Like all the Celtic languages, Welsh very commonly displays word orders other than the canonical VSO, the unmarked surface word order. Traditional Welsh grammar distinguishes between two constructions which are non-VSO: the so-called ‘mixed’ and ‘abnormal’ word orders. Both display what, in a pretheoretical sense, we might call ‘fronting’ of some constituent, although the pragmatic function of this fronting is typically claimed to be very different in each case (see, for example, Fife and King 1991; Watkins 1991; and other papers in Fife and Poppe 1991). In this chapter I consider the derivation of each of these constructions from the point of view of the principles-and-parameters framework of generative grammar.

I will show that the ‘mixed’ construction, which I refer to as the cleft construction in the text, is best analysed in terms of the CP-recursion analysis of Rizzi and Roberts (1989) and Cardinaletti and Roberts (1991). We will see that the clefted constituent is sited in the specifier position of a CP which is itself the complement to a higher complementizer. On the other hand the ‘abnormal’ sentences (a term I keep for the sake of convenience) actually involve adjunction to a matrix CP, and no CP-recursion is involved.

The chapter is structured as follows: section 1 outlines the data and illustrates the major differences between the two construction types. Section 2 discusses some previous generative treatments of one of the types of fronting under discussion, the cleft, and shows that these analyses share a common problem when more data are taken into consideration. Section 3 presents an alternative analysis of both the cleft and the abnormal constructions, and section 4 defends this analysis in detail.

Introduction

This study addresses broadly the issue of how syntax and morphology interact. I approach this topic by examining some of the syntactic consequences of irregular forms that appear in the third-person present tense of the copula in many of the Celtic languages. I suggest in section 2 that the Welsh and Breton copulas are fundamentally different and reproduce a distinction familiar from traditional descriptions of Irish: the Welsh copula is a substantive (i.e. lexical) copula while its Breton counterpart is a functional (or ‘grammatical’) copula, a spellout of tense and agreement features. I then argue in section 3 that, although both languages exhibit irregular forms in the copula's third-person present tense, they behave quite differently syntactically, and that this fact correlates with the distinction between substantive and functional copulas. Only the (Welsh) substantive copula exhibits suppletion that has syntactic effects; the (Breton) functional copula varies morphophonemically in terms of its spell-out and is syntactically inert. I make use of ideas of Pesetsky (1994) to model variation in the morphosyntactic restrictions of the Welsh substantive copula. Both the substantive and functional copula are, I claim, similar in terms of their selectional properties. Their difference, like other kinds of ‘syntactic’ variation between languages, is a formal, morphosyntactic one. In essence my claim is that the variation in the copulas reduces to a formal one between lexical formatives and functional formatives.

Introduction

Recent work in the principles-and-parameters approach to syntactic theory has been concerned with the range of word order variation in the world's languages. It is a tenet of this approach that such variation can be derived from a highly constrained set of simple parameters, interacting with universal principles of natural language. We would like to investigate here how the facts of Irish word order may best be represented within the framework argued for in Chomsky and Lasnik (1993) and Chomsky (1993), and discussed in Marantz (1995). The first section will provide a brief overview of the relevant notions and mechanisms of this framework and the analysis we propose. After this, we will turn to the relevant Irish data, moving on to a discussion of the predictions of our analysis and potential problems it raises. In the final section, we will provide a more detailed refinement of our initial analysis, discussing its import for a feature-driven theory of syntactic variation as in Chomsky (1993).

The framework and an initial analysis

Over the last half-decade, much work in syntax has been devoted to motivating and supporting the claim that all arguments of a verb, and in particular the subject, are base-generated within the maximal projection (VP) of that verb (i.e. the VP-internal subject hypothesis of Fukui and Speas (1986), Kitagawa (1986), Koopman and Sportiche (1991) among many others). This approach entails that in a language such as English, the subject must raise to somewhere within the maximal projection of an Inflectional category to receive abstract nominative Case.

Introduction

In this chapter we present evidence that both Semitic (Arabic and Hebrew) and Welsh clitic systems bear striking similarities to each other, and are significantly different from those of Romance or Germanic. We motivate an analysis of both systems which treats them as base-generated syntactic affixes in Agr. Hence these clitics are not in fact pronouns, that is, XPs.

