Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” - Volume 21, Issue 3, 1997

In Balinese, as in many Austronesian languages of Indonesia, the Philippines and Taiwan, a transitive verb may appear either in a morphologically unmarked construction in which the Patient is identified with the sole argument of a one-place predicate and is therefore arguably the subject or with a nasal-prefixed verb in a construction where the Agent is the subject. This raises the question of whether the language is ergative, accusative or neither. We argue that it could be considered ergative on the grounds that the Patient is identified with the sole argument of a one-place predicate in the unmarked transitive construction, and that certainly the Patient-subject construction is basic.

Although encoding of evidential distinctions has been attested in most language areas of the world, relatively little has been written about evidentiality in sub-Saharan African languages. This paper describes the situation in Lega, a Bantu language spoken in eastern Zaire. The study considers Lega data in light of several models of evidentiality and suggests that their may be two organizational principles operating in different languages: one based on source of information, the other on mode of knowing.

Various claims from the previous literature about the way in which evaluative morphology (particularly diminutives and augmentatives) operates are tested on a large sample of languages. Evaluative morphology is seen as being less morphologically marginal than has been implied in some of the recent literature, but nevertheless as showing some interesting cross-linguistic tendencies.

In Early Greek, ana and ek(s) were still largely independent adverbs. In time, following the normal trend to univerbation ('Today's syntax is tomorrow's morphology'), they became prepositions and -what matters here- preverbs capable of modifying the meaning of a verbal root.As a rule, what took place is univerbation into an unanalyzable verb (fusion). There are, however, some interesting exceptions. The preverb eks- assumed a new shape kse- which preserved morphological transparency, and the conglomerate ksana-(< eks- + ana-) regained independence (fission) as a free adverb: Yesterday's morphology is today's syntax!My paper tries to explain these developments (with parallels in other languages) and to investigate the interplay between form and meaning (constructional diagram-maticity). It is argued that conceptual distance to the semantics of the verb is a determining factor in the behaviour of Greek preverbs. De-univerbation proves to be a very effective means in opposing opacity.

A distinction is commonly made between morphological or surface ergativity, and syntactic or deep ergativity, based on what Dixon has termed the "pivot" behavior (S/A vs. S/O) of a language. Since marked constructions enable an S/A pivot to function even in some deep ergative languages, deep or syntactic ergativity might be interpreted as gradational, depending on the degree to which ergative morphology interferes with the grammar of clause integration and referent coherence. For spoken Kurmanji, a northwest Iranian language with surface ergativity, tentative restrictions on zero-anaphora in conjoined clauses are identified which relate to ergative agreement patterns. These are compared to the distribution of zero-anaphora in other complex constructions involving clause combining. Surface ergativity is found to be one of a variety of factors which may promote re-location of the subject referent in a language in which non-finite structures play a peripheral role, and multi-clause constructions are under pressure to replicate the structure of single proposition clauses.