Skip to navigation – Site map

HomeIssues6PapersThe Morphopragmatics of the Dimin...


The Morphopragmatics of the Diminutive Morpheme (-ba/-wa) in Akan

Clement K. I. Appah and Nana Aba Appiah Amfo


The present paper is concerned with the diminutive morpheme -wa/-ba in Akan. It examines the form, the origin and the various meanings associated with diminutive forms in the language. We attribute the origin of the diminutive to the lexical word for ‘child/offspring’ ɔba, basing our argument on language internal evidence as well as cross-linguistic generalizations. The identified meanings of the Akan diminutive are as follows: small, young/offspring, feminine, member, insignificant/nonserious, affection/admiration and contempt/disdain. Having identified the basic meaning of the diminutive as ‘small’, Jurafsky’s [1996] Radial Category theory provides us with a basis to adequately account for the various meanings; drawing a link, through metaphors and inferences, between the diachronic and the synchronic meanings.

Top of page

Full text

1. Introduction

  • 1 See Schneider [2003] and Dressler and Merlini Barberesi [1994] for fuller bibliographical reports.
  • 2 Research on Diminutives in African languages has mostly concentrated on Bantu languages particularl (...)

1The diminutive has been an object of study for quite a long time, dating back to the nineteenth century (cf. Coleridge [1857], Lewis [1832] inter alia), and this tradition of research on diminutive has continued well up until now.1 In spite of this rich history of research, there has been comparatively little research done on diminutives in African languages and particularly languages belonging to the Kwa sub-group of the Niger-Congo language phylum.2 This paper is an effort to provide more insight into the nature of diminutives in African languages by focusing on the diminutive in Akan, a Niger-Congo language of the Kwa sub-group. It addresses issues relating to the form of the diminutive, its semantic origin, and analyzes the various senses evoked by the use of the diminutive. The analysis of the varying meanings of the Akan diminutive is modelled after Jurafsky’s [1996] Radial Category theory; and like Schneider [2003: 4], we acknowledge that the meaning derived in on-line interpretation by interlocutors of a given diminutive form depends “on the specific interplay of linguistic and situational factors within a given context.” We do find that the Radial Category model provides a principled and neat way of accounting for both the synchronic and diachronic aspects of the meaning of the diminutive. It also makes it possible to explain the various, sometimes even conflicting synchronic meanings associated with the diminutive.

  • 3 Akan is a cover term for at least eleven dialects. Three of these dialects – Fante, Asante and Akua (...)

2The diminutive in Akan is formally realized as a suffix, the actual form varying between the Twi (Asante and Akuapem) dialects and the Fante dialect.3 Whereas in Fante, the diminutive is almost consistently realized as -ba (as in kakra-ba ‘small’, dwordwor-ba ‘smallish’, Essuman-ba ‘girl’s name, female of Essuman’, adwuma-ba ‘an insignificant piece of work’); in Twi, it is often realized as -wa (as in kete-wa ‘small’, kurokuro-wa ‘smallish’, Takyi-wa ‘girl’s name, female of Takyi’, adwuma-wa ‘an insignificant piece of work’). In spite of the varying phonological shapes of the suffixes in the different dialects, we are convinced that it is the same semantic concept in question here – i.e. both forms represent (are allomorphs of) the diminutive. It is interesting that the words for small in both Fante and Twi, kakraba and ketewa respectively, contain the respective diminutive suffixes. We will return to the issue of allomorphy in the following section as we relate the present form of the diminutive to its origin.

  • 4 Christaller [1933] referred to the Akan language as Twi, with the Asante and Fante as dialects.

3The data used in the present paper comes from a variety of sources. Diminutive forms (i.e. words containing the diminutive suffix) were collected from Christaller’s [1933] dictionary of the “Asante and Fante”4 language and the Akan Dictionary [2006] produced by the Department of Linguistics, University of Ghana. An additional database of diminutives was created by the authors, by soliciting for diminutive forms, including female names, from a number of native speakers. In terms of dialect coverage, the data includes words from the three major dialects, Fante (Fa.), Asante (As.) and Akuapem (Ak.). Where lexical items are unique to specific dialects they will be marked accordingly, otherwise, it should be taken that the lexical item in question is used in all three dialects.

4Various interpretations can be assigned to diminutive forms in Akan. They range from concrete concepts such as small, young/offspring, female, membership, etc., to attitudinal ones like insignificance, disdain, affection and admiration. We will be exploring how these different meaning components interact. The rest of this paper is organized as follows: section 2 examines the form and origin of the Akan diminutive. In section 3, the various meanings associated with the diminutive are identified. As a prelude to section 5 which outlines Jurafsky’s radial category theory, section 4 captures some previous approaches to the analysis of the diminutive. Section 6 is a discussion of the metaphors and inferences that motivate various senses of the diminutive in general and the Akan diminutive in particular. We end that section by proposing a structure for the semantics of the diminutive in Akan. Section 7 is the conclusion.

2. The form and origin of the Akan diminutive

5As mentioned in the introduction, the phonological shape of the Akan diminutive varies between the three major dialects; Fante consistently uses the suffix -ba while the Twi dialects mostly use -wa, even though in some cases, both dialects (but especially Akuapem) employs the suffix -ba. Typically, when the diminutive form is used in reference to an offspring, and also when it indicates membership, then -ba is maintained in all three dialects as shown in (1), even if there is an additional attitude conveyed, as may be the case when (1c) is used in context.



ɔheneba (As./Ak.), ɔhenba (Fa.)

‘prince/princess, lit. child of a king’


ɔkraba (As./Ak.), agyinamoaba (Fa.)

kitten, i.e. offspring of a cat’



‘a young white or lightly coloured person’



‘a kid, i.e. offspring of a goat’



‘a church member, lit. child of a church’



‘citizen, lit. child of a nation’

6In many other cases, however, the suffix is represented as -wa, as the Twi examples in (2) indicate.




‘a young noble man’



‘a child of one to seven years’



‘a small fish hook’



‘a small house, cottage’



a trifle’

  • 5 Cf. studies regarding the auxiliary verb san ‘return’ and the comitative verb n(y)e ‘be with’, Amfo (...)

