Publishing Partner: Cambridge University Press CUP Extra Publisher Login

New from Oxford University Press!


Style, Mediation, and Change

Edited by Janus Mortensen, Nikolas Coupland, and Jacob Thogersen

Style, Mediation, and Change "Offers a coherent view of style as a unifying concept for the sociolinguistics of talking media."

New from Cambridge University Press!


Intonation and Prosodic Structure

By Caroline Féry

Intonation and Prosodic Structure "provides a state-of-the-art survey of intonation and prosodic structure."

The LINGUIST List is dedicated to providing information on language and language analysis, and to providing the discipline of linguistics with the infrastructure necessary to function in the digital world. LINGUIST is a free resource, run by linguistics students and faculty, and supported by your donations. Please support LINGUIST List during the 2017 Fund Drive.

Review of  The Making of a Mixed Language

Reviewer: Michael W Morgan
Book Title: The Making of a Mixed Language
Book Author: Maarten Mous
Publisher: John Benjamins
Linguistic Field(s): Sociolinguistics
Subject Language(s): Mbugu
Language Family(ies): Unclassified Central Narrow Bantu
Issue Number: 15.2005

Discuss this Review
Help on Posting
Date: Sun, 4 Jul 2004 23:34:05 +0900
From: Mike Morgan
Subject: The Making of a Mixed Language: The Case of Ma'a/Mbugu

AUTHOR: Mous, Maarten
TITLE: The Making of a Mixed Language
SUBTITLE: The Case of Ma'a/Mbugu
SERIES: Creole Language Library volume 26
PUBLISHER: John Benjamins
YEAR: 2003

Michael W. Morgan, Kobe City University of Foreign Studies

The subject of mixed languages (which Myers-Scotton (2002) proposes calling 'split languages', to avoid the long-standing negative connotation in linguistics of the term 'mixed language'), has been much in vogue over the past ten years, especially in the field of contact linguistics. This is
witnessed by the discussion of mixed languages in most general survey texts on the subject of contact linguistics (e.g. Thomason and Kaufman, pp. 214-238; Sebba, pp. 264-269; Myers-Scotton, pp. 246-271; Winford, pp. 168-207), and in a growing number of articles and collections of articles
dedicated specifically to mixed languages (for example, Matras and Bakker (2003) is a collection of ten articles on the subject).

The book under review here is a very important contribution to this discussion. In it, a detailed description of Ma'a/ Mbugu is presented. This description includes both the historical, socio-cultural and socio-linguistic background, and the explication of the lexicon and grammar of the
language(s) in question. In addition to its inherent interest for those studying mixed languages, the book under review will also be of interest to those interested in Bantu linguistics in general, providing a detailed treatment of a previously under-described East African Bantu language.

The Making of a Mixed Language consists of nine chapters, plus a variety of appended matter (maps at the beginning of the book, a Mbugu-English etymological lexicon with English index at the end). Due to length limitations, in this review I will spend more time discussing the first chapters (pp. 1-93) and less on the actual language description (pp. 95-213; the description largely speaks for itself. I will also have a fair amount to say about the actual language material presented in the book, both the texts (p. 9, p. 201-213), sample sentences and to the lexicon (231-318).

Chapter One:
In this chapter, Mous introduces us to the Mbugu and their language(s). As a people they speak two languages, both referred to by the cover term Mbugu (or Mbughu). The first is Normal Mbugu (= cha kawaida), which is very closely related to Pare (Chasu) (Smith (2003, p. 7) gives it a
lexical similarity of 72 percent) and can be classified in the Shambala group of Bantu Zone G.20. (Aksenova and Toporova (1990) classify it (p. 252), following Bastin 1978, as F.35, perhaps confused by the similarity to Mbugwe, which is classified as F.34.) The second, which is the 'mixed
language' in question, is Inner Mbugu (= cha ndani), also known as Ma'a both among the people themselves and in the literature. Although Mous doesn't go into it here, Smith (2003, p. 6) states that ''Ma'a is 23 percent lexically similar to cha kawaida and 17 percent lexically similar to Pare.'' Thus, lexically at least, we are dealing with twodistinct languages.

