Trilingual Journal of African Languages and Cultures Revue trilingue des langues et cultures africaines Dreisprachige Zeitschrift für afrikanische Sprachen und Kulturen The expression of diminutivity in Central Ring Grassfields Bantu

Studies on the expression of diminutivity in Bantoid languages of the Cameroonian Grassfields have tended to focus on the role that noun class derivation plays within the familiar Bantu paradigm. A closer look at individual branches of Bantoid, however, reveals a more complex picture, which rather suggests a division of labour between derivational strategies and compounding and/or periphrasis. This contribution zooms in on the languages of the Central Ring (CR) branch of Grassfields Bantu, presenting an overview of diminutivisation strategies found here: the notorious transfer to gender 19/6a, which is at times, accompanied by the addition of a semantically bleached suffix –CV, and periphrasis in associative constructions headed by nouns with inherent diminutive meanings such as ‘child’.


Introduction
While diminutives have been studied extensively for their forms and meanings both from a universal perspective (Jurafsky 1996, Bakema & Geeraerts 2000, Grandi & Körtvélyessy 2015 and in Bantu specifically (Gibson, Guérois & Marten 2017), they have been largely neglected in studies on Grassfields Bantu languages, beyond the general statement that gender 19/6a is employed for this purpose (Hyman 1979: 24, Hyman 1980: 234, Tamanji 2009: 31, Akumbu & Chibaka 2012: 54, Möller 2012: 12, Asohsi 2015: 68, Voll 2017. Diminutives are grammatical "elements which make a semantic contribution pertaining to size" (Gibson et al. 2017: 344) in that they primarily express 'physical smallness' (Schneider 2003: 10). Other -derived -semantic functions include young age, insignificance or incompleteness, as well as relation or descent (Jurafsky 1996). Moreover, "diminutives can also be used to convey perspectives and subjective viewpoints, as well as to encode pejorative meanings along the lines of disdain or contempt, or ameliorative meanings encoding affection and admiration" (Gibson et al. 2017: 344). This study investigates the forms and functions of diminutives in the Central Ring (CR) branch of Grassfields Bantu, drawing primarily on data from six of the seven CR languages, i.e. Babanki, Kom, Kung, Kuk, Men and Oku. 2 After a brief overview of the expression of diminutives in Bantu in section 2, section 3 discusses the morphological strategies of diminutivisation in CR. Section 4 sketches a prominent alternative strategy of diminutivisation attested in CR, i.e. periphrasis by an associative construction headed by the noun 'child' or 'tiny item'. A conclusion is provided in section 5.

Diminutives in Bantu
Diminutives in Bantu are "thought to have been historically expressed as part of the noun class system, and several noun classes have been reconstructed as including diminutive meaning" (Gibson et al. 2017). In this regard, the Proto-Bantu class 12 prefix *ka-is most often associated with diminutives while the class 13 prefix *tu-acts as the corresponding diminutive plural class marker /9, Meinhof 1910[1899, Meeussen 1967, Maho 1999, Demuth 2000. Synchronic use of gender 12/13 for diminutives in Bantu is seen in Chindamba (1) and Kimbundu (2).
(3) Venda (S53, Poulos 1986: 289, 1990: 38, Gibson et al. 2017 ku-thavha 'small mountain' (class 20) ku-thavh-ana 'very small mountain' (class 20 + -ana) (4) Tsonga (S53, Poulos 1999: 206, Gibson et al. 2017 In fact, Proto-Bantu *pi-might have a Proto-Benue-Congo ancestor in *pi- (de Wolf 1971: 170-1) which is assumed to have been grammaticalised from a prior independent noun pi or bi 'child', reflexes of which are attested in other branches of Niger Congo (Kähler-Meyer 1971: 347-348). The diminutive prefix either replaces the 'original' noun class prefix in most languages or it is added on top of it in a few cases (Maho 1999). Diminutive classes in Bantu are typically used for secondary classification (Meeussen 1967, Maho 1999. For this purpose, "a noun typically found in a different class is used in the diminutive class for a specific semantic effect" (Gibson et al. 2017: 359). However, there are also instances where nouns are primarily members of the diminutive class without necessarily being physically small. While Bantu languages predominantly employ their noun class system for the expression of diminutivity, other strategies such as derivational suffixes and compounding processes are also used (Maho 1999, Gibson et al. 2017.

