Affixation is a morphological process that adds phonological material to a word in order to change its meaning, syntactic properties, or both. Some examples of affixation in English are given in (1).
a. fond fondness
b. start restart
c. car cars
In this article, an overview of different types of affixation processes is given, followed by a discussion of how to distinguish affixation from other processes. Then, a number of theoretical problems related to affixation are presented.
Types of Affixation
The phonological material added in affixation is called the affix. The affix is attached to the base. Different types of affixation can be distinguished on the basis of the position of the affix with respect to the base or on the basis of how affixation affects the meaning of the base. In this section, the emphasis is on the former. The latter is discussed in the section on delimitation issues.
Suffixation is the most common type of affixation. In suffixation, the affix is added to the end of the base. Examples are (1a) and (1c) above. In (1a), the suffix-ness is added to the adjective fond to produce the noun fondness. In (1c), the suffix -s is added to the noun car to produce the plural of the noun. In most languages, suffixation is the most widespread form of affixation. In languages such as Turkish and Finnish, it is the only type of affixation.
In many respects, prefixation is the mirror image of suffixation. In prefixation, the affix is attached to the beginning of the base. An example is (1b) above, where the prefix re- is added to the base start. Prefixation is less widespread than suffixation, but some languages, e.g. Khmer, only have prefixation. In languages that have both suffixation and prefixation, the former usually has a larger range of functions. Thus, in English, all inflection is expressed by suffixation. In word formation, we find both, but prefixation only rarely changes the syntactic category of the base, as when the verb enrich is formed from the adjective rich.
In most languages, affixation only involves suffixes and prefixes. Infixation, in which the affix is attached inside the base, is much rarer. An example is the infixum- in Tagalog, illustrated in (2).
a. bilih (‘buy’) bumilih (‘bought’)
b. gradwet (‘graduate’) grumadwet (‘graduated’)
The infix in (2) is used to express the past tense of the verb. It attaches itself after the first consonant or cluster of consonants of the base. This is its anchor point. In general, the anchor point of an infix is at most one syllable removed from the left or right boundary of the base.
Whether other types of affixation exist depends on the theoretical framework adopted. Thus, circumfixation is the simultaneous addition of a prefix and a suffix. The mere observation of a form such as enrichment is not sufficient evidence for the existence of a circumfix in English, because the form can be analyzed transparently as a result of suffixation of -ment to the verbal base enrich, which is in turn the result of prefixation of en- to the adjective rich. A better candidate for a circumfix is Dutch ge…te, exemplified in (3).
(3) steen (‘stone, rock’) gesteente (‘type of stone/rock’)
The change in meaning in (3) can be expressed as a generalization to a type. It can be argued that ge…te is a circumfix, because neither *gesteen nor *steente are possible Dutch words. A theory that excludes circumfixes would have to postulate one of these nonexisting forms as an intermediate form in the derivation.
As affixation is the most salient morphological process, it is no surprise that many of the foundational discussions in morphology are directly related to affixation.
A first set of issues concerns the function of affixation. Traditionally, inflection and derivation are distinguished. Inflection adapts a word to its syntactic environment. A prototypical example is the agreement of a verb with its subject in person and number, as illustrated for sleep in (4).
a. The child sleeps.
b. The children sleep.
Derivation is a type of word formation that involves affixation. It forms new lexemes, marked by a different syntactic category and/or a different lexical meaning, as illustrated in (1a-b). Another contrast is the one between inflection and stem formation. An example of the latter can be observed in Dutch kinderstoel (‘highchair’), a compound of kind (‘child’) and stoe (‘chair’). The form kinder is different from the singular ( kind) and the plural ( kinderen). The suffix -er can be said to produce a secondary stem used in compounds.
Another set of delimitation issues concerns the nature of the affix as a phonologically dependent, morphological unit. Affixes contrast on the one hand with clitics, and on the other with stems used in compounding. Clitics are phonologically dependent particles, i.e. particles that must attach to a host word much like an affix, whose distribution is, however, determined by syntactic rules rather than morphological rules. An example is the Italian clitic pronoun -ti (‘you’) in incontrarti (‘to meet you’). While these clitics are generally not considered as affixes, Turkish case and number endings as in (5) are more of a borderline case.
Ankara ve I˙zmire gideceg˘im
‘Ankara and Izmir I go’,
i.e. I go to Ankara and Izmir
In (5), Ankara does not have any case ending, whereas I˙zmir is followed by the dative ending -e. Nevertheless, the dative ending applies to the coordinated noun phrase Ankara ve I˙zmir. Therefore, it can be argued that the dative marker is not an inflectional affix but a clitic. Similar observations can be made about the genitive marker in the queen of England’s hat.
