 1: IL as Dynamic systems

4. Discussion 2: EFL Pedagogy?

Page 3: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and QuestionsIssues and Questions

Dynamic systems (DS) … linear vs nonlinear

Is a classroom (with ss and T) dynamic?

HOW?- - - - -

Page 4: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and QuestionsIssues and QuestionsDynamic systems (DS) … Dynamism: the systems change over time Adaptation and learning: complex systems change in

response to environmental conditions to retain coherence

Heterogeneity of elements/agents: interacting parts may be of different kinds and components of different systems

Openness: energy or influence comes from external sources or systems

Nonlinear: causes of changes are not necessarily proportional to their effects

(from Larsen-Freeman & Cameron, p. 22-36)

Page 5: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and QuestionsIssues and QuestionsDynamic systems Dynamism: changes in time

Adaptation and learning: complex systems change in response to environmental conditions to retain coherence -

Heterogeneity of elements/agents: interacting parts may be of different kinds and components of different systems

Openness: energy or influence comes from external sources or systems

Nonlinear: causes of changes are not necessarily proportional to their effects

Classroom diff targets, mats,

activities, moods

T & ss interact, e.g., T spends more or less time on certain learning

Diff people, materials, activities, TBs, tests

Ss and T bring experiences and language from outside

Low salience, require more time to learn

Page 6: L2 endstate and_dynamic_l2_interlanguage.edited

Traditional disagreement to Traditional disagreement to understanding social phenom.understanding social phenom.

Social scientists and educators have been interested in social phenom to

1. Understand it, and/or

2. 2. To develop models of efficient or optimal conditions for working, managing, producing or learning.

Page 7: L2 endstate and_dynamic_l2_interlanguage.edited

Traditional disagreement to Traditional disagreement to understanding social phenom.understanding social phenom.

Perhaps the biggest challenge is to resolve the macro-micro relationship.

Or, How do we understand individuals and society/group?

Page 8: L2 endstate and_dynamic_l2_interlanguage.edited

Traditional disagreement to Traditional disagreement to understanding social phenom.understanding social phenom.

Social theorists have traditionally seen this as an either/or problem.

SOCIETY OR INDIVIDUAL (totality of individuals) (socialized, society microcosm)

Page 9: L2 endstate and_dynamic_l2_interlanguage.edited

Traditional disagreement to Traditional disagreement to understanding social phenom.understanding social phenom.

Structuralism Soc inds (social determinism, Marx) Inds Soc (methodological individualism,

classical economics [Hayek])

Response to this approach?

InteractionismInteraction of ind and soc inds & soc

(Foucault “discourse”, Bourdieu “habitus”)

Page 10: L2 endstate and_dynamic_l2_interlanguage.edited

Synthesis? – Sawyer’s DSSynthesis? – Sawyer’s DS

Thesis: Marx (ind = soc)

or Hayek (soc=ind’s)

Antithesis: Interaction determines Soc and inds (Foucault, Bourdieu)

Synthesis: Sawyer, R.K. (2005) Social emergence: Societies as complex systems. Cambridge: CUP.

Page 11: L2 endstate and_dynamic_l2_interlanguage.edited

Language as a DSLanguage as a DS

Revision:

Interaction:•Syntax, semantics, phonology in communication

Individual:•Words, morphemes, phonemes

Page 12: L2 endstate and_dynamic_l2_interlanguage.edited

Usage-based and form-Usage-based and form-focused language acquisition focused language acquisition

“Handbook” Article (general overview) in 2 halves

dealing with L2 learning:

1. Systematic influences of frequency, contingency, semantic complexity (multiple cues) and broader aspects of salience and syntactic category

2. L1 entrenchment leads to interference, overshadowing, blocking, and perceptual learning, i.e., the prevention of input from becoming intake for certain “fragile features” of L2

Page 13: L2 endstate and_dynamic_l2_interlanguage.edited

OverviewOverview

Page 14: L2 endstate and_dynamic_l2_interlanguage.edited

QuestionQuestion

Leo: It seems to me that the associative learning of linguistic construction puts emphasis on learners’ experience of encountering/using the language and accumulation of prior experience.

