Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” - Volume 17, Issue 2, 1993

In this article, a discussion is offered of the Ancient Greek particle dé in terms of boundary-marking. "Boundary" is treated as the manifestation in discourse of the interaction between topicalization and discourse structure. The marking of boundaries, therefore, subsurnes such functions as topic marking and intersentantial (interparagraph) connection. Dé is described as a boundary-marker and is shown to have a function in various types and at various levels of discourse. While in the oldest Greek (Homeric epic) dé is used to mark the segmentation that results from the online "continuative" production of spoken discourse, in later, written discourse the particle is used for a variety of functions: from local, intrasentential subject topic switch ("switch-reference") to the setting of 'frames' in discourse, and from the marking of boundaries that are "content-oriented" to rhetorically highly marked segmentation.

This article addresses the question of whether tense markers in Modern Tamil were historically aspectual in function, as Zvelebil (1962) has claimed. The methodological approach employed is that developed by Hopper (1979a, 1979b, 1982) for the analysis of foreground and background in narrative discourse. Narrative texts representing three historical periods — Old Tamil (5th c), Middle Tamil (12th c), and Modern Tamil (20th c.) — are analyzed for correlations between foreground-background marking and the distribution of alleged 'tense' forms. On the basis of the grounding analysis, it is discovered that the forms function aspectually in Old Tamil, analogous to the functioning of aspectual auxiliary verbs in the modern language. The overall diachronic picture which emerges is one in which surface forms and categories undergo change, while underlying functional contrasts remain remarkably stable over time.

The term 'inverse' has traditionally referred to voice systems characterized by alternations of verbal voice marking, alternations that depend on the relative ontologicai salience of the two core arguments of a transitive animate verb, the logical subject and logical object. In typical inverse languages, speech-act participant (SAP) arguments (1,2 person) ontologically outrank non-SAP arguments (3d person), a fact that is grammatically encoded by 1:3 and 2:3 predications assigning one verbal voice ('direct') while 3:1 and 3:2 predications assign the other voice ('inverse'). 3:3 predications are potentially ambiguous, a problem addressed in some inverse systems by 3d person arguments with relatively low ontologicai salience being assigned a special case, the obviative (4th person).The present work addresses the question whether inverseness may be evinced through formal means other than alternations in verbal voice marking. It is argued that this occurs in a Tanoan (Kiowa-Tanoan) language, Arizona Tewa (AT). In AT transitive animate predications, an opposition in paradigms of person-marking verbal prefixes occurs such that one pronominal paradigm is assigned in case of a direct predication (logical subject ontologically outranks logical object), while the other paradigm is assigned in case of an inverse predication (logical object ontologically outranks logical subject). In effect, then, AT has separate direct and inverse pronominal paradigms; these encode the voice alternations, rather than oppositions of verbal voice marking per se. It is argued that an inverse analysis is both appropriate for AT and, in addition, applicable to at least some other Tanoan languages, such as Picurís and Southern Tiwa.

This paper introduces the concept of functional type, a particular combination of functions which are realized in a single clause, and applies this concept to data on voice from Javanese and English, comparing the English passive and two Javanese constructions with passive-like structural and functional properties. It is shown that what is conceived of as general passive function can be divided into a number of different functional types, each one associated with particular discourse properties of the agent and patient and a particular combination of structural representations in Javanese and English.

Data from language acquisition, psycholinguistics, and diachronic studies have all shown that the lexicon has a clear internal structure, which includes relationships among lexical items based on phonetic and phonological characteristics, semantic features, morphology, and frequency of use. In the absence, however, of direct evidence from grammar, such lexical structure has even recently been deemed irrelevant to linguistic theory. In this paper, I use evidence from German grammar, specifically gender assignment, to support a model of lexical structure like that proposed particularly within Natural Morphology. German gender assignment has been shown to be largely predictable on the basis of phonological shape (e.g. final and initial segments or clusters), semantic features, and morphological features — all factors considered to be part of the lexicon's internal structure by Bybee and others. In this way gender assignment reflects lexical structure. Moreover, frequently used vocabulary tends to violate such rules, as Bybee's view of lexical structure would predict. By so doing, German grammar exploits almost exactly the structure of the lexicon which has been proposed based on data from areas other than grammar in its narrow sense.

Why do we say ON Thursday but AT 10 o'clock? Or why do we say AT night but IN the morning? One common answer to such questions is to dismiss the problem: this is the way we speak because this is the way to speak; it is all arbitrary, conventional, idiosyncratic.It is argued that such answers are unilluminating and unsatisfactory. Prepositions such as ON, AT, or IN have their meanings, and the choice between them is motivated by these meanings. There are also certain conventions of use based on cultural expectations; the meanings and the cultural expectations interact and their interaction produces results whose "logic" may be difficult to detect — especially if one looks in the wrong direction, that is, that of "truth conditions" regarding external situations. In fact, however, the problem is not insoluble, and if it is approached with the understanding that meaning is all in the mind and that it is a matter of conceptualizations rather than "truth conditions", the hidden "logic " behind the choice of prepositions for temporal adverbials can be explained.The paper argues, and tries to demonstrate, that the prepositions AT, IN, and ON mean different things, and that the patterns of their use in different types of temporal phrases are determined by their meanings.