Home / Archive / 2015 / Paul Solomiac: Phonologie et morphosyntaxe du dzùùngoo
Document Actions



Solomiac describes the Western Mande language Dzùùngoo spoken by the ethnic group of the Dzùùn, which are part of a complex known as Samogo. Dzùùngoo is spoken in two areas in southwestern Burkina Faso, which form islands of Mande languages almost entirely surrounded by speakers of Gur languages (the Senufo languages Sicite, Nanerige, the southern Central-Gur language Curama or Turka, the not further classified Gur language Northern Tusian) and the Kru language Siamou or Seme. Additionally Bolon, a Western Mande language of the Manding cluster is another neighbor of Dzùùngoo, and speakers of the Atlantic language Fulfulde live near Dzùùn settlements. The Manding variety Jula is used as lingua franca all over the Kénédougou and Comoé provinces in which the Dzùùn settlements are situated. The neighborhood of so many languages is likely to entail different scenarios of contact, hence the present grammar is not only interesting to Mandeists but also to scholars working on the other language families and those whose main subject is the study of language contact.


Dzùùngoo is classified as one of the varieties that form together the Duun language cluster within the north-ouest subbranch of the Western Mande languages. Its closest relatives are the languages Duungo or Duungoma and Banka, both spoken in Mali, and Kpeego, a language spoken by a professional and endogamous group of smiths who live among different ethnic groups such as the Turka, Tusian and Siamou. Dzùùngoo is spoken by some 12,000 speakers living in about 10 villages, and counts several varieties. The present grammar is based on one of them, the variety of the town of Samogohiri.


The first chapter contains, next to a description of the geographical, sociolinguistic and economic situation of the speakers of Dzùùngoo also an estimation of the vitality of the language. Based on the criteria suggested by the UNESCO group of experts working on language endangerment the author attests it a good vitality, especially because of a persistent transmission of the language to the next generation.


Chapters 2 to 5 deal with the phonology of the language, introducing the syllable structure and presenting the sounds which are possible as syllable onset and those that are possible as nucleus. Dzùùngoo has seven oral and five nasal vocalic phonemes, for the mid-close nasal vowels do not have phoneme status. It is interesting that Dzùùngoo shares the reduced repertoire of nasal vowels with the Senufo languages of the area, for instance Supyire and Senar [1], whereas its closer Western Mande relatives have 7 nasal vowels. Additionally to nasal phonemes the language presents a floating or autosegmental nasal element N that surfaces as nasal or entails nasalization only in some particular contexts. Thus, the floating N has an assimilating effect on a following flap [ɾ], which is realized as [n]; it nasalizes a preceding oral vowel, when it is found at a morpheme border and when the first consonant of the second morpheme is different from a stop; when the floating N precedes a stop, it is realized as a homorganic prenasal segment.

The author observes that there are three different vowel lengths: extreme short, short and long. Short vowels are opposed to long vowels, the latter resulting from the loss of intervocalic consonants [l], [d], [ɾ], [g], or [ɣ], as shown by the comparison of several Dzùùngoo items with Jula cognates. Dzùùngoo is in this respect a typical western Mande language, for vowel lengthening due to the loss of an intervocalic consonant is an ongoing process in related languages such as Bambara (Manding), where wágati ‘time’ is alternatively pronounced wáati, and wála ~ wáli ‘or’ can be articulated wáa. However, there is no opposition between extreme short and short vowels, since the former occur only in two precise environments: firstly, when the identical vowels precede and follow ɾ and n, and secondly, in items with the shape CVɾa, CVna and CVnã, where V represents an extreme short [i] when the preceding consonant is a non-back consonant (other than velar or labiovelar) and an extreme short [ũ], when the word-initial consonant is a velar or a labiovelar.


Different morphophonological phenomena are very interesting in Dzùùngoo, among others the traces of a vowel harmony. It consists in the assimilation of the height degree of the root vowel when the imperfective morpheme is suffixed (téí ‘ask, pfv’, térō ‘ask, ipfv’, but vjɛ̀ì ‘be afraid, pfv’, vjɛ̀rɔ̄ ‘be afraid, ipfv’), and the partial assimilation of back vowels when the high front vowel suffix –i, representing the nominalizing morpheme, is added (ɔ + i > ɛi as in tɔ̀ ‘know’ + -i > twɛ̀j ‘knowing’). What is interesting in the last example is the labialization of the initial consonant as an instance of secondary articulation which is the trace of the feature [+back] of the root vowel. From a historical point of view the secondary articulation coincides with the loss of an intervocalic C which it is still present in the contact language Jula, where the nominalizing suffix is –li ~ -ni ~ -ri. The secondary articulation is one of the features Dzùùngoo and its close relatives share with the surrounding Senufo languages (Kastenholz 2002).