7An adequate phonological account of why the diminutive suffix is -ba in some instances and -wa in others have so far proved elusive, as these forms do not appear to follow a consistent phonological pattern. The reason for the choice of allomorph could be semantic; that is, the Twi dialects, particularly Asante, maintains, with few exceptions, an animacy distinction; reserving -ba for animate entities and employing -wa for inanimate entities. (However, see discussion on female names in section 3.3). We do consider this issue far from closed though. Intuitively and empirically, there is no doubt that both -wa and -ba suffixes are allomorphs of the diminutive. Another reason that can be advanced here is that in grammaticalization studies, it has been attested that Fante generally tends to change at a slower rate than Twi, and it still has a number of older forms and features as compared to Twi.5 We, therefore, consider -wa a (phono-)semantically conditioned allomorph of the diminutive morpheme -ba.

  • 6 Details of the components of these two groups and the semantics of their ‘members’ is the subject o (...)

8Structurally, the diminutive can be recognized in two groups of words, namely Groups A and B. For Group A words, the base word can be clearly delineated from the diminutive morpheme, and the result will be a recognizable lexical item (mostly a noun) plus the diminutive suffix. On the other hand, the base words in Group B are synchronically fused with the diminutive suffix such that they are no longer recognizable as full lexical items with distinguishable meanings in the language. That is, even though we see aspects of diminutive meaning running through the Group B examples, we cannot nonetheless make legitimate sense of the remaining part of the word without the diminutive suffix. The phenomenon observed in Group A allows for a relatively high level of productivity whereas those in Group B can be said to be lexicalized.6

Group A

  • 7 DIM stands for diminutive.













‘baby/small bird’



‘small bag’

Group B



‘a kind of small musical instrument’


dwordworba (Fa.)




developing breast (of a teenage girl)’



‘a small calabash with a cover’

  • 8 The full form including the prefix is used mostly for human entities and in utterance initial posit (...)
  • 9 Baule, just like Akan, descends from the Central Tano group of the Kwa branch of the Niger-Congo ph (...)

9As regards the origin of the diminutive suffix, we postulate that it originates from the word for child/offspring which is (ɔ)ba.8 We do so for two reasons: first, there is an obvious phonological identity between the diminutive suffix and the word for child, and we do not assume this to be a coincidence, as we observe the meaning ‘child’ and its associated concepts in the various interpretations of the diminutive forms. Second, the diachronic association of the diminutive with the word for child has been borne out in a number of (particularly African) languages (cf. Greenberg [1959]) notably from the Niger-Congo family. Heine et al. [1991] reports that the Proto-Bantu noun *-gana ‘child’ is the source of diminutive markers in South-eastern Bantu languages like Venda, Tsonga, Sotho and Zulu. Following Timyan [1977], Heine et al. [1991: 94] indicates that in the Kode dialect of Baule,9 what appears to be the diminutive suffix is possibly a “grammaticalized form of the noun ba (pl. mma-mu ‘child’). Notable among all of these associations between ‘child’ and the diminutive suffix is Heine et al’s suggestion that the Ewe (one of Akan’s closest relatives) diminutive suffix -vi is derived from the word for child vi. Jurafsky [1996] also makes similar observation and cites other languages from Asia and elsewhere which demonstrate this phenomenon. His examples which attest to the child-diminutive relationship are reproduced below.



vi << vi ‘child’

Gbeya (Niger-Congo)

be << beem ‘child’

Londo (Bantu) i-Luyana( Bantu)

nwana-< < nwáná ‘child’
ana << ana ‘child’


Mandarin (Chinese)

-er << er ‘son’
(only bound, cf. erzi ‘son’)

Cantonese (Chinese)

-dzai << dzai ‘son’

Fuzhou (Chinese)

-kiaŋ << kiaŋ ‘child’


te - << te ‘son, child’


- sa << -sa ‘child’
bisa ‘his son, child’)

Classical Tibetan (Tibeto-Burman)

-bu - -U << bu ‘child’

Eastern Kayah (Tibeto-Burman)

-phú << phú ‘child’
(only bound, cf.
vεphú ‘my child’)

Ainu (Isolate)

Thai (Kam-Tai)

-po << po ‘child/son’

lûuk- << lûuk ‘child’



-pīl << pil ‘child’

Awtuw (North New Guinea)

-yӕn << yӕn ‘child’

Tboli (Austronesian)

ngá << ngá ‘child’

(Jurafsky [1996: 562, ex. 36-38])

10Having established the origin of the diminutive suffix, we outline the various meanings that are associated with the diminutive in the following section.

3. Meanings

  • 10 See Schneider [2003], Jurafsky [1996], Booij [2007], for discussion.

11Cross-linguistically, diminutives are associated with the basic meaning of ‘small’.10 Even so, the range of meanings expressed by the diminutive in each particular language is not confined to smallness. The following subsections look at the various meanings and interpretations we have identified with the Akan diminutive; for each group addressed in each subsection, we give some examples which illustrate the meaning in question.

3.1. Small

12The meaning ‘small’ is one of the basic (if not the basic) meaning that many researchers have associated with the diminutive in a number of languages. Jurafsky [1996: 534], for instance, defines the diminutive as “any morphological device which means at least ‘small’.” Schneider (2003: 10) says that “[P]rototypically, diminutives express smallness”. Dressler and Merlini Barbaresi [1994: 85] in apparent recognition of the fundamental significance of the meaning ‘small’ of the diminutive, refers to ‘smallness’ as its “morphosyntactic denotation” in contrast to other features such as endearment which they consider as its connotation. The situation is no different in Akan, as the data in (6) show.




‘a pen knife’






‘a small money box’



‘small house, cottage’



‘small fish hook’

13In Akan, the concept of ‘small’ is associated with a good number of diminutive forms in the language, either solely or in combination with other concepts such as feminine, young, insignificance, etc.

3.2. Young

14The meaning ‘young’ is closely related to the meaning ‘small’ as well as the source of the diminutive, which is ɔba ‘child/offspring’. One’s offspring, up to a certain extent, is expected to be small. We need to bear in mind that the feature ‘small’ is not absolute; and it is so in comparison to a certain norm. Again, this meaning component has the potential of combining with other senses of the diminutive. In (7b), for instance, one can talk about a conflation of the diminutive meanings ‘young’ and ‘feminine’.



abarimawa (Twi)

‘boy, lad’


aketesiaba (Fa.)