Diminutives in Central Ring Grassfields Bantu
Diminutives in Central Ring (CR) primarily express physical smallness, but can also encode an offspring relationship, young age, inferiority and/or deficiency. As in other Bantu languages, a shift from one class to the diminutive class in CR also results in an interpretation of the noun concerned as physically small, that is, as falling short of the prototypical size of the referent class member. Therefore, the entities referred to by diminutives are regarded as smaller members of the category. Thus, fətsôtə̀ (19/6a) in Men refers to a brook, creek or rivulet which is smaller in size than what is perceived as standard for its derivative source noun, i.e. tsò (9/10) 'river'. 3 This is not to say that the referents of diminutive forms are small by an absolute standard. Speakers represent a referent as small for a particular communicative purpose, i.e. "smallness is not necessarily perceived, but in fact ascribed" (Schneider 2003: 11). Derived diminutives may undergo semantic specialisations, e.g. Men ēkyû (3/6a) 'bed' derives the diminutive fəkîə (19/6a) 'stool' and Babanki kətíʔətíʔə́ 'ear' derives the diminutive fətíʔətíʔə́ 'mushroom (sp.)', as motivated by similarity in form. As seen in (6)  Diminutives may also entail pejorative or derogatory meanings. Thus, fəŋgəŋtə̂ 'small house' (< əŋgəŋ (3~5/6) 'house'), to a Babanki speaker, can express a negative attitude toward the referent, reflecting the view that the owner could have built a much bigger house. It could also mean that someone else will eventually build a bigger house than the diminutivised one. Linked to pejorative connotations is the notion of inability or incapability of a person or group of people and things. For example, Babanki fəẁìʔtə̂ (< wìʔ (1/2) 'person') normally means 'small person', but can also be used to designate someone who has not achieved much, be they physically small or big. Such pejorative connotations can be adduced for the rest of CR, e.g. in Kung fəfûɔ 'small and feeble thing' (< kəfúɔ (7/8) 'thing') (Kießling 2019: 149). When used on body parts, diminutives can also be a form of insult, e.g. Babanki fəʃɨ̀lə̀ 'small eye' (< əʃɨ́ (5/6) 'eye') does not really mean that the eye is physically small but could be a way to simply humiliate the person concerned. In the next two subsections we focus on the morphological strategies of diminutivisation in CR, i.e. the transfer to gender 19/6a (3.1), and concomitant suffixation (3.2).

Shift to gender 19/6a
The most common strategy of diminutive formation in CR is derivation by which a noun is shifted to gender 19/6a, as described for Babanki (Akumbu & Chibaka 2012), Oku (Yensi 1996), Men (Möller 2012), Kuk (Kießling 2016) and Kung (Kießling 2019: 149). 4 Class prefixes of gender 19/6a which are used for diminutivisation might attach to nouns of class 1 or 9 which come without a class prefix as shown in (7) or they replace a pre-existent class prefix as in (8). 5 Remarkably, stem alternations which characterize the base forms of gender 1/2 and 9/10, e.g. consonant alternations such as v ~ɣ and w ~ ɣ, are absent in their diminutive counterparts. Babanki ətó (5/13) 'hut', pl. tətó kəẁʉ́ (7/6) 'foot', pl. àwʉ́f ətótə̀ 'tiny hut', pl. mətótəf əẁʉtə̀ 'small foot', pl. məẁʉtə5 All CR languages employ a noun class system of the Bantu type with either 12 (Babanki, Bum, Kuk, Kung) or 13 (Kom, Men, Oku) agreement classes (Akumbu 2019: 2). All of them distinguish class 19 marked by fV-and its corresponding plural class 6a (also the class for liquids) marked by m(V)-. The addition of noun class prefixes of 19/6a in diminutive function on top of the original ones, establishing a secondary layer of class prefixes, seems to be restricted to instances where the original noun class prefix has started to merge with the root. A possible candidate is Men fēɣɔíntə̂ 'small children' where the diminutive prefix is added onto an alleged class 2 prefix ɣ- (Möller 2012: 12). 6 Nouns borrowed into CR can also be diminutivised by being shifted to gender 19/6a, as shown in (9), which proves that this strategy is indeed very productive in CR.
