Atlas of Pidgin and Creole Language Structures Online

1. Introduction ⇫ ¶

Mixed Ma’a/Mbugu is not a creole language but a mixed language. It has featured in studies on creole languages since Goodman’s (1971) article on “the strange case of Mbugu” in the collection on Pidginization and creolization of languages edited by Dell Hymes. The language figures prominently in Thomason & Kaufman’s (1988) seminal work on language contact, following Thomason’s (1983) study on the language. These analyses were based on data from earlier studies by others, dating back to Meinhof (1906). My work, Mous (2003a), is based on much more extensive fieldwork and revealed that there are actually two languages involved: a mixed language (Mixed Ma’a/Mbugu) and an ordinary unmixed Bantu language (Non-mixed Ma’a/Mbugu). The two languages share one and the same Bantu grammar (Mous 2003a: 5-6) and differ only in lexicon. This constellation of languages has a double and parallel lexicon (Mous 2003a: 10). I use the term “language” for each of the two lects even though they share one grammar. Other terms are equally problematic. The two lects differ substantially in their core vocabulary and are not mutually intelligible (unless one has learned the other lexicon); moreover, the purpose of the mixed language is to create a different language; hence the term dialect is inappropriate; the term register has the wrong connotation because it suggests a choice depending on context, which is not the case as I show below. I argued that the mixed language was at least partly created as a marker of ethnic identity, a non-Bantu identity. Truly mixed languages such as Mixed Ma’a/Mbugu are extremely rare. By a truly mixed language I mean a language in which 90 % or more of the lexicon has an origin that is different from the grammar language (see Bakker & Mous 1994: 5). Mixed Ma’a/Mbugu is a “symbiotic” mixed language; it is parasitic on a non-mixed language. A mixed language such as Michif is not symbiotic (see Bakker, this volume). I believe that Mixed Ma’a/Mbugu provides a window on how a non-symbiotic mixed language such as Michif could have arisen (see Mous 2003b). The inclusion of a chapter on a mixed language such as Mixed Ma’a/Mbugu in a book on pidgins and creoles is proper for several reasons. First of all in order to understand the differences between mixed languages, pidgins and creoles. Secondly because the double and parallel lexicon has been proposed as one of the processes in the development of a creole (see Lefebvre 1998), and thirdly in order to place pidgin and creole studies within a more general context of extreme consequences of language contact.

Ma’a/Mbugu is spoken in Tanzania, where the language is known under the name Mbugu, or Ki-Mbugu with the Swahili prefix ki to indicate the language, contrasting with Wa-Mbugu for the people. The Mbugu people speak two languages: one is a mixed language and the other is a Bantu language, and the two languages share one and the same grammar and differ only in lexicon. The speakers themselves recognize this relationship between the two languages by calling them Normal Mbugu (for the non-mixed variety) and Inner Mbugu for the mixed variety when they speak the non-mixed variety; whereas in the mixed variety, they speak about Normal Ma’á and Inner Ma’á. In the linguistic literature, the name Mbugu is sometimes confined to the non-mixed variety and Ma’a to the mixed variety. I will use the name Ma’a/Mbugu and differentiate between Mixed Ma’a/Mbugu and Non-mixed Ma’a/Mbugu where I need to differentiate the two. This article concentrates on Mixed Ma’a/Mbugu. The Mbugu people live in the Usambara Mountains, where the dominant language is Shambaa (or Ki-shambaa, Sambaa, Shambala), a Bantu language (Lang-Heinrich 1921, Besha 1989, Riedel 2009). It is difficult to estimate the number of Mixed Ma’a/Mbugu speakers. Ethnologue estimates the number of speakers to be 7,000; see Lewis (2003) for a careful justification of this estimation. I have no means to confirm this or to question it. Speaker numbers are difficult to estimate in Tanzania because the census has not included any questions about language or ethnicity for decades. Such questions have been considered politically incorrect since independence and the national unity politics of former president Nyerere. In the case of the Mbugu there is the additional complication that the people live scattered and are not confined to certain villages. The Mbugu are concentrated in roughly three areas in the Usambara Mountains: Magamba, Rangwi and Bumbuli, but also in these areas most Mbugu people have Shambaa neighbours.

