A typological portrait of Mano, Southern Mande

3.1 Phoneme inventory

The consonant and vowel inventories of Mano are provided in Tables 1 and 2, respectively. Three features of the phonemic inventories deserve special attention: labial-velar consonants, the syllabic nasal and nasal consonants.

Labial-velar consonants are identified by Güldemann (2008a: 156–158) as a typical feature for languages of the North-Western area of sub-Saharan Africa (which is also referred to as the Macro-Sudan Belt) and are common to Mande languages but extremely rare outside Africa. Mano has two of these consonants: voiced/gb/(gbɛ́ɛ́ ‘to cry’), and voiceless/kp/(kpɛ̄ɛ̄ ‘neck’).

In many languages of Sub-Saharan Africa, such as Baoulé (Atlantic), nasal consonants are not fully phonologized, so that nasal stops are considered allophones of oral stops (Creissels 1994: 132). The same applies to nasal vowels, which in many languages do not correspond to nasal phonemes, but rather are part of combinations of the type VN or VNV (Creissels 1994). Mano, among many other Mande languages, has nasal vowels on the phonological level. In contrast, the phonological status of nasal consonants in Mande is somewhat debatable. For half of the Southern Mande languages, such as Goo (Aplonova and Vydrin 2017), different varieties of Dan (Vydrin 2017; Vydrin and Kessegbeu 2008), or Mwan (Perekhvalskaya 2017), it is claimed that nasal stops are not phonologized and receive their nasal quality in nasal contexts, especially in combinations that have nasal vowels. In Mano, as in Tura (Idiatov and Aplonova 2017) and Beng (Paperno 2014), evidence suggests that nasal vowels do not trigger the nasalization of the consonants on the surface level, but rather that nasal consonants are phonologized. For a discussion of this, see Khachaturyan (2015: 10–11).

The Mano system also includes a syllabic nasal /ŋ/ which can function as the nucleus of a syllable. It appears in the position of the only stem syllable or of two identical stem syllables ŋ(ŋ) (ŋ̄, first person singular pronoun of the past series, ŋ̀ŋ́, first person singular negative auxiliary) or in the CVŋ position, where the syllabic nasal is preceded by a vowel (wōŋ̄ ‘tail’, pèŋ̀ ‘forge’). The syllabic nasal always has a distinctive tone: mɛ̀ŋ̀ ‘dream’, mɛ̀ŋ́ ‘behind’. In the former case, ŋ(ŋ), it is realized as a nasal stop that assimilates to the following consonant according to the place of articulation and, in certain cases, causes assimilation by nasality: ŋ̄ ló [n̄n̄ó] ‘I went’. In the CVŋ position, it takes the form of a nasal vowel: a back vowel [ṵ] if the previous vowel is back or central (wóŋ́ [wóṵ́] ‘tail’) or a front vowel [ḭ] if the previous vowel is front (pèŋ̀ [pèḭ̀] ‘forge’). Syllabic nasals are not uncommon in Sub-Saharan African languages: Creissels mentions a few, including Swahili, Igbo and some varieties of Manding (Creissels 1994: 57–64).

For the representation of Mande phonology, I adopt the orthographic notation that is common for Mande (or, broader, African) linguistics: y stands for /j/, ŋw stands for /w̰/, kw for k^w, gw for g^w, kp for k͡p and gb for g͡b. Ŋ with a tonal diacritic corresponds to the syllabic nasal and ŋ without a tonal diacritic corresponds to a nasal velar sonorant.

3.2 Suprasegmental phonology

Most of the languages of Sub-Saharan Africa and all Mande languages have tone (with the exception of some Mande varieties in contact with non-tonal languages, such as some varieties of Jalonke, Lüpke and Storch 2013). Mano has three level tones, low (kɛ̀ɛ̀ ‘year’), mid (kɛ̄ ‘in order to’), and high (kɛ́í ‘refuse’). Extra-high tone is marginal in the system, as it appears only in ideophones, expressive adverbs, and interjections (sɔ̋ɛ̋ 'all and all', ɛ̰̋ 'interjection', etc.) In contrast to Manding languages and some other languages with two tones, including Kpelle, downstep is not attested in Mano. Rich tonal inventories of three or more tones are typical for the Macro-Sudan belt, so Mano is no exception here (Hyman et al. 2018). The tone-bearing unit is a vowel or a syllabic nasal; long vowels and diphthongs are considered combinations of phonemes. Complex tonal contours appear only on VV sequences. All tonal combinations are possible (see Table 3). Unlike some Southwestern Mande languages, such as Kpelle (Konoshenko 2014), no tonal spreading occurs in Mano, and grammatical tonal marking is hosted by specific words rather than phrases. Furthermore, Mano does not have stress accent, which is often (but not always) the case for tonal languages (Hyman 2009). Utterance-level prosodic operations include the lowering of the tone on final vowels in affirmative utterances and a general downdrift.

The syllable seems to be of limited analytical salience (on that point, see Hyman 2011). By contrast, it is often claimed that Mande languages, including Mano and other Southern Mande languages, such as Guro (Kuznetsova 2007), possess “metric feet”, mono- or polymoraic units which have the following properties. First, specific phonotactic restrictions apply: vowel harmony, restrictions on the internal structure and on the inventory of intervocalic consonants. The domain of application of grammatical tones is the metric foot. This means that the rules of tonal assignment in grammatical contexts apply to the entire metric foot, rather than to a syllable or a mora. Finally, as a matter of statistical observation, non-derived lexemes tend to possess only one metric foot. Metric feet appear to be a special case of distributional asymmetries between the stem-initial and non-stem-initial syllables that are widely attested in the languages of the Macro-Sudan Belt (Hyman et al. 2018).