6 Even more remarkable is the fact that this diminutive is derived from the plural form ʌɣɔín, not the singular váin. From this perspective, however, it is dubious whether the segment ɣ is actually to be analysed as plural prefix, since the prefix in the plural form ʌ-ɣɔín 'children' is ʌ-, while the segment ɣ seems to belong to the root. This is probably due to the special nature of the noun 'child' -not only in Men, but in a number of Ring languages -in that number distinction is not only expressed by change in NPx but also by suppletion of roots or rather by two distinct forms of one and the same root whose allomorphic relation has become intransparent by idiosyncratic fusions. While 19/6a is employed for secondary diminutive classification, there are nouns that are primarily members of this gender in CR, without necessarily being physically small. Even those 19/6a nouns whose referents might be regarded as physically small in comparison to some absolute standard do not appear to be derived from any other non-diminutive class, as pointed out for other Bantu languages (Gibson et al. 2017: 359 The nouns in (10) are primarily members of gender 19/6a since there is no evidence of their derivational shift from another source gender on the synchronic level. Note that the absolute size of referents of primary 19/6a nouns such as 'bird', 'tree', or 'knife' is, in principle, no argument against their potential historical origin in derived diminutives, since the contemporarily productive derivational process is also not guided by orientation towards any absolute standard of size, but rather depends on the size which is perceived as prototypical of a given class member. Therefore, it might be that these nouns which are today -and most probably already in proto-(C)R times -primarily affiliated to gender 19/6a actually represent historical diminutives derived at pre-Ring times from a non-diminutive root which has disappeared from (C)R.

Suffixation
Shifting nouns to gender 19/6a for diminutivisation is, at times, accompanied by the addition of a -CV suffix. CR languages differ with respect to the scope of application of the -CV suffix across the lexicon (lexical coverage), the optionality of its presence in diminutives, the available forms of its (lexically conditioned) allomorphs (-t, -l, or -n) and the degree to which they undergo phonological reduction, as briefly summarized in The suffix is obligatorily present in derived diminutives in Babanki but lexically conditioned in Kom, Kung, Kuk, Men, and Oku such that some nouns take a suffix in forming the diminutive while others do not, e.g. Kuk fəbyí 'smallish goat', Oku fəẁîl 'small person'. In Kung there is also variation with respect to application of the suffix, e.g. kəbê 'thigh' derives the diminutive singular fəbê-lə̀ which varies freely with fəbê, whereas the diminutive plural m̄bê-lə̀ does not allow for omission of the suffix in *m̄bê. 7 Concomitant suffixation for diminutive derivation in CR is exemplified in (12).
