Jenifer Larson-Hall
jenifer@unt.edu
Research on age effects has shown unequivocally
that "younger is better in immersion contexts when the individuals in
question have continuing second language input. This study asks whether the
same effects hold true when immersion is for a limited time only. This study
examined 14 adult Japanese speakers of English who had lived in the
The participants were evaluated on several tasks, including an American English [r/l/w] phoneme discrimination task, production of words beginning in [r] and [l], and sentence production. An instrumental investigation of voice onset time was also included. In addition, each of the Japanese participants conversed with the investigator for several minutes and a qualitative evaluation of the transcripts of these conversations was investigated for insight into morphosyntactic ability. Taken as a whole the results indicate that the early learners' performance was superior to the later learners, arguing for the importance of neurobiological factors. However, Japanese living abroad but without early experience were in some aspects superior to those who had never lived abroad, also arguing for the importance of input for adults.
John Field
Current
Drawing partly upon problem-solving theory, this paper offers a model of second-language listening which attempts to reconcile the two perspectives. It cites evidence from a research study in which the gating method was used to investigate the decoding processes employed by second language listeners. The study provides insights into the limited extent to which second language listeners succeed in recognising known words when they occur in connected speech, and reveals L2 decoding to be highly strategic. It testifies to a ‘perseverative effect’, also observed at higher levels of processing, where L2 listeners show themselves resistant to revising an existing hypothesis, however provisional and approximate.
Evidence suggests that a considerable degree of uncertainty is also present in L1 decoding. The difference between the native and the non-native decoder may be found partly in the listener’s confidence as to the reliability of that evidence, but also in the approach adopted to the forming and revising of assumptions.
Paola Escudero
escudero@uva.nl
In this talk, I will present the Linguistic Comprehension (LC) model, a new proposal for explaining the development of sound perception and word recognition in different groups of language learners. Importantly, this proposal is compatible with current psycholinguistic models and improves on the adequacy of phonetic and phonological models.
The LC model assumes that perception and recognition grammars represent the linguistic knowledge which underlies the perception of sounds and the identification or recognition of words respectively. Following Escudero & Boersma (2003), a perception grammar contains cue constraints which connect auditory input with phonological representations. Following Boersma (2001), a recognition grammar contains lexical constraints which ban the access of lexical representations, and faithfulness constraints which preserve the phonological form of the perceived word. In both types of grammars, the constraint rankings follow Stochastic Optimality Theory (Boersma 1998) and combine the phonetic/phonological properties of language input with their frequency distributions.
Regarding language acquisition, the model provides a formal account of the perceptual and lexical development of babies raised in monolingual and bilingual environments as well as that of adult L2/L3 learners. The perceptual component is based on a combination of distributional and lexicon-driven learning, while the lexical component is based on perceptual frequency and message-driven learning. It is proposed that these four learning mechanisms are instantiated by the Gradual Learning Algorithm (Boersma & Hayes 2001) at different stages of the acquisition process.
The proposed model will be exemplified with the acquisition of vowel perception and word recognition by monolingual (Dutch) and bilingual (Dutch-Portuguese) children, and by adult L2-learners/bilinguals acquiring an L3 (Dutch-English-Spanish, Portuguese/Spanish-English-Dutch). Thus, the comparative developmental path predicted by the model will be tested against the results of empirical studies and computer simulations.
Finally, I will discuss the various implications for L1 versus L2 acquisition as well as implicit versus explicit language learning and teaching.
Shawn Loewen
s.loewen@auckland.ac.nz
Incidental focus on form draws
learners’ attention to linguistic items as they arise spontaneously, without
prior planning, in meaning-focused L2 classroom interaction. Such a combination
of form and meaning is argued to be beneficial for L2 learners. While previous
studies have investigated the effectiveness of incidental focus on form in
promoting second language learning through measures such as uptake and individualised test scores, the present study considers
effectiveness by examining learners’ production of targeted forms both prior
and subsequent to the focus on form episodes (FFEs).
Learners’ prior use of linguistic structures targeted in FFEs
may provide some indication of learners’ knowledge of the forms, given that pretesting the forms is not
possible due to the spontaneous nature of incidental focus on form.