Such being the case, in addition to the factors discussed in the article, do we need to consider other factors like social and environmental factors?

Page 15: L2 endstate and_dynamic_l2_interlanguage.edited

Language learningLanguage learning

Factors that affect both L1A and L2A

Page 16: L2 endstate and_dynamic_l2_interlanguage.edited

Frequency Frequency

Learners need to learn the probability distribution P(interpretation / cue, context):

“the probability of an interpretation given a formal cue in a particular context, a mapping from form to meaning conditioned by context” (p.374)

Page 17: L2 endstate and_dynamic_l2_interlanguage.edited

Frequency Frequency

Input serves to create form-function mappings that (in optimal situations) lead to a concordance-type of retrieval system

Page 18: L2 endstate and_dynamic_l2_interlanguage.edited

Cue and outcome learning is not just frequency-based,

… but also depends on the contingency of the cue-outcome rel’ship

Page 19: L2 endstate and_dynamic_l2_interlanguage.edited

ContingencyContingency

Stimulus-response

Reliability of the language form to predict interpretation that determines its acquisition

Page 20: L2 endstate and_dynamic_l2_interlanguage.edited

Contingency Contingency

P = P(O/C) – P(O/-C) = a/(a+b) – c/(c+d)

Shows probability of the outcome of cue (P/O/C) minus the probability of the outcome without cue (P(O/-C)

Outcome No outcome

Cue a b

No cue c d

Page 21: L2 endstate and_dynamic_l2_interlanguage.edited

ContingencyContingency

P = P(O/C) – P(O/-C) = a/(a+b) – c/(c+d)

∆P = 1 presence of cue increases likelihood of outcome (+

assoc)

∆P = 0 cue may equally outcome or no outcome

∆P = -1 presence of cue decreases likelihood of outcome (-

assoc)

Page 22: L2 endstate and_dynamic_l2_interlanguage.edited

ContingencyContingency

Not always clear 1:1 mappings with cue and grammatical interpretation

Ex homophone morphemes “s” (or allomorphs of “s”)

Contingency analysis: plural “s” cue is made less clear (some plurals don’t require “s”) and thus reduces ∆P, suggesting they will not be readily learnable

Page 23: L2 endstate and_dynamic_l2_interlanguage.edited

ContingencyContingency

The fuzziness and complexity of mappings, such as plural “s” or “the” explain why their hard to acquire

Page 24: L2 endstate and_dynamic_l2_interlanguage.edited

Cues rarely exist in isolation,

… so learner has to decide from multiple cues which ones are predictive …

Page 25: L2 endstate and_dynamic_l2_interlanguage.edited

Multiple cues, the PCM, and Multiple cues, the PCM, and Cue-competition Cue-competition

Step 1: select most valid cue using statistical contingency analysis

Step 2: introduce cues thereafter on the basis of their potential to decrease error

Probabilistic Contrast Model (PCM) MacWhinney’s Competition Model

Page 26: L2 endstate and_dynamic_l2_interlanguage.edited

Multiple cues, the PCM, and Multiple cues, the PCM, and Cue-competition Cue-competition

Early Language learners only focus on 1 cue at time

Children1. earlier - focus on the cue that has the highest

validity measured by its

availability (freq) x its reliability (∆P)

high availability is the natural cue for children, over reliability

2. later - children can factor other cues into the interpretation equation, like reliability

Page 27: L2 endstate and_dynamic_l2_interlanguage.edited

Question: Question: usage = input + output)?usage = input + output)?

Glavine:

In 2.3, I feel confused about the idea mentioned by Ellis: “the cue that children first focus upon is that which has the highest overall validity as measured by its availability times its reliability.” Does this idea refer to input, or linguistic development?

