A diachronic perspective on ‘prosodies’ in Central Chadic languages (Afroasiatic)

1.1 Aims of the paper^[2]

The paper aims to illustrate, first of all, that ‘prosodies’ - widely accepted as being characteristic typological features of Chadic languages (and those of the Central Chadic branch in particular) - can be reconstructed to have operated already and extensively on the level of the reconstructed proto-language ‘Proto-Central Chadic’ (PCC). The considerable age and distribution of such prosodies in Central Chadic languages had been assumed since the early 1980s (see Wolff 1981, 1983) and was more recently made one of the basic hypotheses on which Richard Gravina (2014) based his pioneering PhD thesis at the University of Leiden. An alternative approach to historical Central Chadic phonology and lexical reconstruction was very recently proposed in two book-length studies by the author (Wolff 2022a, 2024). In these studies, some of Gravina’s basic assumptions and reconstructions are being seriously questioned. The author proposes a unified theory of Central Chadic historical phonology, which integrates various typological and likely genetic features that are reconstructed as characteristic for these languages and which Gravina had passed by, namely

1. a particular type of minimal vertical vowel system (see Section 2.1);

2. an early root-and-pattern structure for simple roots (see Section 2.2);

3. a tendency of certain reconstructed proto-language segments to prosodise, i.e. to allow certain articulatory features to disassociate from the original host segment and become floating as supra-segmental prosodies and thereby free to re-associate with other host segments in the phonemic chain of the word (see Sections 2.3, 3, and 4).

Secondly, and again at variance with Gravina’s study, the paper makes explicit reference to and illustrates in much detail the historical underlying agglutinative morphology of the proto-language, many of whose reconstructed pre-, in- and suf- fixes have diachronically petrified and fused with simple lexical roots to form words in the present-day languages that in synchronic perspective appear to be monomorphemic. Following an earlier suggestion by Schuh (2002), we consequently distinguish between phonological and morphological prosodies. This distinction rests on the observation that both root-internal and affixal segments may prosodise and thereby become the source of a prosody (see Section 5).^[3]

Thirdly, while Gravina restricted his studies primarily to the effects of pal (Y-prosody), which he reconstructs as phonemic for PCC, while not accepting lab (W-prosody) as being of equal phonological status, our own research established a total of four relevant prosodies, namely pal, lab, nas, and, for the first time, glot.

Fourthly, the paper aims to show how prosodies contribute to the emergence of a number of innovative segmental phonemes, both vowels and consonants, in modern Central Chadic languages, which are not reconstructed for the common proto-language and, therefore, require a historical explanation as to their diachronic development (see Section 6).

Fifthly and more or less in passing, the paper aims to show how shared prosodic ‘colouring’ of plain segments, governed by the regime of prosodies, creates ‘harmonising’ effects within the phonological word. Such harmonisation effect is particularly significant in the case of shared vowel qualities that may encompass the whole word. Our suggested unified prosodic approach questions the validity of alternative analyses, which claim that in a number of Central Chadic languages we are dealing with manifestations of a ‘vowel harmony’ system. Our studies show that, at least in terms of diachronic evolution, we are not dealing with different sets of vowels, which would harmonise across morpheme boundaries between stems and affixes, as is the case, for instance, in more narrowly defined ATR-type vowel-harmony systems. Rather, we are dealing with harmonising effects due to the operation of a limited number of prosodies across the phonological word.

1.2 The notion of ‘prosody’ in Chadic linguistics

It needs to be pointed out that in Chadic linguistics, at variance with more general usages in theoretical linguistics, the notion of ‘prosody’ is quite narrowly defined in terms of specifically referring to phonological processes of palatalisation (pal, Y-prosody), labialisation (lab, W-prosody), (pre-) nasalisation (nas, N-prosody), and glottalisation (glot, ʔ-prosody). In this section, therefore, we sketch out the typological nature and linguistic history of these Chadic prosodies. Qualifying as a ‘prosody’ means that the domain of a given linguistic feature exceeds that of a single segment; its maximal domain is the morpheme or the phonological word. The notion of prosody was introduced into Chadic linguistics in Nigeria in the 1960s and early 1970s in application to initially only very few languages of the sub-branch A of Central Chadic.^[4] It was first used, to the best of the author’s knowledge, by the late Carl Hoffmann (1925–2007) in an influential yet unpublished conference paper of 1965 on the phonological analysis of the Higi language. A mimeographed copy of that paper was circulated among members of the West African Linguistics Society (WALS) including the present author, who eventually passed it on to the late Daniel Barreteau (1950–2007) to be consulted by him for his extensive work on Central Chadic-A languages in Cameroon. The prosodic approach immediately gained currency among linguists of the former Summer Institute of Linguistics (SIL) in Nigeria, as witnessed by the works of Roger Mohrlang on Higi and James Hoskison on Gude in the early 1970s and is being successfully applied in the work of SIL linguists in Northern Cameroon and Chad to this day (cf. Roberts 2001).