2. Sociohistorical background ⇫ ¶

The history of the Mbugu can be reconstructed with difficulty using oral history and linguistic evidence. The scenario that I consider the most likely (see Mous 2003a) is one in which the Pre-Mbugu originated in Kenya, where they spoke an Eastern Cushitic language related to the now extinct Yaaku and to Dahalo. Possibly together with movements of Maasai people, the Pre-Mbugu settled in the Pare Mountains and stayed there long enough to shift to the Bantu language Pare (also known as Chasu or Chathu). The variety of Pare spoken by the Pre-Mbugu can be called Pare/Mbugu. Several groups of them moved out of the Pare Mountains and settled in the Usambara Mountains, but in different periods and following different routes. The late groups had completed the shift to Pare/Mbugu by the time they left the Pare Mountains and spoke only Non-mixed Ma’a/Mbugu.¹ Another Pre-Mbugu group came through the Maasai plains where they had been dominated by the Maasai for a while and also absorbed a Southern Cushitic group. This group developed a mixed language by adding a parallel vocabulary borrowed from Maasai and Southern Cushitic. It is not clear where and when this mixed language developed. A basis for the added parallel vocabulary was formed by the set of remembered items from the original Eastern Cushitic language that they lost. (It is conceivable that the long initiation period was a factor in the creation of the extra, mixed language. It is not uncommon that adolescents learn a secret initiation language during their initiation period. The longer the initiation period the more elaborate such a language can be. These initiation languages are parasitic on the common language and consist of an extra parallel lexicon, similar to the relationship between the mixed and the non-mixed Ma’a/Mbugu. However, nothing is known about the Mbugu initiation. Thus, the relevance of a long initiation period for the creation of the mixed language remains speculative.) The need for the mixed language was felt as an expression of their identity as non-Bantu. The Mbugu are different from the Bantu people among whom they live in terms of culture, means of economy (cattle rather than agriculture), complexion, and language (by means of the mixed language). The persistence of the non-mixed variety in addition to the mixed one can be explained by the fact that some Mbugu people speak only the non-mixed variety; yet these people are considered fully Mbugu. The mixed variety is needed for a different identity; the non-mixed variety for communication with fellow Mbugu.

3. Sociolinguistic situation ⇫ ¶

Tanzania’s policy for national unity has been quite successful in securing Swahili as a functioning national language. This entails that nearly everybody in Tanzania speaks Swahili and that children learn Swahili often even before they go to school. Swahili is the dominant language in all official and semi-official domains. Swahili dominates education even if the official language of instruction is English rather than Swahili in secondary schools. Since the Shambaa are the dominant population in the Usambara Mountains, the Shambaa language is strong in that area and there is unbalanced multilingualism: Mbugu people speak Shambaa but not the other way around. Pare (or Chasu, Asu) is yet another Bantu language spoken in the area. There is a sizeable Pare population in the Usambara Mountains as the results of a slow but steady migration from the Pare Mountains to the Usambara Mountains. The Pare language and the Non-mixed Ma’a/Mbugu language are very closely related. The grammar of Ma’a/Mbugu is very close to Pare and not to Shambaa, the dominant language of their present environment.

Since the Mbugu people speak two languages which differ only lexically, the question arises: When do they speak Non-mixed Ma’a/Mbugu and when Mixed Ma’a/Mbugu? This is a complex question which occupied me constantly while doing fieldwork. In terms of language use, the term register is inappropriate for the Mbugu situation because the choice of language variety is not determined by topic of conversation or social context. At home one of the two varieties dominates. For some families this is the mixed variety; for other families this is the unmixed variety. Some Mbugu speak only the non-mixed variety; nobody speaks only the mixed variety. There is code-switching but this is of the unmarked type (Myers-Scotton 1993). When speakers switch from one variety to the other, the switch is complete and clear. When a speaker opts for the mixed variety, s/he will opt for the mixed lexicon for (nearly) all possible choices. At one occasion when I was taping an interview in normal Mbugu, the narrator switched unconsciously to the mixed variety when a passer-by greeted him in that variety.

