Empty Categories in Tuvaluan: NP-trace, PRO, pro, or variable?

1. Introduction.¹

Besnier's (2000) Tuvaluan is the third book on a Polynesian language in the Routledge "Descriptive Grammars" series, following Maori (Bauer 1993) and Rapanui (Du Feu 1996). The series editor, Bernard Comrie, states that the aim of this series is to provide comparable descriptions of various languages that can serve as a common ground for theoretical and descriptive linguists and make a truly fruitful interaction between them possible. To achieve this goal, the series adopts the Questionnaire (originally published as Lingua 47) as a standard format. Hence, the organization of this book is based on this format.

As a whole, this book is not for general readers who are interested in Tuvaluan. Those who are interested in pedagogical grammar are referred to Besnier (1981). In fact, the reader may need some basic knowledge of Polynesian languages in general to fully appreciate the information this book provides. The author covers a wide range of phenomena and presents a thorough and detailed description for each topic, including discourse-related information. Thus, those who know what to look for, particularly those who are familiar with Polynesian languages, will find this book quite useful. Unfortunately, the book's organization does not make it easy to find where to look for the kind of data one wants. One topic is often discussed in several different sections. Furthermore, the index is organized according to Tuvaluan lexical items, and not grammatical categories. This makes it hard to get an all-inclusive picture of a certain phenomenon, although the author provides detailed cross-reference. In a nutshell, this book is not particularly user-friendly, although its contents, namely the data, are stimulating.

Given the objective of the Descriptive Grammars series, it would be fair to give a theoretical linguist's review of this work. Besnier's data, which are classified and discussed in a descriptive framework, present a number of interesting problems to a theorist who looks at them from an analytical point of view. They definitely stimulate the interest of a theory-oriented reader, offering a number of topics for future research. In this article, I focus on one of such issues, namely the treatment of empty categories. Six phenomena that involve a syntactic gap are discussed in the book: zero-pronominalization, relativization, ko-clefting, raising, equi-deletion, and topicalization. Being a descriptive work, the book does not use the term "empty category," nor does it pay particular attention to the nature of the gap in various constructions. However, as we will see shortly, classification of empty categories plays a significant role in analyzing the relevant data. In the following discussion, I "translate" the terms used in the book into the terms of Transformational Grammar. However, it should be emphasized that the gist of each problem remains the same no matter which theory is used to analyze the data. Thus, I believe that the reader should find no difficulty in projecting each problem into his/her own theoretical framework.

2. Organization of the Book.

The book consists of five chapters, an introduction, references, and an index. The introduction covers the geographical and social aspects of Tuvalu as well as a general description of the language. Tuvaluan belongs to the Samoic-Outlier subgroup of the Polynesian family and is spoken by approximately 9,000 speakers in Tuvalu, formerly known as the Ellice Islands. The basic word order is VSO, although various other orders are also permitted. It is a prepositional language. Zero-pronominalization is possible and preferred for both subjects and objects. Case-marking is ergative. Verbal agreement is limited to some intransitive verbs. Wh-words remain in situ although many of them appear in clefted positions. Passive and antipassive do not exist.

Chapter 1 (1-263) discusses various syntactic phenomena in Tuvaluan. As the series title suggests, the discussion is basically descriptive and does not include explanatory considerations. Thus, it is inevitable for a reader with a formal linguistic background to have some questions about the data. I will discuss...