Table of Contents

Introduction. In this chapter, I analyse a category of compound words in Cinyanja that are made up of a verb and noun which are grammatically or semantically related as exemplified below. In the examples above, the compounds comprise two or more words; cipha ‘killer’ and dzuwa ‘sun in (1a), mponda ‘one who steps’ and matiki ‘tickeys’ in (1b) and cigona ‘it sleeps’ and mubawa ‘in the bar’ in (1c). These verb-noun (henceforth V-N) compound words as shown in (1) above, despite having a complex structure, are genuine compound words which include phrasal material and inflectional morphology. In this chapter, I claim that Cinyanja allows “reification” of lexical complex expressions which can be re-introduced into the syntax as unanalyzable roots and undergo further morpho-syntactic operations. The chapter first discusses the structure of VN compound words in Cinyanja followed by a review of literature on noun class (NC) prefixes and agreement. The phrasal structure of Type 2 and Type 3 compounds is discussed. This is followed by a discussion of V-N compounds based on the syntactic context that externally, they behave like nouns. Based on the Lexical integrity Hypothesis (LIH) I show that, although the internal structure of the compounds is syntactic, it is not accessible by syntax. Finally, I propose that Cinyanja V-N compounds can be reified as roots and re-enter the syntactic derivation as unanalysable units. Finally, a conclusion is provided. 3.1 V-N compoundsVN compounds form the bulk of complex expressions in Cinyanja and are particularly interesting because they share a similar structure with synthetic compound words in English. The compounds have a deverbal noun as the first member followed by an internal argument. The term ‘V-N compounds’ will be used here as it anticipates the findings of section 3.3 where it is shown that they behave as genuine compounds. In this study, V-N compounds are classified into types 1-3. For each type, I show both the singular and plural forms. The singular forms are all classified as (a) while the plural forms are classified as (b). 3.1.1 Type 1: Nominal part of the compound is a non-derived NType 1 V-N compounds comprise a verb and a non-derived noun in the nominal part. The compounds in this category can be schematically presented as below;Type 1: CM + V + FV + N The data below is representative of Type 1 compound words.In (2-7), the plural compounds categorised as (b), have a plural marker prefixed to the entire compound word. Apart from example (2) which denotes something inanimate, and has its plural in class 4, examples (4-8) refer to humans and have their plural in class 2. In this type of VN compound words, the V constituent is made up of the verb stem and the final vowel -a, which Nurse (2008) refers to as a neutral vowel. Additionally, the nouns in type 1 compounds do not take a class prefix in the singular or plural form. They have a zero prefix and resemble the famous English example of truck-driver (see Lieber, 2009c; Padrosa, 2010). In most cases, the nouns in this category are mass nouns.3.1.2 Type 2 VN compounds.In the Type 2 VN compounds, the nominal part of the compounds includes NC prefix. This can be summarised as;Type 2: CM+V+FV+CM+NSome of the data that form this type of compounds are presented below.From the compounds listed above, a number of general observations can be drawn. The first is that the verb in the compound is either followed by a plural noun as in (9-14 and 16-17) or a mass noun as in (15). Concerning the interpretation of these compounds, some display mixed properties of being transparent and having lexicalised meanings as in examples (13) mpalamatabwa ‘plank scraper’ and example (14) mkondamaungu ‘pumpkin lover’. In these compounds, semantic meaning can be construed from the compound elements, thus, the compounds are endocentric, with ‘plank scraper’ being the hyponym of scraper and ‘pumpkin lover’ been a hyponym of lover. In terms of the NC prefix attached to these compounds, it is considered the head of the compound because it attaches to the V+N unit (see Mphasha, 2006). This is similar to the English synthetic compounds if the compound is [truck +drive] + er. It can be argued that the affix –er is the head and suffix of the category N The V element in type 2 VN compounds is bimorphemic and comprises the verb root and a final vowel (FV) –a which is mostly attached to verbs in these compound words. In rare cases, the verb may take an applicative morpheme –ir- between the root and the final vowel as shown in example (16). This shows that the verb can take inflectional and derivational morphology.The noun element in VN compounds presented above thematically corresponds to the direct object of the verb. In general, most Type 2 nouns are bimorphemic. They have number/gender marking represented by a class prefix and a stem. Most of the nouns in this category refer to humans and objects while a small number may refer to animate organs as in (10) mawere ‘breasts’ in citsekamawere ‘last born child’.In VN compounds, the noun object is the internal argument of the verb. What is interesting about this type of compound words is that usually, the internal argument takes the plural form contrary to Booij’s (2005) argument about the position of inflectional affixes. According to Booij, plural marking of the compound should be on the entire compound not on a particular compound constituent, as a result, it must be externally located. The examples of Cinyanja compounds are in tandem with Booij’s argument because plural marking of the compounds is additive. The plural marker is added to the entire compound as shown in (18a) and (19a) as opposed to subtraction of the singular NC prefix as in (18b) and (19b). 