Mbugu is spoken in roughly three pockets in the Usambara mountains of North East Tanzania, with some dialectal variation and varied preference for Inner and Normal Mbugu. Although Mous does not give us figures for the population, Smith (2003, p. 11) estimates, of a total Mbugu population
of around 7000, the number of Inner Mbugu speakers at 2870 and Normal Mbugu speakers at 3920. As Mous notes, speakers of Inner Mbugu also invariably speak Normal Mbugu. In addition, Mbugu tend to learn, in order of average proficiency, three other languages: Shambaa, Swahili and

In this chapter Mous also gives us our first glance at the nature of the relationship between Normal and Inner Mbugu, giving us a short sample text in Inner Mbugu (with interlinear Normal Mbugu version), and a discussion of the properties of the parallel lexicon. As Mous states, the
grammatical differences between Inner and Normal Mbugu are minimal; ''Inner Mbugu is in fact a parallel lexicon to the Bantu language (Normal) Mbugu'' (Mous 2003, p. 10).

In this brief overview to this parallel lexicon, Mous gives numerous sets of examples. Although the individual items in the parallel lexicons are distinct, there are important, structural correlations between the two. Thus, parallel
nouns normally belong to the same noun class; verb derivations are generally parallel in both languages. In examining the parallel lexicon, Mous feels, we must not ignore the role of conscious lexical manipulation in the emergence of Ma'a and the parallel lexicon. He states that ''[t]he strategies ... used in the formation of Inner Mbugu are common strategies for conscious lexicon creation'' (p. 12): borrowing from a wide range of languages, the use of archaic words, truncation of a word with added dummy suffix -e, substitution of remarkable sounds, metathesis, circumlocution, and semantic extension. The purpose it would seem, according to Mous is ''to make a language that sounds different for identity purposes'' (p. 12).

Mous also gives us in the first chapter a survey of earlier descriptions and discussions of Ma'a (starting in 1885 with a short word list), a description of the sources of his data (three fieldwork periods in 1992-1993), and an overview of his aims and new insights in this book. The first aim is the
presentation of new data, which as I discuss later is one of the major plusses of the book. A second aim is to show the intrinsic relationship between the mixed language and the normal Bantu language. The nature of this relationship has often been ignore, due to the two languages being studied in isolation from each other. In previous works (Mous 2001),
Mous has argued that the two languages Ma'a/Inner Mbugu and Normal Mbugu are (ethno)-registers of the same language. The relationship is one of common grammar and parallel lexicon; in terms of grammar Inner Mbugu might be said to be parasitic on Normal Mbugu (NOT Mous' characterization). A third aim of the book is the recognition of a variety of
lexical sources of the Inner Mbugu lexicon. Previous studies have assumed a simplistic, Cushitic-only source. Mous' significant contribution to our understanding, however, is to show the presence of a significant contribution from Maasai, the presence of MULTIPLE Cushitic source languages (rather than a single, proto-Ma'a language), in addition to
a variety of Bantu sources. Finally, Mous attempts to show correlations between clan settlement, language knowledge, route of migration and dialect differences in Ma'a.