The remainder of this article presents a brief sketch of Mixed Ma’a/Mbugu. For more details and evidence the reader is referred to Mous (2003a), see also Mous (1994) for a similar sketch.

4. Phonology ⇫ ¶

There are five basic vowels in both the mixed and the non-mixed variety: /a, e, i, o, u/. There is no length opposition; when long vowels occur, they are disyllabic either on the basis of the fact that they have falling or rising tones or because they are the result of two morphemes coming together.

The Ma’a/Mbugu consonants are represented in Table 1. The velar fricative /x/, the lateral fricative /ɬ/, the glottal stop /ʔ/, and the prenasalized velar fricative occur only in the lexicon of the mixed language. The voiceless prenasalized sounds are common among the Bantu languages of the area (Pare and Shambaa). The lateral fricative is considered deviant in the area and it is viewed as a typical feature of the mixed variety. The Southern Cushitic languages have the lateral fricative.

The palatal stops are rare and have no prenasalized counterpart. Even though there are minimal pairs showing the phonemic status of /r/ and /l/ there is also variation between /r/ and /l/ in a limited number of lexemes. There are complexities in the spirantization of velar stops: /g/~/ɣ/ in some lexemes in both varieties (and phonemic opposition shown by minimal pairs) and /k/~/x/~/h/ in some lexemes, but only in the mixed variety.

Ma’a/Mbugu is a tone language with two register tones like the Bantu languages in the area. Contour tones are sequences of High and Low. Tones occur on vowels and on the syllabic nasal /m/ that occurs as nominal prefix.

In the rest of this article and in the database I use an orthography that has <x> for the voiceless velar fricative /x/, <gh> for the voiced velar fricative /ɣ/, <hl> for the voiceless lateral fricative /ɬ/, <ch> for the voiceless palatal stop /c/, <sh> for the voiceless palatal fricative /ʃ/, <ny> for the palatal nasal /ɲ/, and <’> for the glottal stop /ʔ/. The voiced palatal glide /j/ is written as <y>, and <j> is used for the voiced palatal stop /ɟ/. Prenasalized stops are written as digraphs: <mb> for [^mb], etc.

5. Noun phrase ⇫ ¶

As Mixed Ma’a/Mbugu has a Bantu grammar, its nouns are classified in a typical Bantu noun class system with singular/plural class pairs. Agreement within the noun phrase and on the verb (subject and object) is with the noun class. Classes are overtly marked with noun class prefixes. The noun class system is practically identical to that of Pare. The noun prefixes and the agreement prefixes are presented in Table 2. Class 14.2 is an intrusion due to loans from Swahili. Pare has the forms in 14.1, which are reflexes of Bantu noun class 14.

The singular/plural pairings are 1/2, 3/4, 5/6, 7/8, 9/10, 11/10, 11/6+11, 12/14.1, 14.1/6, 14.2/6+14.2, 15/14.1, 15/6 where 6+11 means that the class 6 prefix for plurality is added to the singular prefix of class 11 rather than replacing it as is the norm. The contents of the noun classes are like those in Pare and most other Bantu languages. Class 16 is the locative class with no lexical members; addition of the pronominal prefix hé to a noun stem renders it locative, e.g. hé-lu-kándo ‘on the wall’, cf. lu-kándo ‘wall (class 11)’. In order to be used nominally, verb stems are placed into class 15. What is of historical interest is that some nouns in the mixed variety lexicon appear to have frozen suffixes, borrowed together with the item from a suffixing source language (Cushitic or Nilotic).

