Due to the recognition of the centrality of the lexicon for SLA theory (see the 1987 thematic issue of SSLA, edited by Susan Gass), the last few years have witnessed an increased interest in understanding lexical knowledge. As Gass (1999) reminded us, learning vocabulary in a second language is a complex task that involves much more than learning sound-meaning pairings; it also involves learning how lexical information is morphologically expressed and syntactically constrained. The present issue provides a natural sequel to the 1999 SSLA thematic issue, “Incidental L2 Vocabulary Acquisition,” by addressing some of the questions raised in that volume, in particular the questions related to the intimate relationship between syntax and semantics at the lexical interface. This issue is devoted to the L2 acquisition of verb meaning and argument structure crosslinguistically, and it explores in detail the nature of linguistic systems that L2 learners acquire in this particular domain. The six central articles offer a coherent approach to the topic, using linguistic theory to help us understand the characteristics of learner grammars. Until recently, linguistic approaches to SLA have placed a strong emphasis on understanding the acquisition of functional categories, for example, and the acquisition of the lexicon has received less attention. Understanding how the lexico-syntactic interface is mentally represented, and how it evolves during the second language acquisition process, is crucial for developing an adequate theory of L2 knowledge in general, as well as for informing theories of the lexicon.

In English, manner-of-motion verbs (walk, run) and directed motion verbs (go) can appear with a prepositional phrase that expresses a goal (goal PP) as in John walked (ran, went) to school. In contrast, Japanese allows only directed motion verbs to occur with a goal PP. Thus, English allows a wider range of motion verbs to occur with goal PPs than Japanese does. Learnability considerations, then, lead me to hypothesize that Japanese learners will learn manner-of-motion verbs with goal PPs in English from positive evidence, whereas English learners will have difficulty learning that manner-of-motion verbs with goal PPs are impossible in Japanese because nothing in the input will tell them so. Forty-two intermediate Japanese learners of English and 21 advanced English learners of Japanese were tested using a grammaticality judgment task with pictures. Results support this prediction and provide a new piece of evidence for the previous findings indicating that L1 influence persists when an argument structure in the L2 constitutes a subset of its counterpart in the L1.

This article presents two experiments that examine the effects of the native language (L1) on the second language (L2) acquisition of argument structure. The linguistic focus is on agentive verbs of directed motion (march, walk) and change-of-state verbs (break, melt) in Spanish and English. Agentive verbs of directed motion undergo a transitivity alternation in English when there is a prepositional phrase (The captain marched the soldiers to the tents) but not in Spanish (*El capitán marchó a los soldados hasta el campamento). Two experiments are reported that examine whether Spanish and Turkish learners of English at the intermediate level undergeneralize the transitivity alternation with manner-of-motion verbs, and whether English learners of Spanish overgeneralize the alternation. In both experiments subjects performed a picture judgment task and a grammaticality judgment task. Results confirmed that the L1 constrains the acquisition of argument structure: There were overgeneralization errors with manner-of-motion verbs in the Spanish study and there were undergeneralization errors with these verbs in the English study. Learnability implications are discussed.

The English locative alternation relates sentences of the type John loaded hay onto the wagon to those of the type John loaded the wagon with hay. Some locative verbs occur in both of these patterns, others in only one or the other. It is known that there are differences among languages with respect to which verbs are possible. The present research focuses on the constructional meaning of the locative alternation and on the constraints governing verbs that can participate in the alternation. One characteristic of the “ground-object” locative is that the object tends to be viewed as completely affected. This is known as the holism effect. Additionally, English has certain narrow constraints on the verbs that can occur in the two constructions. This study investigates whether native speakers of Korean learning English develop knowledge of the holism effect in the English locative and knowledge of the narrow constraints. English native speakers and Korean learners of English participated in a forced-choice picture-description task. Native speakers of Korean also judged an equivalent test instrument in Korean. The primary results are these: When given a ground-object structure, both learners and English native speakers preferentially chose a ground-holism picture. We interpret this as a reflection of the holism effect: Learners, like native speakers, have knowledge of this aspect of the constructional meaning of the locative. English native speakers also show their knowledge of the narrow conflation classes by rejecting ground-object structures containing verbs that are not permitted in this structure, even if the picture would be appropriate. Korean learners show no effect for narrow verb class. We interpret this as showing that the learners have not achieved native-speaker knowledge of the narrow classes. Korean uses a different basis for verb classification.

This paper reports on an experimental study that investigates the acquisition of Japanese unaccusative verbs by English-speaking learners. Following Levin and Rappaport Hovav (1995), it is assumed that unaccusativity is syntactically represented but semantically determined. The experiment is devised specifically to examine whether L2 learners are sensitive to syntactic and semantic properties associated with unaccusative verbs in Japanese, which contrast with the properties of unergative verbs. In particular, the experiment involved picture tasks with two structures: the takusan construction as a syntactic test and the -teiru construction as a semantic test. Overall results of the experiment show that L2 learners generally know the properties investigated; that is, that subjects of unaccusative verbs originate in object position, and semantic notions such as telicity and change of state are aspects of meaning relevant to the classification of unaccusativity in Japanese. Based on these results, it is argued that the mapping of verb arguments to syntactic positions is not random, but rule governed, for most of the L2 learners in the present study.

This study investigates the acquisition of the unaccusative-unergative distinction in L2 Japanese by English learners. The aim is to establish whether learners of Japanese are sensitive to the lexical-semantic characteristics of verbs in similar ways as learners of Romance languages who were found to follow the Split Intransitivity Hierarchy (Sorace, 1993a, 1995a). Two groups of learners participated in the study, one consisting of learners who had not had any previous exposure to Japanese outside the classroom, and the other consisting of learners at the end of a 9-month period of continuous stay in Japan. A control group of native Japanese speakers also took part in the experiment. Subjects were tested on their knowledge of the different behavior of unaccusative and unergative verbs with respect to quantifier floating (Miyagawa, 1989); the native group was also tested on Case drop (Kageyama, 1993). The results show that both the native and the nonnative speakers are conditioned by the Split Intransitivity Hierarchy in their judgments on unergative verbs; however, their judgments on unaccusative verbs do not pattern according to the predictions. It is argued that this difference stems from the ambiguity of the Japanese input on unaccusative verbs, which are characterized by syntactic optionality.

The distinction of two types of intransitive verbs—unergatives (with underlying subjects) and unaccusatives (with underlying objects)—may not exist at early stages of L2 acquisition, both being syntactically represented as unergatives. This idea, referred to here as the Unaccusative Trap Hypothesis, provides an elegant developmental account for a variety of seemingly unrelated syntactic phenomena in L2 English, Japanese, and Chinese. Target language input, structural constraints on natural language linking rules, and linguistic properties of a learner's L1s shape stages in the reorganization of the lexical and syntactic components of interlanguage grammars. Although nonnative grammars may initially override the structural constraints postulated as the Unaccusative Hypothesis (Burzio, 1986; Perlmutter, 1978) and the Uniformity of Theta Assignment Hypothesis (Baker, 1988), at later developmental stages some may still achieve conformity with the norms of natural languages.

In this article I discuss some key points that are raised in the papers in this special issue. The first issue is that a variety of theoretical tools are necessary for a complete understanding of the issues raised in these papers. Second, although the methodology that is used in studies of second language knowledge has improved, it is clear we still lack an agreed-on set of protocols that will permit reliable comparisons across studies. In spite of these challenges, we are making progress in using complementary theories of the role of lexical representation, predication, and crosslinguistic variation to get a full picture of this complex area of lexico-morphosyntactic knowledge.