‘a young woman’



‘a child of one to seven years’



‘a baby’



‘a young nobleman’

3.3. Feminine

15The meaning component ‘feminine’ can be seen in some diminutive forms such as aberewa ‘old lady’ (8a). Even though the morphological breakdown of this word may not be clearly apparent, one can argue that it is made up of the following morphemes: a-bere-wa ‘NOM-to ripen-DIM’. If this is correct, then the concept of old woman is presented as ‘a small mature person’. Thus we see the interaction between the diminutive senses ‘small’ and ‘feminine’.

  • 11 In certain contexts, this word could be used derogatorily in reference to a female individual. This (...)




‘old woman’



young woman (cf. Aketesiaba, Fa.)’



‘a European woman’



female name (feminine Egyir)’



‘maid servant’

16Feminine meaning or gender (for the relevant languages) as one of the meanings of the diminutive is not uncommon. Jurafsky [1996: 536], for example, cites some examples from Hebrew, Hindi and Berber. As indicated earlier, the feminine concept communicated by the diminutive is usually done in combination with other diminutive-associated concepts. For instance, the words ababaawa/abaayewa (As./Ak.)/aketesiaba (Fa.) ‘young lady’ communicate both youthfulness and femininity; thus one who is youthful yet not feminine or feminine yet not youthful cannot be referred to by any of these afore-mentioned labels.

17Finally, we would like to comment on the suffixes that are found in a number of female Akan names. Here again, there are significant dialectal variations. A number of Fante names end in the suffix -ba (occasionally -wa/-ma) as demonstrated in Table 1, below. Asante has a mixture of -wa, -waa and maa, and Akuapem has -bea.

  • 12 This phonological observation is limited to the context of female Fante names; we do not have enoug (...)
  • 13 This is not to suggest that women have absolutely no significant roles to play in Akan traditional (...)

18It seems to us that in Fante names, the diminutive -ba (and what appears to be its intervocalic variant -wa/-ma)12 conflates the meanings of feminine and small. As Booij [2007] mentions, the link between ‘small’ and ‘feminine’ may have a physical component to it; women are expected to be generally smaller in size (in comparison to their male counterparts), do ‘smaller’ things, eat smaller portions of food, etc. There is also a cultural dimension to it, where femininity is associated with nonseriousness, one of the attested pragmatic interpretations of the diminutive (cf. Dressler and Merlini-Barbaresi [1994]). This is borne out by the fact that in traditional Akan societies, women are excluded from serious discussions which lead to the taking of important decisions. Serious issues are not mbaasem ‘female matters’.13

Table 1: Akan Female Names



















Acquaaba (Akwaaba)






Quainuwa (Kwenuwa)









19In Asante, however, we see the occurrence of the suffixes -wa, -waa and maa in the female names and -bea is the default Akuapem female name suffix. Our conclusion is that Asante and Akuapem make use of suffixes from two sources for female names: (1) from (ɔ)ba, the origin of the diminutive and (2) from ɔbaa, the Akan word meaning ‘female’. The Akuapem equivalent of ɔbaa is ɔbea, and we see the root of the word retained in the exact same form, this time as a suffix, in the corresponding female names. We arrive at -waa from -baa through the same, yet not fully understood, process by which the diminutive suffix becomes -wa from (ɔ)ba. For the male names which end in nasals (i.e. a nasal consonant or a nasalized vowel), the suffix then becomes -maa. This again, in another way shows the cultural conflation of the diminutive concept and femininity. It is interesting that the words for ‘child’ and ‘female’, from which these suffixes derive, are phonologically very similar, and this has probably resulted in the present situation where Akan speakers make almost equal use of the two forms in the creation of female names from their male counterparts. Indeed, the semblance of diminutive morphemes to morphemes that form female names from male versions is well documented. Booij [2007: 223], for instance, cites the Dutch example in (9) and observes that the extension of diminution to feminine gender may also come under the evaluative use of the diminutive to the extent that women are physically smaller than men.



‘boy’s name’


‘girl’s name’

3.4. Member

20Sometimes a diminutive form may indicate membership of a group, as the examples in (10) show. The entities referred to by these forms “are categorised as members of the class designated by the base word” (Schneider [2003: 11]).




‘church member’






‘men, children of men, i.e. humanity’

21The words so marked in Akan do not indicate ‘small’, as suggested by Schneider. However, within a paradigm, situated in the appropriate context, they may be considered as ‘low ranking’ constituents of the body designated by the base, as demonstrated by the data in (11):




‘ordinary church member’



‘church elder’






‘head of state’

3.5. Insignificant / Nonserious

22Insignificance is one of the pragmatic or evaluative (cf. Booij [2007]) interpretations associated with the diminutive. Such words when used in contexts refer to ideas, events or activities which are deemed to be of little importance. Mpεpεwa, for example, refers to ears of grain which are inadvertently left behind by harvesters in fields. They are considered as leftovers, and are not significant to the farmer who has probably had a bumper harvest. Likewise, adwumawa is used in reference to a job which is of little worth and possibly involves little effort. The speaker who uses these forms may thus express an attitude of lack of appreciation for the entity in question.




‘ears of grain left behind by harvesters’



‘a little thing, trifle’



‘an insignificant piece of work’

23As the data in (12) show, this sense of the diminutive relates mostly to non-human, inanimate entities. We, however, found one example (Christaller [1933: 171]) which refers to a human being – ɔhenewa ‘a small, petty king, prince, chief’ (formed from ɔhene ‘king/chief’).

3.6. Affection / Admiration / Disdain / Contempt

24Other evaluative meanings which may sometimes be associated with the diminutive are affection, admiration, disdain and contempt, as exemplified by (13-16).





