Words, Affixes, and Clitics as Prototype Categories

Table of Contents

1. Introduction

2. Words and Affixes
2.1. Definition of Word and Criteria for Wordhood
2.2. Variety of Affixes
2.3. Properties of Affixes

3. Clitics
3.1. On the Notion of Clitic
3.2. Types and Properties of Clitics

4. Comparison of Categories
4.1. Affixes and Clitics
4.2. Words and Clitics

List of References

1. Introduction

It is hard to find a discipline which would not be concerned with the notions of category and categorisation in some ways. These issues are of particular interest for psychology, philosophy, and linguistics. These sciences treat these subjects differently, but for all of them they have long been a major focus of a number of their investigations (Cohen 2005:2).

The notion of category is closely connected with the name of the Greek philosopher Aristotle. His conceptualisation of category, being defined in terms of necessary and sufficient criteria, which has clear-cut boundaries and members with an equal membership status, has been valid and relevant for the most part of the 20th century (Taylor 2003:20-21; Aarts 2004: 1-3).

The emergence of cognitive science in 1970 made it necessary to reassess the Aristotelian definition of category (Aarts 2004:3). Plenty of research held at that time showed that not all categories conformed to the classical theory of categorisation, but that there are also those with a prototype structure. The boundaries of these categories are not clear but fuzzy, and their members do not all have an equal status (Taylor 2003: 53).

These findings have had a considerable impact on linguistics, and many linguists were forced to look at linguistic categories anew and reconsider their structure (Taylor 2003:2). Although it is common knowledge today that a great many linguistic categories are, indeed, prototype categories (Kemmer 1992:145), a number of linguists still perceive grammatical categories as being classical in their nature (Taylor 1994:328). These linguists are reluctant to accept the idea that prototypicality might be relevant to grammar and that grammatical categories, like all other categories, can also display prototype effects.^[1] Bouchard, for example, states: “[F]uzzy categorization is involved only in the processes dealing with perception or beliefs about the ‘external world’ and […] it is not […] involved in the functioning of Grammar” (Bouchard 2004:484-85).

The purpose of this paper is to provide some evidence for the existence of prototype categories in grammar, and to demonstrate that prototypicality does apply to this level of language. To complete this task, three grammatical categories, namely words, affixes, and clitics will be closely examined and analysed. First, it will be attempted to define these categories as well as to suggest the criteria for category membership. Then, these categories will be compared and contrasted, and it is not until this comparison that their prototypical nature can reveal itself and becomes evident.

2. Words and Affixes

2.1. Definition of Word and Criteria for Wordhood

In the linguistic literature there are a great many definitions for the concept of word. The most productive and oft-quoted one is given by Bloomfield, who defines word as “a minimum free form, […] a form which may be uttered alone” (Bloomfield 1926:156).

For an item to be referred to as a word, certain criteria must be fulfilled. These, according to Tokar (2012:14-17), comprise, first of all, the isolatability criterion, i.e. a word’s ability of being used independently, movement criterion, and meaning criterion, which defines a word as “any combination of sounds that is associated with some meaning” (Tokar 2012:14). Further criteria include the uninterruptability criterion^[2] and orthographic criterion. The latter represents a word as a material which has spaces on both sides in writing (Tokar 2012:19).

2.2. Variety of Affixes

An affix is defined as “a grammatical element that is combined with a word, stem, or phrase to produce derived and inflected forms” (Encyclopedia Britannica). In general, affixes can be differentiated in two basic ways: affixes that differ in terms of their function and those that are defined according to their position with reference to the stem (Tokar 2012:56). The first group of affixes includes derivational affixes and inflectional affixes. While the function of derivational affixes is the formation of new words, such as un- and - ness in unhappiness and - ly in beautifully (Velupillai 2012:90), the function of inflectional affixes is purely grammatical (Velupillai 2012:90). Typical examples of inflectional affixes are the noun plural suffix - s and the past tense marker - ed (Velupillai 2012:90).