The consonant system presents some pecularities, for instance the presence of the labiovelar voiced and voiceless stops /gb/ and /kp/, further the labiovelar nasal /ŋm/. Next to the alveolar stops /t/ and /d/ and the palatal stops /c/ and /ɉ/ the system shows the affricates /ts/ and /dz/.


In chapter 6 the author presents the tonal system of the language, using Keith Snider’s Register Tier Theory (RTT). This theory makes use of a register tier, a pitch tier and a tonal node tire in order to account for such phenomena as downstep and a floating high tone. Three tones are identified as tonemes: High, Mid and Low, and contrary to the usual rule postulated in autosegmental theories, the author posits a right to left assignment of tones to tone bearing units (moras), adding some rules. According to the first rule the last (i.e. the leftmost) tone is assigned to all remaining moras, if the number of moras exceeds the number of tones. According to the second rule, a non-initial (i.e. another than the rightmost) tone can be associatied to a mora when the moras have identical extreme short vowels separated by a [ɾ] or [n], or when their vowel is an extreme short vowel V in CVɾa, CVna and CVnã structures. Both rules account for a case like gìrìbī ‘sulfur’, where the rightmost M is assigned to the last mora and the remaining L to the first two moras. However the case of fàkūrū ‘hyena’ shows the contrary assignment, since the M tone is the initial one in the right-to-left assignment logic. A discussion of such contradictory cases would have been welcome. The overall impression is that the system is rather complicated, since a left-to-right assignment takes place in particular circumstances such as compounding and derivation. Unfortunately, in this part the author presents the different rules without illustrating them step by step with examples, so that it is really difficult to follow.


The morphology of Dzùùngoo is identified in chapter 7 as situated in the middle of a continuum ranging from the agglutinating type to the isolating type of languages.

The author describes the lexical categories found in the language in the voluminous chapter 8. After an introduction to the chapter he first describes minor categories such as adpositions, conjunctions, particles and interjections, and then presents the more important categories of adjectives, different sorts of nominals, adverbs, nouns, and verbs. In Mande languages the distinction of nouns and verbs is not an easy task. In Dzùùngoo, morphosyntactic and distributional criteria help do distinguish the major categories. Thus, verbs are defined as the items that show an aspectual distinction in form of perfective and imperfective suffixes added to their basic form and that can be predicates of a clause. Nouns and other nominal categories take the definite and plural morphemes.


Adpositions comprise mainly postpositions and some prepositions. The postpositions form two groups: lexical postpositions and clitics, which are phonologically bound to the preceding nominal. Even if borrowed items are found in almost all categories, the category of conjunctions shows the highest portion of words originating in other languages, especially in Jula. Solomiac illustrates (p.122) that this borrowing entails a serious restructuring of certain constructions. Thus, Dzùùngoo conditionals are made by means of predicate particles in post-subject position, without any kind of conjunction. Under the influence of the contact language Jula the conditional conjunction combined with the indicative replaces the Dzùùngoo construction, but sometimes both markings, i.e. the conjunction (of Jula origin) and the original Dzùùngoo post-subject particle are used altogether in the same sentence. Thus contact between languages of the same family lead to the same observation as contact between Gur and Mande, for instance the Senufo languages Minyanka in Mali or Senar in Burkina Faso with Bambara or Jula: conjunctions are likely to be borrowed, and double marking is attested next to structures of the contact language entirely mirrored in the borrowing language (cf. Dombrowsky-Hahn 1999, Traoré 2015).


Under the heading ‘particles’ the author subsumes different sorts of morphemes: clause final particles marking negation and interrogation, discourse particles indicating intensity, emphasis and doubt, etc., TAM morphemes, copulas used in nonverbal predication, and morphemes required in dependent clauses of complex sentences, such as for instance the conditional predicate particles.

Adjectives are characterized semantically by denoting qualities, they differ from verbs by their inability to take verbal flexional morphemes and from nouns by the impossibility to function as head of a noun phrase. However, like in Manding adjectives do not show a uniform distribution. Some function only as predicates, some can only be modifiers, another group can function as both. However, there are two subgroups of the latter according to the way they are used as predicates: the adjectives of one group are connected to the subject by means of the same copula as the one occurring in predicate nominals and predicate locatives, whereas those of the other group do not need such a support, thus resembling verbal predicates.