Chapter Two:
In this chapter Mous gives us the historical and geographical background of the Mbugu people and language(s). This background is based on a collation of oral histories, historical information from the parallel lexicon, a
comparison of Inner Mbugu lexical items with: Shambaa/ Shambala, Pare, Mbugwe, Taita Bantu (Sagala and Davida), Maasai, Gorwaa (and Iraqw), and Old Kenyan Cushitic (Eastern Cushitic and Dahalo), and a reconstruction of a chronology through a study of lexical domains. Mous closes the chapter with a reconstruction of possible scenarios: 1) the parallel lexicon developed in the Usambaras when two groups, one
speaking Mbugu-Pare and the other speaking a mix of Maasai, Gorwaa, etc. come together and form one people due to cultural similarities. 2) The parallel lexicon developed shortly after their stay among the Maasai, with a large portion of the population already being Mbugu-Pare speakers
who remember words from an ancestral Old Kenyan Cushitic language, but with a sizeable minority from Gorwaa, fusing to form a servant group of the Maasai. This scenario is felt by Mous to be more likely, and stays close to the oral histories. 3) A once Cushitic-speaking group in the Pare mountains shifted to Pare. Part of this group left for the Maasai plains, where they formed a servant group of the Maasai and the parallel lexicon emerged shortly thereafter. There was also an influx of Gorwaa. Those that remained in the Pare mountains shifted completely, and then moved to the Usambara mountains. Later the Maasai plains group also moved to the Usambara mountains and merged with the other group there. Normal Mbugu began to influence lexical restructuring in Inner Mbugu. The presence of a majority of Normal Mbugu-speaking women among the Maasai-dominated group determined the grammar, but language loyalty to Inner Mbugu remained high.

Chapter Three:
In this chapter Mous reconstructs the linguistic history of Ma'a, based largely on the synchronic situation. What little evidence we have from the history of Ma'a/Mbugu, shows that little has changed in the language situation: earlier text (there is only one: Copland (1933/34)) and word-lists show few differences from present-day Ma'a. This is true not only of the lexicon but also of the phonology.

For example, the lateral fricative (hl), uncommon in East Africa but characteristic of Southern Cushitic, is the most characteristic sound of Ma'a. Various word lists from the period 1885-1965 use various orthographies, but in general we see consistent attestation of a(t least one) lateral fricative. This may have been inherited from a former Cushitic language: for some items Mous gives a Dahalo or Iraqw source with lateral fricative. In addition, in the earlier word lists we have evidence of a second lateral affricate (which Mous argues is due to the influence of Gorwaa/Iraqw), which at present has merged with the fricative. At present the lateral fricative is used as a sign of Ma'a identity, which has lead to introduction into lexical items without etymological laterals.

Another feature of Inner Mbugu, operating in the process of deriving lexical items from Normal Mbugu, is the truncation rule. Forms often occur with and without truncation. This variation is attested by truncated forms in the earlier documents which now exist only in untruncated form. Mous
argues that the truncation rule is another example of a typical lexical manipulation rule, and thus unsuitable as a chronological landmark in the history of Ma'a.

Remnants of non-Bantu grammar in Ma'a are of particular importance for any theorizing as to the model of emergence of the mixed language Ma'a. Elements of Cushitic grammar in Inner Mbugu are crucial for proponents of gradual bantuization (such as Thomason). As for syntax, Inner Mbugu
syntax is identical to Normal Mbugu syntax; that leaves morphology. Apart from closed classes of function words (pronouns, demonstratives, etc.), which Mous class as lexicon rather than grammar, we have few non-Bantu
structural elements. Even for the closed classes of function words, while the forms in the Ma'a parallel lexicon are clearly non-Bantu and in many cases indications are towards Cushitic, the system (e.g. the distinction of three rather than four degrees of demonstrative distance, the lack of a
gender distinction in pronouns) is clearly Bantu. In addition, none of the non-Bantu elements is fully productive in Ma'a. Also, non-Bantu elements could be either retentions or innovations; a theory of gradual bantuization would require that they be retentions - and that they be from one and the same source language (the original proto-Ma'a). Of the previously proposed Cushitic elements, Mous argues that they fall into several groups: Falsely claimed Cushitic grammar: 1) elements which are in fact Bantu in origin (e.g. amplification suffix -sha also found in Shambaa; the so-
called directional preposition he, which is in fact the normal reflex of Bantu locative class 16 in Pare as well as Normal and Inner Mbugu), 2) elements that are typological (e.g. the considerable use of adjectives when in fact the
number of adjectives is similar in Normal Mbugu, as well as in Bantu Pare; the presence of a typical Southern Cushitic five-term color system as opposed to the Bantu three-term system, when in fact the extra two terms are not necessarily color terms: the term given as 'yellow' meaning 'young,
unripe' and the term for 'green' occurring as the root for 'grass', and the former also occurs in Normal Mbugu and appear to be non-Cushitic in origin), and 3) ghost Cushitic grammar (e.g. left-over Cushitic gender marking, juxtaposition of possessor following the possessed -- neither is attested in Mous' data).