Our central theoretical claim is that this type of clitic system is non-trivially connected to the (full or residual) VSO nature of the languages in question. Adopting and adapting a proposal for English auxiliaries in Chomsky (1993), we propose that weak/clitic pronouns must check features with an Agr head with strong nominal features. However, we argue that the nature of VSO systems is such that Agr heads with strong features are largely absent. It follows that weak/clitic pronouns cannot be licensed in a VSO system. The functional role of such pronouns – which we will argue to involve licensing pro – is then carried by the Agr heads themselves. We thus tie together two apparently unrelated properties of these languages, namely word order and the nature of the clitic system. We also explain the pervasiveness of agreement marking that these languages show; where a Romance or Germanic language has a pronoun, these languages have agreement, hence it is not a surprise to find agreeing prepositions, for example. Moreover, the apparent preference for enclisis that these languages show is a consequence, in our terms, of the fact that the apparent clitics are really affixes; enclisis thus follows from the Right-Hand Head Rule (Williams 1981b).

Introduction

The goal of this chapter is to defend an extended version of the Visibility Criterion (Chomsky 1986a) whereby the link between Theta Theory and Case Theory is generalized. We will argue that the Case-theoretic requirement in the formulation of Visibility should be replaced with a requirement of co-indexation with a functional head; the relevant functional heads being Agreement (Agr) and Tense (T). We will argue that Tense involves two functional heads which are composed into a single chain via the mechanism of selection. This allows the incorporation of aspectual information into the licensing condition. By aspectual we mean the indirect relationship between the utterance time and the time of the event denoted by the verb.

Scottish Gaelic (SG) is particularly relevant to both the extended version of Visibility and to the composition of tense information into a single chain because of two factors: firstly, co-indexation with Agr in SG correlates with word order; and secondly, composition of tense information is transparently reflected in the morphological form of tense and aspect particles. Given these two factors, the extended version of Visibility predicts that certain word orders in SG will be forced or ruled out. This phenomenon is particularly clear with respect to the syntax of measure phrases in the language, which we show to be sensitive to aspectual information.

Generalized Visibility

Typically, argument NPs have to be licensed in two ways: they must be licensed by Theta Theory and they must be licensed by Case.

Preliminary remarks

This book grew out of a conference on Comparative Celtic Syntax held at the University of Wales, Bangor, on 25–7 June 1992. Earlier versions of seven of the ten chapters collected here were given at that conference. The idea behind the conference was to bring together researchers working on the syntax of the Celtic languages from a ‘principles-and-parameters’ perspective (the assumptions behind this perspective are outlined below in section 2.1), and, in particular, to provide a forum where comparative work on Celtic syntax could be presented. The comparative work was intended to be both internal and external to the Celtic family. Hence, one goal of the conference was to encourage those working on Celtic to make comparisons with non-Celtic languages, and to bring relevant phenomena and analyses of Celtic languages to the attention of those working on non-Celtic languages. Although the precise contents differ from the conference, and this volume should not be taken as a conference proceedings, we have compiled this collection with the same general goals in mind.

This introduction is intended to provide the background to the chapters that follow, both for those who are unfamiliar with the principles-and-parameters framework and for those who are unfamiliar with the Celtic languages. In this section, we briefly sketch the historical, geographical and social situation of the languages. Section 2 provides background to the principles-and-parameters framework. This section is of most relevance for readers who may be familiar with the languages but who are less familiar with this framework.

INTRODUCTION

Having looked at the notion of context in detail in the preceding chapter, we can now return to the question of speech acts and locutionary agency. We shall begin with reference – the relation that holds between linguistic expressions and what they stand for in the world (or the universe of discourse) on particular occasions of utterance. We shall then take up a particular kind of reference, deixis, which depends crucially upon the time and place of utterance and upon the speaker's (more precisely, the locutionary agent's) and the addressee's roles in the utterance-act itself.

We shall then consider the grammatical categories of tense and aspect, neither of which is universal, but both of which, together or separately, are found in many unrelated languages throughout the world. As we shall see, tense, unlike aspect, is a referential (and more specifically deictic) category.

Another grammatical category that is closely connected with tense (and in some languages is found independently of tense) is mood. As the term ‘mood’ would suggest, there is a historical association between the grammatical category of mood, as this is traditionally defined, and what is referred to as modality in modern logic and formal semantics. There are, however, important differences between the way in which modality (and mood) are handled, typically, in present-day formal semantics and the way in which mood and modality have been described in traditional grammar.