‘dearly beloved child’







‘a stubborn person’





a despicable person’



‘a small, petty king, prince, chief’




‘a contemptuous way of referring to one who behaves like an older person, usually used in reference to females’

25Such meanings have been characterized by some writers as connotative or associative meanings (cf. Schneider [2003]). For instance, some diminutive forms may express smallness plus an attitude. Some scholers (for example Bybee [1985]) have suggested that a diminutive form must of necessity include the semantic feature ‘small’. The diminutive forms we have identified in Akan which express any of the above-mentioned attitudes such as dɔfowa ‘lover’ (13b) or abomfiawa ‘despicable person’ (15), do not necessarily convey smallness as part of the communicated meaning. We do side with Strang [1968: 136] on her observation that there is diachronic link between smallness and attitude, and that, through a grammaticalization process, diminutives have taken on a range of meanings “from affection through condescension to contempt.”

26It is worth noting that whatever attitude expressed by a diminutive form is the result of the interaction between the meanings of the base and the diminutive morpheme, and, in many cases, the linguistic and extra-linguistic contexts. Attaching the diminutive suffix to a base which basically means love can only evoke feelings of affection. Aniεdem-ba ‘brave + dim’ for instance, although put under admiration, may suggests an attitude of admiration or disdain depending on the communicative situation; the latter in a context where the referent is being bullish, and the former in a situation where the referent has exhibited bravery. The morphological make-up of aperewa is as follows: a-pere-wanom-to fidget-dim’. Since the base verb pere ‘to fidget’, has no positive connotations, adding the diminutive suffix to it still causes the resultant word to remain non-positive.

27The thought expressed in the preceding paragraph is consistent with Schneider’s [2003: 96-102] discussion of the English diminutive morpheme -let, in such words as, kinglet, princelet, dukelet, lordlet, bosslet, godlet etc., where he indicates that when these diminutive forms are used in reference to the substantive holders of these titles, they “are usually depreciatives, expressing a negative assessment of the referent and conveying contempt”. This use of the diminutive presents the referents as incompetent, unimportant and petty. The same can be said about the Akan example in (15b). However, these negative associations do not apply when the referents is a child. Schneider cites the following lines from a nativity play (from A.S. Byatt’s novel Still Life) where three children play the three Magi.

Three diminutive Kinglets, carrying a Kelly lamp, a silver sugar castor, Mrs Ellenby’s Chinese enamelled cigarette box, bowed, wobbled, kneeled (BYA 40)

28Schneider [2003: 100] makes the following observation:

  • 14 Emphasis added.

The adjectival modifier diminutive indicates that the actors are very young, the form kinglets conveys that they are not real kings. At the same time, this diminutive expresses affection towards the child actors and show that the person who uses this form is moved by watching them. Thus, while a deficit is expressed, this deficit is not evaluated in a negative way, since the relevant adult norms do not apply in the given situation.14

29This confirms the fact that the interpretation of the diminutive may depend on extra-linguistics considerations.

30The following sections provide a description of Jurafsky’s [1996] Radial category theory, and demonstrate how Akan fits into this model. Before then, however, we provide a summary of previous approaches to the analysis of the diminutive. It is worth pointing out that these sections draw very much on Jurafsky [1996].

4. Previous approaches to the analysis of the diminutive

  • 15 Schneider [2003: 1] also identifies this lack of attention to the pragmatics of the diminutive as o (...)

31Two main approaches to the characterization of the wide range of the semantics of the diminutive are identified. They are the Abstractionist and the homonymy approaches. The typical abstractionist view is represented by Chao’s [1947: 35], cited in Jurafsky [1996: 537] characterization of the Cantonese tone-shift diminutive as “that familiar thing one often speaks of.” Jurafsky critiques the abstractionist approach (cf. Chao [1947], Shetter [1959]), as often retreating to vague abstractions, when confronted with the seemingly unlimited range of the meanings expressed by the diminutive. Referring to Grimm’s [1967], characterization of the general meaning of the diminutive as ‘taking away something of the force of a word”, Jurafsky notes that “[T]here is much in Grimm’s abstraction that seems correct, particularly in modelling the approximation, resemblance, and hedging senses.” That notwithstanding, he [1996: 537] notes three instances of the use of the diminutive which Grimm’s abstractionist approach fails to account for. The first is “the individuating and deictic exactness cases” such as “the use of the diminutive on words meaning ‘now’ or ‘here’ to mean ‘exactly now’ or ‘exactly here’.” The second is the “intensification sense where the diminutive modifies words meaning ‘small’ to produce words meaning ‘very small’ (e.g., Latin parvulus small-DIM ‘very small’)”. The third, which, according to Jurafsky is the most problematic aspect of this abstractionist approach, is “its failure to cover any of the more pragmatic senses of the diminutive, such as the common affectionate or pejorative uses.”15

32The other abstractionist approach considered by Jurafsky relies on single abstract concepts such as ‘small’ or ‘child’ in characterizing the diminutive. This is illustrated by Wierzbicka’s [1984] argument that “metaphors from ‘small/child’ are the basis of the affection and contempt senses of Polish diminutives.” Whereas, Jurafsky agrees with the intuition that the concepts ‘child’ and ‘small’ are fundamental, he [1996: 538] observes that they are not enough to help account for the wide range of functions of the diminutive. He argues that “[W]ithout metaphorical, inferential, or abstractive extensions, ‘small’ cannot model the individuating or exactness sense, or the use of the diminutive to mark an ‘imitation’ of a natural object.” As he puts it, “[I]t is hard to imagine a definition which referred to ‘small’ in an abstract enough way to cover, for example, Spanish boca ‘mouth’/boquete ‘hole’; indeed a boquete can be larger than a boca.”

33The homonymy approach to the characterization of the multi-functionality of the diminutive, rather than building a single generalized abstract meaning for all senses of a diminutive morpheme, models each sense as a separate lexeme. That is, the ‘small’ sense of the diminutive is a separate lexeme from the ‘child’ sense. For this approach, it is assumed that the fact that synchronically each lexeme is composed of the same phonological material is “coincidental”. As Jurafsky observes, although the homonymy approach has the advantage of avoiding vague and insupportable generalizations, from a diachronic perspective, it is simply the wrong model for accounting for semantics of the diminutive, in the face of abundant evidence of the extension of the meaning of the diminutive over time. Further evidence against the homonymy approach comes from the fact that “the same varied and complex senses of the diminutive occur again and again across languages.” Therefore, Jurafsky [1996: 538] argues, “[I]f the different senses of the diminutive were unrelated, there would be no reason to expect similar groupings of senses in different languages.” Aside from its failure to account for the diachronic facts, “the homonymy approach fails to model the complex overlapping between senses that often occur. For example, the affectionate, contemptuous, and child-related senses of the diminutive are often present in words with the approximative, small, or individuating/partitive meanings [1996: 538-9]. And as we can see from the Akan examples, ‘feminine’ and ‘small’ or ‘feminine’ and ‘contempt’ sometimes overlap in a single diminutive form.