Following Hoffmann’s (1965) lead by bringing the notions of palatalisation (pal), labialisation (lab), and (pre-)nasalisation (nas) prosodies into play, it became possible to reduce the suspiciously large inventory of 93 consonants in Higi to 26 basic consonants, which could come under the ‘colouring’ effects of pal, lab, and nas, either as a single prosody or as a combination of any of these prosodies. Hoffmann’s pioneering approach to Higi phonology stimulated follow-up studies on the same language by Mohrlang (1971, 1972) and Barreteau (1983), who extended the prosodic approach to also include the analysis of vowels. It was soon discovered that in particular pal (Y-prosody) also played an important role in the morphology of certain other Central Chadic-A languages, such as in Gude (Hoskison 1974, 1975) and in Ga’anda (Newman 1977). Since then, it has become established wisdom in Chadic linguistics that at least a number of Central Chadic-A languages tend to be typologically characterised by a pervasive and complex regime of so-called morpheme- and word-level prosodies – minimally in the sense of operational pal and lab. Recent advances in comparative Chadic linguistics show that we are dealing with a much more widely spread phenomenon in Chadic as a whole.

1.3 Observing the Saussurean Firewall: synchronic versus diachronic perspective

Given the considerable age of the Chadic language family within the Afroasiatic phylum, and quite likely also due to various still unidentified scenarios of intra- and extra-Chadic language contact, there is massive typological diversity among Chadic languages as a whole, and particularly so among Central Chadic languages. For instance, synchronic inventories of apparently contrastive vowels in Chadic range from one to seventeen (cf. Wolff 2017), displaying various intermediate stages in the historical development from ‘minimal’ via ‘intermediate’ to ‘maximal’ vowel systems (to adopt terms suggested by Schuh 2017). More generally speaking, phonological and morphological features that we identify with the common proto-language may be still operative in certain modern Central Chadic languages, while in other present-day languages these features have become obsolete and have left little if any traces of their previous existence behind. Occasionally, such ‘archaic’ features can only be detected on the basis of very abstract phonological analyses and by applying the method of internal reconstruction. Consequently, but rarely, we also find some of the prosodies under review being synchronically still productive in the phonologies of present-day Central Chadic languages, while in the majority of languages they can only be reconstructed as having been operative in periods historically preceding the modern phonological systems and, therefore, being of only diachronic relevance. For instance, while in Central Chadic Moloko (Friesen et al. 2017) the synchronic system with only one phonemic vowel */a/ appears to closely resemble the situation in the proto-language PCC, for some other Central Chadic languages like Mandara (Wolff and Naumann 2004) and Lamang (Wolff 2015) the restrictions regarding the distribution of vowels in the synchronic system can be related back to effects from diachronically older stages: for this we need to apply the methods of internal reconstruction and deep-level abstract phonological analysis. Nonetheless and synchronically, the modern languages are best being described as operating multi-vowel systems rather than minimal vowel systems. Again, other Central Chadic languages appear to show synchronically no traces of the proto-system, at least as far as reliable phonological analyses of sufficient abstraction are available for these languages. Therefore and depending on the individual language under research, a prosodic approach may be legitimate and be called for on both sides of the ‘Saussurean Firewall’, which separates synchronic from diachronic analyses and descriptions.

In the present paper, the focus is on the diachronic development of phonological units and processes that can be assumed to have been operational on the level of Proto-Central Chadic (PCC) and whose effects can still be observed in specific synchronic reflexes in modern Central Chadic languages. We, therefore, make sure to explicitly identify all historically reconstructed diachronic representations of data by a preceding asterisk symbol ‘*’. Data without preceding ‘*’ are associated with a synchronic perspective.