In addition, we have in Ma'a a number of other possible non-Bantu structural elements: 3) non-agreement of some modifiers (demonstratives, and possessives when they immediately follow the noun, which are deviant from the Bantu standard, but not necessarily attributable to a Cushitic source), 4) instability of certain noun-class prefixes for certain nouns
(optional application of noun prefix to some nouns of non-Bantu origin, and Mous argues that noun prefixes are the last feature added to the parallel lexicon), 5) presence of a verb 'to have' (though Mous argues that the Inner Mbugu verb lo 'to have' -- of unclear and apparently non-Cushitic
origin -- and the Normal Mbugu preposition na 'with' are not really that different as they might seem, both being used, for example, with the locative class he to express 'there is'), 6) recurrent endings in nouns and verbs (these include: a) the causative suffix -ti, which is the non-Bantu
element most closely approaching productivity, b) a number (about eight) of Non-Bantu disyllabic noun endings with high-tone -e-, some of which seem to possess restricted productivity, c) a locative clitic -za -- clearly Cushitic in origin -- parallel to the Normal Mbugu and Bantu locative suffix -eni), and 7) remnants of a Cushitic gender marker in adjectives (though the masculine marker ku- has been generalized).

Mous concludes this chapter by discussing a number of controversial issues in the linguistic history of Ma'a, and the challenge presented by mixed languages to the field of historical linguistics. The first issue is the question of: language shift or gradual bantuization. Thomason (1995)
classifies Ma'a as a Category 1 mixed language, wherein in terms of sociohistorical characteristics we have gradual long-term change, with or without language shift and in terms of linguistic processes we have a near-total replacement of L1 grammar (a gradual borrowing of Bantu grammar). She proposes (Thomason 1997) a long process of gradual, incremental bantuization with only subsequent, recent lexical manipulation. Mous, who previously (Mous 1996) opposed this view, is now at least partially in
agreement. He argues that the gradual bantuization occurred in a situation of language maintenance with extensive bilingualism. Eventually, however, unlike Thomason, Mous sees a shift to Normal Mbugu, opening the way to the extension of the parallel lexicon. Thomason presents two types of evidence for her theory: 1) fossilized non-Bantu grammatical features from an earlier Cushitic stage, and 2) chronologically ordered Bantu structural interference/ changes occurring at different times. As argued above, many
of the elements of non-Bantu grammar are not, in Mous' opinion, evidence for gradual bantuization. Also, sound changes that Thomason sees as Bantu-induced, are seen by Mous as being differently motivated by the requirements of parallel lexicon building and conscious lexical manipulation. The borrowing of the Bantu noun class system by Ma'a is a complicated phenomenon, and here Mous goes into a detailed excursus on its possible development. Of importance here, however, is the fact that whatever the scenario, when it comes to the prefix system, Ma'a is a kind
of Mbugu, and the present noun class system cannot have come about by gradual bantuization without a concomitant shift to Mbugu.

Next Mous treats the issue of the role of code-switching in
the development of Ma'a. Myers-Scotton (2002) argues that
Ma'a arose as a case of arrested turn-over in code-switching
matrix language. Code-switching does indeed occur between
Inner and Normal Mbugu (as well as with Swahili as the
embedded language), as shown by Mous' texts, and indeed on
another level Ma'a itself might be viewed as a case of code-
switching. However, the switch from Normal to Inner Mbugu
and vice versa cannot be adequately described in terms of
Myers-Scotton's Matrix Frame Model, since normally a switch
is recognized by a switch in grammatical features, while
there is no switch in grammatical framework in a switch
between Inner and Normal Mbugu. Since the grammar is all
Normal (Bantu) Mbugu, all instances of Inner Mbugu are
instances of embedded Islands, and it is impossible to
recognize when a speaker switches languages. In Mbugu we
have two systems which are NOT linguistically independent,
although they ARE separate. In the final analysis, Mous
feels that there is no evidence for such a decisive role for
code-switching in the development of Ma'a.