Furthermore, learners’ subsequent use of targeted forms in classroom
interaction may provide evidence of the effectiveness of FFEs,
particularly as the structures occur in unplanned, on-line communication. In
the present study, 4.5 hours of naturally-occurring, meaning-focused L2 lessons
were observed in three different classes of young adults in a private language
school in
Jenefer Philp
j.philp@ auckland.ac.nz
Shawn Loewen
s.loewen@auckland.ac.nz
Rod Ellis
r.ellis@auckland.ac.nz
While previous research provides strong support for the benefit of some kind of form-focused instruction in classrooms (Doughty & Williams, 1998b; Ellis, 1994, 2001; Long, 1991, 1996; Long & Robinson, 1998; Nassaji & Fotos, 2004; Norris & Ortega, 2000; Spada, 1997), there is continued debate concerning the efficacy of different types instruction. “PPP” (“presentation, practice and production) instruction, a meaning-oriented approach structured to include explicit instruction, structured practice and independent learner output, has been widely adopted in many teacher training programs. This study investigates the relative effects of PPP on adult learner’s acquisition of generic article “a/an” and third person singular “s” (3sgS). Two intact classes of 19 adult ESL students were given identical focused tasks over two half hour sessions which elicited production of the generic article and third person singular s. The students received explicit instruction on the use of the generic article, with negative feedback on oral production, yet only incidental implicit negative feedback following nontarget like production of 3sgS. A control group, received routine instruction without focus on either of the two forms targeted. A pre-test, an immediate and a delayed post-test involved a grammaticality judgment test, an elicited imitation test and a metalinguistic knowledge test of the two structures.
Pre-test performance for both groups revealed low explicit knowledge of both forms, particularly for the generic article. The PPP group exceeded the control group on post-test measures. Gain scores were highest, however, for the generic article, on which students had received explicit instruction with feedback. The study suggests positive effects for PPP on particular forms, and greater effect for explicit rather than implicit instruction within this context. The presentation discusses both the pedagogical and theoretical implications of this study, particularly how the interaction between effectiveness of instruction, targeted structure and the learners’ prior knowledge of that structure.
Iliana Panova McGill University, Canada
iliana.panova@mail.mcgill.ca
Observational classroom-based research on corrective feedback and learner repair in communicative L2 contexts has shown that morphological errors tend to be recast and that rates of repair following recasting of these errors are low (Lyster, 1998; Pica, 2002). Similarly, the use of stimulated-recall methodology has found that adult L2 learners have difficulties in accurately perceiving the focus of recasts provided in response to their non-target morphological forms (Mackey, Gass & MacDonough, 2000).
This study reports on teacher feedback patterns associated with non-target use of article and plural morphology by adult L2 learners at a low proficiency level in an ESL classroom. The analysis is based on transcripts of 15 hours of audiotaped classroom interaction. Results show that inaccuracy in the nominal domain is characterized by high rates of omission and minimal overuse of morphological markers. Non-target morphology was either ignored or recast by the teacher, and the levels of learner repair following teacher recasts were minimal. Following a recent theoretical proposal related to morphological variability in L2 production (Prévost & White, 2000), this paper will discuss the theoretical and pedagogical implications involving the use of recasts as an appropriate and effective response to morphological errors in early L2 acquisition.
John Williams
The University of
Chieko Kuribara
The University of
What kinds of representations are constructed after limited exposure to a new language? Do learners construct grammatical representations at all, and if so, to what extent do they utilise L1 grammatical knowledge? Or is initial learning driven by memory-based learning algorithms that are simply sensitive to the statistical properties of the input? In this experiment native English-speaking participants were exposed to sentences combining English lexis with Japanese syntax. Japanese differs from English as to the headedness of VP, and the availability of scrambling. In the exposure phase participants made (semantic) plausibility judgements on 174 training sentences. This was followed by a grammaticality judgement task. All stimuli were presented auditorily and visually under computer control. Most of the training sentences reflected Japanese canonical word order (e.g., S[SOV]V, “John-ga Bill-ga apple-o ate that said.”), whilst 25% of them were scrambled (e.g.. OS[SV]V, “That book-o John-ga Mary-ga bought that said.” ). The test sentences varied as to the headedness of VP, the inclusion of wh-phrases, and properties of scrambling (e.g. landing-site, distance, and number of applications).