Note that for reasons of methodological stringency and clarity, we relate the transcriptions of available synchronic data from present-day languages directly to the level of the reconstructed proto-language (PCC), i.e. without using tentative group-level or sub-group levels of sub-classification to be associated with intermediate steps of reconstruction and, thereby, with certain hypothetical diachronic developments.^[5] The reason is to avoid circular reasoning and aprioristic interference from suggested sub-classifications, which might influence linguistic reconstruction. Sub-classification must be based on reliable reconstruction and not vice versa. Hitherto, all available sub-classifications of (Central) Chadic languages must be considered doubtful, as long as they are based on only selective sound correspondences, which often do not apply as ‘regular’ to all languages of a postulated language group and, thereby, allow to question the validity of the group as a whole within the classic dendrogram model. (For critical issues regarding the application of the Neogrammarian-type comparative method to [Central] Chadic languages see Wolff [2022a: 38–41, 158–186; Wolff 2024: 1–26; Wolff 2023]).

Arguably, some modern Chadic languages from the different branches of the family conceal the minimal vowel system origin of their presently much richer synchronic vowel inventories. They no longer show traces of a regime of prosodies, nor do they share the inherited morphosyntax type of the proto-language. Why and how they have lost such inherited linguistic features is a question of individual language histories. This, however, is not the topic of the present paper, which focuses on explaining and illustrating how the regime of prosodies that is reconstructed for PCC has affected the historical development towards the modern languages. The idea is to trace synchronic phonetic realisations ultimately back to the reconstructed segmental structure of lexical items from the proto-language. Regarding the presentation of linguistic data, therefore, we use basically four levels of analysis and discussion (for illustration, we refer to ex. 8 further below), namely

1. The synchronic level of (near-) phonetic transcription of the data as found in the database, which are always quoted in bold, e.g. Sukur ʃu ^m but ‘hair’.^[6]

2. The hypothetical diachronic prosodic representation, which identifies the relevant prosodies by raised symbols preceding the slant lines of the reconstructed phonemic representation. This level represents the post-prosodification stage of the individual proto-language, e.g. Proto-Sukur * ^nwy /sbt/.^[7]

3. The level of the hypothetical diachronic segmental reconstruction including all tentatively reconstructed affixal elements, i.e. the pre-prosodification stage, e.g. Proto-Sukur *ma-sbt-y-kʷ. ^[8]

4. The ultimate historical level of the Proto-Central Chadic simple root (without any of the potential affixal elements), e.g. PCC *s(a)b(a)ta ‘hair’.

1.4 Composition of the paper

Following this introduction, the paper will discuss in Section 2 some salient structural features of Proto-Central Chadic, namely its minimal vowel system (Section 2.1), its root-and-pattern structure (Section 2.2), and prosodies in their specific application to Central Chadic languages (Section 2.3). In Section 3 we will identify the reconstructed historical segmental origins of the suprasegmental features of pal (palatalisation), lab (labialisation), nas (prenasalisation), and glot (glottalisation) and provide illustrative examples for their ‘floating’ across a given word. Section 4 illustrates cases of multiple prosodies governing the same word. The examples display instances of rightward spreading, leftward anticipation, and combined spreading and anticipation of prosodies. Section 5 focuses on the agglutinative morphology of Proto-Central Chadic and illustrates how morphological prosodies emerge and may cover the whole segmental chain that makes up the phonological word. Section 6 discusses the emergence of innovative synchronic phonemes that were not reconstructed for the proto-language. Section 7 contains a summarising conclusion.

2.1 A minimal vertical vowel system

In terms of phonological typology, Central Chadic languages are, first of all, characterised by a minimal vertical vowel system based on tongue height that we reconstruct for Proto-Central Chadic (PCC) and which was possibly inherited from Proto-Chadic (PC):

[-low] *ə

[+low] */a/

The two vowels are reconstructed as being short; vowel length was not phonemic. Both represent central vowel qualities, whence we refer to such a system as ‘vertical’ system. However, both reconstructed vowels differ in phonological status: */a/ is phonemic, while *ə functions as highly predictable Svarabhakti (intrusive vowel) and is referred to as non-phonemic ‘pro-/epenthetic’ vowel. A regime of two prosodies operating in the proto-language (and also still operating in a few present-day languages), namely pal and lab, provide the reconstructed (and, in a few cases, synchronically underlying) minimal vowel system with a much richer set of variants in phonetic surface vowel realisations that allow for the various transcriptions by different linguists, see Table 1.

For a cross-linguistic survey of minimal vowel systems that typologically embeds the Central Chadic situation in a global perspective, see Wolff (forthcoming).