In terms of their position, affixes are classified into four groups: prefixes (e.g. un - of untrue), suffixes (e.g. -y of cloudy), interfixes (e.g. - o - of morph o logy) and infixes^[3] (e.g. hanger s -on, abso- bloody - lutely) (Tokar 2012:56). Velupillai (2012:91) also mentions circumfixes, which are affixes comprising two parts, such as ge- and - t in the German word geliebt (Velupillai 2012:91).

2.3. Properties of Affixes

To be classified as an affix, an item must possess certain features usually associated with affixes. These features are the following (Taylor 2003: 204):

- Affixes cannot occur independently of the stem to which they attach.
- Affixes generally cannot be stressed.
- Affixes are often integrated into the phonological shape of the word of which they are a part.
- Affixes are highly selective with regard to the items to which they attach.

Referring back to the Aristotelian definition of category,^[4] it can be said that the categories of words and affixes seem to be fairly classical in their nature. Their boundaries are well-defined, and it does not seem to be difficult to decide whether an item is a word or an affix. An affix, for example, is always a bound morpheme (Velupillai 2012:90), whereas one the distinctive features of a word is its capability of being used independently.

However, the situation changes when one, in addition, considers the category of clitics. These linguistic units that share properties of both words and affixes (Spencer 2012:1) put into question the nature of these categories and give reasons to view them rather as categories with a prototype structure.

3. Clitics

3.1. On the Notion of Clitic

Clitics refer to “intermediate linguistic units, which grammatically behave like words in that they combine with other words or phrases to make phrases, but are phonologically bound to an adjacent word traditionally referred to as the ‘host’” (Russi 2008:3).

In most cases, clitics are defined in phonological terms (Zwicky 1994: xiv), and function as reduced counterparts of full words, such as auxiliaries, personal pronouns or articles (simple clitics).^[5] However, there are also clitics, usually called special clitics, whose phonological nature is not so easy to identify.^[6]

Due to some important properties they share with words and their resemblance to affixes,^[7] clitics represent a fascinating subject in linguistic theory and are of particular interest for a number of linguists. Nevertheless, the question whether a clitic should be seen as a genuine linguistic category or rather as a subcategory of either affixes or words, or be defined as an umbrella term^[8] and not as a linguistic category at all is still open to debate (Spencer 2012: 5, 327-328).

3.2. Types and Properties of Clitics

Clitics differ in terms of their position, with regard to their linguistic nature and according to their syntax and phonology. As for their position, clitics can be proclitics and enclitics. A proclitic precedes “the word to which it is attached”^[9] (e.g. the French definitive article les) (Spencer 2012:4), and an enclitic follows “the word to which it is attached” (Zwicky 1977:8). Typical examples of enclitics are clitic auxiliaries and pronominal clitics.

As far as their linguistic nature is concerned, clitics are very heterogeneous. A distinction is made between clitic auxiliaries (e.g. ’ll < will; ’s < is, has), pronominal clitics (e.g. ’em < them) (Spencer 2012:93), clitic conjunctions and prepositions (e.g. - que in Latin) (Spencer 2012:32), interrogative clitics (e.g. - ne in Latin) etc.^[10] (Spencer 2012:90).

Finally, clitics also differ in terms of their phonology and syntax. Simple clitics, as pointed out earlier, are those clitics that have a regular phonology, and they are used as reduced versions of full words. Special clitics are clitics with an obscure phonology. These clitics might or might not have a full form equivalent, and if they have such an equivalent, the phonological resemblance is not always obvious.^[11] For this reason, special clitics are better seen as “lexically listened items rather than as phonologically derived forms” (Spencer 2012:42). Apart from the difference in phonology, simple clitics and special clitics also differ syntactically. Whereas both the syntactic function of simple clitics and their place in the sentence are the same as those of their full form counterparts,^[12] special clitics obey different principles of ordering (Zwicky 1977:5-6). Sometimes they may even break the established syntactic rules of the language. For instance, a Spanish pronominal clitic lo is placed directly before the verb, although the word order in Spanish is Subject-Verb-Object: e.g . Maria lo compro - Maria bought it (Spencer 2012: 42).