Metric feet in Mano can only occur as one of the following structures: V (ē ‘3sg.pst’), CV(ŋ) (sɔ̄ ‘cloth’; sɔ̄ŋ́ ‘price’), CVV (sɔ̰́ɔ̰́ ‘teeth’, ɓɛ́ì ‘friend’) and CVCV(ŋ) (ɓɛ̀lɛ̄ ‘respect’, ɓɛ̀lɛ̄ŋ̀ ‘tree sp.’). In the CVV structures, a combination of back and front vowels is disallowed, as well as nasal and non-nasal vowels, and vowels should be identical in CVCV structures. The intervocalic consonant should be either /n/, /l/, or /ŋ/,/l/ is not allowed with nasal consonants /m/, /n/, /ɲ/, /ŋw/ in the initial position (yɛ̀lɛ̀ ‘manner’, wòlò ‘tear out’) while /n/ is not allowed with /ɓ/, /l/, /y/ and /w/ in the initial position (mɛ̄nɛ̄ ‘snake’, ŋwánà ‘truth’), all other consonants in the initial position freely combine with both /l/ and /n/ in the intervocalic position. As a rule, grammatical tone applies only to one metric foot. Thus, the verb pɛ́lɛ́ ‘to wash’, which consists of one metric foot, according to the aforementioned parameters (CVCV structure; identical vowels; /l/ as an intervocalic consonant), takes the form with low tones on both vowels in the imperfective: pɛ̀lɛ̀ ‘wash:ipfv’ (1). However, the verb lóní ‘to count’ (2) consists of two metric feet of the CV structure, because the vowels are non-identical which violates the rule of vowel harmony. The grammatical tonal replacement applies accordingly: in the imperfective form, the grammatical tone applies only to the second vowel, which then takes the mid tone, lónī ‘count:ipfv’. The same pattern of change applies to verbs of CV structure with lexical high tone: ɓí ‘to shine’ and ɓī ‘shine.ipfv’(3).

An interesting feature of Mano phonotactics is that combinations of back and front vowels within one metric foot, which are generally disallowed, are nevertheless possible after an (optionally) labialized velar consonant: k^(w)ɔ̀ī ‘firewood’, k^(w)ōí ‘behind’, and k^(w)ɔ̀nɛ̀ ‘eggplant’. (The last word, k^(w)ɔ̀nɛ̀ ‘eggplant’, could potentially be analyzed as containing two monosyllabic feet. And yet the intervocalic consonant /n/ is characteristic of metric feet with the CVCV structure, so the interpretation in terms of a single metric foot is also possible.) It is unclear why a velar consonant should provide an exceptional environment for the violation of the general rule of vowel harmony. In the CVV combinations of this type, V₂ is usually /i/, with one exception: k^(w)ɔ̀ɛ́ ‘near’.^[2] During fast speech, these combinations are pronounced with both front vowels and a strong labialization: [k^wɛ̀ī] ‘firewood’, [k^wēí] ‘behind’, [k^wɛ̀nɛ̀] ‘eggplant’, and [k^wɛ̀ɛ́] ‘near’. This type of process is especially salient in the Guinean dialects of Mano which are in contact with Guinean Kpelle in which a similar process is attested (Konoshenko 2017: 290). In Liberian Mano and Liberian Kpelle, the default variant seems to be with a back – front combination and the labialization of the velar consonant is much weaker (e. g., in Liberian Mano [kɔ̀ī] ‘firewood’), which suggests a micro-areal phenomenon.

4.1 Pronominal paradigms

A distinctive feature of most Southern and Southwestern Mande languages is their rich pronominal paradigms that include several syntactic series of personal pronouns, as well as several series of auxiliaries which index a subject's person and number. Personal pronouns typically distinguish between non-subject pronouns that occur in most argument positions (an inalienable possessor, 6a, direct object, 6b, or an argument of postposition), possessive pronouns that express alienable possession (ex. 6c) and emphatic pronouns used in non-argument positions (ex. 6d). Besides these three positions, Mano also has high tone pronouns, a functional parallel of the high tone nominal forms mentioned above, which are used with demonstratives and with focalization markers (example 6e), and the inclusory pronouns (example 6f), the most typologically unusual of all, which will be discussed in Section 4.2. Mano does not have pronouns that are used in the syntactic position of the subject but has auxiliaries agreeing in person and number with the subject (see Section 6.2). Personal pronouns in Mano are listed in Table A1 of the appendix.

4.2 Inclusory constructions and pronouns

Inclusory constructions are a type of conjunction strategy that consists of an adjunction of two sets. These involve the whole set of participants, or the superset, which are usually represented by an inclusory pronominal, and a subset of participants that is expressed by a pronoun or a full NP (Lichtenberk 2000; Haspelmath 2004: 25–26), as in Mano below:

In (7), kò ‘we’ refers to the entire set of participants, while ɓī ‘you’ refers to a subset. Another subset, implicitly referred to by kò ‘we’, is either the speaker, or the speaker with some other participants. The literal meaning of the construction is ‘we [including] you (sg.)’ and two possible interpretations: either you (sg.) and I, or you (sg.) and we. The whole group of those whose picture is to be taken is indexed again by the non-subject pronoun kō in kō fòtóò ‘our picture’.