3.1.3 Type 3: Nominal part of the compound is a locativeThe type 3 VN compounds are typically composed of a verb and a locative in the nominal part of the compound. Thus, Type 3 VN compounds have the following structure;Type3: CM +V + FV+ LOC+ (CM) NThe data in this category is as presented below.The compounds in this category refer to humans, objects and things. The majority of the compounds have class 16 or class 18 locatives. While examples where a class 17 locative prefix ku- may precede a noun are possible, I have not come across any among the compound words in this study. It is observed that compounds with the initial prefix m- and refer to humans have the class 2 plural prefix –a added to the compound word as a whole. Crucially, the compounds with the class 7 prefix ci- that refer to humans such as cigonamubawa‘ drunk’ in (23) also take class 2 plural prefix –a, not their official class 8 plural prefix zi-. However, compound words with the initial prefix ci- such as cipondamuthengo ‘consultation fee’ in (21) and citsegulapakamwa ‘money paid to start marriage negotiations’ in (22) that refer to things and objects rather than to humans, neither take class 8 plural prefix zi- nor class 2 plural prefix –a.The verb element in Type 3 compounds is also bimorphemic and consists of the verb root and a final vowel (FV). The final vowel takes the shape of -a in almost all the VN compounds just like in Type 1 and Type 2 compounds. Another notable feature of the verb in these compounds is that it does not inflect for tense, and the final vowel seems to be a categoriser for verbs. The verb is followed by a locative in the nominal part. To the compound combination, an appropriate prefix is attached by a class marker that is deemed appropriate for each of the nominal compounds. This suggests a syntactic structure of VP in these compounds.The noun constituent either has an overt NC prefix as in (24) m-khala-pa-mw-ala, ‘stone sitter’ or lacks an overt NC prefix as in (23) ci-gona-m-bawa ‘a drunk’. The absence of the class prefix can be attributed to the characteristic property of class 5 nouns in Cinyanja and has nothing to do with the compounds. Most of the nouns in this category refer to humans and objects and are usually singular, for example, cigonamubawa, ‘drunk, musesapakhomo, ‘maid’ or mass noun as in (21) cipondamuthengo ‘‘consultation fee’ To conclude, for Type 1, a traditional compound analysis would be available, where the verb and the noun simply form a complex morphological entity which can then be subject to further derivations/inflections. However, in order to understand the properties of Type 2 and 3 VN compounds, an understanding of the nature of NC prefixes in Bantu and Cinyanja specifically is required. This will be discussed in the next section.3.2 NC prefixes and agreementIn Bantu languages, nouns are classified according to number and gender denoted by the NC prefixes. The gender class is mainly associated with semantic properties of nouns as animate, size, plural, quality or location. Prefixation of an NC marker to a stem, on the other hand, helps to determine the agreement properties of a syntactic construct, which is an inflectional feature and the change in meaning of the noun stem which is a derivational feature. This section discusses Bantu noun class prefixes, whether they are inflectional or derivational. I argue that the noun class prefixes in compound words are inflectional and they control agreement like other noun class prefixes attached to individual nouns.3.2.1 NC prefixes: derivation or inflectional?This section is meant to explore the nature of the noun class prefix and eventually determine whether it is inflectional or derivational. Generally, Bantu noun class prefixes express grammatical information of number and gender. In some Bantu languages such as isiZulu, the NC prefix is optionally preceded by an augment vowel. This augment mirrors the vowel in the NC prefix as in the isiZulu example umuntu ‘person’, izipho ‘gifts’ (Taraldesen, 2010; Halpert, 2012:3). In most Bantu languages, compound constituents are also usually preceded by an NC marker. This is shown in the Cinyanja examples below.In (32a) above, the initial NC prefix ci- marks the compound word ciphamaso ‘hypocrite’ as a member of noun class 7 while in (b), the initial NC prefix m- marks the compound word mpalamatabwa ‘carpenter’ as a member of noun class 1.