As for the properties of clitics, it can be said that clitics are generally unstressed, need a host to attach to, and they attach accidently, which means that the host of a clitic can be a word of whatever category (Spencer 2012:18).

4. Comparison of Categories

4.1. Affixes and Clitics

What makes clitics an interesting, but a rather challenging grammatical category is that clitics are often mixed up with function words, such as articles, auxiliaries, personal pronouns (Spencer 2012: 92-93) and affixes, in particular inflectional affixes^[13] (Spencer 2012:7).

It appears to be especially difficult to distinguish clitics from affixes, because both of them are bound morphemes (Zwicky & Pullum 1983:502), both of them need a host (in case of affixes, a stem) to attach to (Zwicky & Pullum 1983:503), and both affixes and clitics do not generally bear stress. Luckily, there is a set of criteria developed by Zwicky and Pullum that makes this distinction possible. These two linguists, being motivated by a desire to identify the linguistic nature of the English contracted negator n’t, came to conclusion that clitics and affixes can and must be differentiated. The first criterion that they developed is concerned with the host selectivity and is viewed as the most important criterion in the distinction between affixes and clitics (Spencer 2012:40). According to this criterion, clitics are elements that “exhibit a low degree of selection with respect to their hosts, while affixes exhibit a high degree of selection with respect to their stems” (Zwicky & Pullum 1983:503). The contracted form n’t, which is an inflexional affix (Zwicky & Pullum 1983: 512), attaches only to auxiliary verbs, and therefore is very selective. The host of the enclitic ’ s (the auxiliary has) can be any word, no matter what category it belongs to.^[14]

[...]

^[1] Possible explanation for this could be the fact that “[g]rammatical prototypes”, as Suzanne Kemmer (1992:147) points out, “are […] not so easily accessible to psycholinguistic investigation as lexical ones, since […] they are far more abstract or schematic”.

^[2] According to this criterion, a word is a unit “that does not allow the insertion of a modifying element within its boundaries” (Tokar 2012:19).

^[3] These affixes are characteristic of a non- standard usage of English (Tokar 2012: 56).

^[4] See Introduction

^[5] Typical examples of simple clitics are the reduced auxiliary is in The girl over there’s reading (Spencer 2012: 15) and the reduced du, das, einen and dem in Wennde ins Kino gehen willst, mußte die Karten ’n Tag vorher im Vorverkauf holen (Nübling 1992:1)

^[6] More on simple clitics and special clitics later in this paper

^[7] See the final chapter for more on the differences between clitics and affixes, and between clitics and words.

^[8] According to Zwicky (1994: xiii), umbrella terms are “names for problems, for phenomena that present ‘mixed’ properties of some kind, not names of theoretical constructs”.

^[9] Note that in the context of clitics, the verb to attach should not be taken literally. Whereas most affixes, when they attach to a word, integrate into its shape, there are a number of clitics which are divided from their hosts by means of an apostrophe or a blank space. Such clitics are also seen as being attached to their hosts, because they cannot function independently and need something to lean on.

^[10] In Romance languages, one also finds adverbial clitics, and discourse clitics in Russian and Greek. An example of a discourse clitic in Russian is the enclitic to (Spencer 2012:33-35).

^[11] For example, a full form of a Serbian/Croatian special pronominal clitic je (her) is njû (Spencer 2012: 24).

^[12] The German pronominal clitic ’s and its full form es both function as objects in Er hat’s gewusst and Er hat es gewusst (Spencer 2012: 27).

^[13] Compare : T he Queens of England where -s is an inflexional plural affix and the Queen’s hat where ’ s is a possessive clitic (Spencer 2012:2).

^[14] a) The ball you hit’s just broken my window [verb]; b) The drive home tonight’s been really easy [adverb] (Zwicky 1983:504); c) The guy you were talking to’s already left [preposition] (Spencer 2012:15).