Main findings are:
Ø All subjects accepted canonical and scrambled structures that had been received in training at above-chance levels.
Ø Some of the participants (40%) reliably accepted types of grammatical clause-internal scrambling that they had not encountered during training.
Ø L1 head-initial VP influenced acceptance rates.
Ø Structures that violate the constraints appropriate for operator movement (such as topicalisation and wh-movement) but not for scrambling, tended to be rejected.
Ø Performance could not be simply explained in terms of linear word-order similarities between test and training items, ruling out a statistical learning account.
Based on these findings, we conclude that after limited exposure in a meaning-focussed task the participants had acquired a notion of movement, which was likely to be guided by grammatical knowledge transferred from L1.
Renita Silva
rhsilv@essex.ac.uk
Harald Clahsen
There has been much debate on the mechanisms by which morphologically complex words are processed. The Dual Mechanism Model (Pinker, 1999) posits that there is a dissociation between rule-based forms which are decomposed (e.g. regulars), and associative forms which are stored fully in the lexicon (e.g. irregulars). This model has been based on various studies in L1 showing dissociation between regular/productive processes on the one hand and irregular/unproductive processes on the other. In light of this, Stanners et al. (1979) found that the reaction time to a target word can be facilitated by the presentation of a prime word earlier and specifically that in inflections, regular forms produce full priming while irregulars do not. Looking at derivation, Clahsen (2004) found that adjectives in German with the productive prefix un- showed full priming, while only partial priming was found for the unproductive prefix in-.
Although the Dual Mechanism model can account for much of the extant empirical evidence in L1, it remains to be shown whether it can also account for L2 behaviour. Ullman (2001) proposes an additional aspect in that there are two different memory systems such that the construction of rules is dependent upon procedural memory and associations are dependent upon declarative memory. Based on the Declarative/Procedural Model, Ullman (2001) postulates that L2 will rely on declarative memory while L1 will make use of both memory systems.
In this study, masked priming experiments investigated the occurrence of priming in the regular past tense and productive deadjectival nominalisations, and whether the L1 and L2 would behave the same. The results indicate that native speakers show full priming for both inflection and derivation indicating that these regular/productive forms are decomposed. However, the L2 showed no priming for inflection and reduced priming for derivation which is supportive of the Declarative/Procedural Model.
Muiris Ó Laoire
David Singleton
CLCS,
In discussion of cross-linguistic influence a debate continues regarding the critical factor prompting a native speaker of language x to resort to language y when using language z. Does this arise from a perception that language y is typologically closer to language z than any other available language (the psychotypological perspective - see e.g. Kellerman 1983; Ringbom 1987; Singleton forthcoming) or from the fact that language y is, in common with language z, a non-native language (the L2 factor perspective - see e.g. Hammarberg 2003; Williams & Hammarberg 1998). Singleton & Ó Laoire’s recent studies (forthcoming a and b) of French being learned as an L3 by Anglophone learners with Irish as their L2 or as a second L1 indicated a marked predominance of the psychotypological factor in lexical transfer patterns. The privileged transfer source was English, which is explicable in terms of learners’ recognition that, in lexical terms, English is closer to French than Irish is. The present paper will report on a study of learner German involving two groups of English-speaking participants for whom German is their L3 and who again have Irish as their L2 or as a second L1. The focus of the study in this instance will be on two areas where Irish morphsyntax is closer to German than to English, i.e. morphological inflection and word order. It is expected that there will markedly more evidence of Irish-based transfer in this case than in the earlier studies.
This paper is situated within the
recent wave of
While such findings are seen to reflect an important role for learning context in L2 acquisition, there is a need to consider whether naturalistic exposure simply speeds up the acquisition process, or whether it genuinely differentially impacts on L2 acquisition by giving rise to language acquisition and usage which simply do not happen in the foreign language classroom. That is to say, it remains unclear to what extent classroom learners who are not subject to naturalistic exposure in the target language community for a prolonged period may simply require more time before such development emerges in their L2 speech repertoire.
