Many Swahili words are built up by using roots and affixes. Affixes may be described in terms of the class or category of the word they combine with, and the category of word found by the root and affix combination. Whereas roots do not change, many affixes do. Swahili morphology is summarily described under the main headings: pronouns and pronominal prefixes, verbs and noun classes. (Safari & Akida, 1991)
3.1 Free and bound morphemes
Whether a morpheme is bound or free, can be defined by considering their occurrence. Morphemes that can occur on their own are said to be free. Bound morphemes need to be attached to other morphemes. In English, for instance, nouns are free- and determiners are bound morphemes (Fromkin, 2000). Both bound and free morphemes occur in Swahili, but there are more bound morphemes than free morphemes. In English, certain word classes are always bound (like nouns); in Swahili one word class can contain both bound and free morphemes:
The nouns in example (2) (Mohammed, 2001) are free morphemes. The nouns in example (3) are bound morphemes. The word stems -toto and -su require a prefix that gives the word a certain number and class (Givon, 2001). Adjectives, like -dogo, meaning 'small', also need a number- and class prefix (m-, wa-, ki-, vi-, etc.). This leads to clauses like m-dogo, which can be used to express that it is small for a child (ibid.).
3.1 Pronouns and pronominal prefixes
Pronouns in Swahili are divided into personal, possessive, demonstrative, generalising and interrogative pronouns (Myachina, 1981). Personal pronouns have a morphemic structure built out of roots plus the appropriate marker.
These independent pronouns can stand on their own and function as an object or a subject. But they are only used in selective contexts: for instance the word ni 'am' as in 'I am an African' Mimi ni Africano. Next to that, they can also emphasize the subject of the sentence. In (2i') and (2ii'), mimi underlines the subject (Benjamin, 1998):
A pronoun can never replace a pronominal prefix, if a construction requires a pronominal prefix. A pronominal prefix must be included whether or not an independent pronoun is used. *Mimi tapita posta is an ill formed Swahili sentence.
Besides subject and object markers, tense markers can be attached to Swahili verbs as prefixes. The essential component of any finite verb is (in the following order): subject prefix - tense marker - verb root. If an object prefix is inserted, the verb root always follows. A negation marker of the verb always precedes the subject prefix (Safari & Akida, 1991).
The object agreement is an optional item. If there is a specific direct object, the object agreement is obligatory; if the direct object isn't specific, the object agreement can be deleted (ibid.). Mood is always the final suffix. It can either be indicative -a, subjunctive -e, or negative -i. When the suffix is indicative, the word describes ongoing actions or states, habitual actions of the present, actions and states in the past or future and imperatives. The subjunctive mainly expresses desires, possibility, necessity and requests (ibid.). In Swahili clauses, the infinitive (a prefix) is rarely used; other inflectional prefixes are used to modify the clause. We can conclude that Swahili is a bare verb language. In English, bare stems are also used without the infinitive 'to' like in 'I can sing' (Deen, 2003).This puts the two languages in the same class: they are both bare verb languages instead of root infinitive languages.
In Swahili, every noun is assigned to a specific noun class. The noun classes are generally marked by a class prefix. Swahili nouns are inflected for gender and number by a characteristically Bantu prefix system (Vitale, 1981). Gender is grammatical and affixes mark a noun for membership in a noun class. These are, for the most part, not definable on semantic grounds. There is a certain amount of discrepancy as to which of several numbering systems should be used in the classification of the gender system.
Table 4: Original Swahili noun classes (Safari & Akida, 1991)
The original Swahili classes 12 and 13 have no reflexes in present-day Swahili. In modern Swahili, nouns are categorised in far less noun classes as shown below:
Table 5: Modern Swahili noun classes (Mohammed, 2001)
Note that classes 15, 16, 17 and 18 have not been illustrated above because they do not have plural forms Nouns in classes 1/2 denote only humans (but not all humans are in class 1/2). Class 14 refers to abstract characteristics. Class 15 has verbal infinitives and classes 16 - 18 are locatives. For the remaining classes the semantic base is less obvious. For example: class 3/4 contains words denoting plants and trees, class 9/10 contains names of animals, and class 6 contains liquids (Brown & Ogilvie, 2009).
Variation in morphology
As we have seen, Swahili nouns take various class prefixes according to the class they belong. The noun class system is not as strict as implied. There are many words that do not fit in one of these classes based on a semantic characterization. There is variation which is illustrated in atypical, flexible, irregular, and un-prefixed nouns. A variation for the use of the noun classes is nominal derivation, by shifting nouns from one class to the other. For example, shifting nouns into class 7/8 denotes diminutive -kitoto 'a small child'- while class 6 can be used to express a group of individuals, and not only the plural form: fisi (class 10) 'hyenas', mafisi (class 6) 'a pack of hyenas' .
But not only the classification of the noun class can vary. There is also variation in the noun prefixes. Sometimes two noun classes share the same prefix, but do not share the same noun class. For example, the regular prefix in Classes 1 and 3 is m- . Some agreements are the same for both classes, and some are not. This can depend on the phonological form of the prefix rather than the syntactic-semantic characteristics of the noun class (Alcock & Ngorosho, 2002).
M. Brenzinger (1992) Language Death. Walter de Gruyter & Co., Berlin