Finally, Mous discusses the challenge presented by mixed
languages to historical linguistics and to language
classification. In short, Mous does not see mixed languages
as a threat to the comparative method, especially when we
distinguish between grammar and lexicon, and between
inherited and borrowed lexicons. Issues of language
classification are secondary, and are determined by whether
we give preference to grammar (in which case Ma'a is Bantu)
or lexicon (in which case Ma'a is Cushitic, or, as Mous
shows, multiple in origin). Although not a threat, Ma'a and
other mixed languages do have much to offer to historical
linguistics, especially as to the nature of borrowing. Ma'a
borrowing is unusual in that: it is additive and replacive
simultaneously; the basic vocabulary is affected HEAVIER
than the marginal vocabulary; and borrowed words are often
deformed. All three of these features are the result of
lexical manipulation, which as we have seen is given pride
of position in Mous' account of the development of the Ma'a
parallel lexicon. Finally, the study of Ma'a and other mixed
languages has led to the proposal of a variety of socio-
historical models of how they emerged.

Chapter Four:
In this chapter Mous discusses the phonology of Mbugu. The
main difference between the two languages is that Inner
Mbugu possesses a few additional consonants not present in
Normal Mbugu. These are: the glottal stop /'/, the lateral
fricative /hl/, the velar fricative /x/ and the reanalyzed
velar fricative /nhx/. Aside for exceptional items (such as
Normal Mbugu a'a 'yes') these items do not occur in Normal
Mbugu. In addition, /ch/ is extremely rare in Inner Mbugu.
Otherwise, the two systems present basically a single

All voiced stops are optionally realized as implosives (a
feature common to languages of coastal East Africa).
Prenasalized stops (absent for palatial consonants) are
considered to be unitary consonants. There are a number of
phonological processes at work in Mbugu. Occasionally we see
variation between lateral approximant /l/ and the alveolar
trill /r/. Also a spirantization of velars (also found in
Pare) is not uncommon; however, though /k/ is realized
as /x/ and /x/ is realized as /h/, all three remain
phonemic. This spirantization also affects the voiced velar
(whereby /g/ becomes /gh/). All cases of spirantization
occur only in certain words.

Syllables are always open, with /y/ being the only possible
coda, and then only after /a/. The syllable structure is
typically Bantu and is true for both languages. It can be
summarized as: (N).(C(G))V.(C(G)V (where N = syllabic
nasal /m/ nominal prefix, C = consonant, G = glide, V =
vowel). The minimal word is CV; in Ma'a many verbs consist
of only CV. The typical shape of the verb root is, however,
CVC, and given the many affixes, an inflected verb might
take the form: CV-CV-CVC-VC-V. The canonic shape of nouns is
prefix + stem; thus: (C)V+CV(CV(CV)). Exceptions are class
9/10 nouns with nasal prefix (disappearing before consonant
stems). In class 9/10, monosyllabic nouns occur only in

