CC BY
22
MPI EVA, Leipzig — JGU, Mainz — ILI RAN, St. Petersburg
RESOLVING ALIGNMENT CONFLICTS (plenary lecture)
The question how to constrain possible alignment patterns have long concerned typologists, who noted, in particular, that case shows greater predisposition for ergative alignment as compared to agreement, as well as suggested a number of generalizations concerning preferences of certain syntactic constructions (control properties, reflexives, coordination, etc) for certain alignment patterns (see, e.g., [Croft 2001]). Yet such generalizations have been plagued by availability of counterexamples, which challenge their validity [Bickel 2010]. In my view, one way to account for both recurrent patterns and counterexamples is to acknowledge that alignment properties may be due to different factors.
Thus, while certain constructions may have certain alignment preferences for functional reasons these preferences may conflict with the basic alignment pattern of a language. For example, imperatives are known to show a general preference for accusative (or active) alignment to the dispreference of ergative alignment [Dixon 1994]. In this context it is instructive to consider the behavior of imperatives in ergative languages. As it turns out, the imperatives in these languages either retain the ergative pattern, at least partially (as, e.g., in some Daghestanian languages, which retain the ergative-style verb agreement in imperatives), or switch to accusative alignment (as in Dyirbal and some other languages discussed in [Dixon 1994]), or else imperative form can be blocked (as in Ket, where only the conjugation classes with ac-cusative-style agreement have regular imperative forms). Such alignment changes can be taken as evidence for alignment preferences of certain constructions. Similar alignment conflicts are also observed in the ditransitive domain, where it is conventional to distinguish between indirective (T=P 4 R), neutral (T=P=R), and secundative (T 4 P=R) alignment ([Haspelmath 2005]; cf. [Comrie 1982; Dryer 1986; Siewierska 2004]). Also in the ditransitive domain, certain constructions
may for functional reasons show either indirective preference (e.g., incorporation, antipassives), or secundative preference (e.g., formation of verbal reciprocals). Therefore, for example, formation of antipassives is more consistent in languages with indirective alignment (where antipassives, if available, always target T/P arguments) than in languages with secundative alignment. In the latter languages an antipassive may either target T (in accordance with the functional preference, as, e.g., in Northern Paiute), or R (in accordance with the alignment profile, as, e.g., in Eastern Ojibwa), or else be blocked altogether (as, e.g., in Tzotzil). In my paper I will discuss further examples of alignment conflicts both in monotransitive and ditransitive domains. Generally, it will be shown that when a construction-specific functional bias is consistent with the language profile (i.e. the basic alignment pattern) more cross-linguistic consistency is found, while there is room for more variation when the two factors are in conflict.
References
Bickel 2010 — B. Bickel. Grammatical relations typology // J. J. Song (ed.). The Oxford Handbook of Language Typology. Oxford: OUP, 2010. P. 399-444.
Comrie 1982 — B. Comrie. Grammatical relations in Huichol // P. J. Hopper, S. A. Thompson (eds.). Studies in Transitivity [Syntax and Semantics 15]. New York: Academic Press, 1982. P. 95-115.
Croft 2001 — W. A. Croft. Radical Construction Grammar. Oxford: OUP, 2001. Dixon 1994 — R. M. W. Dixon. Ergativity. Cambridge: CUP, 1994.
Dryer 1986 — M. S. Dryer. Primary objects, secondary objects, and antidative // Language 62, 4, 1986. P. 808-845.
Haspelmath 2005 — M. Haspelmath. Argument marking in ditransitive alignment types // Linguistic Discovery 3, 1, 2005. P. 1-21.
Siewierska 2004 — A. Siewierska. Person. Cambridge: CUP, 2004.