Dzùùngoo has an endangered numeral system, which seems to be mixed vigesimal, sexagesimal and octogesimal, for, next to the bases 1-10, the bases 20 (corresponding to the lexeme for ‘person’), 40, 60, 80 and 800 are used to form more complex numbers applying the operations of addition, multiplication and substraction. Due to its seeming complexity it is being given up in favor of the expanding Jula decimal system, a process that was accelerated by literacy activities. Although similar systems exist in the languages of the region (for instance, the Senufo language Supyire has the bases 1-5, 10, 20, 80 and 400, cf. Carlson (1994); Western Karaboro has a similar system, with 200 instead of 80 as a base), no effort was made nationwide to maintain them.

The pronominal system shows some peculiarities. Thus, Dzùùngoo has a logophoric pronoun. Reflexive pronouns are distinct for the first and second person singular, all other persons show one and the same form of reflexive pronoun. A reciprocal morpheme is added to the reflexive pronoun to express reciprocal events.


Like other Mande languages and languages of other families spoken in the region, Dzùùngoo has different types of adverbs. Among them there are ideophones and onomatopoeia, some of which have the same meaning as the verbs they are associated with. The author further introduces another group of adverbs that he calls ‘quasi nouns’. It assembles items that can be used as nouns and fixed postpositional phrases, mostly with place and time adverbial meaning.

The section on nouns and verbs considers derivation and compounding of the respective category. Verbs are presented according to their valency.


Next to sections on the definite – indefinite distinction, qualification and coordination of nouns, chapter 9, entitled ‘nominal operations’ contains a thorough analysis of the genitive construction. Like Bambara, Dzùùngoo shows a distinction between inalienable and alienable relation. In both the order is possessor – possessum, the former without any connecting morpheme, the latter with the connecting morpheme –re, which resembles the clitic postposition –re used to mark locative predicates, among others also those expressing possession and volition. The distribution is very similar to the marked and unmarked genitive constructions in Bambara. Inherent possession such as body parts or bodily emanations is unmarked, possessed concrete objects such as artifacts are encoded in the marked construction; visible and audible emanations (shadow, words) of a non-human are unmarked, those of a human are marked; in locative relations the function of the place determines the choice of the construction the unmarked construction is used when the place is the possessor (as in the yams of Saraba), the marked one when the place is the possessum (the village of the monkeys). In interpersonal relations such factors as domination, reciprocity, kinship or alliance relation are decisive. A distinction vis-à-vis the genitive construction in Bambara concerns alliance relations (i. e. one’s husband, one’s wife, one’s nephew): in Manding they are usually encoded in an unmarked genitive construction, except for one’s wife as determined item, which can occur in either construction depending on the respective variety. In Dzùùngoo all alliance relations are expressed as an alienable relation, showing the connective. An entire subsection deals with the relations of nominalized verbs and their arguments.


In the first part of chapter 10 headed “verbal operations” Solomiac describes the verb phrase in the basic clause structure NPs p NPo V as discontinuous, since it comprises a postsubject particle p and the verb V itself, separated by an object noun phrase in the case of transitive verbs. The verb V occurs in one of three different forms: a basic stem, a perfective form and an imperfective form, the latter two showing different suffixes. They combine with different particles p (among them Ø) to yield various tense, aspect and modality categories. A particle p can further be followed by one of three auxiliairies seemingly originating from verbs: bè´ ‘go’ and nà´ ’come’, and the verb bà´ ’become’, producing the categories of conclusive or functioning as directionals. Some remarks are in order in this context. First, the perfective category shows the perfective form of the verb in the affirmative but the basic verb form in the negative. Dzùùngoo is not unique to show such a distribution. Thus, according to Winkelmann (1998) in Tiéfo (Cɛfɔ), an endangered Gur language spoken in southwestern Burkina where three different verb forms are attested, too, the basic and not the perfective verb form is used in the negative perfective. Winkelmann cites other similar cases found by Jungraithmayr (1988) and explains the lacking perfective marking by the fact that an activity that has not taken place is inevitably not accomplished and hence not marked as accomplished (= perfective).


