In this paper the concept of “lexical unit” as proposed by Cruse (1986) is introduced in the context of L2 vocabulary learning to replace the rather vague notion of “word.” In the first part of the paper, this concept is defined in terms of lexical semantics and then applied to the L2 learning task. The second part describes two experimental studies in which the learning of different types of lexical units is examined. In the first study, totally new lexical units are compared with multiword items that are made up of familiar forms. In the second experiment, different types of new senses of familiar forms are compared. Both studies underscore the importance of knowledge of form—but not of previously learned meaning—for the learning of new meanings for familiar forms.

This study investigates the effects of simplification, type of linguistic item, and second language experience on learners' intake of linguistic items contained in written input. Learners at two levels of language experience were exposed to one of the following four conditions: a simplified or unsimplified reading passage with the present perfect tense form or a simplified or unsimplified reading passage with the present subjunctive form. To measure learners' intake, a repeated-measures analysis of variance was performed on the raw scores obtained on a task consisting of a pre- and posttest. The tests were a multiple-choice recognition assessment task. Significant main effects were found for type of passage and language experience, and there was a significant interaction between type of passage and task. Results suggest that simplification does not have a facilitating effect on learners' intake and that learners at different levels demonstrate a different pattern of performance while internalizing written input. Implications for pedagogical written materials and future research are also discussed.

Findings from a longitudinal study of bilingual children acquiring Cantonese and English pose a challenge to the noun phrase accessibility hierarchy (NPAH; Keenan & Comrie, 1977), which predicts that object relatives should not be acquired before subject relatives. In the children's Cantonese, object relatives emerged earlier than or simultaneously with subject relatives, and in their English, prenominal relatives based on Cantonese emerged first, with object relatives followed by subject relatives. These findings are discussed in light of findings on the typology and acquisition of relative clauses (RCs) and the underlying processing motivations of the NPAH. Prenominal object relatives in the bilingual children's Cantonese and English have the same word order as main clauses and can be analyzed as internally headed RCs. The reconceptualization of RCs as attributive clauses (Comrie, 1998a, 1998b, 2002) is supported by children's early RCs lacking a strict grammatical relationship between the head noun and the predicate. Furthermore, as observed by Diessel and Tomasello (2000, 2005) for English, bilingual children's earliest RCs consist of isolated noun phrases (NPs). The early object relatives produced by bilingual children are therefore essentially NPs with the linear order of a main clause, resulting in a configuration that is conducive to early production.We thank Yasuhiro Shirai for organizing the workshop on second language acquisition of RCs at Cornell University in January 2006, where this paper was presented, as well as for his valuable comments on earlier versions of this paper. We thank all the participants at the workshop and two anonymous SSLA reviewers for their comments. Additional feedback offered by William O'Grady, Salikoko Mufwene, and John Whitman is greatly appreciated and hereby acknowledged. We thank our children for their contributions to this paper and all of the members of our research team who have contributed to this work—in particular, Uta Lam for her technical assistance. This research has been fully supported by the Research Grants Council of the Hong Kong Special Administrative Region, China (project Nos. HKU336/94H, CUHK4002/97H, CUHK4014/02H, and CUHK 4692/05H) and direct grants from the Chinese University of Hong Kong (01/02, 03/04).

Although many theoretical issues revolving around implicit and explicit knowledge in second language (L2) acquisition hinge on the ability to measure these two types of knowledge, few empirical studies have attempted to do so. However, R. Ellis (2005) did develop a battery of tests intended to provide relatively separate measures. This study aims to validate the use of Ellis’s test battery in an English as a foreign language context and to investigate the extent of Chinese first-year university students’ implicit and explicit L2 knowledge. Test scores loaded on two factors, as in R. Ellis (2005), thus demonstrating construct validity for the tests as measures of implicit and explicit knowledge in a population of Chinese university-level learners of English in a Chinese (as opposed to English as a second language) context. These learners were found to have higher scores on measures of explicit knowledge than on those of implicit knowledge because of the instruction they had received and their English learning environment.

