卓群布農語內在動貌之句法研究

Abstract

本論文旨在探討卓群布農語內在動貌的句法表現。主要發現可分為四個層面：第一，受詞-事件照應(object-to-event mapping)在卓群布農語中屬於數詞的特性。此特性透過內在動貌(Asp)與帶有數詞之指示詞組(DP)間的呼應(Agree)來呈現。當此呼應關係存在，該事件解讀為終點事件(telic event)；當此呼應關係不存在，該事件則解讀為無終點事件(atelic event)。而動貌最大投射(AspP)的位置則介於輕動 詞最大投射(vP)與動詞最大投射(VP)之間。第二，在卓群布農語中，受詞-事件照 應與事件結構兩者雖有互動，但保有相互獨立的關係。第三，本文認為中綴-i(n)- 應為內在動貌標記，而非Zeitoun 等(1996)所認為的完成貌(perfective aspect)標記 或Jeng(1977, 1999, 2000)所提的過去式(past tense)標記。最後，內在動貌與事件 結構在卓群布農語裡和語意以及詞彙存有依賴(dependent)關係，並非如MacDonald(2006, 2008b)所認為的是一個自主(autonomous)的系統。

This thesis investigates Takituduh inner aspect under MacDonald’s (2006, 2008b) event feature approach from three perspectives. The first research question concerns the syntactic nature of inner aspect in Taikituduh. The second research question is whether event structure and the object-to-event (OTE) mapping are two independent properties of Takituduh inner aspect. The last attempt of this thesis is to determine the syntactic status of the inflectional infix -i(n)- in Takituduh. With regard to the first question, this thesis claims that the OTE mapping is a property of numerals instantiated via the (non)existence of an Agree relation between Asp and an DP that involves a numeral. Takituduh events are telic only when such Agree relation is approachable. Besides, the maximal projection AspP, is structurally between vP and VP but absent in Taktituduh statives. As for the second question, the answer appears positive in that, although the OTE mapping is observable in both Takituduh accomplishments and achievements with respect to tense entailments and the modification of event structure properties, only accomplishments are subject to the modification of both the beginning and the end. Moreover, for achievements, the end is invisible to the munga’-modification. Based on MacDonald’s (2008b) proposal that the beginning and the end of events are grammaticalized as the interpretable event features <ie> and <fe> on different heads, this thesis argues that, in Takituduh, these two distinct properties are distinguished in the way that <fe> can be modified and aspectually calculated only when it is within the extended aspectual domain of interpretation defined by AspP. The aspectual domain of interpretation is extended only if Asp is in an Agree relation with a DP that contains a numeral. Concerning the last question, it is observed that the OTE mapping disappears when Takituduh events are expressed by predicates in the -i(n)- form. To account for this fact, this thesis proposes that -i(n)- is the overt form of inner Asp. It differs from the defective zero aspect in that its aspectual domain of interpretation is extended by default. There are two follow-up claims to this major proposal. First, Takituduh intransitive activities, which are supposed to be without an inherent endpoint, hence involving no <fe>, become accomplishments in the -i(n)- form. This phenomenon is explained by the argument that all Takituduh eventives involve <fe> on V, thus inherently telic. The prima facie activities are due to the lack of an Agree relation between Asp and a DP that involves a numeral. In addition, when a stative verb/adjective root is combined with -i(n)-, it alternates between an achievement and an inchoative stative verb. Regarding this issue, this thesis proposes that this alternation is derived from Asp-V(-A) incorporation in L-syntax in the sense of Hale & Keyser (1993) and MacDonald (2008b). With the overt Asp -i(n)- bearing <ie> being added in L-syntax, if there is no <fe> present on empty V, the result of Asp-V incorporation is the inchoative stative alternate; otherwise, <fe> is present on empty V, which ultimately adjoins to Asp, giving rise to the event feature configuration of achievements. The argument that -i(n)- is inner Asp is compared with others in the literature. All in all, for four reasons, the infix -i(n)- is better considered as inner Asp instead of the perfective aspect marker (Zeitoun et al. 1996) or the past tense marker (Jeng 1977, 1999, 2000). Firstly, -i(n)- has the L-syntactic nature of encoding telicity (Travis in prep). Secondly, the perfect aspect marker -i(n), assumed to be outer Asp, has scope of interpretation over -i(n)-. In addition, -i(n)- can occur with the future tense prefix na-. Lastly, the Head Movement Constraint (Travis 1984) and the Mirror Principle (Baker 1985) reveals that the structural position of -i(n)- is lower than the causative morpheme p-, as v, because -i(n)- is closer to the V root. The structural lower position of -i(n)- strongly suggests that -i(n)- is not the perfective aspect marker, outer AsP, or the past tense marker, T, both of which are structurally higher than v. The analyses in this thesis have two theoretical implications. First, aspectual predicate types may be cross-linguistically universal in terms of MacDonald’s (2006, 2008b) event feature configurations, but they are parametric with respect to how they derived, in L-syntax or S(yntactic)-syntax. Moreover, MacDonald (2008b)’s claim that inner aspect is an autonomous system within natural language is questionable. In Takituduh at least, inner aspect is lexically dependent (e.g., <fe> only appears on V) and semantically contingent (e.g., <fe> is obligatorily on dynamic V roots, but optionally on stative roots).