1 The main units of derivational analysis

Derivation Versus Conversion

Conversion involves change of syntactic category with no overt marker of that change, as for example in English when the verb kick becomes a noun kick ‘an instance of kicking’ or a noun saddle becomes a verb saddle ‘to put a saddle on.’ There is no question that conversion is a sort of derivation in the sense that already existing lexemes are used to form new lexemes. Rather, the issue here is whether conversion is a form of derivation distinct from affixation. If conversion were simply a species of affixation, the noun kick and the verb saddle might have the structures in (2):
(2)

Debate has centered on whether conversion is best analyzed as attachment of a zero affix or as something else, for example, simple relisting of items in the mental lexicon (or listing of category-less items in the lexicon, as frequently advocated by adherents of distributed morphology The zero-affixation analysis predicts that items that have undergone conversion should display the same properties as items derived by overt affixation such as determination of a clearly circumscribed semantic category (e.g., agent or instrument) or inflectional class (e.g., masculine, feminine, neuter in languages that distinguish grammatical gender). To the extent that items that have undergone conversion do not behave as overt affixes do—instantiating several semantic categories or falling into multiple inflectional classes, the zero-affixation analysis becomes less attractive. The debate is by no means settled: as the articles in Bauer and Valera illustrate, which analysis is supported may depend on careful examination of the facts as well as on the typological characteristics of the languages examined¹².

2. Formal Means of Derivation

Derivation may be accomplished by any formal morphological means including affixation, reduplication, internal modification of bases, and subtraction. By far the most common morphological processes used for derivation is affixation, and in particular prefixation and suffixation, with suffixation apparently more frequent than prefixation. Infixation is well-represented as well, but circumfixation is rather rare in the languages of the world. Reduplication is also well-attested. Processes of internal modification include internal vowel and consonant changes as well as root and pattern or templatic systems. Formal processes of truncation or subtraction appear to be rather rare. We deal with each of these formal techniques in turn. A good resource for typological patterns of derivation is Štekauer et al.

2.1 Affixation

Prefixation involves the attachment of a bound morpheme before a free or bound base (3a). Suffixation involves the attachment of a bound morpheme after a free or bound
Circumfixation is a form of affixation that involves simultaneous bound morphs, one prefixally and one suffixally crucially, in cases of circumfixation neither the prefix plus the base nor the suffix plus the base can be shown alone to contribute a recognizable part of the meaning of the derived form. For example, in the Tagalog example in (4) ka- before the base and -an after the base together create collective nouns.
Infixation involves the insertion of a bound morpheme into a base. For example, in Karok (a language isolate of California), an infix <eg> inserted after the onset of the first syllable of the base creates the intensive form of the verb (5a), and in Hua, a Trans-New Guinean language, an infix <’a> inserted before the last syllable of the base forms a negative
While other affixes are simply attached to one or the other of the edges of the base, the position of an infix must be specified with respect to some phonological landmark or ‘pivot’ in the base, for example, the initial or final vowel, consonant, or consonant cluster, the stressed syllable, and so on. Much attention has been devoted in recent research to determining the nature of those

2.2 Reduplication

Reduplication uses repetition of all or part of the base as a means of word formation. Samoan, for example, uses full reduplication of a verb to form a related noun:
Partial reduplication repeats a portion of the base, frequently but not always a prosodic constituent like a syllable, foot, or minimal word:
n Mokilese, for example, reduplication copies a heavy syllable, that is, a syllable that either ends in a consonant or contains a long vowel. Researchers have worked to define the scope and limits of reduplication as a morphological phenomenon, to identify subtypes of reduplication, and to account in theoretical terms for restrictions on reduplication such as whether the repetition always involves some sort of prosodic constituent or where the site of reduplication can be. is an excellent survey of the literature. More extensive theoretical treatments can be found in

2.3 Templatic Derivation

So-called templatic or root and pattern morphology is sometimes treated as a species of affixation, called ‘transfixation,’ but in recent years it has typically been analyzed as a distinct formal means of word formation rather than as a complex form of affixation. Templatic morphology involves the intercalation of segments according to a template or abstract pattern of consonants (C) and vowels (V). For example, in Arabic and Hebrew roots typically consist of three consonants that constitute the semantic core of words. Vowels may be interspersed among the root consonants in different patterns to express distinctions that are often inflectional. The example in (8) shows words based on the root ktb, all of which have something to do with writing¹³:
In the Arabic example in (8), the root consonants occur in a fixed order and modifications to meaning can be effected by interspersing different vowels, or by varying the arrangement of consonants and vowels. The pattern CVCVC is associated with the verb ‘write,’ but the pattern CVCCVC is a causative verb ‘cause to write’ and the pattern CVVCVC can have a reciprocal meaning (‘corresponded’ = ‘wrote to each other’). Templatic word formation is most familiar in the Semitic languages, but it is attested in other languages as well, as the examples.

2.4 Subtraction

Derivation more often than not involves some incremental addition to a base, but it can also involve truncation or deletion of material from a base. For example, some verbs in the Uto-Aztecan language Tohono O’Odham derive a perfective verb from an imperfective one by deleting a final consonant English uses a process of deletionto formhypocoristics,for example, Tom from Thomas or Pat from Patricia. It is increasingly believed that subtraction is in fact a form of templatic morphology, one in which a longer base is made to fit a smaller template, resulting in the deletion of

2.5 Internal Changes

Derivation may be effected by internal modifications to a base that may include changes to vowels or consonants, changes in stress or tonal patterns, or combinations of these. Internal vowel changes are often referred to as rules of umlaut or ablaut. Historically, umlaut involves the fronting (and sometimes raising) of a base vowel triggered by the front vowel in an adjacent suffix. The suffixal vowel may subsequently change (10a) or be lost entirely, leaving the base vowel as the sole exponent of the morphological distinction, as in the German plural example in
ther vowel changes are generally referred to as ‘ablaut’or ‘apophony.’ For example, female terms are derived from male terms in Manchu by fronting vowels, as illustrated in
Internal modification of bases may also involve changes in consonants, a process that is sometimes called consonant mutation. McLaughlin (2000, p. 335), for example, illustrates a process of diminutivization in the West Atlantic language Seereer-Siin.
n this language diminutives are formed by prenasalizing a consonant in the base.