Adjectives and other modifiers follow the head noun and agree in noun class with it, e.g. luhige lu-kuhlo [11.door 11-nice] ‘the nice door’. The adjectival agreement prefix is identical to the noun prefix. Reduplication of the adjective is used to express intensity. Many adjectival concepts are expressed in verbs, for example hlagha ‘be broad’, ‘aku ‘be white’. Degree is expressed by a verbal construction involving ‘to surpass, be more’, (see line 9 of the glossed text). Nominal modification requires the connecting pronoun that consists of a class-agreeing pronominal prefix and the stem á:

In the Non-mixed Ma’a/Mbugu, possessives and demonstratives agree in noun class with the head noun using the pronominal class agreement prefixes. In this respect the mixed variety is markedly different: When used attributively possessives and demonstratives are invariant and do not agree with the head noun (cf. 2a-b). When used predicatively, possessives do show agreement (cf. 2a) but demonstratives do not (cf. 2b).

Possessives distinguish person and number but not gender, in both varieties. Demonstratives make three distinctions in both varieties: ‘this (near me)’, ‘that (near you)’, ‘that (distant)’.

Word order in the noun phrase is head-initial with modifiers following the head noun: adjectives, numerals (lines 1, 3, 24 in the glossed text), genitives (line 2 in the glossed text), demonstratives (example 2a), possessives (example 2b), relative clauses (example (9). However, the demonstratives can also occur before the head noun (lines 3, 12, 13 in the glossed text).

Personal pronouns (see Table 3) need not be used since person is indicated on the verb (pro-drop); they are used for contrastive emphasis. Instead of a third person pronoun the distal demonstrative is used with noun class expression in the non-mixed variety: válá (class 2) ‘that (far away), they, them’, but the invariant distal demonstrative hú for singular and plural and all classes in the mixed variety. The personal pronouns of Non-mixed Ma’a/Mbugu are identical to those in Pare.

Adpositions are rare. There is a form mpaká ‘until’, which is a borrowing from Swahili where it is a noun meaning ‘border’ and also functions as adposition ‘until’. The genitive/connective chá for class 7 is used to render a noun phrase temporal (line 2 of the glossed text). There is an invariable form na which functions as a coordinator ‘and, with’ (lines 3, 15 of the glossed text) or it makes the phrase locative (lines 3 and 7 of the glossed text) or temporal (line 24); the invariable ní is used as a preposition ‘by’ (see example 6) and the invariable form kwa (borrowed from Swahili) is used in the function of ‘with’ (see example 5).

6. Verb phrase ⇫ ¶

The Ma’a/Mbugu verb has all the complexities of a Bantu verb. Except for the imperative, all verb forms have a subject prefix expressing person/number and class. The object can, but need not, be referred to with an object agreement prefix. Between the initial subject prefix and the pre-stem object prefix TAM prefixes occur. There are some twenty different of such “tense” prefixes, which can also occur in combination. These prefixes form a system of “tenses” together with an inflectional suffix to the verb form. The tense system is rich and contains categories such as expectational, narrative, background, conditional, habitual, hortative, present unfinished, imminence, but also past, perfect, present and the other more common categories. This complex system is identical for both varieties. Parallel lexical verbs also have the same Aktionsart; for example, the verb ‘to be tired’, which is kwaha in the mixed variety but ghata in the non-mixed one, requires the same tense prefix áa (glossed with pst) to express a present state, cf. (3).

Ma’a/Mbugu has the usual Bantu verbal derivational suffixes such a causative (7), applicative, reciprocal (5) and passive (6). The presence of a causative/intensive suffix is a feature that the language shares with Shambaa. The applicative is used to render a dative (4) or a locative (see line 4 of the glossed text) constituent object. The (negatively) affected object in (4) is expressed in the object prefix m- in the verb. The passive is often followed by the perfect, as it is in (6).