(12) CR diminutive derivation in 19/6a and suffixation 8 Base Kung īɣa᷇ ŋ (5/10) 'root', pl. səɣa᷇ ŋ kəbê (7/4) 'thigh', pl. ībê sʌf (9/10) 'maize', pl. səsʌf fəɣa᷇ ŋə̂ 'small root', pl. m̄ɣa᷇ ŋə̂ fəbê(lə) 'tiny feeble thigh', pl. m̄bêləf əsʌblə̀ 'tiny feeble maize plant', pl. m̄sʌblə7 The datasets on which these claims are based vary with respect to individual languages. The datasets for Kuk and Kung are quite limited comprising some 20 diminutives which have been checked with two consultants, respectively. Regarding Kom and Oku, 25 items were taken from secondary sources and checked systematically with two consultants each. The Men dataset is a bit larger including some 40 items from various sources. While the Kuk dataset stems from elicitation exclusively, Kung and Men data are based on elicitation and narrative discourse. 80 Babanki items were provided by the first author and checked by two other native speakers. In all cases diminutives have not been checked for their potential range of variation across different individuals. In Babanki, Kom, and Oku, all three suffix allomorphs can occur in both the singular and plural and it is possible for some words to take -lə̀ or -nə̀ in Babanki without any semantic difference. In Kom and Oku, it is possible to leave out a suffix and still obtain the diminutive meaning only by transfer to gender 19/6a. In Kung and Kuk suffixation for diminutivity is more restrictive than in Babanki, Kom and Oku in two respects. First, only one suffix allomorph, i.e. -(l)ə̀ (along with various types of reduction) has been observed so far. Second, the distribution of this suffix is constrained by semantically intransparent lexical criteria, i.e. some diminutives require the suffix obligatorily, others apply it optionally, while some lack it altogether. Of all CR languages Kuk is the one with the most limited use of the suffix. In all 10 examples given in Kießling (2016), only one, fəʒúglə̀ 'smallish snake', requires the suffix -lə. Another one, fəɲàmə̏ 'small animal', contains the reduced form of the suffix, i.e. -ə, and in a third one, fəsʌb(lə) 'smallish maize cob', it is optionally present.
Remarkably, two of these suffixes, namely -tV and -lV, which accompany diminutive derivation of nouns in CR resemble the verbal extensions -tV, and -lV commonly found in Bantoid and in Grassfields (Watters 2003: 245, Hyman 2018 to derive attenuative meanings in verbs (Akumbu & Chibaka 2012: 137, Tamanji & Mba 2003, Mba & Chiatoh 2003, Harro 1989, Mba 1997, i.e. a reduced degree of quality in states and intensity in actions and events. Semantically, the effect is parallel to diminutivisation with nouns. While the suffix -nV serves diminutive function, it is not attested in the attenuative. On the other hand, -kV is attested in the attenuative but has not been found in diminutives. In Babanki, for example, -tə and -kə function as attenuative suffixes often combined with a frequentative or iterative function, as shown in (13) Depending on the situation in individual CR languages, attenuation is often also linked to repetitive and frequentative notions, due to the common experience that distributive repetition and parcellation tends to entail a diminution of intensity. Thus, in Men, the cognate suffix -te has attenuative function which is often combined with a frequentative, iterative, distributive or pluractional notion, as shown in (15) In Kom some lexemes seem to distinguish frequentative and attenuative, e.g. tʃʲá 'kick' allows for a contrast of the frequentative tʃʲá-lɨ́ 'kick repeatedly' vs. the attenuative tʃʲá-tɨ́ 'kick a little', whereas others derive polysemous stems with -tɨ̀, e.g. mzì-tɨ̀ (< mzì 'swallow') 'swallow a bit; swallow repeatedly' (Jones 2001). In Kung, the suffixes -nə and -lə are marginally attested in attenuative function: mwàe-nə̀ 'twinkle' (< mwàe 'shine'), ɲɔʔ-lə̀ 'roast a bit' (< ɲɔʔ 'roast'), zʊʔ-lə́ 'make warm' (< zʊʔ 'heat'). In Kuk, the widespread pluractional suffix -kə is marginally attested in lɨ́m-kə́ 'wait for a long time' (< lɨ́mə́ 'wait for').
The application of these verbal extensions to nominal bases, obviously motivated by the functional parallelism of diminution and attenuation, thus represents an instance of morphological strategies DOI 10.15460/auue.2020.93.1.203 A&Ü | 93 / 2020 Akumbu & Kießling | The expression of diminutivity crossing word class boundaries. In a diachronic perspective, this might be analysed as a spillover of verbal derivational morphology into the nominal domain which could have been triggered or promoted by nominalisation of verbs extended by the attenuative, as is suggested by examples such as the Men diminutive fətsəḿtə̀ 'small dream' (< tsəm (9/10) 'dream') which coexists with a verb tsəmtè 'dream' obviously including the attenuative suffix -tə.