Prototypically, the historical underlying minimal (vertical) vowel system will operate with either one phonemic vowel */a/ or two */a, ə /. The phonological status of *ə (‘schwa’) is in doubt and depends on one’s view: Some linguists for individual languages describe schwa as phonemic, some as non-phonemic. In addition, this diachronic underlying minimal vertical vowel system is complemented by two high vowels [i] and [u], which occur on the proto-language level as much as in phonetic surface realisations of modern languages and thus give the impression of a triangular system in which front and back-rounded vowels play a role, see Table 2.

However, the high vowels are not reconstructed as vowel phonemes in PCC, because they are consistent conditioned allophones of the reconstructed approximants */y/ (IPA /j/) and */w/ in syllable nucleus positions.^[9] In individual modern Central Chadic languages, the historical underlying minimal vowel systems tend to be hidden behind a plethora of surface vocalic realisations with occasionally more than ten phonetic vowels, such as IPA [a, ε, æ, e, ɔ, o, ə, ɪ, ʊ, i, u, œ, ø, y].^[10] Such rich surface vowel inventories are the result of the ‘colouring’ effects of lab (W-prosody) and pal (Y-prosody) on the historical underlying minimal vowel system. Y-prosody accounts for raising and often simultaneous fronting (*a → [æ∼ɛ∼e], *ə → ɨ∼ɪ∼i]), W-prosody for raising and consistent back-rounding (*a → [ɔ∼o], *ə → ʊ∼ʉ∼u]). For details see Wolff (2022a, 2024.

2.2 An early root-and-pattern structure

In terms of root and word structure, PCC operated an early and simple root-and-pattern system, which it apparently inherited – via Proto-Chadic (PC) – from its ultimate genetic Proto-Afroasiatic predecessor. In root-and-pattern systems of the Afroasiatic type, the ‘root’ is made up of consonants (incl. the approximants *y and *w) and carries the basic semantics of the word. Vowels are added (forming ‘patterns’) and may, depending on their quality and position in the root, modify the basic meaning (and specify the grammatical category) of the word. For PCC, only */a/ could be used for forming vocalisation ‘patterns’ by being inserted in root-medial inter-consonantal slots.^[11] The formations resulting from such ‘a-vocalisation’ patterns are here referred to as ‘root types’. Present-day Central Chadic languages may show reflexes of more than one root type per reconstructed lexical item, i.e. display parallel historical underlying forms for the same lexical item: PCC *m(a)ta ‘to die’, for instance, has reflexes in modern languages of the two possible root types *mta and *mata.^[12] Approximants (*/y/, */w/) that form part of roots may end up in syllable-nucleus position, where they are consistently realised phonetically as high vowels [i] and [u]. Consequently, separate high vowel phonemes are not reconstructed for PCC (see Section 2.2 above). Potential inter-consonantal vowel slots that are not filled by */a/, [i] (← */y/) or [u] (← */w/) may be filled by epenthetic *ə, which is considered to be of non-phonemic status on the level of the proto-language (for details see Wolff 2022a, 2024).

2.3 Prosodies as typologically characteristic and inherited features

From beyond the Central Chadic branch in whose languages prosodies abound, corroborative yet somewhat rare evidence is provided also for West Chadic languages (Schuh 2002) and for East Chadic and Masa languages (Roberts 2007, 2009). These studies strongly suggest that prosodies are an early feature of Chadic languages to be reconstructed for the common proto-language (PC), in particular with regard to lab (W-prosody) and, most of all, pal (Y-prosody).

In addition to these two widely accepted prosodies, prenasalisation of obstruents (N-prosody) was already reported for Higi (Barreteau 1983; Hoffmann 1965; Mohrlang 1971, 1972) and for languages of the formerly so-called Bura-Margi group (Hoffmann 1987). Very recently (Wolff 2022a, 2024), glottalisation (glot, ʔ-prosody) was discovered to show a behaviour similar to that of the other three prosodies and, therefore, also to play a role in the history of certain Central Chadic languages. Both N-prosody and ʔ-prosody, however, are restricted to having effect only on consonants and approximants and apparently may only affect one segment in the word, while Y- and W-prosodies may affect both consonants and vowels and more than one segment in the word. nas and glot can be used to explain the presence of synchronic prenasalised obstruents and glottal consonants in modern Central Chadic languages that are not reconstructed to have been part of the proto-language’s phonemic consonant inventory (see Section 6). The ultimate origin of prenasalised obstruents, which are widely distributed among modern Chadic languages, as well as the origin of some rare glottal consonants that appear in a few modern Chadic languages only, had hitherto remained unsolved issues in historical Chadic linguistics.