  • 16 Including diachronically motivated studies – Jurafsky’s [1988] account of Cantonese, Heine et al.’s (...)
  • 17 With a long history going back to Bréal [1897], cited in Jurafsky [1996]. Studies on change that le (...)
  • 18 See Jurafsky [1996: 539-541] for discussions of these lines of research.

34The foregoing shows, Jurafsky [1996: 539] argues, that both the strict abstractionist and the strict homonymy positions lack “the theoretical machinery for defining a polysemous semantic category, since they are forced to stake out some arbitrary position between abstraction and homonymy, pointing out some generalizations and avoiding others.” In response, two lines of research (polysemy-based account of the diminutive in individual languages16 and studies of universal tendencies in semantic change)17 have emerged, providing the background for accounting for what Jurafsky [1996] calls “the astonishing cross-linguistic regularity in the semantics of the diminutive as it extends beyond the meaning ‘small’” as well as the development of the various senses over time.18 These two lines of research are fundamental to the radial category approach which we discuss in the next section.

5. The Radial Category Theory

35In accounting for the varying semantics of the diminutive beyond the putative basic meaning of ‘small’ and the development of the various related senses, the radial category approach combines tenets of the two research paradigms mentioned in the previous paragraph (see also fn 19 and 20). On the one hand, it exploits the emphasis placed on the diachronic relation between the senses of polysemous morphemes by the unidirectionality hypotheses, with ordering constraints specifying which types of senses are derived from which others. On the other hand, it accounts for the synchronic relations between senses of the diminutive by focusing on the mechanisms of semantic change such as metaphor and inference.

5.1. The radial category

36The radial category is characterized as a graphic representation of a polysemous category which has internal structure consisting of a central sense of prototype together with conceptual extensions, represented by a network of nodes and links. “Nodes represent prototypes of senses, while links represent metaphorical extensions, image schematic transfer, transfers to different domains, or inferences” [1996: 542]. The radial category is seen both as a synchronic object, describing the motivated relations between senses of a polysemous category and as a diachronic object, capturing the generalization of various mechanisms of semantic change.

37Figure 1 is Jurafsky’s proposed universal structure for the semantics of the diminutive, determined on both synchronic and diachronic evidence. This is consistent with the expectation that the radial category will serve both diachronic and synchronic purposes. In this figure, nodes are labelled with names of senses and arcs with mechanisms of semantic change which include Metaphor (M), Inference (I), Generalization (G) and Lambda-abstraction (L).

Figure 1: Jurafsky’s proposed universal structure for the semantics of the diminutive

Figure 1: Jurafsky’s proposed universal structure for the semantics of the diminutive

38On the synchronic level, the radial category accounts for the various senses of the diminutive in an elegant manner by taking into account the kinds of metaphors and inferences, which relates the senses of the diminutive and thus motivate the senses themselves. This way, the radial category is able to account for the apparent paradoxical senses of the diminutive such as cases where the same morpheme marks both ‘intensification’ and ‘attenuation’. On the diachronic level, the centre of the radial category serves as the basic (historical) sense from which the semantic paths that the diminutive has taken over its development may be mapped.

5.2. On the predictive power of the radial category

39The radial category approach agrees with the observed unidirectionality of semantic change (cf. Heine et al. [1991]), showing that “the meanings of the diminutive in a particular language will develop diachronically from central senses toward senses on the edge of the category” (Jurafsky [1996: 543]). The universal radial category (Figure 1), for example, shows domain shifts from the central physical domain of ‘size’ to the domains of ‘gender’ ‘social power’ and ‘conceptual centrality’. Based on this, some cross-linguistic predictions can be made in respect of diminutives and their distribution. That is, languages will share common prototypes from which they may diverge in choosing coherent subsets of the universally-sanctioned lines of development, yielding connected graphs beginning with the prototype. This leads to the ultimate prediction that the central meaning of the diminutive, ‘child’, is historically prior and metaphorically and inferentially motivates the other senses (cf. Wierzbicka [1984]). As we show in §3, the diminutive in Akan originates from the word for child – ɔba and the closely related sense small, as predicted by Figure 1. It is from these that the various senses of the diminutive in Akan develop through regular semantic extension mechanisms like metaphor, inference and generalization (cf. Figure 3). For example, the insignificant sense of the diminutive in Akan can be shown to derive from the small sense by mean of the metaphor value is size, by which speakers conceptualise the value of an item in terms of size so that ‘big things’ are considered significant. In this context anything that is small, is considered insignificant by inference (I). This can be shown by the extracted graph in (Figure 2).

Figure 2: The path for the sense insignificant (extracted from Figure 1)

Figure 2: The path for the sense insignificant (extracted from Figure 1)

5.3. The radial category and semantic change

  • 19 We will not comment further on this because that will take us too far afield, considering the focus (...)

40The radial category has the virtue of being able to show, by means of various semantic extension mechanisms, how multiple senses of a morpheme develop and how they relate. Three such mechanisms are observed. One is metaphor, which is the means by which meaning shifts between domains by mapping the sense of one domain unto another domain. Another is inference or context-induced reinterpretation, which is the mechanism by which a morpheme acquires a new meaning through the conventionalization of what used to be a derived implicature of its old meaning. The third is generalization or bleaching which is the process by which a sense is created by dropping specific features of meaning and arriving at a meaning which is more general and less informative. In addition to these well documented mechanisms of semantic change (cf. Heine et al. [1991], Traugott & König [1991], Bybee et al. [1994]), Jurafsky [1996] argues that to account for certain senses of the diminutive such as approximation and exactness, a new mechanism – lambda-abstraction, which gives rise to quantificational and second-order meanings from propositional ones is needed19. In the following, we show the extent to which the senses of the diminutive in Akan may be accounted for by these mechanisms of semantic change.

