Introduction

1.1 General remarks

   Chichewa is a language of the Bantu language group in the Benue-Congo branch of the Niger-Kordofania language family. It is spoken in parts of east, central and southern Africa. Since 1968 it has been the dominant language in the east African nation of Malawi where, until recently, it also served as that country’s national language. It is spoken in Mozambique (especially in the provinces of Tete and Niassa), in Zambia (especially in the Eastern Province), as well as in Zimbabwe where, according to some estimates, it ranks as the third most widely used local language, after Shona and Ndebele. The countries of Malawi, Zambia, and Mozambique constitute, by far, the central location of Chichewa. Because of the national language policy adopted by the Malawi government, which promoted Chichewa through active educational programs, media usage, and other research activities carried out under the auspices of the Chichewa Board, out of a population of around 9 million, upwards of 65 percent have functional literacy or active command of this language. In Mozambique, the language goes by the name of Chinyanja, and it is native to 3.3 percent of a population numbering approximately 11.5 million. In Tete province it is spoken by 41.7 percent of a population of 777,426 and it is the first language of 7.2 percent of the population of Niassa province, whose population totals 506,974 (see Firmino 1995). In Zambia with a population of 9.1 million, Chinyanja is the first language of 16 percent of the population and is used and/or understood by at least 42 percent of the population, according to a survey conducted in 1978 (cf. Kashoki 1978). It is one of the main languages of Zambia, ranking second after Chibemba. In fact, out of the 9.1 million people of that country, it is esitimated that 36 percent are Bemba, 18 percent Nyanja, 15 percent Tonga, 8 percent Barotze, with the remainder consisting of the other ethnic groups including the Mombwe, Tumbuka, and the Northwestern peoples (see Kalipeni 1998). The figures show that at least upwards of 6 million people have fluent command of Chichewa/Chinyanja.

   As indicated, the language is identified by the label Chinyanja in all the countries mentioned above except, until recently, in Malawi. It is commonplace to see many publications or former school examinations making reference to the language as Chinyanja/Chichewa. The factors that led to such a multiplicity of labels will not be spelt out here. The relevant details are readily available elsewhere (see Mchombo website, http://www.humnet.ucla.edu/humnet/aflang/chichewa/).

1.2 General features of Chichewa

   In its structural organization, Chichewa adheres very closely to the general patterns of Bantu languages. Its nominal system comprises a number of gender classes characteristic of Bantu in general. The noun classes play a significant role in the agreement patterning of the language. Thus, modifiers of nouns agree with the head noun in the relevant features of gender and number, as will be illustrated below (see section 3 below). In its verbal structure, Chichewa is typical of Bantu languages in displaying an elaborate agglutinative structure. The verb comprises a verb root or radical, to which suffixes or extensions are added (cf. Guthrie 1962) to form the verb stem. The extensions affect the number of expressible nominal arguments that the stem can support. In other words, verbal extensions affect the argument structure of the verb (Dembetembe 1987; Dlayedwa 2002; Guthrie 1962; Hoffman 1991; Mchombo 1999a, 2001, 2002a, b; Satyo 1985). To the verb stem are added proclitics which encode syntactically oriented information. This includes the expression of Negation, Tense/Aspect, Subject and Object markers, Modals, Conditional markers, Directional markers, etc. The structural organization of the verb will be discussed in detail below. Motivation for the suggested structural organization will be provided.

   With regard to phonological aspects, Chichewa is a tone language, displaying features of lexical and grammatical tone. Basically, Chichewa has two level tones, high (H), and low (L). Contour tones also occur but then only as a combination of these level tones, usually on long syllables (Mtenje 1986b). In its segmental phonology, Chichewa has the basic organization of five vowel phonemes. The verbal unit manifests aspects of vowel harmony. This will be illustrated in sections that focus on the structure of the verb. In its syllable structure, Chichewa has the basic CV structure common in Bantu (Mtenje 1980). These issues will be taken up in the next chapter. At this juncture, attention will be turned to the noun classification system and related issues.

1.3 The classification of nouns

   A major feature of Bantu languages is the classification of nouns into various classes; another is the elaborate agglutinative nature of the verbal structure. The latter will be reviewed in detail in subsequent chapters. With regard to nominal morphology, Chichewa displays the paradigmatic case of nouns maintaining, at the minimum, a bimorphemic structure. This consists in the nouns having a nominal stem and a nominal prefix. The prefix encodes grammatically relevant information of gender (natural) and number. This plays a role in agreement between the nouns and other grammatical classes in construction with them.

