Introduction

In this chapter we look in more detail at the phonological properties of clitics. For many linguists the most important feature of clitics is the fact that they can't exist as autonomous words because they lack inherent stress, and therefore need a host (with stress) to ‘lean on’ phonologically. While this is an important aspect of clitichood, it can't be a defining property, otherwise we would never be able to define clitics in languages that lack a stress system. In Section 4.2 we therefore generalize the idea of ‘lacking inherent stress’ to ‘lacking inherent prosodic prominence’. This means that we can define clitics in languages that have a tone system or an accent system but no stress system, for instance, Japanese.

In Section 4.3 we see how the placement of a clitic cluster is often a complex interaction of syntax, prosody and morphology, but in at least one familiar system, that of Pashto, it seems that it is just prosody that determines the placement. Another question arises with regard to the internal structure of the clitic cluster. In Bulgarian the stress of the host interacts in rather complex ways with the organization of the cluster.

Here we should anticipate arguments we will be advancing later in the book about a number of well-known clitic systems. It is frequently claimed that the Romance languages, Greek, Albanian and Macedonian have clitic systems (especially pronominal clitics). In these languages the pronominal elements attach to a verb form (either the lexical verb or an auxiliary verb). In Section 4.4 we will show that in some cases the pronominals appear to violate the normal stress pattern of the language, while in other cases they appear to be incorporated into the stress domain of the verb. Where the pronominal appears to be incorporated into the lexical stress domain, it thereby takes on some of the characteristics of an affix, whereas if it remains outside the normal domain of lexical stress, it is easier to treat it as different from an affix, i.e., as a clitic.

Do clitics exist?

To conclude, we summarize what we see as the crucial issues surrounding the notion of clitic. We will conclude that the notion is often a useful one in description but that it's difficult to justify setting up any universal category of clitic or clitichood.

First, we have to conclude that traditional typology is not of much help in identifying a universal category of clitic. The most promising way of establishing a typology of clitics in the classical sense would be to set up an implicational scale of some sort. When we set up such a scale, we find some property that implies the existence of some other property. For instance, in morphology if a language distinguishes a trial number then it will also have a dual and a plural distinction and so we can set up a scale along the lines trial → dual → plural. Cases such as Polish show that there is little chance of finding a set of implicational scales for clitics and affixes. The Polish floating inflections show typically clitic properties of wide scope and (completely) promiscuous attachment. On the other hand, unlike special clitics in most languages, they don't really show any special syntactic behaviour. They can appear in almost any position in the clause, much like the (non-clitic) unstressed pronominals in that language or the Russian conditional mood marker by, which makes them look more like function words than clitics. However, they trigger idiosyncratic allomorphy on their hosts in the manner of a typical, highly morphologized affix. Indeed, they are more affix-like in this regard than are the regular English verb suffixes -s, -ed, -ing. It’s therefore difficult to imagine any sense in which we could define a set of implications along the lines ‘if an element displays clitic property Pc then it will not display affix property Pa’, or conversely ‘if an element displays affix property Pa then it will not display clitic property Pc’.

Introducing clitics

Consider the sentence Pat's a linguist. If we transcribe the sequence Pat's phonetically we can see that it forms a single syllable, /pæts/, which expresses both the subject of the sentence ‘Pat’ and the verb. A more transparent way of representing the structure would be to pronounce the sentence as Pat is a linguist. But in the sequence Pat's the verb is has been shortened to just a single consonant and has been attached to the noun Pat in the manner of a suffix. In one sense, then, /pæts/ is composed of two words, even though it's pronounced as a single word. Put differently, the 's is phonologically just the final part of /pæts/ but in terms of the sentence structure it functions as the main verb of the sentence.

This type of behaviour makes 's a typical instance of a clitic. In the present case the clitic is a form of a word which is phonologically attached to another word, its host. The 's corresponds to another form of the same word which doesn't show the same reliance on a phonological host. In Pat is a linguist we can pause between Pat and is or even insert another word: Pat, apparently, is a linguist. Similarly, if we want to confirm that Pat really is a linguist we can put the main emphasis or accent on the word is and say Pat (really) IS a linguist. None of these things is possible with the clitic form 's. The full form is therefore behaves like a genuine word, while the clitic 's behaves more like a suffix. A clitic which attaches to the right edge of its host, like a suffix, is called an enclitic. A clitic which attaches to the left edge of its host is called a proclitic. (There are also endoclitics which attach inside their host, in the manner of an infix, though these are much rarer.)

This book is an exploration of a very intriguing collection of linguistic beasts. Clitics is a fascinating subject. To study them adequately you really need to be concerned with all aspects of linguistics, from detailed phonetics to the analysis of discourse and conversation. Much of the interest they provoke is precisely because they sit at the interface between sound structure, word structure and sentence structure (not to mention the lexicon and language use). Partly for this reason, pinning down the notion of clitic is a little like trying to catch minnows with your bare hands. This has made it more difficult than usual to organize and arrange our material. In many cases it's hard to understand the point of, say, a phonological analysis of a set of clitics until you've seen the syntactic analysis of them, but without the phonological analysis it's difficult to understand the syntax. For this reason we've written some of the book cyclically, introducing some concepts at a basic level and returning to them later in the book to explore them in more detail. We've tried to reduce unnecessary overlap as far as possible but the reader will sometimes see similar examples cropping up in different places. In addition we've tried to perform a balancing act over choice of language examples. On the one hand it's important to see as much of the richness of clitics in the world's languages (and for us it's interesting to encounter or revisit many of those languages), but on the other hand it's much easier to understand some of the complexities of an unfamiliar language if you see it in several different contexts. For this reason we’ve tended to concentrate on languages that have been discussed a good deal in the literature, or with which we’re more or less familiar, or both.