Several Southern and Southwestern Mande languages have inclusory constructions. Although inclusory pronouns are especially common in Austronesian languages (Lichtenberk 2000; Bril 2004; Lee 2011) and languages from several language families spoken in Australia (Singer 2001), they are also attested in European languages, such as French dialects and some Slavic languages, including Polish and Russian (Tesnière 1951). What makes the Mande family typologically outstanding is that inclusory pronouns in some of the Mande languages, including Mano, are represented by a separate series^[3] (see Table A1 of the appendix). This property does not exist outside this language family (for more detail on inclusory constructions in Mande, see Khachaturyan 2019).

4.3 Expression of plurality

In Mano, when a noun that has a non-human referent is unmarked for number, it can correspond to one or many entities in the world. Thus, the form gbá̰ ‘dog’ can denote one or several dogs. The number of referents can be specified by specialized number determiners, number words (gbá̰ vɔ̀ dog pl ‘dogs’) or numerals (gbá̰ dōó ‘one dog’, gbá̰ pèèlɛ̄ ‘two dogs’). The typological literature claims that this unmarked form expresses general number (Corbett 2000: 9–10). In contrast, for human referents, an unmarked form is usually interpreted as singular (with the exception of non-referential noun phrases): gwɛ̄ɛ̄kòlò ‘old man’; NOT ‘old men’; but gwɛ̄ɛ̄kòlò vɔ̀ ‘old men’. Mano pronouns also distinguish between singular and plural forms. This means that Mano referents split into two categories depending on their position on the Animacy Hierarchy (Smith-Stark 1974): pronouns, as well as noun phrases that denote humans distinguish between singular and plural, while an unmarked form of non-human referents has a general interpretation.

Plurality in Mano is most often expressed by the number words vɔ̀ and nì. The vɔ̀ marker combines with all referents except kinship terms and pronouns and its semantic interpretation is additive plurality (‘X’ + ‘X’ + ‘X’…). Nì combines with all types of referents without restriction, but the semantic interpretation is not the same for all nouns (see below). This means that the Animacy Hierarchy again plays a role, the division occurring in a different position of the hierarchy, separating kin from other human referents (on that point, see Dahl and Koptjevskaja-Tamm 2001).

The nì marker is typologically rather rare semantically, as it combines the values of associative plurality and distributive plurality, which together I call “non-additive plurality”. The former indexes a set of objects that includes the referent of the host noun and some other objects connected to it (Daniel 2000: 4). Typically, the associative marker is used with human referents (Sèé nì 'Se and his family/friends'), but it also can be used with other types of referents (yíízólóyàà nì 'jaundice and other sicknesses', see also example 8b). The associative marker is typically used with proper names, pronouns and kinship terms, but can also be used with other nouns. Distributive plurality, in turn, indexes “…the separation of members of a group, whether entities, events, qualities or locations. Each is considered distinct in space, sort or time” (Corbett 2000: 111). In the distributive value, the marker is often combined with mass nouns (9b), but it can also be used with discrete entities (mōtōò nì ‘motorbikes here and there’). While both values, associative and distributive, are known from descriptive and typological literature, their combination in one marker appears to be cross-linguistically rare.

The following examples contrast additive and non-additive plurality. Example (8a) illustrates the usage of nɔ̀ɔ̀ ‘catfish’ with vɔ̀ triggering an additive reading. Example (8b) shows the usage of the same noun with nì and this triggers an associative reading. Example (9a) illustrates the incapability of the mass noun wèŋ̄ ‘salt’ to combine with the additive marker vɔ̀. Finally, in (9b), the same noun combines with the marker nì, triggering a distributive reading.

While number words are the most common means to express plurality, a few nouns express plurality (with or without a number word) by an irregular inflection and this can be considered a case of suppletion (Veselinova 2006; Corbett 2009). These nouns include names of humans and kinship terms with a most general meaning (mī/mīī ‘person’, mīā/mīà ‘persons’; lēē ‘woman’, lōà/lōō ‘women’; gɔ̰̄ ‘man’, gɔ̰̄à̰ ‘men’; nɛ́/nɛ́fú ‘child’, nɔ́ɔ̀ /nɔ́ɔ̀ɓé ‘children’; and mìì ‘parent’, mìà ‘parents’) and the noun pɛ̄ ‘thing(s)’, pɔ̄ɔ̀/pɔ̄ɔ̀ɓé ‘things’. Suppletion displays some patterning: the labialization of mid-low or mid-high vowels and an additional formant: -à/ -à̰ or ɓé; some part of the nominal stem remains unvaried – minimally, the initial consonant. As a consequence, nominal suppletion in Mano could be characterized as “weak” (Dressler 1985) or “partial” (Corbett 2009).

Finally, plurality can also be expressed in adjectives. Of Mano adjectives,^[4] approximately one-half can undergo reduplication and this includes most non-derived adjectives. While non-reduplicated adjectives can freely combine with the singular/general (10a) or plural forms of nouns (10b), some reduplicated adjectives combine only with the plural forms of those nouns that distinguish between singular and plural forms (10c versus 10d) and trigger a plural reading when combined with nouns that do not distinguish between such forms, which is illustrated in (11). These reduplicated forms express nominal plurality. For some other adjectives, the value of reduplication is to express quality intensity (12) or both intensity and plurality (13).