This paper investigates such a
question based on a three-way cross-sectional comparison of university learners
of French in
phf.bos@let.vu.nl
Relatively high percentages of
Moroccan children that attend primary education in the
Rahim Sarı
The Turkish Armed Forces,
rahimsari@yahoo.com
Most ELT materials follow an integrated skills syllabus that places almost an equal emphasis on teaching and practice of four skills from the very beginning of instruction, presuming that skills development will accordingly take place parallel to the focus given in the learning environment. However, with the input hypothesis placing the receptive skills of listening and reading to the fore and regarding the productive skills as natural outcomes of language acquisition, there emerged an interest to test the validity of the hypothesis in this respect.
The paper examines the results of an experimental classroom research study that compares the speaking and writing performances of an experimental group of young learners under an input-based form of instruction with those of a control group under an integrated-skills form of instruction. The study covered 28 weeks of instruction at a high school prep-class in Turkey. During the treatment, while doing a lot of extensive reading and listening, the experimental group did not participate in any writing activity beyond sentence level in or outside the classroom and were not forced to speak in the classroom. The control group, however, followed their standard course books of an integrated-skills approach with emphasis on grammar, writing and speaking activities. The control group also spent time on grammar and writing activities outside the class.
At the end of the experimental period a speaking and two writing tests were given to the groups. The experimental group scored better in the speaking and writing tests despite the deliberate avoidance of practice with the productive skills during the treatment, which makes the study particular. The results of the study seem to challenge the idea of placing equal emphasis on four skills from the very beginning of instruction and have some serious implications for language teachers, syllabus makers, and curriculum designers.
Ute.Bohnacker@nordlund.lu.se
Many theories of developmental syntax build on the assumption that acquisition proceeds from the bottom up, where (in generative terms) lower structural projections are acquired before higher ones, and lexical projections are targetlike before functional projections are (e.g. Vainikka & Young-Scholten 1994, 1996; Hawkins 2001). On some approaches, learners moreover only have to grapple with the acquisition of the topmost clausal level, CP, which Platzack (1996, 2001), Hulk & Müller (2000) and Bhatt & Hancin-Bhatt (2002) describe as a “vulnerable domain”. Grammatical operations related to this level are acquired imperfectly and comparatively late, and elements associated with the CP tend to be omitted. The vulnerability of the C-domain is contrasted with the “invulnerability” of lower structural domains (VP, IP), which are said to be acquired effortlessly and early by the language learner.
The present paper challenges the alleged existence of universally vulnerable or invulnerable syntactic domains, on the basis of quantified data from 9 native-German adult learners of Swedish. German and Swedish are both V2 languages, but vary in VP headedness (OV/VO) and particle-verb syntax. Oral and written production data were collected after 3, 6, 9 and 15 years of exposure to Swedish. The learners master syntactic operations at high clause-structural levels from early on, in particular non-subject-initial V2 (contra e.g. Pienemann 1998, Platzack 2001). At the same time, the L2ers exhibit a nontargetlike syntax at lower structural levels which persists for many years, in particular in transitive verb particle constructions (Bohnacker, in press). I show that such findings are best explained by adopting a theory of acquisition that assumes full transfer of L1 syntax at higher and lower clause-structural levels (e.g. Schwartz & Sprouse 1994, 1996). I thus argue that L2 theories that postulate the existence of universal developmental paths and universally (in)vulnerable syntactic domains should be abandoned.
botwinik@post.tau.ac.il
L2ers of English accept omission of subcategorized prepositions (null prep, henceforth) in questions (Qs) and relative clauses (RCs), (1). If Qs are derived obligatorily via movement, null prep Qs violate the ECP indicating a wild intergrammar (Klein 1993, 1995).
(1) a. Which exam is the student worrying?
b. Here’s the exam that the student is worrying.
Dekydspotter, Sprouse and Anderson (DSA) (1998)) argue that interrogatives in L2 acquisition are construals, not involving movement but rather licensing of a pro by the verb with an incorporated P. Postulating that null prep grammar is a generalized procedure in L2 development, they predict that it should be attested in the acquisition of any L2. In response, Klein (2001) argues that P‑incorporation is untenable, as null prep is attested in adjunct questions (e.g. Which cafeteria did she eat her lunch?), and proposes an Op-movement analysis of null prep Qs, triggered by P-stranding in the input, predicting that null prep should not be attested in a non-stranding L2.
