Linguistic universals and syntactic change

LINGUISTIC UNIVERSALS AND SYNTACTIC CHANGE Brian Joseph Ohio State University A crucial problem facing historical linguists is how to account for the direction of certain changes. The direction of change taken by two constructions, Object Raising and Object Deletion, between Medieval and Modern Greek, is documented here, and is shown to be explainable by reference to substantive universal constraints holding on the form these constructions can take in natural language, as evidenced by data from a variety of languages. We conclude that universale guided the direction of these Greek syntactic changes, and in general can rule out certain conceivable changes as impossible. Universals thus lead to a more restrictive, and hence stronger, theory of syntactic change.* Amongthe primary concerns oftheoretical linguists are the problems ofcharacterizing what constitutes a 'possible' and conversely an 'impossible' human language. Successful answers to these problems allow for the formulation of a more restrictive theory of language than was previously available. Most current research into linguistic universals and universal grammar is directed to these ends. By studying the ways in which all languages are alike and the ways in which they differ, one can better understand the notion of 'possible human language', and therefore better determine those elements or combinations of elements which cannot occur in a human language. The specification 'combinations of elements' is necessary, for it may be the case that a language may have either feature X or feature Y, but not both. Historical linguists face a similar problem. They study the movement from one 'possible human language', i.e. an attested or reconstructed stage of a language, * The contents of this paper are drawn largely from my doctoral dissertation (Joseph 1978b). I profited greatly from the guidance and advice of my advisers—Judith Aissen, David Perlmutter , and Jochem Schindler. The research and work herein were supported by the Arthur Lehman Scholarship Fund, an NDFL (NDEA Title VI) grant for Modern Greek (both administered through Harvard University), and a post-doctoral fellowship offered by the Izaak Walton Killam Memorial Scholarship Committee of the University of Alberta. Pre-Modern Greek forms are given in standard transliteration, while Modern Greek forms are given in a rough morphophonemic transcription—however, d is the voiced interdental fricative, and g is the voiced velar fricative. Citations from pre-Medieval texts are identified by traditional abbreviations, and can be found in standard editions of the works cited. Medieval texts are identified by the following abbreviations : Belis. II Belisarios II (Wagner 1874 : 348-78) Boustr. Chronicle of Boustronios (Sathas 1873, Vol. 2) Doukas Historia Byzantina of M. Ducas (Migne 1866, Vol. 157) Erotop. Hesseling & Pernot 1913 Hermon. Legrand 1890 Lyb.Lybistros and Rhodamne (Wagner 1881 :242-349) Makh.Dawkins 1932 Monemb. Chronicle of Monembasias (Baletas 1947:41-7) Quadrup. Fable of the quadrupeds (Wagner 1874:141-78) Rim. Alex. Holton 1974 Spanos Mass of the beardless man (Legrand 1881 :28-47) 345 346 LANGUAGE, VOLUME 56, NUMBER 2 (1980) to anotherone, i.e. alater stage ofthatlanguage. However, thehistoricallinguistmust try to determine why a language changed in the way it did, rather than in some other way. Given the range of conceivable changes which a language might undergo, which are possible changes and which are not? As with synchronic theory, successful answers to this question can lead to the formulation of a more restrictive theory of language change. These two pursuits complement each other. Since languages change from one possible form to another, the constraints imposed by linguistic theory on possible synchronic grammars provide the upper bounds within which changes can occur.1 Conversely, showing a conceivable historical change to be in fact impossible advances the understanding of synchronic grammars, for a conceivable synchronic state which could have arisen is thus ruled out. In this paper, an attempt is made to define further the notion of 'possible diachronic change' with respect to syntactic change. Substantive universal constraints which hold in synchronic grammars are used to explain the direction taken by certain changes in syntax between Medieval and Modern Greek. It is claimed that Greek could not move from one possible language state to another conceivable one because the latter is ruled out by linguistic theory as an 'impossible human language'. The...