Both Normal and Inner Mbugu are tonal, with two basic tones:
High and Low. Tone is distinctive on every syllable of a
word, and there are no restrictions on patterns, except that
verb roots are either High (H) or Low (L) regardless of the
number of syllables. Tone attestation is, however, not even,
and not the same in the two languages. For example, in
disyllabic noun roots, in both Inner and Normal Mbugu HH is
the least attested type, but it is relatively more common in
Normal Mbugu. On the other end, while it is not uncommon in
Normal Mbugu, LH is the overwhelming type for Inner Mbugu.
This is due to the common use Inner Mbugu makes of -e plus L
(LL)H tone pattern to transpose Normal Mbugu verbs. Also
monosyllabic noun roots must be High in Inner Mbugu, but not
in Normal Mbugu. Verbal roots in Normal Mbugu are either C
or CVC, and H and L are equally distributed in the former
and L only slightly more common in the latter. In Inner
Mbugu, on the other hand, the structure of verbal roots is
much more complicated, as is the distribution of tones. In
general, ALL monosyllabic verbal roots are H (a restriction
which also applies to monosyllabic noun roots). For
disyllabic roots, HL and LL are well represented, as would
be expected for a standard Bantu verbal system. There are a
few exceptional HH roots, and LH roots with final -a. This
leaves a number of LH verbal rots not ending in -a, which
are atypical for Bantu.

Chapter Five:
This chapter discusses the verbal system of Mbugu. The
differences between Normal and Inner Mbugu are negligible,
and the system is typically Bantu, as are the forms
themselves (many occurring in Pare or Shambaa). The verb
consists of a stem with a series of inflectional prefixes,
and an inflectional vowel suffix. Inflectional prefixes
consist of the normal Bantu categories: subject prefix
(agreement for 1sg, 1pl, 2sg, 2pl, plus each of the 16 noun
classes, and obligatory except for imperatives and
infinitives), one or two object prefixes (all the above
agreement classes, plus reflexive), a variety of 'tense'
prefixes (20 prefixes, including categories of tense (Mous
provides attestation for 18 tenses, plus a few combinations
of tense prefixes) proper, mood, polarity, etc., and
occurring between the subject prefix and the object prefix
(es)), negative prefix (occurring initially before the
subject prefix) and a high-tone prefix (marking the boundary
of predicated domain). Mous' exposition of verb forms is
based on his corpus, and as such is probably incomplete when
it comes to inflectional paradigms. It does however present
a rich, very Bantu-like, picture of the Mbugu verb system.

In this chapter Mous also discusses the copula and the verbs
'to be'. A copula is required in both Normal and Inner
Mbugu, unlike some other Bantu languages such as Swahili.
Also, both Inner and Normal Mbugu possess the verb 'to be'
and locative verbs 'to be at a given place', although the
two languages are not entirely parallel in this area (as we
have seen in the discussion above of the verb 'to have' in
Inner Mbugu).

Mous ends the chapter with an exposition of seven verbal
derivation suffixes present in Mbugu, all of which are
common Bantu derivational suffixes (intensive, causative,
applicative, stative, reciprocal, reciprocal-applicative,
and passive), as well as a few non-productive extensions.

Chapter Six:
In this chapter Mous discusses the noun class system in
Mbugu. Both Inner and Normal Mbugu possess a Bantu noun
class system, wherein every noun falls into one of sixteen
noun classes (mostly occurring in singular-plural pairs),
defined according to agreement phenomena. Noun classes are
largely overt, with membership in each class shown by the
presence of a specific class prefix. The noun class system
is identical to that of Pare, except for class 14.2 which
are due to Swahili influence. In general, class membership
is the same in both varieties of Mbugu, with equivalent
nouns of the Inner Mbugu parallel lexicon occupying the same
noun class as their Normal Mbugu (and Pare) pairs.