   Let us look at the system of noun classification in Bantu languages. Typical examples of a nouns are provided by the following:

   Of interest is the question of the basis for this classification of nouns. This is an issue that still awaits a definitive response. The formal structure of the noun, which does have some bearing on its class membership, has relevance to the regulation of the agreement patterns of the languages. In brief, noun modifiers are marked for agreement with the class features of the head noun, and these features are also what are reflected in the SM and the OM in the verbal morphology. This can be illustrated by the following:

   In these sentences, the words in construction with the nouns are marked for agreement with that head noun (the actual agreement markers in these examples are chi and u; the i vowel in chi is elided when followed by a vowel, and the u is replaced by the glide w in a similar environment). Chichewa is a head-initial language; hence, the head noun precedes its modifiers within a noun phrase. The formal patterns that yield the singular and the plural forms are, traditionally, identified by a particular numbering system now virtually standard in Bantu linguistics (Bleek 1862/69; Watters 1989). Consider the following data:

   These classes show part of the range of noun classification that is characteristic of Bantu languages. The full range of noun classes for Chichewa is presented in table 1.1 below; the class numbers used in the examples reflect the classes listed in that table. The singular forms of the first group above constitute class 1, and its plural counterpart is class 2. These classes tend to be dominated by nouns that denote animate things although not all animate things are in this class. In fact, it also includes some inanimate objects. The next singular class is class 3, and its plural version is class 4. This runs on to classes 5, 6, 7, and 8. There is also class 1a. This class consists of nouns whose agreement patterns are those of class 1 but whose nouns lack the m(u) prefix found in the class 1 nouns. The plural of such nouns is indicated by prefixing a to the word. For instance, the noun kalúlu ‘hare’ whose plural is akalúlu typifies this class. Each of these classes has a specific class marker and a specific agreement marker. Beginning with class 2, the agreement markers are, respectively, a, u, i, li, a, chi, zi. Class 1 is marked by mu (or syllabic m), u, and a, depending on the category of the modifier.

   Consider the following:

   In this, the numeral módzi ‘one’ is marked with the agreement marker m but the verb has a for the subject marker. The u is used with demonstratives and when the segment that follows is a vowel. This seems to apply to most cases, regardless of whether the vowel in question is a tense/aspect marker, associative marker or part of a stem, such as with possessives. The possessives could themselves be analyzed as comprising a possessive stem to which an associative marker is prefixed (cf. Thwala 1995). Consider the following:

   In this sentence, the w is the glide that replaces u when a vowel follows regardless of the function associated with that vowel.

   Although most of the nouns are bimorphemic, there are a number of cases where a further prefix, which may mark either diminution or augmentation, is added to an already prefixed noun. This is shown in the following:

   In this sentence, the pre-prefixes ka for singular and ti for plural, are added to nouns to convey the sense of diminutive size. These pre-prefixes then control the agreement patterns (cf. Bresnan and Mchombo 1995), which provides the rationale for regarding them as governing separate noun classes. In fact, in other Bantu languages, for instance Xhosa and Zulu, the nouns have a pre-prefix that is attached to the “basic” prefix (cf. Dlayedwa 2002; Satyo 1985; van der Spuy 1989). In Xhosa, for instance, nouns consist of a pre-prefix, basic prefix, and a noun stem. The pre-prefix and basic prefix are involved in the agrement patterns. One significant point to be made is that locatives also control agreement patterns. Consider the following:

   This gives such locatives the appearance of being class markers. It has been argued that locatives in Chichewa are not really prepositions that mark grammatical case but, rather, class markers (for some discussion, see Bresnan 1991, 1995).

   At this stage it would be useful to provide the full range of noun classes for Chichewa. This is presented in table 1.1. Note that some classes are not present in this language. For instance, Chichewa lacks class 11, with prefix reconstructed as du in proto-Bantu.

Table 1.1 Noun classes in Chichewa

   Some of the classes have prefixes which are starred. These classes consist of nouns which, normally, lack the indicated prefix in the noun morphology. Samples of class 5 nouns are provided above. Most of the nouns in classes 9 and 10 begin with a nasal but there are no overt changes in their morphological composition that correlate with number. The number distinction is reflected in the agreement markers rather than in the overt form of the noun. Examples of class 9/10 nouns are: nyǔmba ‘house(s),’ nthenga ‘feather(s),’ mphîni ‘tattoo(es),’ nkhôndo ‘war.’ Class 15 consists of infinitive verbs. The infinitive marker ku- regulates the agreement patterns, just like the diminutives (classes 12 and 13) and locatives. The infinitives are thus regarded as constituting a separate class although, just as is the case with the locatives, with minor exceptions, there are no nouns that are peculiar to this class. The minor exceptions to locatives have to do with the words pansi ‘down,’ kunsi ‘underneath,’ panja ‘outside of a place,’ kunja ‘(the general) outside,’ pano ‘here (at this spot),’ kuno ‘here (hereabouts),’ muno ‘in here.’ With these, the locative prefixes pa, ku, and mu are attached to the stems -nsi, -nja, and -no, which are bound. The agreement pattern regulated by the infinitive marker ku- is exemplified by the following:

1.4 On the status of prefixes

   At a more general level of analysis the question arises with respect to the status of the nominal prefixes. Are they morphological units that combine with the stem in the morphological component of grammar, or are they syntactic elements that form a phonological word with the stem? In an analysis of Shona, a Bantu language spoken in Zimbabwe, Myers (1991) argues that the prefixes in nominal structure are syntatic determiners which form a phonological word with the stem. The structure of the noun could be represented as below, for constructions with the diminutive or the locative.