According to the unified theory proposed by the current author (Wolff 2022a, 2024), all prosodies in Chadic languages have their origin in reconstructed non-vocalic segments of the proto-language, which underwent either partial or complete desegmentalisation and prosodification (see Section 3). This means that certain articulatory features such as pal, lab, nas, and glot could divorce from the original segment and become floating over domains larger than a single segment such as syllable, morpheme or word, re-associating with one or more segments elsewhere in the segmental chain of phonemes in the morpheme or word. When and why some segments prosodise and when and why they do not remains an open question; so far, no conditioning factors have been identified. According to our most recent studies (Wolff 2022a, 2024), prosodic effects in Central Chadic are fourfold and of two types:

Type 1 prosodies potentially affect all segments. These are lab (W-prosody) and pal (Y-prosody); they are the most common prosodies across Chadic languages.

Type 2 prosodies affect obstruents and approximants only. These are nas (N-prosody) and glot (ʔ-prosody). Prenasalised obstruents are found across practically all Chadic languages. ʔ-prosody, on the other hand, is comparatively rare; it is also not confined to languages of the Central branch. It explains the sporadic synchronic emergence of certain glottal(ised) segments that as such cannot be reconstructed for the common proto-language.

The first three prosodies, i.e. W-, Y- and N-prosodies, had already been discovered in the 1960s and early 1970s to play a role in descriptive work on synchronic phonological systems of Central Chadic-A languages (see Section 1.2 above).

The comparative relevance of both W- and Y-prosodies for pinning down underlying and diachronic vowel correspondences in eventually all Central Chadic languages was discovered soon after (Wolff et al. 1981, Wolff 1981, 1983). Prenasalisation as prosody had also been used for comparative purposes by Hoffmann (1987) in application to the languages of the then so-called Bura-Margi group. ʔ-prosody, however, had not been identified as existing and relevant for comparative Central Chadic until just recently (Wolff 2024).

That, and how W- and Y-prosodies are essential in order to establish sound correspondences between Central Chadic languages, and how they contributed to vocalogenesis in the history of Chadic languages, had been made the topic of a number of papers by the present author since the early 1980s (for a synopsis, see Wolff 2017).

Today and almost without exception, Chadicists acknowledge the usefulness of the prosodic approach – at least for some if not most of the languages belonging to the Central Chadic branch.^[13] Therefore, we consider Richard Gravina’s (2014) PhD thesis from Leiden University a pioneer study on historical Central Chadic phonology in terms of (a) its comprehensive approach that would cover the languages of all Central Chadic sub-branches, and (b) – in principle – taking into account prosodies as they were proposed in earlier literature on the subject. Of the three prosodies suggested in earlier literature on the subject, namely Y-, W-, and N-prosodies, Gravina’s selective focus is on the widely spread Y-prosody. He underestimates the equally important role of W-prosody, and he completely disregards N-prosody. He reconstructs a unique – and in universal typological terms rare – 3-vowel-system *a, *ɨ, *i, which contradicts all previously proposed assumptions about and tentative reconstructions of Proto-(Central-)Chadic vowels. He reconstructs Y-prosody as a separate phonological unit on a par with vowels and consonants and claims that “Proto-Central Chadic did not have … a labialization prosody” (Gravina 2014: 326), because he dismisses the widely occurring instances of W-prosody as secondary “reanalysis of the labialization component of a lost Proto-Central Chadic labialized velar consonant” (Gravina 2014: 326). He maintains that “[t]he labialization prosody developed quite recently” and only in a sub-set of Central Chadic languages (i.e. in his “Vowel Prosody languages”; Gravina [2014: 328]). This insinuates that we are dealing with an areal feature within Central Chadic with much lesser distribution and relevance than Y-prosody. Thereby he treats the two major Chadic prosodies, namely pal and lab, as phonologically and historically essentially different. In the view of the present author and based on more than 40 years of working on the historical linguistics of Central Chadic languages, both assumptions respectively reconstructions misrepresent the diachronic evolution of Central Chadic historical phonology.^[14]

With judgements based on the same set of data, Gravina’s approach and analyses are seriously questioned in Wolff (2022a, 2024, who provides an alternative theoretical and methodological approach as well as competing analyses of the same primary data. Basically, Gravina’s analysis of a pre-existing reconstructed Y-prosody that was different in nature from the source(s) of W-prosody is challenged by a theory that accounts for all prosodies in a unified and consistent manner (see Section 3.). As a matter of fact, this unified theory is quite in line with what Gravina himself proposes for W-prosody only, namely that a prosody has a segmental origin in the proto-language. Our own comparative work, however, suggests that both Y- and W-prosodies (plus N- and ʔ-prosodies) share parallel historical developments, because they all stem from reconstructed host segments in the common proto-language.

