El Estado y los mecanismos de integración social
B. Efecto redistributivo del gasto social (2002)
In their sketch of Squliq Atayal grammar, Ogawa & Asai (1935) pointed out the existence of ga (later known as topic marker), identifying it as a marker linking subject and predicate, like ga and wa in Japanese.25 Egerod (1966, 1969, 1980) viewed ga as
25
The difference between the nominative case ga and the topic marker wa in Japanese was not known at that time since this identification was first made by Kuno (1972). According to Iori (2005[2001]), it was rarely known that traditional Japanese grammarian Daizaburou Matsushita had proposed similar distinction in the early 20th century.
the particle of precentral exposure. Information-structural notions had not been touched upon until Rau (1992) identified ga as a topic marker and claimed that subject conveyed old information while predicate conveyed new information. Despite her empiricism-based proposal and morphosyntax-orientation, Rau (1992) indeed opened the window to the study on the interaction between syntax and information structure in Squliq Atayal. Egerod (1993) further compared ga with qu’, showing that the former indicated topic/comment sentence type while the latter subject/predicate sentence type.
L. Huang (1994) was a pioneering work of adopting statistical evidence even though the statistical part only played a minor role in the whole paper. She counted the frequency of actor voice (AV) and non-actor voice (NAV) in transitive clauses of one story, finding out that the ratio was 30:52. Thus, it was claimed that NAV clauses were actually the canonical transitive clauses while AV clauses with two arguments should be regarded as antipassive clauses. This paper only provided the basic descriptive frequency without further proving whether the ratio was statistically significant or not.
Adopting the notion of transitivity in Hopper & Thompson (1980), Rau (1994) analyzed the story line in a folktale and claimed that “non-referential or indefinite O’s show correlation with the verb morphology, case marking, and word order characteristics of ‘intransitive’ clauses in Atayal (p.509)”. That is to say, the clauses with a non-referential or indefinite patient might be intransitive to some extent. This idea was further explored in Y. Chen (2007). Under the same hypothesis of transitivity, Y. Chen (2007) studied the correlation between transitivity and foregrounding/backgrounding with the assistance of statistics. One of her findings was that word order might affect transitivity: the word order verb-patient-actor had an indefinite patient while the word order verb-actor-patient had a definite patient. Despite the close relationship between definiteness and information types, these two studies did not really go into the discussion on information structure.
understanding of topicality in Squliq Atayal by adopting inferential statistics, which had the power to infer the overall fact from limited samples. Rau (2000) made use of VARBRUL, the linguistic software designed for binary logistic regression, to study the factors behind the variation of preverbal and postverbal subjects.26 She discovered that postverbal subjects were associated with old information and topic continuity while preverbal subjects were with new information and topic discontinuity: “[R]ecent second mentions, resumptives, simple NPs, bound personal pronouns, zeroes, and interrogative pronouns favor the VS order whereas text-structural boundary, first mentions, modified NPs, free pronouns, and demonstrative pronouns favor the SV order (Rau 2000:227).”27 Nevertheless, the scope of her study was confined with VARBRUL, which only allowed a dependent variable of two categorical values—that is, preverbal subject or postverbal subject. It did not suffice for the demonstration of the overall information-structural properties of different grammatical functions. In addition, the so-called “preverbal subject” was a misnomer because preverbal subjects in her study were grammaticalized topics (GTOPs) in reality, the properties of which will be presented later in the present thesis.
S. Huang & Tanangkingsing (2011) was a study more faithfully following Givón’s method, and it aimed to make a comparison of the transitivity phenomenon in three Formosan languages, Kavalan, Squliq Atayal and Tsou. In their study on Squliq Atayal, the objects of AV clauses and the subjects of NAV clauses were nearly indistinguishable in terms of their definiteness and the perfectivity of these clauses. The essential difference between them was their topic persistence: the former had lower topic persistence while the latter had higher topic persistence. The assumption behind their
26 Logistic regression is a statistical method to test whether a categorical dependent variable is influenced by the (either categorical or continuous) independent variables the researcher assumes. Binary logistic regression only permits a dependent variable of two categorical values (Field 2009, Gray & Kinnear 2012, Pallant 2007, Yang 2004).
27“Resumptives” in Rau’s study indicated subjects reintroduced into discourse after a lapse (Rau 2000:219).
study was that the AV transitive clauses were extended intransitive clauses in nature, wherein there were oblique-marked NPs (i.e. the AV objects in the present thesis), whereas the NAV clauses were canonical transitive clauses.28 Put differently, they compared the patients of these two types of clauses to examine which one had higher transitivity--that is to say, if it did, it was a real transitive clause. In addition to a small methodological flaw that this paper adopted a loose one-tailed probability, its focus was confined on the topicality of patient, without presenting a general picture of how information was manifested in a sentence.
In the lately published introductory grammar of Squliq Atayal, L. Huang & Hayung (2016), it was briefly discussed that topic sentences aimed to prepose known information to the sentence-initial position usually for the sake of contrast and comparison.
The abovementioned studies all broadened our knowledge of the information-structural notions in Squliq Atayal. Nevertheless, both their foci and methods were quite different from each other so that they either individually or as a whole did not delineate a general picture of how information structure was constituted and how it really interacted with syntax.