7. Sentences ⇫ ¶

The basic sentence word order is SVO; see for example line 7 in the glossed text. The initial constituent can be a topic as it is the case with ‘my name’ in the first line of the text, which is followed by a verb form with the 1sg subject only expressed in the subject prefix, followed by the name as a secondary predication or predicative complement. The next sentence begins in a time adjunct followed by the verb form (with subject prefix), the locative complement and a second time adjunct.

Locative phrases are sometimes marked as such. In line 4 the locative complement needs to be marked as such by the extra locative class prefix hé‑. In lines 3 and 7 the phrase is rendered locative by the preposition na. But inherently locative phrases like aré ‘there’, aŋá ‘home’ (line 25), aná ‘high’ (line 13, but na aná in lines 7 and 15) and the place name Kinhko in line 2 have no locative marking. There is also a locative enclitic zá ‘inside’ (lines 12 and 20).

Nominal clauses have the invariable copula ní (lines 6, 8, 16, 23 and 26) but when it is inflected for tense the inflected verb kwa ‘to be’ is used (line 18). There is a verb of possession in the mixed variety, lo (line 10) which is equal to ‘to be with’ in the non-mixed variety (line 10 would be ni-na ikoti [1sg-with coat] ‘I have a coat’).

Relative clauses may use a relative pronoun that consists of the pronominal class-agreeing prefix plus the root ó. The pronoun follows the head noun, agrees with it and marks the beginning of the relative clause. There are no differences between the two languages in this respect, cf. (9).

Alternatively the verb form can start in such a relative pronoun as in line 10. However, not all relative clauses are marked as such. Line 5 has an object, ‘thing’, which is followed by a relative clause. In this instance the relative clause is a normal clause following the head noun immediately and consisting of a normal verb form plus an adjunct.

Infinitival complements contain the nominal verb form with the class 15 noun prefix. Line 8 shows such an instance of an infinitive or nominal verb used as complement of the verb ‘to want’ (and followed by a headless relative clause with a relative pronoun referring place). Other examples of an auxiliary plus infinitive are in lines 9 and 15. The infinitives in lines 8 and 9 contain an object prefix.

Sentences can become more complex in a number of ways. Relative clauses are often used to link clauses. Line 6 of the glossed text shows such an instance with the continuation of the first clause consisting of a copula plus a headless locative relative clause, and such a headless locative relative clause can also be found in lines 8, 17 and 25. Connectives are also used: kubá ‘that’ (line 13) or kwa kubá (line 22), mirá ‘but’ (line 15), kwató ‘thus’ (lines 16, 19 and 21).

8. Lexicon ⇫ ¶

The lexicon of the mixed variety is on a par with that of the Non-mixed Ma’a/Mbugu with forms sharing semantics and formal morphological properties such as noun class membership, Aktionsart and presence of verbal derivation: If a verb is derived by a causative, it is so in both varieties, even if this causative is lexicalized, e.g. with ‘to cure’, which is ‘úti with lexicalised causative ti in the mixed variety and kija with the lexicalised causative -ija in the non-mixed variety. Example (3) shows that the parallel verbs have the same Aktionsart requiring the past marker áa for present (resultative) meaning.

The lexical items in the mixed variety come from a variety of sources: from the original East-Cushitic language (in Figure 1 indicated as “other Cushitic”), from Maasai, from a recent Southern Cushitic source (in Figure 1 indicated as “pre-Iraqw” and a few items from Bantu languages from the nearby Taita Hills in Kenya. Some deviant lexical items in the mixed variety are manipulated forms from the non-mixed variety (in Figure 1 indicated as “NMb”). Such forms are the clearest indication that the creation of the mixed language involved lexical manipulation.

Figure 1. The origin of the deviant lexicon in the mixed variety

Loans from Shambaa and Swahili are always shared by the two varieties. This shows that the formation of the mixed language was complete when the people settled in the Usambara Mountains and that the mixed language is not expanding. On the contrary, the extra lexicon is slowly crumbling. However, children acquire the mixed variety at the same time as they acquire the non-mixed one. New concepts, which are mostly loans from Shambaa or Swahili, are not represented in the mixed variety in a special way.