|Full Title:||Patterns in Domain-restricted Properties|
|Start Date:||18-Sep-2013 - 21-Sep-2013|
|Meeting Email:||click here to access email|
Cross-linguistic comparison shows that the properties that have been used to characterise languages are (mostly) not ‘holistic’. That is to say, they either do not or only rarely apply in all (relevant) domains of a given language, but rather seem to be instantiated to a greater or lesser extent. Take, for instance, the classical morphological types proposed by 19th and early 20th century philologists and anthropologist-linguists (principally, the Schlegels, Humboldt, Schleicher and Sapir): we no longer expect to identify languages which are isolating, fusional/inflectional, agglutinating or polysynthetic through and through.
A question that has not received systematic attention, however, is whether systems showing mixed properties exhibit distributions that are systematically sensitive to specific features (e.g. nominal vs. verbal, finite vs. non-finite, inflectional vs. derivational, specific classes of verbs or nouns etc.). In other words, can we identify patterns in and/or restrictions on the partial (i.e. domain-specific) presence of a property? Is the distribution of mixed properties predictable? Is it productive to think in terms of the hierarchically organised ‘scales’ that have been identified for phenomena like relative-clause formation, argument encoding and diatheses (cf. i.a. Silverstein 1976, Keenan & Comrie 1977, Dixon 1979) in other domains too? Might it be possible to think in terms of parametric settings reflecting varying degrees of ‘myopia’, with the same phenomenon being permitted crosslinguistically in domains defined by more or less specific features?
As an example, consider null arguments. Several distributional splits can be observed here. One is whether nominals can be null regardless of their grammatical function. Thus topical subjects and direct and indirect objects can all remain unpronounced in many Asian languages, whereas in typical null subject languages like Italian, only topical subjects can be null (cf. so-called ‘topic drop’ or ‘radical pro-drop’ vs. ‘subject drop’, Huang 1984; Neeleman and Szendröi 2008). In the subject domain, we observe a second split with respect to referentiality: in ‘full’ null subject languages of the Romance type, both referential and non-referential pronominal subjects can be left unpronounced, whereas in Icelandic, only non- or quasi-referential subjects can be null and in German, it is exclusively non-referential subjects which are null (cf. the so-called ‘semi null subject languages’, Platzack 1985, Rizzi 1986). Further splits distinguish between the omissibility of first- and second- vs third-person subjects (cf. so-called ‘partial null subject languages’; Holmberg, Nayudu & Sheehan 2009, and Holmberg 2010), and whether null-subject phenomena extend beyond the clausal domain or not (cf. Rodrigues 2004). Apparently, then, null-argument phenomena seem to be more or less sensitive to properties like information structural and referential prominence, with a purely formal consideration (distribution across phrases of different types) also potentially coming into the picture.
|Linguistic Subfield:||Historical Linguistics; Morphology; Syntax; Typology|
This is a session of the following meeting:
46th Annual Meeting of the Societas Linguistica Europaea
|Calls and Conferences main page|