Abstract
The typological status of Formosan languages, Philippine languages, and other socalled “Philippine type languages” has been one of the great mysteries in the study of the syntax of the world’s languages. They have often been assumed to be typologically unique in possessing the so-called “focus system”, a type of voice system that is characterized by the use of various verbal affixes to indicate the thematic role of the NP bearing the nominative (or absolutive) case in a sentence.1 They have also been assumed by some linguists to be typologically so unusual that notions such as transitivity, actancy structure, etc., which are commonly used in syntactic description, are either irrelevant or inapplicable to the study of these languages. This dissertation provides a small-scale preliminary comparison of the syntactic typology of Formosan and Philippine languages. It aims to provide a clear statement of the typological status of Formosan and Philippine languages. In order to achieve this goal, I examine the clause structure in Formosan and Philippine languages from a broad typological perspective. Specifically, I compare the morphosyntactic features exhibited in Formosan and Philippine languages with those found in languages that have been unambiguously identified as having an ergative actancy structure. Based on the results of the cross-linguistic comparison, I offer a clear statement of the typological status of Formosan and Philippine languages. This dissertation consists of eight chapters. Chapter 1 is the introduction, Chapter 2 provides the theoretical orientation, Chapter 3 describes and evaluates previous analyses, while Chapters 4–7 cover the discussion of transitivity and ergativity in individual Formosan and Philippine languages. Chapter 8 is the conclusion. Following the introduction in Chapter 1, Chapter 2 gives a brief orientation to the theoretical framework employed in this study. All data analyses are conducted within a revised version of Dixon’s Basic Linguistic Theory. In order to discuss the typological status of Formosan languages and Philippine languages, some notions that are crucial to the discussion of clause structure are introduced in this chapter. Syntactic notions covered in this chapter include: (i) core arguments vs. peripheral arguments (or adjuncts), (ii) valency vs. transitivity, (iii) canonical transitive, passive, and antipassive, (iv) actancy structure, and (v) the various uses of the term “ergative,” such as “syntactically ergative,” “discourse ergative,” and “lexically ergative”. In discussing “transitivity,” I provide a critical review of various transitivity tests that have been previously discussed in the syntactic literature. Following the discussion of syntactic notions that are relevant to the study of transitivity and actancy structure, Chapter 3 provides a critical review of previous analyses of so-called “Philippine type languages”. In order to provide a basis for the review, I first discuss verbal clause patterns in these languages, including a comprehensive coverage of the reflexes of PAN *-um-, *-en, *-an, *Si-, (and PMP *maR- 3 and PMP *maN-), which characterize verb forms in Formosan and Philippine languages. I then proceed with a brief introduction and critical review of seven different types of analyses that have been provided in the literature. These include: (i) the “passive” analyses, (ii) the “focus” analyses, (iii) the “ergative” analyses, (iv) the “active” analysis, (v) the “fluid voice” analysis, (vi) the “hybrid” analysis, and (vii) the “symmetrical voice” analysis. The following chapters provide case studies of two Formosan languages (Kavalan and Atayal) and two Philippine languages (Central Cagayan Agta and Dibabawon Manobo). In each case study, I re-examine the clause structure in the target language in terms of morphosyntactic and semantic criteria, and address at least the following two questions. First, what constitutes the canonical transitive construction in each language? Second, what kind of actancy structure does each language have (accusative, ergative, or split ergative)? In order to facilitate the discussion of transitivity and actancy structure in these languages, I provide a discussion of some of the basic morphosyntactic facts for each language, such as word order, construction markers (including ligatures, prenominal case-marking elements, demonstratives, etc.), and their pronominal system. Two additional questions are addressed in the case study of the languages. First, what are the constraints that condition the relative order of pronouns? Second, are the pre-nominal monosyllabic forms determiners or nouns? Based on convergent evidence from morphosyntax and semantics, I conclude that only one, rather than both, of the dyadic clause patterns constitutes the transitive 4 construction in each of the languages under investigated. Specifically, dyadic -an clauses are transitive construction in Kavalan. Dyadic –un clauses, dyadic –an clauses, and dyadic s- clauses are transitive constructions in Atayal.2 Dyadic -an clauses, dyadic -ãn clauses, dyadic i- clauses, and dyadic i- -ãn clauses are transitive constructions in Central Cagayan Agta.3 Dyadic -on clauses, dyadic -an clauses, and dyadic i- clauses are transitive constructions in Dibabawon Manobo. As for the other dyadic clause pattern (i.e., dyadic -um-/(-)m- clauses in Kavalan; dyadic (-)m- clauses in Atayal; dyadic ma-, mag-, maN- clauses in Central Cagayan Agta; and dyadic -um-/-og-, or maN- clauses in Dibabawon Manobo), formerly analyzed as transitive constructions in previous analyses, these are treated as extended intransitives or pseudo-transitives, a type of intransitive clauses. Based on the observations that (i) the S of an intransitive clause and the O of a transitive clause have the same morphological marking, whereas (ii) the A of a transitive clause has a distinct morphological marking in each individual language, I conclude that each individual language has an ergative case-marking system. I note that even though neither Atayal nor Dibabawon Manobo has a (productive) verbal agreement system, it is still possible to conclude, based on the case-marking system alone, that both of these languages also exhibit a pure ergative actancy structure. As for Kavalan and Central Cagayan Agta, verbs in these two languages can carry an optional genitive clitic pronoun or pronominal-related form that agrees with the A of a transitive clause in person and number,4 but NOT with the S of an intransitive verb NOR with the O of a transitive verb. This suggests that both Kavalan and Central Cagayan Agta have an ergative agreement system. Because Kavalan and Central Cagayan Agta exhibit ergativity in both their nominal case-marking system as well as in their verbal agreement system, I conclude that both Kavalan and Central Cagayan Agta have a pure ergative actancy system. In addition to the discussion of transitivity and actancy structure, I also provide a detailed discussion of constraints that determine the relative order of pronouns in Squliq Atayal and Dibabawon Manobo in Chapters 5 and 7, respectively. The order of pronouns in Squliq Atayal has been previously described as solely determined by the PERSON hierarchy (either 1 > 2 > 3 or 2 > 1 > 3). However, as shown in Chapter 5, section 5.2.2, PERSON alone CANNOT adequately account for the pronominal order facts in Squliq Atayal. Instead, it should be used in conjunction with other constraints. I propose a set of three constraints that are required to adequately account for the order of pronominals in Squliq Atayal. As for the order of pronominals in Dibabawon Manobo, I provide a set of rules that appears to better account for the facts than those appearing in previous descriptions of the language. The categorical status of pre-nominal monosyllabic forms, which are commonly analyzed as “determiners” in previous analyses, is a focus of discussion for each of the languages in the case studies. I show that some of the pre-nominal monosyllabic forms are “AUXILIARY NOUNS” that carry the feature [+extension], i.e., nouns that require a dependent predicate or a dependent complement clause. Specifically, in Squliq Atayal, at least two of the pre-nominal forms, qu/ and squ/, might be auxiliary nouns, at least in some cases. In Central Cagayan Agta, all the pre-nominal monosyllabic forms are auxiliary nouns, while in Dibabawon Manobo, at least two of the pre-nominal forms, to and kan, are auxiliary nouns. As for Kavalan, in the textual data that I examined, there is no single instance of the pre-nominal monosyllabic forms that is followed by an element that is unlikely to be a noun. It is still uncertain whether these forms might also be auxiliary nouns in Kavalan. Following the four case studies in Chapters 4–7, Chapter 8 provides a summary of the discussion in Chapters 1–7 and suggests directions for future research.