While adjective reduplication is not uncommon cross-linguistically, its typical value is the intensity of the quality that is expressed by the adjective (as in Hindi, Singh 2005; or Turkish, Sofu 2005). However, adjective reduplication in Southern and certain Southwestern Mande languages typically expresses plurality, a semantic property that is proper to noun phrases and may be expressed in adjectives when they are targets of agreement (Vydrine 2004). The plural value of adjective reduplication may be an areal feature that Southern Mande languages share with Kwa: the same phenomenon is observed in Akan (Winkler and Obeng 2003) or Baoulé (Creissels and Kouadio 1977: 274).^[5] Moreover, the adjective may be the only locus of the expression of plurality. This is because Mande, like Mano in particular, has few nouns that have plural forms. Thus, in cases such as (11b), the apparent controller of the agreement is unmarked for number.

In addition to reduplication, plurality in adjectives can be expressed by a tonal form. This applies to those adjectives that are reduplicated but whose non-reduplicated counterpart does not exist. Thus, the adjective in (14), yɔ́sɔ́yɔ̀sɔ̀ 'dirty', does not exist in a non-reduplicated form (*yɔ́sɔ́). For adjectives such as the one in (14), the basic tonal pattern that is High–Low changes to High–High, so that the overall morphological process could be considered a case of partial reduplication: of a tonal pattern but not a segmental pattern.

In addition, three adjectives are pluralia tantum, which only occur with plural referents: dòló ‘similar.pl’, kɛ̀nɛ́ɛ̀ ‘small.pl’, and kélékélé ‘small.pl’ (15). Other adjectives occur in Mano that mean ‘small’: ɓéŋ́, tííkpé, nɛ́ɛ́, and péétí, so the adjectives kɛ̀nɛ́ɛ̀ and kélékélé could be considered suppletive plural forms.^[6] The adjective gbɔ́kò ‘1. numerous, 2. elderly, 3. fat’, when reduplicated, retains only one meaning: gbɔ́kògbɔ́kò ‘fat.pl’. The adjective gbṵ́ṵ̀ has the meanings ‘1. other, 2. certain, 3. different.pl’, it is in the latter meaning that it behaves similar to a plurale tantum.

Thus, number agreement applies between nouns and adjectives. Those nouns that do not distinguish between number forms, as in (11, 14) and plural-marked nouns (10b, 10d), combine with both plural-marked and plural-unmarked adjectives. In contrast, when nouns do distinguish between number forms and are unmarked for plurality, they combine only with plural-unmarked adjectives (10a, 10c and 15).

4.4 Inalienable vs alienable possession

Mano resembles most other Mande languages in that it has two types of possessive construction, inalienable and alienable. In the case of a pronominal possessor, a non-subject pronoun is selected, as in (16a). Example (16b) illustrates a construction that consists of a possessor juxtaposed to the possessum in absence of any formal marking. The second type consists of the possessor expressed by a possessive pronoun (17a), optionally preceded by a noun phrase (17b). Non-subject and possessive pronouns are clearly marked with a tonal contrast: non-subject pronouns have the mid-tone (ŋ̄) and possessive pronouns have low tone (ŋ̀).

An alienable possessor can be added to combinations of the former type:

The existence of the distinction between alienable and inalienable possession is far from being cross-linguistically unusual (Nichols 1988; 1992; Dahl and Koptjevskaja-Tamm 1998), but several aspects of the distinction need to be clarified.

First, the class of inalienably possessed nouns in Mano is much broader than the typological prototype, which includes body parts and kinship terms, spatial terms, and some significant parts of objects and basic objects of material culture, such as cattle (Nichols 1988). Mano nouns that are inalienably possessed include, of course, kinship terms (dàā ‘father’, lēkè ‘younger sibling), but also other terms of social relations that are not related to kinship: bèá ‘contemporary, person of the same age group’, tɔ́ɔ́má ‘namesake’, gèlè ‘enemy’, gɔ̰̀á̰ ‘equal’, kélá ‘witness’, and yókò ‘enemy’. By comparison, the noun nɛ́ ‘child’ is alienably possessed. Body parts, excretions, physiological liquids are also inalienably possessed (zò ‘heart’, sèē ‘nail’, and líà ‘blood’), together with some parts of objects (tō ‘source’, gbá̰á̰ ‘ceiling’, and gbɔ̀lɔ́ ‘frame’). Not surprisingly, spatial terms, some of which give rise to postpositions, also belong to the inalienably possessed class (yàlà ‘place’, wì ‘lower part’). This class also includes the names of measures, groups and assemblies (ŋwūnà ‘type’, kùù ‘number’, kɔ̀pāà ‘handful’, and kpíní ‘envelope’), attributes, sicknesses (ɲɔ́nɔ́ ‘taste’, tɔ̀nɔ̄ ‘benefit’, fàŋá ‘force’, zùō ‘paralysis’, tɔ́ ‘name’, and tɛ̰́ɛ̰́ ‘totem’), property, constant or provisory (fɔ́nɔ́ ‘salary’, kɛ̄pɛ̀ ‘heritage’, sāmā ‘gift’, sālà ‘salary’, wɔ̀lɔ̀ ‘sleeping place’, and kɔ́nɔ́ ‘food’), some names of dishes (ɓáá ‘(rice) porridge’, fìlí ‘pilaf’) and names that designate the age or sex of an animal (gɔ̰̄ ‘male’, mū ‘female’, and ɲɔ̀nɔ́ ‘young’). Finally, another subclass of inalienably possessed nouns includes names of actions that occur in N + V complex predicates: súò ‘calling (of someone)’, súò kɛ̄ ‘call sb.’.^[7] Even so, nouns that denote objects of material culture are never inalienably possessed.