This study examines the development of speech rhythm in second language (L2) learners of typologically different first languages (L1s) at different levels of proficiency. An empirical investigation of durational variation in L2 English productions by L1 Mandarin learners and L1 German learners compared to native control values in English and the learners’ L1s showed that the L1 groups followed comparable developmental paths in their acquisition of vocalic variability and accentual lengthening. However, the two L1 groups diverged in the proportion of vocalic materials in their L2 utterances, exhibiting L2 acquisition patterns that are consistent with direct transfer from the L1. The results support a multisystemic model of L2 rhythm acquisition in which the various linguistic-systemic properties that contribute to speech rhythm are acquired at different proficiency levels and depend on different acquisition processes with respect to L1 influence and universal effects. We conclude that theories of L2 phonology need to be able to accommodate the multisystemic nature of L2 prosodic acquisition. Additionally, L2 phonological acquisition theories, and SLA theories more generally, should take into account the nonuniform manner in which the various prosodic properties of the interlanguage reflect L1 transfer effects as well as universal constraints on acquisition.

This experimental study on English, Spanish, and Turkish as second languages investigates the interaction of universal principles and L1 knowledge in interlanguage grammars by focusing on verbs that participate in the causative/inchoative alternation (such as break in English). These verbs have the same lexico-semantic composition, but differ crosslinguistically as to how they encode the alternation morphologically. Results of a picture judgment task show that, as in L1 acquisition, L2 learners of Turkish, Spanish, and English with different L1s rely on a universal mechanism when learning transitivity alternations. L1 influence plays a prominent role in the morphological realization of the alternation. These findings suggest that UG and L1 knowledge may not affect all linguistic domains in the same way at a given stage of development. It is proposed that transfer is subject to modularity in interlanguage grammars.

The purpose of the present paper is to examine the hypothesis that there are fundamental differences between using language in different situations or for different purposes, and that these differences may be accounted for in terms of the requirement of the task to be solved on the basis of knowledge represented in a particular way.

This paper reports on a test of the validity of Pienemann's (1998) Processability Theory (PT). This theory predicts that certain morphological and syntactic phenomena are acquired in a fixed sequence. Three phenomena were chosen for this study: attributive adjective morphology, predicative adjective morphology, and subordinate clause syntax (placement of negation). These phenomena are located at successive developmental stages in the hierarchy predicted by PT. We test whether they actually do appear in this predicted hierarchical order in the L2 of Danish, Norwegian, and Swedish learners. The three languages mentioned are very closely related and have the same adjective morphology and subordinate clause syntax. We can, therefore, treat them as one language for the purposes of this study. Three analyses have been carried out: The first follows Pienemann's theory and is concerned only with syntactic levels; the second is a semantic analysis of the acquisition of number versus that of gender; the third analysis studies the various kinds of mismatches between the inflection of the noun, the controller, and the adjective. The results are the following: The first test supports PT as it has been described by Pienemann. The second analysis shows that there is an acquisitional hierarchy such that number is acquired before gender (in adjectives), and the mismatch analysis raises questions about the fundamental assumptions of the theory.

In a paired-associate learning (PAL) task, Dutch university students (n = 129) learned 20 English second language (L2) idioms either receptively or productively (i.e., L2-first language [L1] or L1-L2) and were tested in two directions (i.e., recognition or production) immediately after learning and 3 weeks later. Receptive and productive performance was affected by direction of learning. This finding parallels findings from PAL experiments on L2 individual-word learning. On a productive test, productive learners had a sizable advantage over receptive learners, whereas on recognition, receptive learners outperformed productive learners. Two idiom characteristics, imageability (capacity to evoke a mental image) and transparency (overlap between literal and figurative meaning), as assessed in a norming study by an independent sample (n = 80), qualified these findings. Indicating the importance of dual coding in idiom learning, imageability predicted performance, and receptive learning was particularly inefficient for low imageable idioms. Transparency was a weaker predictor of performance and only affected recognition.This study is based on the first author's MA research. We would like to thank the four anonymous SSLA reviewers for their helpful comments on this manuscript. Any errors or omissions remain our own.