Compounding usually combines items that occur as independent words, e.g. bookshelf. Words such as philanthropic and anthropomorphic share many properties with compounds, but their constituent parts do not occur as independent words (in English). They are often called neoclassical compounds. In fact, anthropo is the Ancient Greek word for human being, a sense it also has in English words. Although it is not a word in English, it does not behave like an affix either. It has a stem-like meaning and can appear in different positions in a compound. Other elements of the same general type, e.g. pseudo, occur almost exclusively as a prefix, often with a reduced meaning more typical of an affix than of a stem, e.g. pseudocultured. It is very difficult to draw a clear borderline between stems and affixes among such neoclassical elements. The term affixoid is sometimes used to refer to items for which it is difficult to determine whether they are stems or affixes. An example from outside neoclassical word formation is German reich (‘rich’) in ertragreich (‘productive, profitable’). In English, -ful in successful is similar.
Affix or Process?
One of the central questions in morphology is whether affixation should be seen as a rule applying to two elements, a base and an affix, or as a process applying to a base. Charles Hockett (1954) calls the first position Item & Arrangement, and the second Item & Process. Although all morphological processes can be described in either framework, some descriptions are more natural than others.
In a strict Item & Arrangement position, such as defended by Rochelle Lieber (1992), stems and affixes have separate entries in the mental lexicon. The suffix -ness is described as a noun that requires an adjective to its left. Given an appropriate concept of headedness (cf. below), this explains why fondness is a noun. The Item & Arrangement approach works particularly well as long as the form of the resulting word is the concatenation of the forms of the stem and the affix, as in fondness. Special provisions have to be made for cases where phonological processes interact with the concatenation, as in (6–8).
a. active activity
b. intend intention
a. a house to house asylum seekers
b. to read a good read
a. live life
b. extend extent
In (6), suffixation triggers a phonological change of the base, stress shift in (6a), and a change of the final consonant in (6b). In an Item & Arrangement account, one might say that these phonological changes are triggered by the affix. In (7), there is no affix, but the relationship between the nouns and verbs is very similar to what we find in cases of genuine affixation such as encase and entertainment. In an Item & Arrangement account, one might either say that (7) is not affixation but something else (e.g. conversion), or that (7) involves zero affixes, i.e. affixes that do not have a phonological realization. The examples in (8) show that phonological changes can occur without an affix. This can be interpreted as phonological rules triggered by a zero affix or by conversion. The general approach in Item & Arrangement is that concatenation of base and affix is taken as central. Cases such as (6–8) are treated as exceptions to be fitted in.
In an Item & Process approach, as argued for by Stephen Anderson (1992), the process of affixation takes priority over the affix. In (7), the word formation process changes the syntactic properties of the input without affecting its form. In (8), the process modifies the form of the input by changing the quality of the final consonant and, in (8a), of the vowel. In (6), the process affects the form of the input by changing the stress position or the voicedness of the final consonant and appending some further phonological material, the suffix. Affixation of the type illustrated by fondness, the prototypical case in Item & Arrangement, is a special case in Item & Process where there is a perfect match between the input and a part of the output of the affixation process.
It is difficult to find conclusive arguments for either Item & Arrangement or Item & Process approaches to affixation. Proponents of Item & Arrangement usually claim that their approach is theoretically more restrictive than Item & Process. Proponents of Item & Process typically argue that processes such as infixation, reduplication, vowel change, consonant change, and conversion are too frequent, especially in non-European languages, to treat them as exceptional compared to ‘pure’ affixation. Linguistic data alone cannot determine as to which of the two approaches is correct, because every Item & Arrangement account can be reformulated into Item & Process and vice versa.
The concept of head of a word only makes sense in an Item & Arrangement-based account of morphology. The head determines the syntactic category of the resulting word as well as other properties such as gender in languages that distinguish them. Thus, fondness is a noun because -ness is the head, and French activité (‘activity’) is a feminine noun because -ité is the head.
Different methods have been proposed to determine the head. Anna Maria Di Sciullo and Edwin Williams (1987) propose that the head is always the rightmost element. This concurs with the observation that suffixes often change the syntactic category of their base, but prefixes, as in (1b), usually do not. Exceptions include cases such as enrich and debug, where the prefixes make verbs out of an adjective and a noun. An alternative proposed by Lieber (1990) is that the last affix to attach is the head. This generalization makes re- the head in (1b) and accounts correctly for enrich and debug, but has problems with prefixes such as counter- in (9).
a. intuitive counterintuitive
b. example counterexample
c. sign countersign
In (9), counter- is the only (hence last) affix attached, but the syntactic category depends on the word it attaches to. The Spanish diminutive, illustrated in (10), is problematic for both.
a. el pintor .‘the painter’
el pintorsito ‘ the little painter’
b. la cancyón ‘ the song’
la cancyonsita ‘the little song’
The final -o and -a in (10) are not a part of the affix, but so-called word markers, whose form is determined in this case by the grammatical gender of the word (-o for masculine, -a for feminine). Although the suffix -sit is the rightmost element and the only affix, it is the base that determines the gender as indicated by the article. In an Item & Process approach, there is no need for a head, because its function is subsumed in the affixation process.