Chapter Seven:
This chapter discusses adjectives and other nominal
modifiers (genitive and relative pronouns, possessives,
demonstratives, and quantifiers). Most of these follow the
head noun and show agreement with the noun class.
Demonstratives and quantifiers may also come before the head
noun. We also have cases without agreement, as with certain
recent loans, higher numerals, etc. As in other Bantu
languages, adjectives are a closed and extremely limited
set; Mous' data attests only 37 items, each with Normal and
Inner Mbugu forms. Genetive and relative pronouns form
parallel systems which are common to both Normal and Inner
Mbugu, generally the pronominal prefix plus -a (for
genitive) or -o (for relative). The systems of possessives
in Inner and Normal Mbugu differ somewhat in that the Normal
Mbugu system is typically Bantu (pronominal prefix plus
lexical root for the possessor), while in Inner Mbugu
agreement is NOT shown when following the head noun
attributively. They agree when used predicatively or when
used attributively but separated from the head noun by
another non-agreeing modifier (e.g. demonstrative). The
systems of demonstratives are remarkably different between
Inner and Normal Mbugu. Both systems have three degrees of
deixis, but while Normal Mbugu has separate forms for each
noun agreement class, Inner Mbugu possesses only one form
for each degree of deixis. As for quantifiers, while one
item ('any') is shared by both Inner and Normal Mbugu, the
other three items ('other', 'all', and 'which') are
adjectives in Inner Mbugu, not quantifiers and thus show
different agreement properties.

Chapter Eight:
In this chapter Mous treats invariables and other words:
that is, the hodge-podge of leftover words. The first group
includes prepositions, and all members are shared by the two
language varieties. The next class is independent
invariables (i.e. invariables which can form a phrase by
themselves), and in general they resemble adverbs. These
include expressions of time, location, manner, etc.
Generally there are separate items in the two lexicons,
though a few items are shared by both Inner and Normal
Mbugu. Next comes the class of independent personal
pronouns, used only for emphasis. Although the forms are
different in the two language varieties, the systems
manifest the same distinctions. Higher numerals form the
next group of invariable modifiers.

Chapter Nine:
The final chapter deals with notes on syntax, code-switching
and texts. The notes on syntax are limited to a few remarks
and do not represent a full syntactic analysis. Mous also
includes a section on set expressions (greetings) in the two
language varieties. The section on code-switching, though
relatively short, is of great interest and is accompanied by
a text with discussion. Finally we have four texts: two in
Inner Mbugu, one in Normal Mbugu, and one in a mix of the

End Matter:
The book closes with Notes, References, a 2300-plus item
Etymological Lexicon, English Index to Lexicon and Index.

One thing missing from Mous' monograph, a lack which will
perhaps be especially hard felt by those new to the notion
of mixed languages, is a discussion of the general notion of
what a mixed language is, a comparison with other candidate
mixed languages, and a discussion of what makes Ma'a/Mbugu
unique. Ma'a/Inner Mbugu stands out as remarkably different
from the other prototypical mixed languages usually
discussed in the survey literature on the subject, Mednyj
Aleut and Michif (both Category 2 mixed languages according
to Thomason (1995)). Mous does, however, discuss some of the
issues of mixed languages in general when he discusses the
various models for the emergence and development of Ma'a.

On the other hand, the present book is a wealth of data and
information concerning the language under question. In fact
it is a complete descriptive grammar of not only the mixed
language Ma'a/Inner Mbugu, but also simultaneously the
''matrix'' language Normal Mbugu, showing the close
relationship between the languages, something that Mous
argues strongly for.

In terms of data, three things stand out: Texts, Sample
sentences, and Lexicon. As for texts, the six texts (seven
if we include the equivalent Normal Mbugu text provided
parallel to the first Inner Mbugu text) presented in the
present book represent the only published Mbugu texts
besides Copland (1933/34), which publishes one Ma'a (pure
Inner Mbugu) text. Texts vary in length from 30 words to
almost 300 words, with a slight predominance of Inner Mbugu
text material. As for sample sentences, Mous provides a
wealth of illustrative examples for all of his points. Over
all, Mous provides us with an equal picture of the two
language varieties, with 124 Inner Mbugu sentences and 122
Normal Mbugu sentences (plus 2 sentences with both). For any
given point, Mous seems to try to provide material from both
language varieties. Exceptional lapses do, however, occur,
as when all three example sentences for logical sequence
-ta-, or both example sentences for imperfective conditional
-he- are from Inner Mbugu, or all three example sentences
for the default tense are from Normal Mbugu.