I checked the above predictions and claims using L2 Hebrew (L2H), a non-stranding language. My findings show that in the acquisition of L2H null prep grammar is not attested, and that interrogatives are derived via movement (Klein 1993, 2001, contra DSA 1998). Based on the observation that null prep in L2H is attested only in RCs with adjunct PP (be-, 'in') and that its rate in adjunct Qs as compared to argument Qs in L2 English (L2E) is significantly higher (not accounted for in Klein 2001), I argue that null prep has two sources: (a) (mis)analysis of the locative P as null D of the N Place, or (b) V-P reanalysis crucially triggered by P-stranding. Both are operative in L2E accounting for the high rate of null prep in adjunct Qs. Only (a) is relevant for L2H, affecting only RCs as they involve Op, but not Qs that employ an invariably overt Q-operator.
Hyun Kyung Miki Bong
This talk discusses the implications of certain experimental results for existing L2 acquisition theories, focusing on their treatment of learnability of functional properties (categories, features, etc), optionality between L1 and L2 structures, and the role of economy principles. The results were obtained from translation, grammaticality judgement, and multiple-choice tests conducted on five different proficiency groups of adult Japanese-speaking learners (JSLs) of English. The tests concerned various types of wh-question formation in the context of complex sentences: e.g. “What did Taroo say (that) Miki bought?”.
The results indicate that (i) the JSLs develop the English wh-question formation by using movement of wh-expressions, not by manipulating the Japanese grammar; (ii) they do not use the L1 and L2 structures interchangeably, but demonstrate variability of those structures among themselves; and (iii) they do not violate economy principles and indeed show some preference toward economical options.
I point out that none of the existing L2 theories that are variants of the deterministic parameter-setting model, according to which a specific parameter is set to match a specific input setting, can capture all these results. For example, the results suggest that the apparent absence of particular functional features from the L1 does not determine the L2 learners’ ability to acquire them, casting doubt on the claims of the deterministic view of learnability, according to which L2 learners cannot acquire functional properties, owing to either the inaccessibility or the impairment/unavailability of those elements.
Instead, I demonstrate that the results that cannot be explained by any previous theories can be accommodated and even predicted by a minimalist weak-deterministic model as proposed in Bong (2005): i.e. the Economical Parameter-Setting model. This model adopts the economy considerations in the Minimalist Program (Chomsky 1995, 2000) and modifies a pre-minimalist model of language learnability and language change (Clark and Roberts 1993).
Christine Dimroth
Max Planck Institute for
Psycholinguistics, The
christine.dimroth@mpi.nl
This contribution presents a particular approach to the question of how the human learning faculty deals with the complex task of untutored second language acquisition. Proponents of a 'learner varieties approach' focus on changes in the form and function of items/structures and ask which internal (and to some extent external) factors shape the course of acquisition processes.
To this end, production and comprehension data are collected with varying degrees of control from mostly untutored learners - real beginners up to highly advanced speakers. The data should include enough context to allow insights into what learners can and cannot do with a given variety – preferably in a longitudinal set-up which captures development. Analyses are guided by an 'unknown language perspective', trying to study the organising principles of developing learner varieties in their own right, rather than in relation to (one out of many) descriptions of the target language.
Quantitative and qualitative changes in the structure of learner varieties are considered to be evidence for learning, even though they do not always bring the learner closer to the target language. For example, when studying the acquisition of finiteness in learner varieties of Dutch, German, and French, researchers have tried to understand what learners do before the first target-like means are available, and how they then build on these earlier structures (Starren, 2001; Dimroth et al., 2003; Becker, 2005). Changes in other areas may then be interpreted in relation to these developmental sequences: Is the integration of temporal adverbs in learner varieties of French determined by the expression of finiteness (Benazzo, 2002)? Is the development of inversion in learner varieties of Dutch related to the acquisition of auxiliaries (Verhagen, 2005)? Because of its focus on forms and functions, the approach might broadly be labeled as functionalist. However, second language learners are not assumed to acquire only what is immediately functional. They might incorporate input properties that are frequent, or salient, without necessarily having a “need” in the first place. But what they do incorporate may have repercussions for the internal organisation of their learner variety as a whole.