Experts believed for a long time that minimal vowel systems and the massive effect of prosodies on both vowels and consonants were exclusive typological features of languages belonging to the Central branch of Chadic. Therefore, it was revealing to learn from Schuh (1998, 2002) about the existence of an almost Central Chadic type minimal vowel system (/a/, /aa/, /ə/) in West Chadic Miya. Schuh (2002) further reported the scattered distribution of morphological Y-prosody, or traces thereof, in a small number of West Chadic-A (Bole) and West Chadic-B (Miya, Duwai, Bade, Ngizim) languages. Likewise revealing was Roberts’ (2007) report of minimal vowel systems and the existence of prosodies in one Masa group language (He’de) and in East Chadic-A languages (Somrai, Gabri, Kabalai), and of prosodies operating also in an East Chadic-B language (Mawa). Reference is to an unpublished paper, in which Roberts presented “evidence of palatalization and labialization operating to varying degrees.” He concluded by saying that

… the only way of explaining the existence of prosodic systems in scattered languages throughout the Eastern branch is to claim that prosodies were characteristic of the Proto-language, and that such systems have been lost to varying degrees in most of the individual languages. And since prosodies still play such a central role in languages of the Central Chadic branch, it might be supposed that prosodies were part of Proto-Chadic itself.

If Proto-Chadic did indeed use prosodies fully, then such a system has seriously eroded in the Eastern branch, despite its retention in some languages like Somrai. (Roberts 2007; quoted with kind permission).

This corresponds to the picture drawn by Schuh (2002) concerning the scattered evidence of Y-prosody across languages of the West Chadic branch. We, therefore, join Roberts in the assumption “that the prosodies may have been characteristic of Proto-Chadic from its early days” (Roberts 2009: 138).

Given the characteristic association of prosodies with historical underlying minimal vowel systems, this could mean that minimal vowel systems were also once characteristic for all of Chadic, i.e. to be reconstructed for Proto-Chadic. This, however, still deserves to be underpinned by more focused phonological analyses of languages of the other branches, besides West-Chadic Miya (Schuh 2002), Masa-group He’de, and East-Chadic Somrai (Roberts 2007, 2009).

It deserves to be pointed out that underlying minimal vowel systems and the salience of prosodies don’t reveal themselves to the superficial look at language data but need highly abstract and deep-level analysis to be detected and recognised as either diachronically pre-existing or synchronically still relevant in a given language.

Under the prosody approach, a widely shared and legitimate common assumption in modern Chadic linguistics is that pal and lab may coexist in the synchronic phonological system alongside vowels and consonants on which they have ‘colouring’ effects. Therefore, and for some languages, reference to these prosodies is legitimate in synchronic descriptions, depending on the language under research. The historical origin of these prosodies can be considered irrelevant for synchronic descriptions and, therefore, usually remains obscure in synchronic descriptions. This follows from a methodologically sound observation of the ‘Saussurean Firewall’ between synchronic and diachronic perspectives and analyses (see Section 1.3 above).

Soon after the notion of prosodies was successfully applied to the synchronic analysis and description of Central Chadic (sub-branch A) languages in the early 1970s, it was also found very useful in the early 1980s for the comparative and historical analysis of Central Chadic-A languages (see Barreteau 1987; Wolff 1981, 1983; Wolff et al. 1981). In particular, lexicalised effects of Y- and W-prosodies were suggested to be responsible for the lack of regular 1:1 sound correspondences among phonetic surface vowels between pairs of closely related languages, which – when the prosodies were not properly taken into account – seemed to completely escape the application of the comparative method as it had been well established by the Neogrammarian school since the mid-19^th century.

3.1 Y-prosody

Y-prosody stems from the only reconstructed palatal segment in PCC, namely the approximant */y/ (IPA j) as well as from allophonic *[y], which often relates to reconstructed */|_/ to /ɗ/. The segmental origin of this prosody can very frequently be identified with a petrified polyfunctional and usually suffixal marker *{-y(a)}, which has survived as synchronic morphological marker in a number of modern languages (see Wolff 2019). Y-prosody may spread in both directions, but mostly by leftward anticipation from suffixal position.

