Philip, J; (2010) Harmony, Head Proximity, and the Near Parallels between Nominal and Clausal Linkers. In: (pp. p. 50). Linguistics Association of Great Britain: UK.

Preview

PDF 2010_Harmony.pdf Available under License : See the attached licence file. Download (143kB)

Abstract

This paper puts forward a notion of harmonic word order that leads to a new generalisation over the presence or absence of disharmony: specific functional heads must cross-linguistically obey this notion of harmony absolutely, while for other categories the presence of harmony is simply a tendency. The difference between the two classes is defined by semantics. This approach allows us both to draw certain parallels between restrictions on word order in nominals and in clauses, and furthermore to explain why other expected parallels should fail to be realised completely, specifically as regards differences in the distribution of relative clauses in the NP and complement clauses in the sentence. Syntactically independent relative clause markers and subordinating complementisers share a striking restriction as regards ordering: relative clause markers are always initial in postnominal relative clauses, and final in prenominal relative clauses (Andrews 1975; Downing 1978; Lehmann 1984; Keenan 1985; De Vries 2002, 2005); similarly, initial subordinating Cs only appear in postverbal complement clauses, while final subordinating Cs are only possible where the complement clause is preverbal (Bayer 1996, 1997, 1999; Kayne 2000). In this paper, I provide new evidence from eighty genetically and geographically diverse languages of a third category sharing precisely the same restriction: linkers in the complex NP. These are syntactically independent, semantically vacuous heads, serving to mark the presence of a relationship between a noun and any kind of phrasal dependent (Rubin 2002; Den Dikken and Singhapreecha 2004; Philip 2009). The class of linkers in the NP therefore includes the ezafe in Indo-Iranian, the associative marker -a in Bantu, as well as purely functional adpositions such as of in English. Like relative clause markers and subordinating Cs, the linker always intervenes linearly between the superordinate head (the noun) and the subordinate dependent. Crucially, relative clause markers, subordinating Cs, and linkers in the NP form a natural class: they are syntactically independent, semantically vacuous words serving purely to mark the presence of a relationship between head and dependent. Any member of this class is a ‘linker’. I propose a theory of disharmony whereby linearisation rules targeting heads with specified semantics can require such heads to appear in a prominent position, either initial or final, irrespective of the general headedness of the language. Linkers, being semantically vacuous, are of course impervious to such rules; they will therefore always conform to the harmonic, or optimal, word order. I propose a theory of harmony whereby the optimal word order is determined by the interaction of three independently motivated harmonic word order constraints: Head Proximity (adapted from Rijkhoff 1984, 1986, cf. Head-Final Filter, Williams 1982), the preference for uniformity in headedness (initial or final), and the preference for clausal dependents to appear in final position (Dryer 1980, 1992). Where the three constraints compete, it is always Head Proximity that takes precedence. I show that the distribution of all three types of linker is fully captured by this proposal. Moreover, this theory of ordering also accounts for another well observed near parallel between clauses and nominals, as well as its exceptions. This concerns a left-right asymmetry in the distribution of clausal dependents: while in OV languages complement clauses appear with near equal frequency in both preverbal and postverbal position, in VO languages they are found uniquely in postverbal position (Dryer 1980; Hawkins 1994; Dryer 2009); similarly, in OV languages relative clauses are distributed relatively evenly between prenominal and postnominal position, whereas in VO languages they are almost always postnominal, with very few exceptions (Mallinson & Blake 1981; Hawkins 1983, 1990; Lehmann 1984; Keenan 1985; Dryer 1992, 2007, 2008; De Vries 2005). The theory predicts these exceptions to be permitted only in languages that are rigidly N-final. Hawkins’ (1983) Noun Modifier Hierarchy suggests that this prediction is borne out; apparent exceptions (cf. Dryer 2008) are found underlyingly to be N-final.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as XMLJSONCSV

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item