Page 28: L2 endstate and_dynamic_l2_interlanguage.edited

Salience of cue and importance of outcome are other factors

Page 29: L2 endstate and_dynamic_l2_interlanguage.edited

SalienceSalience

Strength of conditioning between and cue and outcome depends on relative physical intensity,

Salience = “intensity of the subjective experience of stimuli” (p.379)

Page 30: L2 endstate and_dynamic_l2_interlanguage.edited

SalienceSalienceRescorla and Wagner (1972) verified the following

equation from several empirically rigorous experiments:

dV = ab(L-V)

V = associative strength of outcome to cue; dV = the change in V occurring in each trial of conditioning a = salience of outcome b = salience of cue L = amt of processing given to a completely unpredicted

outcome

Page 31: L2 endstate and_dynamic_l2_interlanguage.edited

SalienceSalience

dV = ab(L-V)low salience cues with low outcome importance

– especially for “everyday communicative survival”, dV on any learning trial will be trivial and may never be acquired

Ex 1. 3rd person “s”, 2. grammatical function words and bound inflections tend to be short and low in stress and thus hard to recognize (NS only hear 40-50% in clipped speech, bottom-up vs connected speech top-down)

Page 32: L2 endstate and_dynamic_l2_interlanguage.edited

SalienceSalience

Children L1 studies: only after critical mass of words are known can they start to perceive grammatical words to right or left of content words

Lenition can occur in language learning to language change:

one’s own l2 life (Chinese prosody) in a generation of language users (a-wight) and over generations to the modification of the

language itself (Latin French)

Page 33: L2 endstate and_dynamic_l2_interlanguage.edited

Interacting factorsInteracting factors

Goldschneider & Dekeyser (2001; G&D)

proposed acquisition order is not just

a frequency effect, but a combination of 5 determinants:

1.   perceptual salience: # of phones in functor + presence/absence of vowel + relative sonority of functor

2.   semantic complexity

3.   morphological regularity: phonological variation and allomorphy (e.g., plural “s” sound) + contractability (multiple forms of cue) + homophony

4.    syntactic category

5.    frequency

Page 34: L2 endstate and_dynamic_l2_interlanguage.edited

Interacting factorsInteracting factors

D&G (2001) Data from 12 studies, with 924 participants were analyzed: 3 factors significantly correlated with acquisition order:

·        perceptual salience r=0.63

·        frequency r=0.44

·        morphological regularity r=0.41

Multiple regression analysis:

3 factors + semantic complexity + syntactic category 71% of the variance in acquisition order (salience having highest predictive power on its own)

Page 35: L2 endstate and_dynamic_l2_interlanguage.edited

L1 (& L2) L1 (& L2) Language Learning DSLanguage Learning DS

Page 36: L2 endstate and_dynamic_l2_interlanguage.edited

Factors specific to L2 learning

Why is Nativelike competence rare for adult L2 learners?

Or,

What prevents input intake?

Page 37: L2 endstate and_dynamic_l2_interlanguage.edited

Limited endstate of usage-based SLA

Associative language learning from naturalistic usage “Basic Variety” of IL:

ok for everyday communication with mostly nouns, verbs and adverbs, but with little or no functional inflection, and few subordinating elements, determiners and prepositions

Page 38: L2 endstate and_dynamic_l2_interlanguage.edited

Interference

Page 39: L2 endstate and_dynamic_l2_interlanguage.edited

Interference

interactions of memories of similar experiences memory failure

retroactive interference: remembering new forget old

Proactive inhibition : prior learning inhibiting new learning

Adding new associations forget others (“Response competition model”):

“when multiple traces are associated with same cue, they tend to compete for access to unconscious awareness” (p.384)

Page 40: L2 endstate and_dynamic_l2_interlanguage.edited

InterferenceInterference

Lado’s Contrastive Analysis Hypothesis (CAH): L1 interferes with L2

Proactive inhibition = negative transfer from L1 to L2 (faux amis or typological diff), but positive edge can be seen with cognates (French and English)

Similarities and differences between L1 and L2 influence the acquisition of grammar, vocabulary, pronunciation

Word learning principles from more general aspects of cognition involving associative learning processes like proactive inhibition in learning an L2 word for something that already has an L1 equivalent

Page 41: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blocking blocking

Overshadowing: 2 cues lead to an outcome, but the more salient cue will be more associated with outcome, thus overshadowing the other AND with Rescorla-Wagner equation, the more a cue is associated with an outcome, the less possible it will be for an outcome have additional associations (i.e., cues);

latent inhibition makes less salient cues more difficult to learn in future

Page 42: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blocking blocking

Blocking: a type of “learned selective attention” due to overshadowing (ie automatically learned inattention)