In example (1) from Podoko (Mandara), we reconstruct the original presence of a very common prefix *{ma-} to the simple PCC root *dz(a)kʷ(a)ɗa ‘fly (n)’. PCC */ɗ/ eventually changed into its frequent allophone *[y], which became the source of Y-prosody (pal), while root-internal */kʷ/ weakened to *[w] in a parallel fashion (yet without creating W-prosody [lab] in this particular example). Y-prosody then affected both the root-initial consonant */dz/ → [ʤ] and the lexical final vowel */a/ → [e]. The pre-prosodification and pre-sound change Proto-Podoko form can be symbolised as apocopated *ma-dzkʷaɗ, which underwent a fairly common deletion of prefixal */a/ and sound changes to intermediate *m-dzway; the post-prosodification synchronic phonetic shape is transcribed as ⁿ ʤəwe in the database. Note that the self-evident word margins in the examples are not explicitly marked.^[17]

(1)

Example (1) directs our attention to a permanent and crucial issue in descriptive linguistics, namely that of distinguishing monophonemic prenasalised obstruents from biphonemic nasal+obstruent clusters. In synchronic perspective and based on the transcription in the database ( ⁿ ʤəwe ), we could be led to maintain the trivial generalisation that the maximal leftward anticipation of Y-prosody was limited by the left-margin word boundary by allowing the synchronic word-initial consonant (presumably /ⁿdz/) to be affected by palatalisation ^y/ⁿdz/ → [ⁿʤ]. However, in diachronic perspective and based on our reconstruction involving a nasal-initial prefix, we would be faced with a nasal+obstruent cluster straddling a morpheme juncture *n-dz, in which not the nasal but only the obstruent was affected by the prosody: *n-dz^y → *[n-ʤ]. This means that in diachronic perspective, the left margin relevant for the anticipation of pal would have been the morpheme boundary rather than the word boundary.^[18]

3.2 W-prosody

W-prosody stems from the reconstructed approximant */w/ as well as potentially from any labialised consonant */Cʷ/ from a reconstructed series of PCC labialised velars. One of the most common sources is a reconstructed petrified suffix *{-kʷ(a)} of still opaque semantics. W-prosody may spread in both directions, i.e. mostly by leftward anticipation from the suffixal position, but also by rightward spreading.

In examples (2a, b) from one Bata and one Daba group language, we see the operation of W-prosody by rounding the root-initial consonant in Sharwa ɓʷah (*ɓaxʷ → ʷ/ɓax/ → ɓʷax; 2a) and affecting the root-internal vowel in Daba ɓoh (*ɓaxʷ → ʷ/ɓax/ → ɓaʷx; 2b), both effects stemming from the disassociation of the feature lab of the original C₂ */ɣʷ/ of the common PCC simple root *ɓ(a)ɣʷa ‘to hide’. Sharwa and Daba differ in the choice of the target segment for W-prosody to re-associate with.

(2a)

(2b)

3.3 N-prosody

N-prosody apparently always stems from a petrified prefix with a nasal initial consonant, of which two have been reconstructed for Proto-Central Chadic, namely *{ma-} and *{na-}, of which *{ma-} has survived as productive synchronic morphological marker in a number of modern languages. Given its prefixal origin, N-prosody will as a rule observe rightward spreading.

In example (3) from Vame (Hurza), we reconstruct again the original presence of the common prefix *{ma-} plus the likewise common suffix *{-kʷa} that both combine with the simple PCC root *ɣʷ(a)pa ‘flour’. Both PCC */ɣʷ/ and suffixal /kʷ/ underwent delabialisation; the root-initial velar became voiceless → [x], while the suffixal velar became voiced → [g]. The prefix *{ma-} desegmentalised completely and prosodised to survive only as N-prosody, which affected C₂ of the root changing historical underlying */p/ → [^mb]. The pre-prosodification Proto-Vame form can be symbolised by *ma-ɣʷp-kʷa; the post-prosodification synchronic phonetic form is transcribed as həᵐbəga . Note how the nasal feature after disassociation from prefixal */m/ jumps the root-initial consonant to re-associate as N-prosody with the original C₂ of the simple root; this rules out an origin of [^mb] from an intermediate nasal+obstruent cluster.

(3)

3.4 ʔ-Prosody

ʔ-prosody stems from either the reconstructed PCC glottal consonants */ɓ, ɗ, ʔy/ or allophonic *[ʔ], of which *[ʔ] may historically relate to almost any reconstructed obstruent. ʔ-prosody, as limited as its occurrence is, apparently may only spread by leftward anticipation and appears to preferably target the root-initial segment. ʔ-prosody is found remarkably often, but not exclusively, in languages of the Kotoko groups.