The role of explicit information (EI) as an independent variable in instructed SLA is largely underresearched. Using the framework of processing instruction, however, a series of offline studies has found no effect for EI (e.g., Benati, 2004; Sanz & Morgan-Short, 2004; VanPatten & Oikkenon, 1996). Fernández (2008) presented two online experiments with mixed results. She found an effect for EI with processing instruction on one target structure (subjunctive in Spanish) but not the other structures (object pronouns and word order in Spanish). Thus, the effects of EI could be related to the target structure or to a processing problem, or both. The present study is a conceptual replication of one of Fernández’s experiments. The target was German accusative case markings on articles with both subject (S)- verb (V)- object (O) and OVS word orders. As shown by Jackson (2007) and LoCoco (1987), learners of German as a second language misinterpret OVS sentences as SVO, ignoring case markings as a cue of who does what to whom. Thus, the goal of the instructional intervention was to push learners to process case markings and word order correctly. The treatment consisted of structured input items (Farley, 2005; Lee & VanPatten, 2003) under two conditions: +/−EI. Following Fernández, the treatment was conducted via computer using e-Prime, and learners’ responses were recorded as they made their way through the items. Whereas Fernández did not find an effect for EI for word order and object pronouns in Spanish, we found an effect for word order and case markings in German: (a) Twice as many learners in the +EI group reached criterion (began to process input strings correctly) compared with the −EI group, and (b) learners in the +EI group began processing word order and case markings sooner than in the −EI group. Even though the processing problem was the same in both Fernández’s and our experiments, we attribute the difference in results to the interaction of particular structures with the processing problem and call for additional research on the role of EI not just in processing instruction but in all formal interventions.

This investigation of NS–NNS interaction in same- and cross-gender dyads on four information exchange tasks revealed that male and female NNSs made and received a comparable number of opportunities to request L2 input and modify interlanguage output during interaction with female NSs, but during interaction with male NSs, these opportunities were significantly lower for female than for male NNSs. In addition, more request-response exchanges were found on tasks in which either the NS or the NNS was given initial control over task-related information. Findings of the study were attributed to cultural similarities and differences in the interactional behaviors of the participants.

Two commonly used test types to assess vocabulary knowledge for the purpose of reading are size and levels tests. This article first reviews several frequently stated purposes of such tests (e.g., materials selection, tracking vocabulary growth) and provides a reasoned argument for the precision needed to serve such purposes. Then three sources of inaccuracy in existing tests are examined: the overestimation of lexical knowledge from guessing or use of test strategies under meaning-recognition item formats; the overestimation of vocabulary knowledge when receptive understanding of all word family members is assumed from a correct response to an item assessing knowledge of just one family member; and the limited precision that a small, random sample of target words has in representing the population of words from which it is drawn. The article concludes that existing tests lack the accuracy needed for many specified testing purposes and discusses possible improvements going forward.

If one would ask me as a regular, though not systematic reader in applied linguistics to mention a major theoretical issue in this field of study, I would choose for the controversy about “behavioristic” audiolingual learning versus “mentalistic” cognitive code learning. This dichotomy is apparent in many theoretical papers, and starting point of various research programs.

There has been little research investigating how mode of input affects incidental vocabulary learning, and no study examining how it affects the learning of multiword items. The aim of this study was to investigate incidental learning of L2 collocations in three different modes: reading, listening, and reading while listening. One hundred thirty-eight second-year college students learning EFL in Taiwan were randomly assigned to three experimental groups (reading, listening, reading while listening) and a no treatment control group. The experimental groups encountered 17 target collocations in the same graded reader. Learning was measured using two tests that involved matching the component words and recalling their meanings. The results indicated that the reading while listening condition was most effective while the reading and listening conditions contributed to similarly sized gains. The findings suggest that listening may play a more important role in learning collocations than single-word items.