Some phenomena are embarrassing, and Solomiac’s labeling seems unfortunate. Thus, the so called unaccomplished (‘inachevé’) category displays a suffix added to the perfective verb suffix. The author describes it as a perfect that lacks the feature of present relevance, however up to now the two suffix meanings seem to be contradictory and would need further investigation. Another unsure category is the one called ‘inchoative’ (pp.138, 244). This term is normally associated with the entry into a state or a beginning activity. The particle under consideration gáá originates in the verb gaa ‘look for’ and conveys the meaning of ‘be about to’, ‘be on the point of’ or ‘having almost V-ed’ and the cited Dzùùngoo example (translated as ... le vieux des gens de Dondaré a failli retourner ‘the old man of the inhabitants of Dondaré almost returned’) do not corroborate the hypothesis of entry into an activity, for the activity itself is averted. The terms commonly used to designate a morpheme expressing a situation that is about to take place but that can be diverted or impeded are ‘prospective’ (Comrie 1976:64) or ‘proximative’ (Heine & Kuteva 2002:311). Similar instances of grammaticalization of a verb with a similar meaning into a proximative auxiliary are among others attested in Bambara:




sà ...






‘They are on the point of dying ...’. (Dumestre 2011:831)


The second part of chapter 10 is about marking of verb phrases in the dependent clauses of complex sentences such as the conditional, participial and other clauses; its third part about coding of main events in narrative texts.

Chapters 11 and 12 deal with nonverbal and verbal predication respectively. Nonverbal predicates are divided according to the category that holds the position of predicate: predicate nominal, predicate adjective, predicate locative and ‘predicate copula’. Predicate nominals in equative clauses (illustrated with n° 363, p.257, Sibiri is a farmer) display a Ø copula when it has present time reference, but it makes use of the auxiliary bà´ when the clause has another temporal reference. When the predicate is a pronoun, a focus particle seems to be obligatory. There are different groups of predicate adjectives: one group does not show a copula, others show the copula ni  [2] which is also present in predicate locatives. Under the heading of ‘predicate copula’ the author subsumes identifying (copula ni) and existential clauses (deictic copula ). Locative clauses are also the pattern encoding predicative possession and mental events, for instance the expression of desire (n° 376, p.260).


In the chapter on verbal clauses one phenomenon merits particular mention. The author considers kúú ‘reverse, cover’ to be a trivalent verb, which takes two objects. In an active clause one object fills the position before, the other the position after the verb. Solomiac considers the object status of both to be evidenced by the possibility of each to function as the subject of a passive clause.
















subject <agent>

object1 <patient>

object2 <beneficiary>

‘The hunter put his hands on his face.’ (n° 400, p.268)

In a passive clause the beneficiary of the active clause becomes the subject-patient, and the patient of the active clause becomes an instrument encoded as postpositional adjunct marked by means of the postposition .









subject <patient>

postpositional adjunct <instrument>

‘A face is covered with hands.’ (n° 401, p.268)

The second possibility displays the patient of the active clause encoded as the subject of the passive clause, hereby putting a particular emphasis on the instrument.









subject <instrument>

object <patient>

‘One covers his face with hands.’ (n° 401, p.268)

According to the author’s analysis in this passive clause (3) the postposition is separated from the noun that complements it by the predicate particle . He stresses that instrumental roles marked by the same postposition show further an unusual behavior, for they can follow the predicate particle p and precede the object when they are topicalized.


I would like to add two remarks in this context. First, it would seem more appropriate to talk about the examples (1), (2), (3) as instances of different possibilities of argument realization of the verb kúú ‘reverse, cover’ than about passivization alone. In fact, the author shows it partially, when he labels the patient argument of (1) the instrument of (2) and (3). Second, there is another possibility to explain the “unusual position” of the postposition which is, in the author’s analysis, separated from its complement by the predicate particle p (the narrative in the examples). I propose to consider as a postposition in some examples in the data, for instance (2) above, but as a verb V1 with the meaning ‘use, make use of’ in a serial verb construction (SVC) in others, for instance (3) with kúú as V2. The literal translation of this example is then: ‘Hands are used to cover one’s face’. Example (4), displaying also as V1 illustrates more obviously Solomiac’s criteria of SVCs (p.285).












V1 make.use.of

object of V2


‘... the granary was filled with hyena’s excrements’

Thus, according to the criteria determined for the SVCs in Dzùùngoo, only the V2 tsyɛ̄ɔ shows a flexional morpheme, here the retrospective, whereas the V1 remains in its basic form. There is only one subject, Fàkūrū byáán hyena’s excrements. The meaning of the SVC, considered as a unit, is different from each of the constituting items; the reader unfamiliar with Dzùùngoo does not know if exists as an independent verb, but in (4) the V1 adds an additional participant. Similar constructions are attested in several languages in the region, among others the Bambara multiverb constructions with the verb kɛ́  [3] and the Supyire SVCs with the verb taha (Carlson 1994, chapter 8, section The Dzùùngoo verb was probably reinterpreted as postposition, for instruments are encoded by the postposition la in other contexts (n° 395 p.266). As a result the verb in SVCs coexists with the postposition ká, both covering the same function.