Page 43: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blockingblocking

Conclusion: “human statistical reasoning is bound by selective attention effects whereby informative cues are ignored as a result of overshadowing or blocking” (p.387)

Learned inattention can be “pervasive and longstanding” and further learning about the cue is attenuated/lessened

Page 44: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blockingblocking

Redundancy of language forms much more import in L2A than in L1A – e.g., children learn adv phrases quite late, unlike L2 learners who quickly learn and tend to focus on use of adv time phrases instead of morph inflections for tense (pace VanPAtten)

Also, plural inflections overshadowed by S-V combo

Page 45: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blockingblocking

Non-acquisition of low salient cues due to overshadowing from L1 which may direct learner attention to other areas for cue to interpretation making redundant all other input and leading to “fossilized” or “stabilized” basic variety of IL

Page 46: L2 endstate and_dynamic_l2_interlanguage.edited

Overshadowing and Overshadowing and blockingblocking

Pedagogical response?

Retune selective attention (F-on-F or consciousness raising, like VP’s PI) to help learners notice the cue and raise it to salience and provide meaningful input

Page 47: L2 endstate and_dynamic_l2_interlanguage.edited

Question … over-optimism?Question … over-optimism?Lily: The author gave various research findings to support other

non-age factors, and claimed that L2 learners are actually being “blocked” by these factors rather than “fossilized”.

However, can these finding continually explain what happens after they are being “blocked”? How come that after explicit instructions and learning, most L2 adult learners still can not fully acquire the target forms whereas L2 children can?

How about automaticity? How come she learned third person singular “-s” when she was six and is now using it without any effort whereas I learned this form when I was 12, and I still have to “think” before I say it?

Page 48: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual learning

Tuning that automatically emerges as a result of exemplars influencing the “organization of the whole system and the dimensions of the underlying psychological space” (p.390)

“Raquel”

Page 49: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual learningExample of perceptual learning is “feature imprinting”: People can decompose complex objects based on their

experience with the component parts, so once trained to see object in a specific way, that perception will be incremented in subsequent processing episodes.

Example 1: training to recognize certain dot formations will influence future dot configuration interpretations

Example 2: Chinese and Japanese speakers see characters as wholes, but learners see them as grouping of strokes or components (p.394)

(Q: also apply to the writing of characters for NS?)

Page 50: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual learning

Sound categories and categorical perception of L1 subject to perceptual learning

Neural state of infants learning L1 is a plasticity, or tabula rasa (blank slate) that later becomes a tabula repleta (full slate) when L1 is learned, causing strong negative transfer , e.g., “l” and “r” in Japanese, and articles for Chinese.

Transfer which requires restructuring existing categories is difficult

Page 51: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual learning

“perceptual magnet theory” (Kuhl and Iverson, 1995) – preexisting categories serve as magnets for items which do not fit categorically in the existing system, and the result is to distort the perception of the new items by attracting them into the vicinity of the original category’s perceptual space. (p.393)

Page 52: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual learning + Question

In normal L1A, input naturally tunes language system, but “sad irony” for L2A is that in normal L2A, “more input compounds the error” (p.394);

Sandy: I am not certain of Ellis’ point when discussing the issue of more input will lead to error for L2 speakers (p. 394). Does this indicate that the sensitivity of L2 acoustic cues will interfere with learners’ speaking and listening? What kind of input does the author refer to?

[… lead to error = lead to “reinforcement” of error?]

Page 53: L2 endstate and_dynamic_l2_interlanguage.edited

Perceptual LearningPerceptual Learning

McClelland presents a connectionist simulation using a self-organizing map network that verifies “diabolically” maintain tendency

[Q: models are programmed with codes, though … so how it represents a closed system = pessimistic if it represents humans and their learning (un)potential]

Page 54: L2 endstate and_dynamic_l2_interlanguage.edited

L2 Morpheme Acq. Order

Transfer effects in L2 morpheme acquisition order

Influence of L1 on SLA, diff L1’s will have different negative transfer effects

Ex article, plural acquisition for speakers of an L1 without these, like Japanese and Korean

Page 55: L2 endstate and_dynamic_l2_interlanguage.edited

L2 Morpheme Acq.Order

Dulay and Burt (1974) study suggests that Japanese and Korean L1 children have more difficulty learning indef article and plural “s” than Spanish and Chinese children

[Chinese …?!]