This article reports an empirical study in which we explored the role played by two forms of feedback—error correction and model texts—on child English as a foreign language learners’ reported noticing and written output. The study was carried out with 11- and 12-year-old children placed in proficiency-matched pairs who engaged in a three-stage collaborative writing task involving (a) spontaneous noticing of linguistic problems while composing, (b) comparison of their texts with the feedback provided, and (c) rewriting of their original output. Results indicate that although the children noticed and later incorporated mainly lexical features into their output, gains in the linguistic acceptability and comprehensibility of their revised texts showed an advantage for error correction over models. Learners in the error correction condition reported more noticing of grammar at the comparison stage, which later emerged in their revisions. The potential effects of both feedback strategies on children’s reported noticing and output production are discussed, and conclusions are drawn for the role of feedback in children’s classroom second language acquisition.

The literature exploring the executive function correlates of bilingualism is vast, but to date, few studies have concentrated on children, for whom the bilingual advantage appears even more inconsistent than for adults. We investigate a highly heterogeneous group of children (in terms of bilingual experience and socioeconomic status) and identify the critical threshold of bilingual experience from which an advantage can be observed at group level. The modeling methods adopted allow the use of fine-grained, continuous factors for age and socioeconomic status, thereby effectively controlling for their effect and isolating the specific effect of bilingual experience. We pioneer the use of Cox Proportional Hazard regression to analyze performance in the Simon task. This allows the modeling of all data points without transformation or outlier removal, and captures both accuracy and reaction time within the same analysis, while being able to handle multiple predictor variables.

This article examines the implications of connectionist models of cognition for second language theory. Connectionism offers a challenge to the symbolic models which dominate cognitive science. In connectionist models all knowledge is embodied in a network of simple processing units joined by connections which are strengthened or weakened in response to regularities in input patterns. These models avoid the brittleness of symbolic approaches, and they exhibit rule-like behavior without explicit rules. A connectionist framework is proposed within which hypotheses about second language acquisition can be tested. Inputs and outputs are patterns of activation on units representing both form and meaning. Learning consists of the unsupervised association of pattern elements with one another. A network is first trained on a set of first language patterns and then exposed to a set of second language patterns with the same meanings. Several simulations of constituent-order transfer within this framework are discussed.

The main purpose of the current study was to examine the relative significance of vocabulary knowledge (VK) and syntactic knowledge (SK) in second language (L2) listening ability, while accounting for the effect of several cognitive and affective factors. A total of 263 English-as-a-foreign-language learners took a standardized listening test (IELTS), as well as a battery of nine linguistic (two aural SK tests and two aural VK tests—covering both breadth and depth of VK), cognitive (two working memory tests and a metacognitive knowledge questionnaire), and affective measures (two L2 listening anxiety questionnaires). Structural equation modeling analysis revealed that both VK and SK were significant predictors of L2 listening ability; however, VK was a stronger predictor with an effect size being almost twice as much as the one for SK (.55 vs. .28). The results also showed that metacognitive knowledge, working memory, and L2 listening anxiety are significant predictors of L2 listening ability.

Voice onset time (VOT) was measured in the production of /t/ in the initial position of 60 English words spoken by native English (NE) speakers and native Spanish (NS) speakers who began learning English before or after the age of 21 years. The subjects rated the words for familiarity, age of acquisition, imageability, and relatedness to word(s) in the Spanish lexicon. The subjects in all three groups showed two well-known phonetic effects: They produced longer VOT values in the context of high than nonhigh vowels, and longer VOT in one- than in two-syllable words. As expected, the NS subjects who learned English prior to the age of 21 years judged the English words to be more familiar and more like a Spanish word than did the subjects who began learning English later in life. Also, many but not all of the NS subjects produced /t/ with shorter VOT values than did the NE subjects. However, regression analyses showed that none of the lexical factors mentioned above or the text frequency of the 60 English words examined affected the NS subjects' VOT values. Thus, variation in the accuracy with which NS subjects produce English /t/ must be accounted for by factors other than the lexical status of the words in which /t/ occurs.