6. The radial category and the senses of the diminutive in Akan

41In section three, we outlined the senses of the diminutive that we found in the corpus on which the present paper is based. They include: small, young, feminine, member, insignificant/nonserious, affection/admiration, and disdain/contempt. In this section we attempt to show which of the mechanisms of meaning change – metaphor, inference or generalization – is at work in the development of these senses. We argue that metaphor and inference are the two main mechanisms responsible for these meaning extensions. These are discussed in turn.

6.1. The radial category and semantic change

42Various metaphors may be employed to map senses of a polysemous morpheme. In the case of the diminutive, Jurafsky notes two types of metaphors that may be used. They are metaphors for gender and metaphors for centrality and marginality. We discuss these in turn.

6.1.1. Metaphors of gender and the diminutive

43Jurafsky observes that the link between women and the diminutive rests on the metaphor in (17), by which women are conceptualized as children.


women are children/small things

44The full import of this metaphor may not be wholly applicable to Akan. It would only, work for Akan with some modification. This is because in Akan women are not seen as ‘children’ per say, as the use of this metaphor suggests. As the discussion in §3.3 shows, women are only sometimes regarded as not fit to handle certain “big things/issues” (cf., fn.16). The metaphor for mapping the relation between female gender and the diminutive in Akan, therefore, may be stated as in (18a).




women are fit for small/nonserious things20


value is size / size is value

45It may be recalled that we referred to Booij’s [2007] observation that the link between the diminutive and female gender may probably be because women are, in relative terms, physically smaller than men. We believe that in Akan, the link may not just be physical, and that underlying the supposed conceptual link between the diminutive and female gender is a tacit value judgement, where the size of the referent is assumed to equal its value (significance). In that sense, the referent of the diminutive will be assumed to be insignificant, as captured by the metaphor in (18b). The foregoing is consistent with Dressler & Merlini Barbaresi’s [1994] treatment of the diminutive in terms of the semantic category small and pragmatic category nonserious.

6.1.2. Metaphors and the use of diminutives for group membership

46The other group of metaphors which Jurafsky [1996] discusses deal with centrality and marginality. They include metaphors like;


social groups are families


category centrality is size

  • 21 See Jurafsky [1996: 547-548] for discussions.

47The metaphor in (19) models membership in social groups as a family, with some being central, and others marginal, and the diminutive may be used to mark either.21 With the metaphor in (20), centrality of a category is linked to size, so that central or prototypical members are large whilst peripheral members are small.

48Whereas the metaphor in (19) works for Akan, as will be shown below, the one in (20) is somewhat off the mark. That is because in Akan, entities that Jurafsky calls central categories (e.g., parents in the family) are not necessarily more central than others (e.g., children). Again, asɔreba ‘church member’, ɔmamba ‘citizen’, etc are not necessarily considered as marginal(ized) members of the ‘groups’ they belong to. We believe, therefore, that the member sense of the diminutive in Akan comes from the conceptualization of social groups as families (19), where the base names the family of which the referent of the diminutive form is a member. This is borne out by the fact that the different senses of the diminutive derive from the core meaning of ‘child’, and a child is born into a ‘group’ – family. It is actually an anomaly in the traditional Akan society to find a child without a family. Indeed, the very notion of ‘child’ conjures a sense of belonging to a larger group – the family. In this sense, the relation between the base and the diminutive may be construed in terms of category centrality but only to the extent that category centrality is viewed in terms of the metaphor in (21) where the central member is seen as a parent and the diminutivized member is seen as a ‘child’.


category centrality is parenthood

49The same metaphor explains the small-type-of sense of the diminutive. That is, if A is a ‘child’ of B, then, up to a certain age, A is a small-type-of B and then also A is related to B.

6.2. Conventionalization of inference

50Conventionalization of inference is described as the process by which a morpheme acquires a meaning that had been an inference or implicature of its old meaning. It has its roots in Grice’s [1975] discussion of conventionalized implicature, where it is argued that the literal meaning of a construction often develops from the institutionalization of a conventionalized implicature on the part of the addressee. Based on the claim that the diminutive originates from a morpheme meaning ‘child’ or signifying ‘child’ in some way, Jurafsky [1996: 551] argues, for example, that the ‘affection’ sense of the diminutive developed through the conventionalization of implicature. As he puts it, “[G]iven this core sense, then, and given the natural tendency to feel affection towards children, a hearer hearing a core diminutive referring to children, will draw the natural inference that the speaker feels affection toward the diminutivized object (child).” In Akan, this ‘affection’ sense of the diminutive can be inferred from the ‘child’ sense of the diminutive. The same can be said about the other pragmatic senses of the diminutive (the bottom half of the radial category – Figure 3), which are arrived at through inferences from the more substantive senses of the diminutive (the top half of the radial category). Our difficulty is with appreciating the extent to which these more pragmatic senses of the diminutive are conventionalized. Unlike the description given by Jurafsky, these senses are not really conventionalized, so that when the respective diminutive is used, the putative conventionalized meaning is evoked. We reckon that they are still context-dependent. The ‘disdain’ or ‘contempt’ senses of the diminutive in Akan are only inferred from the ‘small’ sense. As indicated in §3.6, the diminutive form aniEdemba ‘a stubborn person’ may invoke ‘admiration’ or ‘contempt’ depending on one’s assessment of the prevailing circumstance. These senses are not conventionalized, though they are the results of inference.

6.2.1. Conventionalized inferred meaning and the ‘small-type; sense of the diminutive

51Before we end the discussion, we will want to comment on one type of diminutive use that is really conventionalized. This is what Jurafsky calls the ‘small-type’ sense, where the diminutivized noun is a smaller form of the base with which it shares form and function. They are exemplified by the data in Table 2 (Jurafsky [1996: 552, Table 14])

Table 2: Lexicalized classificatory diminutives






toi 35














‘garbage can’



















mUt UkwE

‘tree, stick’

mUt Ukos.


52Rhodes [1990], cited in Jurafsky [1996], refers to these as classificatory diminutives because the diminutivized objects are classified in the same ontological hierarchy as the larger one. However, as Jurafsky [1996: 552] observes,

these are not just cases where a language marks two objects as being identical except for variation in size; in other words these are not concepts which are ‘-emically’ the same. In each case, the language distinguishes between a smaller version of an object, [which may be] marked with an adjective meaning ‘small’, and the diminutive, which marks a separate concept.