Associative construction for diminutive formation
Diminution is also achieved in CR by periphrasis in associative constructions headed by various nouns of gender 19/6a which encode a diminutive notion either in their lexical meaning as with nouns meaning 'tiny item' or by a combination of their lexical meaning with a diminutive derivation as in the case of fəẁán (Babanki), fəwáe (Kung) and fēwán (Oku), all meaning 'little child'. Other nominals that can be used in head position are fəǹdɛʔ (Babanki) and fɨ̄fúin (Kom) both meaning 'tiny item'. In Babanki both fəẁán 'little child' and fəǹdɛʔ 'tiny item' can be used interchangeably for the same diminutive function. While Kuk and Men also have distinct lexical items for this meaning, i.e. fəfwâtə̀ (Kuk) and fəɲâŋ (Men), it is not clear to which extent they are also used in constructions such as the ones in (16) and which type of division of semantic labour pertains with respect to the usage of 'child'.
(16) CR diminutive derivation using associative constructions Kung mbvə̄ (9/10) 'chicken', pl. səmbvət sàʔ° (9/10) 'trap', pl. sətsàʔ°k əpfəǹdʊŋ (7/8) 'pig' pl. ūpfəǹdʊŋ wāe fə̄ mbvə̄ fə̄ 'small feeble chicken' wāe fə̄ tsàʔ fə̀ 'small trap' wāe fə̄ fəpfəǹdʊŋ fə̄ 'small feeble pig' Oku əbkún (3/6a) 'bed', pl. əmkún kētíɛ (7/8) 'chair', pl. əbtíɛ ntɔǹ (9/10) 'pot', pl. ntɔǹsè fēwán é əbkún 'small bed' fēwán é kētíɛ 'small chair' fēwán é ntɔǹ 'small pot' The syntax of the examples presented above follows the pattern of CR associative constructions, i.e. the preceding head noun (N 1 ) is linked to the following modifier noun (N 2 ) by an associative marker (AM) which agrees with the class of the head noun according to the formula given in (17)  In some CR languages such as Kung and Kuk, the prefix of the head noun (NPx) is dropped, as soon as a modifier follows (Kießling 2016(Kießling , 2019. Thus in (18), the head noun fəwāe 'little child' loses its noun class prefix fə-due to the fact that it is modified by the noun mbvə̄ 'chicken' which is linked to the head noun by the associative concord of class 19 fə. Under certain conditions, some CR languages such as Kung require an additional noun phrase terminal enclitic (ENCL) which indexes the class of the head noun and which is reminiscent of the determiner enclitic in the West Ring languages Aghem (Hyman 2010) and Isu (Kießling 2010). While the West Ring determiner enclitic largely serves to mark the non-focalised status of nouns, the morphosynactic and pragmatic conditions of its distribution in Central Ring, however, remain completely unclear so far.