While it is well established that the class of inalienably possessed nouns can include some non-prototypical items, in Mano, every third inalienably possessed noun could be considered non-prototypical. Thus, the repartition between the two classes is often arbitrary. For example, the noun fìlí ‘pilaf’ denotes a dish consisting of rice mixed with spices and either fish or meat. The noun is inalienably possessed: wìì fìlí ‘pilaf with meat’. However, the word ɓòō ‘sauce’, which is always distinguished by the type of the main ingredient, is alienably possessed. Should the main ingredient be specified, a different construction is used that has a head in the low-tone construct form: béìlɛ́ɛ́ɓòò < cassava + leaf + sauce:cstr > ‘cassava leaf sauce’.

Another interesting feature of the grammar of possession in Mano is that some inalienably possessed nouns are obligatorily possessed, that is, they always occur with a possessor (approximately 75% of all inalienably possessed nouns), and some are optionally possessed, at least in certain contexts.^[8] The distribution into these two classes also seems arbitrary. Let us consider two synonyms, sālà and fɔ́nɔ́ ‘salary’. Both are inalienably possessed: à sālà, à fɔ́nɔ́ ‘her salary’. Nonetheless, sālà can be used without a possessor, while fɔ́nɔ́ cannot:

5.1 Word order

Mano and other Mande languages observe a fixed S–AUX–O–V–X word order (21, 22), with the direct object preceding the verb, while all other arguments, expressed by postpositional phrases, follow it (Creissels 2005; on the SOVX word order as an areal feature of the Macro-Sudan Belt, see Güldemann 2008a). Unlike many languages of Sub-Saharan Africa, the Mande languages do not have double-object constructions in which both the recipient and the theme are expressed as patients of a prototypically transitive verb (Creissels 2018).

The Mande word order S–AUX–O–V–X probably emerged from a reanalysis of prior nominalizations which changed the typologically more common VOX word order to OVX (the analysis is presented in Nikitina 2011 along with an overview of alternative viewpoints). According to the hypothesis proposed by Nikitina, nominalization constructions were reinterpreted as verb phrases as part of finite constructions, and in particular, the patients expressed by the inalienable possessors preceding deverbal nouns were reinterpreted as direct objects. This hypothesis explains the striking parallel between the syntax of verb phrases and possessive nouns phrases, the fact that postpositional phrases display a distinct behavior in nominalization (see below), and the aforementioned absence of double-object constructions. Reanalysis of prior nominalizations into finite clauses is not uncommon cross-linguistically (Yap et al. 2011); in particular, the emergence of a typologically rare word order as a result of reanalysis of nominalizations was claimed for Cariban languages (Gildea 1998).

5.2 Nominalization

Mano has three nominalization forms: zero-marked, gerund (-à, the vowel assimilates to the preceding vowel) and a low-tone form. They are used in all positions that are typical of nouns, such as the arguments of verbs and postpositions and may be used with derivational morphology, such as the adjectival suffix –zɛ̀, as illustrated in (30d). Unlike the former two forms, the low-tone form is used whenever the verb is nominalized with a postverbal argument adjacent to it. The distribution of nominalization forms in different contexts is presented in Khachaturyan (2015: 145–156).

The argument expression in nominalization roughly follows the Possessive-Accusative type suggested by Koptjevskaja-Tamm (1993): while the P argument is expressed precisely the same as the direct object in finite constructions (25, 26), the S argument is expressed like an inalienable possessor (24) and the marking of the A argument is identical to the marking of the alienable possessor (26).

The situation is complicated by the fact that in all contexts, the marking of the direct object is identical to the marking of the inalienable possessor (6a-6b). This blurs the distinction between the Possessive-Accusative and the Sentential type in the case of the direct object. The complication can be explained by the aforementioned hypothesis that verb phrases in Mande derive from nominalizations.

Mano is reminiscent of Austronesian languages and in particular of the Tolai language, in terms of the possessive split in the expression of subjects and the identical coding of inalienable possessors and direct objects (Foley 2014). Unlike Tolai, in which the possessive split is motivated by the degree of agentivity of the subject, it is motivated in Mano purely by argument roles (on a similar subject, see van Lier and van Rijn 2013). In addition, in Mano nominalizations, intransitive subjects are expressed in the same manner as direct objects. In Southwestern Mande, the situation is similar: the intransitive subjects are treated as direct objects in certain grammatical constructions, including nominalizations (Konoshenko 2017: 333). This situation is analyzed as an instance of (split) ergative/absolutive alignment (Vydrin 2011).

5.3 Postverbal arguments and clause-level nominalization

Postverbal and preverbal arguments do not have an identical syntactic relationship with the verb, which can be observed by their behavior in nominalizations. When the verb is nominalized, the nominalized constituent always includes the direct object when the verb is transitive, but not necessarily the oblique argument:

The verb in the example above, túó ‘frighten, to be afraid’, has a postpositional argument headed by the postposition lɛ̀ɛ̄. The postpositional phrase contains the nominalized verb phrase yíí ɓōò ‘pump water’, in which the verb is marked by the suffix -à. The verb ɓō, in turn, also has a postpositional argument, gùlù yí ‘from the pit (lit.: in the pit)’, but this argument is situated outside the nominalized verb phrase and is separated from it by the postposition lɛ̀ɛ̄.