53This class of ‘small-type’ diminutives are lexicalized. “The marker may begin by purely marking size, but eventually the diminutive form becomes frequent enough that it becomes susceptible to lexical drift” [1996: 552]. The Akan examples in (22) are illustrative of the lexicalized ‘small-type’ diminutives.






‘nail’ (AS)





‘canoe’ (FA)

54These are the real examples of lexicalized diminutive forms. That is, they started out as diminutive forms but have taken on a life of their own becoming integrated into the lexical system of the language.

6.3. A note on Generalization

55There is a bit of generalization in the conceptualization of some of the senses of the diminutive. We argued (§ 3.3) that women are assumed not to be fit for some important things. Therefore, there is the tendency for some to generalize this attitude, regarding issues relating to or from women (or even women themselves) as nonserious. The same can be said for the ‘small’ sense of the diminutive as it relates to children. In a society where children are supposed to be seen and not heard, anything from children, who are generally ‘small’, may generally be assumed to be ‘nonserious’. The related-to sense of the diminutive may also be a generalization of the ‘member’ sense and ‘small-type-of’ sense.

6.4. The semantics of the Akan diminutive – a radial category

56The discussions far has shown that the diminutive in Akan fits pretty well into the universal radial category (Figure 1) in that the core sense of ‘child’ ultimately underlies all the senses and extensions of the diminutive in Akan. However, the details of the radial category representing the semantics of diminutive in Akan will need slight tweaking, especially with regard to the metaphors that motivate the various senses of the diminutive. We suggest a radial category specific to the semantics of the Akan diminutive (Figure 3), based on both language internal and cross-linguistic evidence.

Figure 3: Proposed structure for the semantics of the diminutive in Akan

Figure 3: Proposed structure for the semantics of the diminutive in Akan

7. Conclusion

57The present paper has been concerned with the form, origin and meanings associated with the diminutive in Akan. We have suggested that the diminutive affix -wa/-ba originates from the noun ɔba ‘child’, an assertion which is consistent with earlier observations by researchers such as Jurafsky [1996], Heine et al. [1991], etc. The senses of the diminutive form identified in our corpus is varied, ranging from more concrete meanings such as small, young, feminine, member to more evaluative ones like insignificant/nonserious, affection/admiration, and disdain/contempt.

58Jurafsky’s Radial Category model allows for an adequate account of the different categories of meanings as we are able to relate the diachronic meaning to the synchronic ones as well as link the various synchronic meanings. The major language change mechanisms that allow for such linkages are identified as metaphors and inferences. Crucially, we acknowledge that even though the proposed semantic structure for the Akan diminutive, to a large extent, fits into Jurafksy’s proposed universal structure for the semantics of the diminutive, some adjustments will have to be made, taking into consideration certain language internal specificities, to arrive at a workable structure for Akan.

59We hope the analysis of the Akan diminutive, presented in this paper, following Heine et al.’s [1991] examination of the Ewe diminutive will stimulate further research into the form and meanings of diminutives in languages of the Volta Basin.

Top of page


Akan dictionary: pilot project, Accra, Department of Linguistics, University of Ghana, 2006.

Amfo Nana Aba Appiah, “Recurrence Marking in Akan”, Pragmatics 15: 2/3, 2005: 151-168.

Amfo Nana Aba Appiah, “Noun Phrase Conjunction in Akan: the Grammaticalization Path”, Pragmatics 20:1, 2010: 27-41.

Booij Geert, The grammar of words: an introduction to morphology, 2nd edition, Oxford, Oxford University Press, 2007.

Bréal Michel, Semantics, Reprint, New York, Dover, 1964 [1897].

Bybee Joan, Morphology: a study of the relation between meaning and form, Amsterdam, John Benjamins, 1985.

Bybee Joan, Perkins Revere & Pagliuca William, The evolution of grammar: tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press, 1994.

Chao Yuen Ren, Cantonese primer, Cambridge, MA, Harvard University Press, 1947.

Christaller Johann G., Dictionary of the Asante and Fante language called Tshi (Twi), 2nd edition, Basel, 1933.

Coleridge Hebert, “On diminutives in ‘LET’”, Transactions of the Philological Society, 1857: 93-115.

Contini-Morava Elle, “Noun classification in Swahili”, 1995 :

Dressler Wolfgang U. & Merlini Barbaresi Lavinia, Morphopragmatics: diminutives and intensifiers in Italian, German and other languages. Berlin, Mouton de Gruyter, 1994.

Frajzyngier Zygmunt, “The de dicto domain in language”, in Traugott Elizabeth C. & Heine Bernd (eds.), Approaches to grammaticalization, Vol 1, Amsterdam, John Benjamins, 1991: 219-251.

Givon Talmy, “Serial verbs and semantic change: Niger-Congo”, in Li Charles N. (ed.), Word order and word order change, Austin, University of Texas Press, 1975: 47-112.

Greenberg Joseph H., “Africa as a linguistic area”, in Bascom William R. & Herskovits Melville J. (eds.), Continuity and change in African cultures. Chicago, University of Chicago Press, 1959: 15-27.

Grice Paul H., “Logic and conversation”, in Cole Peter & Morgan Jerry L. (eds.), Syntax and semantics, vol. 3: Speech acts, New York, Academic Press, 1975: 41-58.

Grimm Jacob, Deutsche Grammatik, 4 vols. Reprint, Hildesheim, Georg Olms Verlagsbuchhandluch, 1967 [1890].

Heine Bernd, Claudi Ulrike & Hünnemeyer Friederike, Grammaticalization: a conceptual framework, Chicago, University of Chicago Press, 1991.

Heine Bernd & Reh Mechthild, Grammaticalization and reanalysis in African languages, Hamburg, Helmut Buske, 1984.

Hopper Paul J. & Traugott Elizabeth C., Grammaticalization, Cambridge, Cambridge University Press, 2003.

Jurafsky Daniel, “Universal tendencies in the semantics of the diminutive”, Language, 72: 3, 1996: 533-578.

Lewis G. C., “On English Diminutives”, The Philological Museum 1, 1832: 679-686.

