This paper examines the intersection of syntax and semantics in second language acquisition (SLA), a perspective on language learning that has received relatively little attention in the past, in ordertoanswer the following question: Why do English-as-a-second-language (ESL) learners tend to use the same morpheme (e.g., HAVE) for Possessives (P) and Existentials (E), when English has separate forms (HAVE, THERE BE) for these two functions? Working within a functionalist framework, data are analyzed from the longitudinal case study of a Cambodian adult who, despite formal ESL instruction and residence in an English-speaking community, persistently uses the form has for both P and E. Although first language transfer can be invoked as a partial explanation forthis, it alone cannot account for the systematic conflation of P and E in interlanguage. Rather, it is argued that the shared semanticproperties of P/E, together with syntactic, pragmatic, and perceptual characteristics of native language, interlanguage, and second language constructions, make this overlap imminently transferable, especially in untutored or low-level instructed SLA.

Among the structural and lexical factors claimed to account for avoidance in second language learning are (a) cross-linguistic difference, (b) cross-linguistic similarity, and (c) intrinsic complexity of the second language features avoided. This paper examines patterns of avoidance and preference for phrasal verbs or equivalent single-word verbs among Swedish learners of English. We assumed that if the subjects avoided English phrasal verbs, particularly the figurative ones, even though phrasal verbs exist in Swedish, this would indicate that inherent semantic difficulty of second language forms was the main factor contributing to the avoidance behavior. If, on the other hand, the learners did not show any preference for one-word verb forms in English, or indeed favored the phrasal forms, this would support the assumption that avoidance or nonavoidance depends largely on differences or similarities between the native and the foreign language.

A multiple-choice test and a translation test were given to two groups of advanced Swedish-speaking learners of English. Each test consisted of 20 sentences, allowing for the choice of either a phrasal or a synonymous single-word verb. The test answers showed that the Swedish learners avoided neither phrasal verbs in general nor the figurative ones in particular, regardless of whether the verbs were similar to, or different from, their Swedish translation equivalents. Furthermore, the results were compared to the avoidance patterns of a group of advanced Hebrew-speaking learners of English. From the comparison it emerged that the Swedish learners usedsignificantly more phrasal verbs than the Israelis, notably figurative ones. These results suggest that the avoidance is determined more by a systemic incongruence between the first language (L1) and the second language (L2) than by the inherent difficulty of L2 forms.

The study investigates the role of causal reasoning in second language (L2) narrative comprehension. Subjects read and recalled texts that were either high or low in causal coherence. This read-and-recall procedure was repeated. Half of the subjects reported their thoughts during reading. The major finding was that first language (L1) readers recalled the high coherence texts better than the low coherence texts. Advanced L2 readers showed a similar advantage for thehigh coherence texts, but only in their second reading. Intermediate readers' recall was lower and showed no advantage for highly coherent texts. The verbal report data generally supported these patterns. These results indicate that L2 readers have difficulty detecting the higher order causal structure in texts. This structure is a central component of L1 readers' mental representation of the information in a text. As L2 readers' skills improve, their ability to detect causal relations increases, although even advanced L2 readers may need a second reading to do so.

This study investigated the generalizability of Gardner's socioeducational model for second language learning to a highly different social context from that in which the model was developed. Specifically, the study examined Jewish high school students in Israel learning Arabic as a foreign language in a setting characterized by a high degree of intergroup conflict. Four tenets were identified as forming the essential core of Gardner's theory. These we would expect to apply across widely differing contexts in order to support the generalizability of the model. Other elements in the model, as well as the degree of relationship between variables, may be specific to the social context. The main hypothesis posited an expanded version of Gardner's model that included additional constructs—both general and context specific.

The subjects for this study were 484 pupils studying Arabic in the 10th grade in regular high school programs. The instruments consisted of written questionnaires that were administered during two regular class periods. Teachers' marks were collected at the end of the school year. The hypothesis was tested using linear structural relations (or LISREL) causal modeling techniques. Regarding each of the four core elements in Gardner's theory, we concluded that the model was indeed generalizable to the context of this study. Other relationships between variables in the model are interpreted in light of the specific language learning context.