INTRODUCTION

As we saw in the last chapter, words cannot be defined independently of other words that are (semantically) related to them and delimit their sense. Looked at from a semantic point of view, the lexical structure of a language – the structure of its vocabulary – can be regarded as a network of sense-relations: it is like a web in which each strand is one such relation and each knot in the web is a different lexeme.

The key-terms here are ‘structure’ and ‘relation’, each of which, in the present context, presupposes and defines the other. It is the word ‘structure’ (via the corresponding adjective ‘structural’) that has provided the label – ‘structuralism’ – which distinguishes modern from pre-modern linguistics. There have been, and are, many schools of structural linguistics; and some of them, until recently, have not been very much concerned with semantics. Nowadays, however, structural semantics (and more especially structural lexical semantics) is as well established everywhere as structural phonology and structural morphology long have been. But what is structural semantics? That is the question we take up in the following section.

We shall then move on to discuss two approaches to the task of describing the semantic structure of the vocabularies of languages in a precise and systematic way: componential analysis and the use of meaning-postulates. Reference will also be made, though briefly, to the theory of semantic fields (or lexical fields).

A Festschrift for John Lyons is long overdue. He is, without question, one of the most outstanding and internationally famed linguists of the present time, with a distinguished career as Lecturer in the Universities of London and Cambridge, as Professor in the Universities of Edinburgh and of Sussex and now as Master of Trinity Hall Cambridge. He has lectured widely and has published (and is still publishing) many important books and articles. He has deservedly received many honours and awards, including five honorary doctorates and, most significantly, a knighthood ‘for services to the study of linguistics’. As yet, however, no volume has been published in his honour. That omission is now to be rectified.

There are, I believe, two requirements of a good Festschrift. The first is that it should have a clearly recognisable theme, and one, moreover, that is associated with the person in whose honour it is published. The second is that the contributions should be made by scholars who have been closely connected as colleagues, students or friends. I have, I hope, as editor, succeeded in satisfying both of these requirements. The choice of the theme was easy enough. John Lyons has worked mainly in the field of semantics, and is probably best known for his magnificent two-volume Semantics (1977), which is, I believe, the most comprehensive scholarly work on a single topic in linguistics that has ever been published.

It is uncontroversial that an account of semantics for natural language must provide an explanation of the information intrinsic to natural-language expressions. The assumption that such a semantics for natural language involves reconstructing the information natural-language strings convey about the non-linguistic objects they are used to describe has held sway now for some twenty years since Lewis (1972) poured scorn on the representational view of meaning then current - that of Katz (1972). However, this approach to natural-language content is far from unproblematic; and in this chapter I argue for a return to a representational approach. I advocate a proof-theoretic account of interpretation in which natural-language expressions are seen as providing the encoded input to a process of interpretation which builds structure via a process of deduction. The information an individual expression conveys is information about how to build structured configurations which constitute the interpretation of the string in which the expression is contained. The structures that result from this process are linked labelled databases set within a logic framework defining inference over complex databases. In section 1,1 set out problems for the classical form of the truth-theoretic paradigm originally advocated by Montague (1974), arguing that we need both a concept of content which underdetermines truth-theoretic content for individual expressions and a process of interpretation for strings which involves structure building. In section 2, I set out a model of utterance interpretation meeting these requirements.

Introduction

In comparison with English, German appears at first sight to have a very simple tense-aspect system, certainly if we restrict the terms tense and aspect to grammatical categories of the languages, whether expressed by synthetic or periphrastic means. German has two simple forms, the Present (as in ich liebe ‘I love’) and a simple past form, which I will henceforth, following German usage, call the Präteritum (as in ich liebte ‘I loved’). In addition, there is a compound past form, as in ich habe geliebt ‘I have loved’, which I will henceforth, following German usage, refer to as the Perfekt. Other periphrastic forms include the Pluperfect (as in ich hatte geliebt ‘I had loved’), the Future (as in ich werde lieben ‘I shall love’) and the Future Perfect (in German conventionally called the Futur II, as in ich werde geliebt haben ‘I will have loved’), although, as in many other languages, the question arises whether the Future and Future Perfect are more properly to be classified as tenses or moods. Some varieties of German also have Double Perfect and Double Pluperfect periphrastic forms, as in ich habe geliebt gehabt and ich hatte geliebt gehabt; English has no corresponding forms. The focus of interest in this article will be the distinction between the Perfekt and the Prateritum, the two major verb forms used with past time reference.

In some varieties of German, the system is even simpler, through loss of the Präteritum, leaving the Perfekt as the only form with past time reference as part of its basic meaning.