Similar behavior of postpositional phrases in Wan, another South Mande language, led Nikitina (2009) to conclude that postpositional phrases in Wan are situated in the adjunct position to the IP (see also Nikitina 2018). A crucial difference between Mano and Wan is that in Wan, postpositional phrases are typically extraposed from the nominalized verb phrase,^[9] while in Mano, a verb can be nominalized together with its postpositional argument:

It is important to note that in (28), which is similar to (27) in all respects but the position of the postpositional phrase, the form of the verb ɓō ‘take off’ is also different. While it occurs in the gerund form in (27), it occurs in (28) in the special low-tone nominalization form that is used in nominalizations with oblique arguments. Dan-Gweetaa and Kla-Dan, languages closely related to Mano belonging to the Dan language cluster, also have a low-tone form that functions similarly (on Dan-Gweetaa, see Vydrin 2008).

Lexicalized nominalizations often occur with obligatory postverbal arguments, including postpositional phrases. Example (29a) is a finite construction with the verbal complex zò gélé <heart-burn> ‘to be angry’. In (29b), the verbal complex is lexicalized to form the noun zògèlègé ‘anger’. It is important to note the low tone of the head verb gèlè ‘burn:nmlz’. The argument of the inalienably possessed noun gé ‘stomach’ is omitted. Examples (30a) and (30b) are similar with the verbal complex zò dɔ̄ <heart-lay> ‘to trust’ and the lexicalized nominalization zòdɔ̀àmɔ̀ ‘belief, faith’; note the low tone in (30b) on dɔ̀ ‘lay:nmlz’. The argument of the postposition in (30b) is represented by the dummy pronoun à. For the nominalization in (30c), the object of belief is precise, wálà ‘God’. Lexicalized nominalization can further undergo adjectivization with the suffix –zɛ̀, as in (30d), zòdɔ̀àmɔ̀zɛ̀ ‘faithful’.

In Mano, the nominalization form is used whenever the verb is nominalized with a postverbal element, be it a postpositional phrase or an adverb. Furthermore, it can be used even with a specific type of finite predication, namely, a conjoint construction that expresses either a consecutive action, or a simultaneous action (on conjoint constructions, see Section 7.1):

In (31b), the subject of the copula sentence is the construction nù áà pɛ̄ séŋ́ wì, which contains the verb nū ‘to come’ in the low-tone nominalization form and a finite predication áà pɛ̄ séŋ́ wì ‘he breaks everything’ with a third person singular auxiliary áà.^[10] As the nominalized constituent with the low-tone form can contain finite clauses, this form should be considered a marker of clausal nominalization, in contrast to the VP nominalization marked (in certain contexts) by the gerund form, as in (22). While nominalization with clausal dependents has nothing unusual (see nominalization in English with complement clauses: [Learning that everyone should be treated with respect] was hard for Mary), it is worth noting that in Mano, such dependents require the use of specific nominalization marking. To conclude, nominalizations in Mano are characterized by mixed syntax with partly nominal (expression of direct objects as inalienable possessors) and partly verbal properties (the expression of verbal complements and adjuncts; on mixed syntax of Mande nominalizations, see Nikitina 2008; the issue from a broader perspective is addressed in Lefebvre and Muysken 1987).

5.4 Transitivity and lability

Mano does not have any morphological marking of derived (in)transitivity, which can also be observed in several other Southern Mande languages, such as Guro or Beng (Kuznetsova 2011; Paperno 2014). Some other Southern languages, such as Kla-Dan (Makeeva 2011), at least, have causative markers, which is also a typical trait of Western Mande, such as Kakabe (Vydrina 2011) or Mandinka (Creissels and Sambou 2013). To express valency changing, Mano has two procedures: one is analytic, such as the impersonal or reflexive construction, the other is lability, with a morphologically unmarked variation of argument structure, in which the verb can appear in the same form in the transitive or intransitive constructions. A striking feature of Mande, and Mano in particular, is the widespread P-lability, a variation in the argument structure in that the patientive participant occupies the position of the subject. The following example illustrates morphologically unmarked causal versus noncausal alternation in which the verb ‘to (make) fall’ is used in a transitive construction and an intransitive one with no overt morphological marking.

A far more cross-linguistically rare feature is passive lability in which the semantics of the verb have strong agent-oriented components. This is in contrast to an earlier prediction by Haspelmath (1993) that this type of lability cannot occur, it appears to be typical for Mande (Cobbinah and Lüpke 2009; Vydrina 2011; Creissels 2014), but also occurs elsewhere, such as in some Daghestanian and Austronesian languages (for a typology of lability, see Letuchij 2013).

6.1 Verbal forms used in TAMP constructions

Mano verbs have rich suprasegmental morphology. The verbal morphology includes five grammatical tonal forms as well as four affixes. Some of the affixes only occur in combination with a tonal form. The verbal paradigms used in various finite and non-finite constructions have a total of ten forms, including a baer form with lexical tone and no segmental affixes; and three forms in which the only morphological exponent is grammatical tone (the imperfective form, conjoint form, and nominalization form). In addition, Mano has two forms that have grammatical tone combined with a segmental morpheme (real protasis and counterfactual protasis forms, both with the suffix-á) and three forms in which the exponent is a segmental morpheme (gerund and counterfactual, both with the suffix -à, as well as the infinitive, with the suffix ‐pɛ̀lɛ̀). Finally, there are regular contractions of gerunds with the postposition ká ‘with’. No verbs contrast all forms; there is always at least some syncretism. Thus, conjoint forms are always homonymous with nominalization forms; gerunds are always homonymous with the counterfactual forms. Mid-tone CV verbs, such as pā ‘fill’, and high-tone CVCV verbs, such as félé ‘become clear’, have a maximum of seven contrasting forms, while the low-tone CVa verbs, such as nìà ‘turn’, contrast only three forms. Table 4 provides example paradigms.

