Introduction

The AuxRange, the area between subjects and main verbs in VO languages, has traditionally attracted the most attention in research on adverbs, undoubtedly because it is the only place many common adverbs (and negation) may occur in the familiar European languages, because it is where ambiguities show up most clearly, and because it is only here that obvious interactions between adverbs and auxiliary verbs take place. Thus the early writers on adverbs in the generative syntactic tradition made it a major focus (e.g., Keyser 1968, Jackendoff 1972, Ernst 1984), and those who were interested in the behavior of auxiliaries and/or negation have also had to assume or propose some analysis of these elements (Baker 1971, 1981, Emonds 1976, Sag 1980). More recently, adverbs have become a common diagnostic (a) for head movement, the central theoretical issue in the AuxRange in current Principles and Parameters (P&P) theory (Platzack 1986, Pollock 1989, and many others), and (b) for various proposals for alternative subject positions (Bobaljik and Jonas 1996 and many others). Thus it has become increasingly important for there to be a coherent theory of adverb licensing.

The goal of this chapter is to establish that the principles developed in earlier chapters can account for a wide range of data in the AuxRange. In particular, this approach holds that the main syntactic constraints on free-adjunction are the Directionality Principles and, to a lesser extent, Weight theory.

The traditional analysis

The western tradition of describing case systems can be traced back to the Greeks. Ancient Greek, like the other ‘older’ Indo-European languages, was a fusional inflecting language in which case marking could not be separated from number marking, where there was also some fusion of the stem and inflection, and where gender correlated closely with declensional type in the first (feminine) and second (masculine) declension. In other words, Greek was structurally parallel with Latin as illustrated in section 1.1. Given this kind of structure, it is not surprising that the Greek descriptions of case were based on the word rather than on stems and suffixes. The modern concept of the morpheme (the smallest meaningful, or better, smallest grammatical segment) is not very useful in languages where the expression of several grammatical categories has fused representation across the whole language, i.e. what Matthews calls cumulative exponence or cumulation (Matthews 1974/1991).

Cases were described in terms of what are called here case forms, where a case form is an inflected form of a noun. Since the inflection of case and number is fused, it would be more accurate to talk of case/number forms. The term case is from Latin cāsus, which is in turn a translation of the Greek ptōsis ‘fall’. The term originally referred to verbs as well as nouns and the idea seems to have been of falling away from an assumed standard form, a notion also reflected in the term ‘declension’ used with reference to inflectional classes.

I can remember my first encounter with case quite clearly. It was in 1949. The language was Latin and the book was Latin for today. The first sentence was Discipulī, pictūram spectāte and it came with a translation ‘Pupils, look at the illustration.’ I cannot say that I quickly became enamoured of case. There was not much pleasure to be had in memorising paradigms, but eventually there were rewards: the rolling hexameters of Virgil, the cleverly contrived odes of Horace and the epigrammatic prose of Tacitus, all exploiting the genius of a highly inflected language, a language where grammatical functions were expressed in the most highly condensed fashion, a single short suffix on a noun expressing case, number and sometimes gender, a single suffix on a verb expressing tense, aspect, mood, voice and the person and number of the subject.

There were other minor encounters with the language of Beowulf and the language of Njal, but my next significant encounter with case came in 1966. In that year I took up a fellowship to study Australian Aboriginal languages and I was sent to western Queensland to record Kalkatungu, or Kalkadoon in the more familiar spelling, a language which at that time had no more than a dozen fluent speakers. Like most Australian languages Kalkatungu had a well-developed case system. For Kalkatungu there was no available grammar and therefore no paradigms to learn. The paradigms had to be built up by a mixture of elicitation and recording of discourse.

About seven years have passed since I wrote the first edition of Case. It is now time to review what I wrote then and to take account of recent publications in the field. This second edition incorporates a number of significant additions to the data and some revised interpretations of data. It also incorporates a number of improvements and expansions to the discussion of important concepts, taking into account current developments in the field. As in the first edition, I have paid particular attention to traditional and current notions and terminology, not just in case itself, but in the areas of word class, structure, agreement, roles and grammatical relations.

The most substantial revision has been to section 3.3, which deals with abstract case in the Chomskian paradigm. I have updated this section, not without some difficulty. The book as a whole is aimed at students and academics in general linguistics or in languages, but the Chomskian paradigm contains numerous concepts and terms peculiar to itself. Introducing too many of these notions in a short section can lead to obscurity, but introducing too few runs the risk of distortion. Moreover, different authors within the paradigm adopt different approaches and the model is forever changing. Interested readers can follow up the references given in note 4 to chapter 3.

Introduction

This chapter provides a global perspective of case systems and their marking. It is divided broadly into two parts. In section 5.2 the organisation of the core or nuclear relations is surveyed and in section 5.3 the organisation of the peripheral relations is described. As Nichols 1983 points out, it is difficult to maintain a strict distinction between core grammatical cases, which encode S, A and P, and semantic cases like locative, allative and instrumental. A single case may cover A and instrumental function or P and allative (destination) function. Nevertheless there are significant generalisations that can be made about cases that encode S, A and P, even if they sometimes cover peripheral grammatical relations as well.

Organisation of the core

As noted in section 3.5.3, noun phrases bearing core relations are often unmarked altogether, while noun phrases in peripheral relations are marked by inflectional cases, adpositions or both. In languages like this the burden of distinguishing subject (SA) from object (P) or the absolutive relation (SP) from the ergative (A) is borne by cross-referencing bound pronouns or word order. On the other hand there are languages with inflectional case only for the core relations, or for these plus possessor function (genitive), with adpositions for other grammatical relations.

In the majority of languages the core grammar is organised on an SA/P basis (accusative system). In a minority of languages the core is organised on an SP/A (ergative) basis.