In example (4) from Mser (Kotoko-Central), we see the operation of ʔ-prosody stemming from the common suffixal root augment *{-kʷa} at the right word margin, which underwent delabialisation and was reduced to its allomorphic shape *{-ʔa}. This allomorph underwent desegmentalisation and prosodification of its initial consonant to give rise to the floating feature *[^ʔ], which re-associated with the root-initial */s/ at the left word margin. This re-association creates the phonetic surface output of a glottal [s^ʔ], which as such is not reconstructed for PCC, but must be considered a Kotoko group innovation. Since the emergence of glottal [s^ʔ] is historically motivated and is not governed by any synchronic phonological rule of allophony, /s^ʔ/ must now be considered a synchronic phoneme in the languages where it occurs. Note that under the regime of the historical underlying minimal vowel system, the approximant */y/ appeared in syllable-nucleus positions (such as CVV.) as high front vowel [i] and is reflected in modern Mser as phonologised /i/. The PCC simple root is PCC *ɬa ‘to cut’, which in pre-prosodification Proto-Mser *ɬ-y-kʷa carries the reflexes of a reconstructed suffix chain *{-y-kʷa}.

(4)

Note that, given the rather frequent occurrence of consonantal metathesis in Central Chadic, examples like the above would allow alternative analyses by assuming consonant metathesis to have operated: *ɬ-y-kʷa > *s-y-ʔa > *sʔya >  sʔia . Under such alternative analysis, we would explain the database transcription s’ia as reflecting an obstruent+glottal stop cluster /sʔ/.

6.1 Emergence of new vowels

Starting off from the minimal vertical vowel system of PCC with only two vowels (*/a/, *ə) and the presence of the two approximants (*/y/, */w/), ‘vocalogenesis’ (cf. Wolff 2017) affects

1. the approximants */y/, */w/ in terms of their perfectly conditioned allophones [i], [u], which are – over time – eventually phonologised to yield synchronic /i/ and /u/ in modern Central Chadic languages;

2. the only phonemic vowel */a/, which allows for several conditioned allophones, namely

   1. [ə] (possibly conditioned in pre-juncture positions),

   2. [ɛ∼e] as prosodically ‘coloured’ allophones under the effect of pal,

   3. [ɔ∼o] as prosodically ‘coloured’ allophones under the effect of lab,

   4. which may eventually and – over time – become phonologised to synchronic /ə/ and/or synchronic /e/ and /o/ (if not occasionally /ɛ/, /e/, /ɔ/ and /o/) in modern Central Chadic languages;

3. the non-phonemic vowel *ə, which may eventually and – over time – become phonologised to /ə/, like its prosodically ‘coloured’ conditioned variants [ɪ∼ɨ∼i] under the effect of pal and [ʊ∼ʉ∼u] under the effect of lab, which are diachronically phonologised to merge synchronically with innovative /i/ and /u/ (stemming from the phonologisation of *y and *w) in modern Central Chadic languages.

Note that by umlaut stemming from (formerly or still present) adjacent syllables containing /a/, surface phonetic [i] and [u] may be realised as lowered to [e] and [o], which gives rise to more instances of overlapping of allophones.

Across Central Chadic, languages may end up synchronically with vowel systems between only one (always /a/) and eight or ten (/i, e, ɛ, a, ɔ, o, u, ə/, occasionally plus /œ, y∼ø/) short vowel phonemes. (Some Chadic languages will also develop distinctive vowel length and thereby double the inventory of distinctive short vowels. Vowel length, however, is not reconstructed for PCC; see Wolff 2022a, 2024.).

6.2 Emergence of new consonants

Starting off from the reconstructed PCC consonant system with 26 consonants (including a series of labialised velar obstruents, but not counting the approximants *y, *w), the effect of prosodies is likely to enlarge the synchronic inventories of distinctive consonants in modern Central Chadic languages.

1. pal may create single or whole series of palatalised consonants;

2. lab may create single or whole new series of labialised consonants (such as bilabials with simultaneous lip rounding) in addition to the reconstructed series of labialised velars;

3. nas may create single or whole series of prenasalised obstruents;

4. glot may create additional single glottal(ised) consonants.

Further, the frequent allophone *[ʔ] may become phonologised to /ʔ/ in synchronic phonological systems of modern Central Chadic languages. Segmental fusions such as among abutting nasals and plain or labialised velars create synchronic velar nasals /ŋ/, /ŋʷ/. All of these are not reconstructed for Proto-Central Chadic (see Wolff 2022b).