6.2 Auxiliaries

The auxiliaries in Mande languages play an indispensable role in the expression of TAMP. A typologically remarkable feature of the TAMP system in Mano as well as in other Southern Mande languages and Southwestern Mande languages, is the rich inventory of auxiliaries. These not only contribute to the semantics of a particular TAMP construction, but they index the person and number of a subject. As in many other Mande languages, the Mano verb occurs in its lexical form and the auxiliary is the only exponent of the TAMP values in most TAMP constructions. Thus, TAMP constructions in (36a–c) differ only in their marking of the auxiliary: the auxiliary of the past (36a), perfect (36b), and conjunctive series (36c).

Other than expressing TAMP, Mano auxiliaries also index the person and number of a subject. In (36a), the form ē not only expresses the past tense but also the subject's person (third) and number (singular). The auxiliary co-occurs with the full noun phrase of the subject. Instead of treating forms such as ē as subject pronouns inflected for TAMP, an argument has been made to analyze them as TAMP markers (or auxiliaries) that are inflected for person and number and agree with the subject (Hachaturyan 2010). On (subject) agreement in Mande, see Konoshenko (2013; 2015). For more examples of subject indexing in the auxiliaries of African languages and more generally, for a discussion on the criteria used to distinguish free subject pronouns and bound pronominal markers, see Creissels (2005: 51–54). The full paradigm of Mano auxiliaries is given in Table A2 of the appendix.

6.3 Negation

Mano negation is expressed in copulas, auxiliaries and in additional predicative markers that are used in certain negative constructions.

Negation in Mano is asymmetric (in the sense proposed by Miestamo 2005). There are two reasons for this: negation involves a change in the TAMP constructions (constructional asymmetry) and the inventory of negative constructions is reduced with respect to the inventory of affirmative constructions (paradigmatic asymmetry). In particular, in negative non-verbal constructions, Mano only has one copula, wɔ́, while affirmative non-verbal clauses contain either an existential auxiliary (37, 38) or either of the four available affirmative copulas lɛ̄, lɛ́, wɔ́ or gɛ̰̀ (39, 40). While Mano has three affirmative presentative constructions (40), no negative correlate occurs.

In declaratives, only one negative auxiliary combines with different lexical and grammatical means, and that auxiliary is used to form negative habitual, aorist, and perfect constructions, whereas in the affirmative zone, three different auxiliaries are used for these purposes (41). The inventory of affirmative modal constructions does not map into the inventory of negative ones: thus, there is no direct negative equivalent of the prospective construction (see Section 6.4).

In addition to auxiliaries that have negative polarity (the negative series, as illustrated in (41), but also prohibitive) and the negative copula wɔ́, Mano employs negative predicative markers, including gbāā (41d), which is used in negative perfective constructions. It is important to note that another negative predicative marker, gɔ̰̀á̰, is used only in negative conditional constructions. Conditional clauses in Mano use extensive dedicated morphology, see Section 7.2.

6.4 Prospective

In contrast to other auxiliary series, the prospective series is always used with another auxiliary: it most often appears with the conjunctive auxiliary (with which it regularly merges). The series also stands out phonotactically because the markers have the form CVCV or CVCVCV, which is not attested elsewhere in the auxiliary paradigm. The series also has a peculiar polysemy. First, the series is used in quotative constructions (42); second, when the prospective auxiliary is combined with a conjunctive auxiliary coreferential with it, the entire construction conveys the modality of intentionality (43, 44) and prospective aspectual value, whereas a future action or state is expected to occur given the present state of affairs (45, 46).

In the quotative construction, the prospective marker is the sole means of introducing reported speech; there is no speech verb. The marker may optionally be followed by a conjunction kélɛ̀. The quoted predication can take any form; in (42), it is represented by a copula construction.

Quotative indexes are not uncommon in Mande. They are particularly widespread in Manding (on quotative indexes in Mandinka, see Creissels & Sambou 2013: 441–443). In some Mande languages, quotative indexes can be inflected for person and number (Idiatov 2010). Syntactically, in Mano, as in Mandinka or in Tura, another Southern Mande language, quotative indexes are used as quotative predicators, which are defined as a predicative element “similar to quotative verbs that cannot be classified as full-fledged verbs in a given language” (Güldemann 2008b: 15). Idiatov (2010) reported on other Mande languages with quotative indexes, in particular, Jula of Samatiguila or the Ko dialect of Mende (Western Mande) and observed that they are used non-predicatively as clause-linking markers. A polysemy such as the one attested in Mano (quotative, intentionality) has also been attested in Wan, Southern Mande (Nikitina 2018a), while Mandinka adds prospective to the range of meaning that is expressed by the quotative construction (Creissels and Sambou 2013: 443–445). The same type of polysemy is also widely attested outside the Mande family. On the development of intentionality meanings from quotatives, see Güldemann (2008b); on the development of future-related verbal meanings out of intentionality, see Bybee et al. (1994). Given the cross-linguistic tendencies in grammaticalization (quotative > intention > prospective), the Mano prospective series most likely emerged from a fusion of pronominal elements with some quotative index or a verb. However, the identity of that index or that verb and in general the origin of the series remain unresolved questions.

7.1 Conjoint form

The conjoined construction in Mano, which consists of an auxiliary of the conjoint series and a verb in the low-tone conjoined form, is an example of a clause-chaining category that occurs commonly in languages of Sub-Saharan Africa, where in a chain of clauses representing a sequence of events, only one is finite and the remaining clauses exhibit some properties of non-finiteness.^[12] The Mano construction is primarily used in the narrative chain. The narrative is usually divided into micro-episodes, with the first event in the perfective, and the subsequent events in the conjoint construction, and the subjects of the respective constructions are typically (but not always) coreferential.