Results: Differences in rank of order acquis between

monolingual NS children and L2 learners of English from diff L1 backgrounds

Page 56: L2 endstate and_dynamic_l2_interlanguage.edited

L2 Morpheme Acq.Order

Also overgeneralization errors from L1 influence.

Taylor (1975) found that elementary ss made more transfer errors, while intermediate ss made more overgeneralization errors “two distinctive linguistic manifestations of one psychological process” (Taylor, 1975b, p.87)

Page 57: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and QuestionsIssues and Questions

Kate: In factors special to L2, it seems that the major problems of interference, overshadowing and blocking, and perceptual learning result from the fact that L2 is learned after L1.

So, if two languages are learned at the same time, do these problems still exist?

Page 58: L2 endstate and_dynamic_l2_interlanguage.edited

Conclusion System is more than the sum of its parts, it is a

“highly variable dynamic system” (p.397)

“Every usage is dynamically influenced by the interactions among different parts of the system that are unique in time” (p.396)

SLA is not stepwise, but on a multidimensional continuum, e.g., Bayley’s (1994) study of past tense morphology in advanced Chinese learners of English, involving tense, aspects, phonetic saliency, phonological processes that converge with morphological classes, as well as social and developmental factors

Page 59: L2 endstate and_dynamic_l2_interlanguage.edited

L2 IL as DS

Page 60: L2 endstate and_dynamic_l2_interlanguage.edited

L2 IL as DSOntology of learning and IL?

Glavine: According to Ellis, the perceptual systems may alter the sensitivity to stimulus features, and learners can become more sensitive to those which are psychologically significant dimensions of variation among the stimuli. I am thinking about if this perceptual system can be considered as one part, or similar construct of interlagnuage development. If they are similar, how can learners be trained to develop the perceptual system? Only by making mistakes?

Page 61: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and QuestionsIssues and QuestionsComplex Dynamic SystemsComplex Dynamic Systems

1. Do you think N. Ellis means when he says that the IL is dynamic system whose “whole does not equal the sum of its parts”? Do you agree?

Leo: In the section of conclusion, N. Ellis talked about the complex system. According to Larsen-Freeman, SLA can be regarded as a complex dynamic system and learning a language changes the knowledge system, rather than adding to the system. And SLA varies in terms of the nature of the interaction among all the elements, including individual and social factors. SLA is always in movement and never reaches equilibrium. Perhaps this can provide another perspective on the attainment of adult SLA.

Page 62: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and Questions 1. N. Ellis is trying to provide a non-age

dependent explanation for L2 limited endstate. That is, he is providing a number of cognitive learning scenarios that he also thinks apply to SLA. Are you convinced?

2. Some researchers, like VanPatten, think that language acquisition cannot be comparable to other forms of skill acquisition. Do you think language learning uses different learning mechanisms or principles than other forms of learning (e.g., learning Tae kwon Do)?

Page 63: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and Questions Issues and Questions PedagogyPedagogy

1. Lily: If L2 acquisition is proved to be affected by prior knowledge, in particular, our L1, how does this fact challenge teaching? Moreover, how do we deal with the L1-L2 trade-off issue? (Leo Q.7 aslo, how to help ss overcome L1 influence?)

2. How does your awareness of your students’ learning problems (due to your familiarly with Chinese and L1 entrenchment) affect your teaching materials, methods or expectations?

3. Because some educators are pessimistic about L2 endstate (especially for unmotivated High School and College L2 learners), they might lower their expectations or limit their teaching focus to certain aspects of L2 learning. If you didn’t feel you had to teach to a test or teach a fixed curriculum, what would you like to focus on with your students?

Page 64: L2 endstate and_dynamic_l2_interlanguage.edited

Issues and Questions Issues and Questions PedagogyPedagogy

Page 65: L2 endstate and_dynamic_l2_interlanguage.edited

End

Thank you for your attention.