From a wider comparative perspective, lexical items meaning 'child' are quite commonly employed for diminutive functions, even-DOI 10.15460/auue.2020.93.1.203 A&Ü | 93 / 2020 Akumbu & Kießling | The expression of diminutivity tually following a universal path of grammaticalisation (Heine & Kuteva 2002: 65-7). In various Bantu languages such as Cuwabo, Nzadi, Eton and Bafia (Gibson et al. 2017: 358-359), in Kikongo (Huth 1992) and in Sotho-Tswana and Nguni (Güldemann 1999), reflexes of Proto-Bantu *jánà 'child' seem to have been developed to diminutive markers, probably independently of each other. Beyond Bantu, the same process operates in various branches of Niger-Congo, e.g. on Susu díi 'child' (Anderson, Green & Obeng 2018) and on Ewe ví 'child' (Heine & Hünnemeyer 1988, Heine, Claudi & Hünnemeyer 1991, reflex of a Niger-Congo root *bi 'child' (Kähler-Meyer 1971: 347-348) which is assumed to be the ultimate source of Proto-Benue-Congo *pi- (de Wolf 1971: 170-1), the ancestor of Proto-Bantu class 19 *pi-. A remarkable detail about the CR situation is that in none of the CR languages it is simply the noun 'child' which is employed for periphrastic diminution purposes in associative constructions, but rather its diminutive stem in 19/6a. Borrowed words can also be diminutivised in CR by means of the associative construction (19) This preference is probably due to the fact that the ordinary diminutivisation strategy by transfer to gender 19/6a would create no visible effect in contrast to the base form which is already assigned to 19/6a. Alternatively, the simple addition of one of the diminutive suffixes -tV, -lV or -nV is not sufficient in most cases. Thus, diminutives such as *fəɲín-tə̀ 'small bird' or *fəkɔʔ-tə̀ 'small tree' which are simply formed by adding the suffix to the basic 19/6a noun forms (as elaborated in section 3.1), are not acceptable in Babanki. However, precisely this case is attested in Men where nouns primarily affiliated to gender 19/6a such as fēnɨ ᷇ŋ 'bird', fəsɨ́s 'pepper' and fəɲí 'knife' derive their diminutives, i.e. fēnɨ ᷇ŋ-tə̂ 'small bird', fēsɨ́y-lə́ ~ fēsɨ́y-tə́ 'small pepper' and fəɲí-tə̂ 'small knife', respectively, only by additional suffixation. So far, it has been assumed that every noun can be diminutivised through nominal affixation, but this is not always the case. In Babanki, for instance, the nouns in (21) can only be diminutivised by means of the associative construction with fəẁán 'little child' or fəǹdɛʔ 'tiny item' as head noun, since a morphologically derived diminutive in gender 19/6a is not available for them. So, the associative construction appears as a compensatory strategy here. he range of nouns that do not lend themselves to morphological diminutivisation by transfer to gender 19/6a in Babanki is varied, including, but not limited to insects, household items and body parts. While the motivations for these restrictions are still unclear, instances such as kətíʔətíʔə́ 'ear' suggest that morphological diminu-tivisation might be blocked by the presence of semantic specialisations of parallel forms in gender 19/6a such as fətíʔətíʔə́ 'mushroom (sp.)'.
While diminutives are formed by the morphological and morphosyntactic operations outlined above, it appears that augmentatives do not receive a similar treatment in CR. In Babanki, for example, augmentatives are expressed by an attributive usage of inchoative-stative verbs such as ɣɔʔ 'be(come) big' (ɣɔʔkə́ pl.) illustrated in (22).

Conclusion
Diminutivisation in Central Ring languages is generally achieved by a widely attested shift of nouns from various genders to gender 19/6a marked by prefixes fV-/m(V)-which replace the original noun class prefixes. The productivity of this strategy is manifested in its recurrent application to borrowed nouns. Sometimes, diminutivisation in gender 19/6a is accompanied by the addition of a semantically bleached suffix CV. Remarkably, some of the allomorphs of this suffix, i.e. -tV, -lV, resemble the verbal extensions -tV, and -lV commonly used in Bantoid and in Grassfields to derive attenuative meanings in verbs. From a diachronic perspective, this might be analysed as a spillover of verbal derivational morphology into the nominal domain which could have been promoted by nominalisation of verbs extended by the attenuative. Another diminutivisation strategy discussed is periphrasis in associative constructions headed by various nouns of gender 19/6a which encode a diminutive notion either in their lexical meaning as with nouns meaning 'tiny item' such as fəǹdɛʔ (Babanki) and fɨ̄fúin (Kom) or by a combination of their lexical meaning with a diminutive derivation as in the case of fəẁán (Babanki), fəwáe (Kung) and fēwán (Oku), all meaning 'little child'. More finegrained generalisations about the limits of morphological diminutivisation in CR and regularities regarding its division of labour with syntactic strategies will only be possible on the basis of a much more extensive corpus of diminutives which includes data from all under-researched CR varieties, especially from Bum for which diminutive data have not been available at all so far.