Speech verbs in narratives more often than other verbs occur in the conjoint form. Furthermore, switching reference of the subject is a norm, rather than an exception. Example (48) represents a reported dialog, so in (48b) and (48c), the subject of the speech verb is different from each immediately preceding one.

The chain of conjoined constructions is rarely very long, consisting of a maximum of three to four clauses. A good balance between perfective and conjoined constructions creates a stylistic effect of narrative fluidity, and is a sign of a good narrator. However, children typically overuse perfective constructions because often they have not yet fully acquired the art of storytelling. The use of the conjoint construction in the Mano narrative remains an object for future study (for an analysis of aspectual marking used for narrative structuring in Totela, a Bantu language, see Crane 2011).

The conjoint construction is also used to express events that occur simultaneously with some other event:

It is important to note that while the conjoint construction in the narrative is usually used as part of a chain of statements, during conversation, it alone can constitute a turn and even occur in a question. This type of conjoint construction occurred in an example suggested by my language consultant, when a question that contained a conjoint construction was asked by one participant in a conversation (50c) as a direct follow-up of a prior question framed in a regular non-verbal predication (50a).

Thus, while the conjoint construction does not need to immediately follow a clause expressed by a full-fledged aspectual construction (perfective, in the case of a sequence of actions in the narrative, or some other construction, in the case of simultaneous events), it is always directly or indirectly related to some situational landmark. For an analysis of a similar construction, “null tense”, in Wolof, Atlantic, employing the notion of discursive anchor point, see Robert (2010). Givón (2001: 357) classified clause-chaining with the help of forms with reduced finiteness into two types: clause-chaining in which the finite-most predication is the ultimate member of a chain, which corresponds to the SOV word order, and clause-chaining in which the finite-most predication is the first member of a chain, and this corresponds to the VO word order. As observed by Creissels (2018: 804), Mande languages, as exemplified by Mano, put the finite-most clause in the initial position of the chain and challenge this typology by not being VO but rather OV languages, or to be more precise, SOVX languages. Therefore, as Creissels concludes, the relevant parameter in the typology of clause-chaining should not be OV versus VO, but rather verb-final versus verb-medial.

7.2 Conditionals

Mano conditional clauses deserve typological attention due to the richness of dedicated conditional morphology and owing to semantic patterns that do not conform to typological prototypes. First, while Mano verbal paradigms have ten morphological distinctions, three of them are employed solely in conditional clauses, including two dedicated tonal forms, conditional and irrealis, combined with the conditional suffix –á and a counterfactual form with the suffix –à. The contrasts in verbal forms are illustrated by the verb ló ‘go’: lō-á go:cond-cond vs lò-á go:irreal-cond vs ló-à go-cntrfct. In addition, as noted in Section 6.3, of the two negative predicative markers, one, gɔ̰̀á̰∼gɔ̰̀, appears only in conditional constructions.

On the semantic side, Mano makes a formal distinction between real (51), imaginative (52), and predictive conditionals (53), on the one hand, and counterfactual conditionals (54), on the other. In the former, the protasis is formed by a conjunctive auxiliary with the conditional verbal form (é lōá ‘if he goes’), and in the latter, the protasis is formed with a subjunctive auxiliary and an irrealis verbal form (è lòá ‘if (s)he had gone’) (on the semantic classification of conditionals, see Thompson et al. 1997). The repartition of conditional types in these two categories, counterfactual and non-counterfactual, does not correspond to a prototypical realis/irrealis distinction with real situations coded differently from hypothetical and counterfactual situations. Thus, in Nanti, Arawak, real conditionals contain realis marking and are therefore different from all other types of conditionals containing irrealis (such as in hypothetical conditionals) or double irrealis marking (as in counterfactuals; see Michael 2014). While marking counterfactual conditionals as a distinct type is not uncommon across languages (Lazard 1998 mentions a few), these cases often contain the same marker for the protasis and apodosis of counterfactual conditionals. For example, this occurs in Rembarrnga, non-Pama-Nyungan (Verstraete 2005) or in Hausa, Chadic (Bagari 1976). In contrast, in Mano, protasis and apodosis in conditionals, including counterfactuals, are expressed by strikingly different constructions. Yet in Mano, the distinction between counterfactual and non-counterfactual conditionals is far from fixed, because a non-counterfactual construction can be used in a semantically counterfactual context, as in (55) (no examples of the reverse type have been attested), which contributes to blurring the distinction between realis and irrealis. For a critique of the cross-linguistic validity of the notion of irrealis, see Bybee (1998) and Cristofaro (2012).

The existence of dedicated conditional morphology notwithstanding, no sharp boundary occurs between the conditional and other types of constructions. This means that non-counterfactual apodosis in Mano is always expressed by a non-dedicated construction (52, 53). Mano is similar to some Papuan languages (Thompson et al. 1997) in that it lacks a clear distinction between “if” and “when” clauses. Thus, on the one hand, temporal constructions may receive a conditional interpretation, as in (56), with the negative counterfactual protasis expressed by a negative copular construction with the linker yé ‘when’ which is typically used in temporal clauses. The value of counterfactuality in (56) is expressed by the marker of a retrospective shift, kɛ̀ì. On the other hand, formally conditional constructions often receive a temporal, non-conditional interpretation, as in (57).