   In this the prefixes comprise syntactic determiners that combine with the stem at the level of phonology. There is thus no morphological component dedicated to word formation. In fact, in other analyses, mainly couched within the Principles and Parameters Theory, information pertaining to number and gender is factored into separate structural projections, with movement accounting for their subsequent realization within the same overt form (cf. Carstens 1991). These analyses have been countered in the work of Bresnan and Mchombo (1995) on the basis of lexical integrity. Specifically, Bresnan and Mchombo noted that:

morphological constituents of words are lexical and sublexical categories – stems and affixes – while the syntactic constituents of phrases have words as the minimal, unanalyzable units; and syntactic ordering principles do not apply to morphemic structures. As a result, morphemic order is fixed, even when syntactic word order is free; the directionality of “headedness” of sublexical structures may differ from supralexical structures; and the internal structure of words is opaque to certain syntactic processes. (Bresnan and Mchombo 1995: 1)

   Adopting the general strategy that the internal structure of words is opaque to syntactic processes, Bresnan and Mchombo adduce evidence which demonstrates that such syntactic processes as extraction, conjoinability, gapping, inbound anaphora, and phrasal recursivity do not apply to Bantu nouns. This undermines the syntactic analysis of the nominal structure in Bantu proposed by Myers as well as Carstens, and maintains a morphological structure of the nouns. The one area where a syntactic analysis appears plausible is in locative nouns. In these, the agreement patterns appeared to alternate between agreement with the locative or with the class of the basic noun. Such alternative concord is impossible with the diminutives, where only the outer prefix controls agreement. With locatives, on the other hand, the agreement can, sometimes, be with the inner class marker. Consider the following:

   However, this also allows for the following expression with the possessive agreeing with the basic class marker of mpando ‘chair’:

   b. pampando wánga

   Such alternation in the concord seems to indicate that the locative must have a syntactic structure since the opacity of the word to syntactic processes is violated. In the analysis provided by Bresnan and Mchombo the claim was that the locative marker may have indeed originated as a syntactic element but that it has undergone steady morphologization. The alternative concord appears to indicate that the morphologization process is not complete. In brief, the nouns in Bantu satisfy the tests for lexical integrity, indicating their status as morphological words. The noun class markers are not syntactic determiners but morphological units, specifically, prefixes, combining morphologically with the stem to yield the noun.

2
Phonetics and phonology

2.1 The consonant system

   In its consonantal inventory, Chichewa has a range of sounds. These include plosives, nasals, fricatives, affricates, glides, and an alveolar lateral. Although in standard orthography it is claimed that the trill [r] is present, in allophonic variation with the lateral [l], it is a sound that is not common in speech. The rule concerning the distribution of [r] is that it appears after the front vowel phonemes [i] and [e], as in Luganda, a language spoken in Uganda (Katamba 1984). However, the rule is not general. In its formulation in the Chichewa Orthography Rules, it is immediately accompanied by the rider that the rule does not apply when the conditioning environment is created by affixation. Thus, according to the rule, [r] should occur in the following words, as indicated:

   However, [l] should not be changed to [r] when the conditioning environment results from affixation. For instance, one of the forms of the copula ‘be’ is the irregular verb -li. Consider the following expressions:

   The first-person-singular pronominal marker in Chichewa is ndi, and the first-person-plural marker is ti. If either one of these is attached to the copula -li, the lateral [l] would be in the environment for the trill, as shown below:

   According to the rule, the lateral of the copula should be a trill. However, the lateral does not become a trill and even the rules for Chichewa orthography clearly prohibit the change of the lateral to a trill in such environments (Chichewa Board 1990). The reality is that even in the cases where the trill is supposed to be legitimate, in ordinary pronunciation of the words given above, it is the lateral that is used, not the trill. In recent revisions of the orthography of the language, it has been proposed to drop the trill altogether.¹ This move constitutes a major effort to reflect the patterns of speech of the people.²

   Another significant feature of the sound system of Chichewa is aspiration. Plosives and affricates have aspirated counterparts and aspiration, like voicing, is phonemic in the language. The following minimal pairs may help illustrate the point:

   The consonantal system is represented in table 2.1.

Table 2.1 Consonants

	Bilabial		Labio-dental		Alveolar		Palatal		Velar
Plosives	ph				th				kh
	p	b			t	d			k	g
Nasals	m					n			ŋ
Fricatives			f	v	s	z	ʃ	ʒ
Affricates					ts	dz	tʃ	dʒ
Laterals						l
Trill						r
Semi-consonants	w						y

   In ordinary orthography, the following conventions are adopted:

	[] = ny	Chinyanja	‘Nyanja language’
	[ŋ] = ng’	ng’ombe	‘cow’
	[ŋg] = ng	ngongǒle	‘debt’
	[tʃ] = ch	chikoti	‘whip’
	[dʒ] = j	jando	‘circumcision ceremony’
	[ʒ] = zy	zyolika	be upside down (as a bat)3