Florencia Franceschina
Lancaster University, UK
f.franceschina@lancaster.ac.uk
The presence of non-target-like forms
is one of the typical features of interlanguage
grammars. Now, is the presence of non-target-like forms in learner language
reliable evidence of failure to acquire the target language? And conversely, is
the presence of target-like forms reliable evidence of successful learning?
This contribution will discuss how
In this framework a distinction is made between competence (i.e., a speaker’s knowledge of the language) and performance (i.e., how the speaker puts his/her language knowledge to use in specific circumstances). Any given instance of non-target-like performance could be either a reflection of non-target-like competence or a distorted view of target-like competence realised through non-target-like performance. For example, L1 Chinese learners of L2 English tend to make past tense errors even at advanced levels, such as
(1) I talk to my friend yesterday
A learner who produces sentences of this type could be said to lack (full) knowledge of the English Past Tense rule. However, it may also be the case that the learner knows the English Past Tense rule, but he/she has difficulty pronouncing the consonantal cluster at the end of ‘talked’.
This paper will focus on the dilemma of deciding whether learning has or has not taken place in the presence of non-target-like learner performance by reference to the work of two groups of generative researchers investigating advanced L2 learners who have taken different views on this matter. The specific hypotheses that will be considered are the Missing Surface Inflection Hypothesis (Prévost & White 2000) and the Failed Functional Features Hypothesis (Hawkins & Chan 1997).
Although traditionally grammaticality judgements have been one of the most valued types of evidence for acquisition in this framework, it is also accepted that triangulation is important, and it is more and more common to see studies using different types of evidence (including production, comprehension, metalinguistic and online data).
Simona Pekarek Doehler
simona.pekarek@unine.ch
Johannes Wagner
jwa@sitkom.sdu.dk
In a socio-interactional approach to learning (including second language acquisition) learning is understood as inseparably linked to social practices in which a person engages –practices that are locally accomplished and historically shaped, and, most typically, interactionally organized (Firth & Wagner, 1997; Lantolf, 2000; Kramsch, 2002). This view implies an analytic and conceptual focus on activity as the starting point for the study of human functioning, including practical reasoning and language learning.
In this contribution, we will focus
on one specific approach within socio-interactionally
oriented
Such evidence shifts attention from investigating language form as being a context-independent indicator for learning to exploring language as a situated type of behavior; it also shifts attention from learning as a progressive interiorisation and restructuring of cognitive capacities, knowledge or know-how, towards understanding learning in terms of changing the capacity to participate in social activities (i.e. Lave & Wenger, 1991).
Natasha Tokowicz
tokowicz@pitt.edu
Janet G. van Hell
In this talk, we will compare the sensitivity of electrophysiological measures of second language (L2) learning to more traditional behavioral measures. In traditional psycholinguistic investigations, improved accuracy indicates L2 learning (e.g., in the grammaticality judgment or lexical decision tasks). However, recent electrophysiological studies, in which event-related brain potentials (ERPs) are derived from recordings of brain activity during language processing, suggest that behavioral measures of L2 learning might underestimate what is actually known. In ERP studies, learning is indicated by differences in the amplitude of activation in different conditions (e.g., grammatically acceptable and unacceptable) during timeframes that correspond to certain cognitive events. Findings based on electrophysiological measures sometimes deviate from the behavioral measures obtained from the same participants; this discrepancy may derive from the requirement to make an overt response. For example, grammaticality judgment tasks require the learner to indicate overtly whether they believe a sentence to be unacceptable whereas ERP tasks do not require overt responses but rather measure brain activity as it unfolds over time. As such, ERPs may be more sensitive measures of learning than behavioral measures, particularly those that involve yes/no judgments. We will examine several questions about the differences between behavioral and electrophysiological measures of L2 learning. First, what behavioral and electrophysiological tasks measure explicit and implicit learning? Second, electrophysiological investigations have demonstrated that beginning L2 learners are sensitive to L2 structure from very early stages of learning, both in terms of meaning (McLaughlin, Osterhout, & Kim, 2004) and in terms of grammar for some constructions (Tokowicz & MacWhinney, 2005). How can these findings be reconciled with results from behavioral measures that suggest L2 is learned only explicitly (e.g., DeKeyser, 2000)? Finally, is brain sensitivity a precursor to demonstrating behavioral overt sensitivity (i.e., preparedness to learn), or does it change in tandem with overt sensitivity?