Introduction

In this chapter we bring together a number of points raised in earlier chapters and ask to what extent we can draw a clear dividing line between clitics, words and affixes. We will show that the search for clear dividing lines is an arduous one. On the one hand we'll see a variety of ways in which elements that we would normally wish to call affixes may exhibit properties typical of clitics. We'll conclude the chapter with instances in which the divide between clitic and (inflecting) lexeme is blurred. Between those two types of situation we'll examine instances in which one and the same set of elements in a language can behave sometimes like affixes and sometimes like clitics (what we call ‘mixed systems’). In effect, then, we will start by examining elements that are fairly clear instances of clitics and progressively move towards elements that increasingly show properties of words.

However, we set the scene with three case studies of elements that might be considered clitics (or even weak function words) but which are certainly (German infinitive marker, Finnish possessor inflections), or almost certainly (Greek verbal ‘clitics’), affixes. The purpose of these case studies is to alert us to the fact that things aren't always as they seem, and that traditional perspectives, or analyses that initially look very promising, can be misleading.

The next set of case studies considers the extent to which affixes can show clitic-like properties while still remaining affixes. We begin by looking at the clitic-like phonology of English ‘Class II’ affixes. Inflectional affixes are supposed to conform to strict ordering constraints, but in a number of languages the syntactic past history of the clitics has left vestiges in the form of ordering variation. Sometimes the order is significant, as when identical pronominal clitics in Bantu languages occupy one position or the other within the affix string depending on whether they express the subject or the object argument.

Introduction

In this chapter we look at morphological aspects of clitics. There are two aspects to this. First, all investigators are agreed that in many of those constructions that are conventionally labelled as clitic systems the clitics themselves behave like affixes, morphological elements, rather than words. We therefore need to survey these morphological modes of behaviour. We'll start by summarizing a celebrated and still important set of criteria for distinguishing clitics from affixes (what in future we will simply refer to as the ‘Zwicky–Pullum criteria’). Applied to the English negation ‘clitic’ n't, these criteria yield a perhaps surprising conclusion. We then return to a phenomenon briefly introduced in Chapter 3, the clitic cluster. We will find that within the cluster the clitics behave much more like affixes than anything else, no matter how they might behave as a cluster. To set the scene for that demonstration we briefly survey the behaviour of uncontroversial affixes in the verb forms of Classical Nahuatl. We then turn to a more controversial issue. When we speak of clitics as phrasal affixes, we are thinking of a clitic as a morphological object (an affix) that happens to be attached to whatever happens to come at the edge of a phrase, rather than a true affix which seeks out a particular type of word within the phrase. Now, given current approaches to morphology, there are two ways of thinking of affixation. We can treat it in the manner of classical structuralist linguistics and regard it as essentially the addition of a morpheme to another string of morphemes.

Introduction

Scholars from diverse linguistic traditions have observed that the behaviour of clitics is neither that of an independent word nor that of an affix, but enjoys what Klavans (1982) calls ‘dual citizenship’. Among the classical studies on clitic phenomena, two oft-cited and influential papers stand out, namely Wackernagel (1892) and Zwicky (1977). In Section 3.2 we offer a summary of these studies, which represent two early attempts at identifying and classifying the idiosyncrasies of clitic placement. Section 3.3 offers a descriptive overview of a number of well-documented (though not necessarily uncontroversial) placement patterns. Having surveyed the meaning and function of clitics in the previous chapter, we look in somewhat more detail at the distributional properties of clitics and the various positions within which clitics occur cross-linguistically. We first briefly introduce the notion of clitic cluster and then explore the way that clitics or clusters of clitics are positioned within their domain (the clause, the noun phrase, occasionally the verb phrase). Placement patterns vary significantly: while some clitics may show a strong preference for the second position in the clause, others are banned from the edge of the sentence, and yet others must be adjacent to a specific word category. We will also show that even a concept such as ‘second position’ can mean different things in different languages. Clitics very often express functional properties akin to (or even identical to) inflectional categories in a language, which in practice means inflectional properties of nouns and of verbs. A clitic's linear positioning usually associates it in some way with the noun or verb whose properties it expresses, and so the noun and verb serve to define domains for clitic placement. In Section 3.4 we examine these domains. On the other hand, a good many clitics express properties of entire clauses or, in the case of discourse clitics, of utterances. Section 3.5 is devoted to such clause-domain clitics.

Introduction

This chapter will discuss various ways in which clitics interact with syntactic structure, in as theory-neutral a way as possible. This will involve examining the syntactic positions of clitics, the syntactic factors determining them and the syntactic role of clitics in phenomena such as agreement, doubling and climbing. We will see that behaviour generally associated with clitics, especially lack of ‘doubling’ of overt NP arguments and clitic climbing, can be found in otherwise canonical agreement systems. This sometimes has important consequences for theoretical models that attempt to deal with clitic systems.

Distributional idiosyncrasies

We will begin by surveying some of the peculiarities of function words that are not necessarily analysed in the literature as clitics, and which in any case are not special clitics. In Section 6.2.1 we look at properties that align ordinary function words with clitics, and in Section 6.2.2 we look at properties that align function words with affixes.

Clitic-like behaviour

In many cases we find that unaccented function words have properties that make them look more like clitics than full words. The most obvious such property is phonological: unaccented function words are phonological clitics, at least in languages with some sort of accentual system (for instance, word stress). However, there are other respects in which function words might resemble clitics more than lexical words. One very common instance of this concerns word order. In many languages the order of words within the clause or within, say, a noun phrase can be very free.