Matisoff James, “The mother of all morphemes: augmentatives and diminutives in areal and universal perspective”, in Ratliff Martha & Schiller Eric (eds.), Papers from the first annual meeting of the Southeast Asian Linguistic Society, Tempe, AZ, 1991: 293-349.

Rhodes Richard A., “Lexical hierarchies and Ojibwa noun derivation”, in Tsohatzidis Savas L. (ed.), Meanings and prototypes, London, Rutledge, 1990: 151-158.

Schneider Klaus P., Diminutives in English, Tubingen, Max Niemeyer Verlag, 2003.

Shetter William, “The Dutch diminutive”, Journal of English and German Philology 58, 1959: 75-90.

Strang Barbara, Modern English structure, 2nd edition, London, Arnold, 1968.

Sweetser Eve, From etymology to pragmatics, Cambridge, Cambridge University Press. 1990.

Timyan Judith, A discourse-based grammar of Baule: the Kode dialect, PhD dissertation, City University of New York, 1977.

Traugott Elizabeth C., “From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticalization”, in Lehmann Winfred & Malkiel Yakov (eds.), Perspectives on historical linguistics, Amsterdam, John Benjamins, 1982.

Traugott Elizabeth C. & König Ekkehard, “The semantics and pragmatics of grammaticalization revisited”, in Traugott Elizabeth C. & Heine Bernd (eds.), Approaches to grammaticalization, vol. 1, Amsterdam, John Benjamins, 1991.

Wierzbicka Anna, “Diminutives and depreciatives: semantics representation for derivational categories”, Quanderni di Semantica 5, 1984: 123-130.

Top of page


1 See Schneider [2003] and Dressler and Merlini Barberesi [1994] for fuller bibliographical reports.

2 Research on Diminutives in African languages has mostly concentrated on Bantu languages particularly Swahili, (see Schneider [2003] and the references therein). Heine et al. [1991] is an exception; it provides, among other things, an analysis of the Ewe (Niger-Congo, Kwa) diminutive -vi in the context of grammaticalization.

3 Akan is a cover term for at least eleven dialects. Three of these dialects – Fante, Asante and Akuapem are considered as major dialects simply due to their literary status; the latter two (along with others not mentioned here) are often referred to collectively as Twi. In this paper, Twi is used in specific reference to the Asante and Akuapem dialects.

4 Christaller [1933] referred to the Akan language as Twi, with the Asante and Fante as dialects.

5 Cf. studies regarding the auxiliary verb san ‘return’ and the comitative verb n(y)e ‘be with’, Amfo [2005], [2010] respectively.

6 Details of the components of these two groups and the semantics of their ‘members’ is the subject of another study in progress.

7 DIM stands for diminutive.

8 The full form including the prefix is used mostly for human entities and in utterance initial position, otherwise ba is used. Note that in Akan the word ɔba can only mean ‘child’ in the sense of offspring. ‘Child’ in the sense of ‘a person below a certain age’ is abofra (Twi), abofraba (Fa.)

9 Baule, just like Akan, descends from the Central Tano group of the Kwa branch of the Niger-Congo phylum.

10 See Schneider [2003], Jurafsky [1996], Booij [2007], for discussion.

11 In certain contexts, this word could be used derogatorily in reference to a female individual. This, of course, follows from the idea that the job of a maid is by no means prestigious, and may even be despised.

12 This phonological observation is limited to the context of female Fante names; we do not have enough evidence at this point to generalize for other contexts.

13 This is not to suggest that women have absolutely no significant roles to play in Akan traditional societies. Indeed, they sometimes do; in many cases they are the kingmakers, however a female cannot, as a matter of traditional practice, ascend the throne.

14 Emphasis added.

15 Schneider [2003: 1] also identifies this lack of attention to the pragmatics of the diminutive as one of three main problems found with the analyses of diminutives, noting that “diminutives have not, as a rule, been studied from a pragmatic perspective.”

16 Including diachronically motivated studies – Jurafsky’s [1988] account of Cantonese, Heine et al.’s [1991] analysis of Ewe diminutive, Matisoff’s [1991] study of Thai diminutives, Contini-Morava’s [1995] account of Swahili noun classes (including diminutives), as well as Dressler & Merlini Babaresi’s [1994] synchronic account of the diminutive in Italian, German and English.

17 With a long history going back to Bréal [1897], cited in Jurafsky [1996]. Studies on change that leads to grammaticalization view change as unidirectional. This view of change in terms of UNIDIRECTIONALITY leads to the view that predictions can be made about the direction of change along different axes. One view, called BLEACHING (Givón [1975]), DESEMANTICIZATION (Heine & Reh [1984]), or GENERALIZATION (Bybee et al. [1994]), claims that meaning changes from the more informative and specific to the more abstract and vague. Another view of unidirectionality of change “focuses on the tendency of semantic change to proceed from the ‘real’ physical or spatial world or the ideational domain to create more qualitative, evaluative, and textual meanings (Traugott [1982]; Sweetser [1990]; Frajzyngier [1991]; Heine et al. [1991], Hopper & Traugott [2003]).”

18 See Jurafsky [1996: 539-541] for discussions of these lines of research.

19 We will not comment further on this because that will take us too far afield, considering the focus of the present paper. The interested reader may consult Jurafsky [1996: §4.4].


21 See Jurafsky [1996: 547-548] for discussions.

Top of page

List of illustrations

Title Figure 1: Jurafsky’s proposed universal structure for the semantics of the diminutive
File image/png, 105k
Title Figure 2: The path for the sense insignificant (extracted from Figure 1)
File image/png, 9.0k
Title Figure 3: Proposed structure for the semantics of the diminutive in Akan
File image/png, 96k
Top of page


Electronic reference

Clement K. I. Appah and Nana Aba Appiah Amfo, The Morphopragmatics of the Diminutive Morpheme (-ba/-wa) in AkanLexis [Online], 6 | 2011, Online since 27 March 2011, connection on 26 March 2023. URL:; DOI:

Top of page

About the authors

Clement K. I. Appah

Lancaster University, England

Nana Aba Appiah Amfo

University of Ghana, Ghana

Top of page



Creative Commons - Attribution-NonCommercial-NoDerivatives 4.0 International - CC BY-NC-ND 4.0

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search