In chapter 13 the author describes marked structures, among which he considers interrogative and imperative clauses, negation and processes of focusing and thematization. Negation in Dzùùngoo presents some peculiarities within Western Mande. Generally a sentence contains two negation markers: one in post-subject position and a final negation marker. According to the tense, aspect and modality, the negation markers in post-subject position are either suppletive forms of the affirmative morphemes or they are specifically negative morphemes added to the morphemes occurring in affirmative clauses. Different particles are possible as final negation markers: next to a general negation morpheme waa there are others, which have the meaning of ‘never’ or ‘not at all’. Dzùùngoo and several other languages of the Duun cluster share the double negation marking with some of the Senufo languages of the region, for instance Senar and Supyire, and Kastenholz (2002) assumes that Dzùùngoo and its close relatives borrowed it from Senufo. Like in the Senufo languages that show a comparable final negation marking (cf. Dombrowsky-Hahn 2006), it is not always easy to determine the scope of negation or, vice-versa, to know where it is to be placed in complex sentences, if only one of the clauses is to be negated.

Complex structures are described in the last chapter preceding the conclusive one. Next to relative clauses, subordinated and coordinated constructions this chapter encloses serial verb constructions, although, as the author stresses, they present a unique predicate opposed to the other, multiclause structures.

Solomiac’s thorough analysis of a small Mande language based on a huge amount of natural data is an important and highly welcome contribution to Mande linguistics and to the linguistic study of languages in a multilingual area.


Carlson, Robert J. 1994

A grammar of Supyire. Berlin: de Gruyter

Comrie, Bernard 1976

Aspect. An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press

Dombrowsky-Hahn, Klaudia 1999

Phénomènes de contact entre les langues minyanka et bambara (Sud du Mali). Köln: Rüdiger Köppe

Dombrowsky-Hahn, Klaudia 2006

'La négation dans les énoncés simples et complexes en kar (Senufo)'. Journal of African Languages and Linguistics 27,2:127-153

Dumestre, Gérard 2011

Dictionnaire bambara - français suivi d'un index abrégé français - bambara. Paris: Karthala

Heine, Bernd and Tania Kuteva 2002

World lexicon of grammaticalization. Cambridge: Cambridge University Press

Jungraithmayr, Hermann 1988

'Zur Negation in afrikanischen Sprachen'. In: Studia Indogermanica et Slavica. Festgabe für Werner Thomas zum 65. Geburtstag, ed. by Peter Kosta, pp.485-496. München: Sagner

Kastenholz, Raimund 2002

'"Samogo" language islands, and Mande-Senufo (Gur) interference phenomena'. In: Lexical and Structural Diffusion. Interplay of Internal and External Factors of Language Development in the West African Sahel, ed. by Robert Nicolaï and Petr Zima, pp.91-109. Nice: Faculté des Lettres, Arts et Sciences humaines; Prague: Faculté des Etudes Humaines

Traoré, Daouda 2015

Le senar (langue senufo du Burkina Faso): éléments de description et d'influence du jula véhiculaire dans un contexte de contact de langues. Göttingen: Cuvillier

Winkelmann, Kerstin 1998

Die Sprache der Cɛfɔ von Daramandugu (Burkina Faso). Frankfurt am Main: Berichte des Sonderforschungsbereichs 268 "Kulturentwicklung und Sprachgeschichte im Naturraum Westafrikanische Savanne" Band 11. Frankfurt am Main: Johann Wolfgang Goethe Universität

[1] Carlson (1994), Traoré (2015).

[2] A copula ni is known from many Niger-Congo languages; there is evidence that traces of a proto-form *ni are present in several Senufo languages in the identifier morphemes that vary according to the nominal class of the referent (for Supyire cf. Carlson 1994).

[3] We can hypothesize that Dzùùngoo and Bambara kɛ́ are cognates, although has probably lost some of its functions. kɛ́ is among others used in Bambara as independent verb with the meaning ‘do, make’. Here is an illustration of its function as a verb in a multiverb predicate that introduces an instrument participant (active clause):






















‘Pliers, we use them to take iron out of the fire.’ (Dumestre 2011:504)




There are no comments yet.

Do you have any additions or comments?

Submit comment.