Finally we have the Etymological Lexicon, which is an
extensive list of more than 2300 items, and is accompanied
by an English index to aid in looking up items. Again I
spot-checked the lexicon and cross-checked it with the
English index. Sampling the Lexicon for items beginning with
d-, m-, sh-, o- and z-, we get the following picture: 162
items shared by Inner and Normal Mbugu, 213 items in Inner
Mbugu and 176 items from Normal Mbugu (plus 4 items without
indication of language variety). The slight predominance of
items occurring only in Inner Mbugu over those occurring
only in Normal Mbugu probably results from the corpus-based
nature of Mous' data. As we saw above, his texts and sample
sentences are also slightly imbalanced in favor of Inner
Mbugu. Extrapolating to the total corpus of 2300 items, this
gives us 690 (29 percent) shared items, 900 (40 percent)
Inner Mbugu, and 740 (31 percent) Normal Mbugu. Counter
checking against the English index (taking the 145 entries
'day before yesterday' to 'fog'), we get similar figures,
indicating minimal sampling error. Looked at another way, if
these figures are representative, a given Inner Mbugu text
should contain about 60 percent uniquely Inner Mbugu lexical
items and 40 percent items shared with Normal Mbugu.

A final word about the book: this is a high-quality book, as
might be expected from its hefty price. I did, however, note
a few typos. For example, on p. 96 the glottal stop /'/ is
missing from the chart of Mbugu consonants (it should be in
italics, as it occurs only in Inner Mbugu). I didn't keep
notes on other typos, but I did catch about 8-10 in the
course of reading the text.

Aksenova, I.S. and I.N. Toporova (1990). Vvedenie v
Bantuistiku: Imja, Glagol. Nauka.

Copland, B.D. (1933/34). A Note on the Origin of the Mbugu
with a Text. Zeitschrift fuer Eingeborenen-Sprachen 24:241-

Lewis, Scott (2003). Mbugu/Ma'a Project (= SIL Electronic
Survey Reports SILESR 2004-001.). SIL International.

Matras, Yaron & Peter Bakker, eds. (2003). The Mixed
Language Debate: Theoretical and Empirical Advances (=
Trends in Linguistics. Studies and Monographs 145). Mouton
de Gruyter.

Mous, Maarten (1996). Was There Ever a Southern Cushitic
Language Pre-Ma'a? In C. Griefnouw-Mewis and R.M. Voigt
(eds.) Cushitic and Omotic Languages: Proceedings of the
Third International Symposium, pp. 201-211. Ruediger Koeppe.

Mous, Maarten (2001). Ma'a as an Ethnoregister of Mbugu. In
Derek Nurse (ed.) Historical Language Change in Africa (=
Sprache und Geschichte in Afrika (SUGIA) 16/17), pp. 293-
320. Ruediger Koeppe.

Myers-Scotton, Carol (2002). Contact Linguistics: Bilingual
Encounters and Grammatical Outcomes. Oxford University

Sebba, Mark (1997). Contact Languages: Pidgins and Creoles.
St. Martin's Press.

Thomason, Sarah G. (1995). Language mixture: ordinary
processes, extraordinary results. In Carmen Silva-Corvalan
(ed.) Spanish in Four Continents: Studies in Language
Contact ad Bilingualism, 15-33. Georgetown University Press.

Thomason, Sarah G. (1997). Ma'a (Mbugu). In Sarah G.
Thomason (ed.) Contact Languages: A Wider Perspective, pp.
469-487. John Benjamins.

Thomason, Sarah G. and Terrence Kaufman (1988). Language
Contact, Creolization, and Genetic Linguistics. University
of California Press.

Winford, Donald (2003). An Introduction to Contact
Linguistics (= Language in Society, volume 33). Blackwell
Michael W. Morgan holds a doctorate in Slavic
Linguistics, teaches languages (and, when he gets the
chance, linguistics), and conducts research on historical
and comparative linguistics, language typology, sign
languages, various Indo-European languages, Basa Bali (like Mbugu possibly a language historically with ethno-registers), and language education.