NEGATION PATTERNS IN THE KWA LANGUAGE GROUP by LAUREN ELIZABETH SCHNEIDER A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS In THE FACULTY OF GRADUATE STUDIES Master of Linguistics We accept this thesis as conforming to the required standard Dr. Sean Allison, Ph.D.; Thesis Supervisor Dr. Steve Nicolle, D.Phil.; Second Reader Dr. Rod Casali, Ph.D.; Third Reader TRINITY WESTERN UNIVERSITY June 2017 © Lauren Schneider Lauren Schneider MA Thesis Acknowledgments I would like to thank everyone who contributed to this project. A special thanks to my thesis advisor Sean Allison for his guidance along the way. I would like to thank my committee members Steve Nicolle and Roderic Casali whose expertise led me to this paper topic. Additionally, I would like to thank Michael Obiri-Yeboah from the University of California San Diego for providing native speaker data for Gua (South Guang, Kwa). I would also like to thank everyone who gave me feedback at the 48th Annual Conference on African Linguistics. A special thanks to Samson Lotven and Levina Abunya for volunteering to share their language data with me on Gengbe and Krache, respectively. Lastly, I want to thank the Canada Institute of Linguistics for giving me the opportunity to study with them and complete my M.A. in Linguistics. It has been a huge blessing to be a part of this program during the last two years. I am especially grateful for the generosity in financial aid and in covering the costs of attending ACAL 48. i Lauren Schneider MA Thesis Table of Contents Acknowledgments ………………………………………………............................. i Table of contents ………………………………………………............................. ii 1. Introduction ………………………………………………............................. 1 2. Background ………………………………………………............................. 2 2.1 Previous work on negation …………………..................................................... 2 2.2 Kwa language background …………………..................................................... 12 3. Presentation of data and analysis …………………………………………...... 15 3.1 Akan ………………………………………………....................... 15 3.2 Ewe ………………………………………………....................... 19 3.3 Guang languages ………………………………………………....................... 24 3.4 Ghana-Togo Mountain languages 4. Conclusions ………………............................................... 48 ………………….……………….......................................... 64 4.1 Summary …………………..................................................... 64 4.2 Jespersen’s cycle in Kwa …………………..................................................... 68 4.3 Concluding remarks …………………..................................................... 70 Abbreviations …………...………………………........................................ 71 References ……………………………………....................................... 72 Appendix ……………………………………....................................... 75 ii Lauren Schneider MA Thesis Abstract There currently exists extensive literature written on the topic of negation but it has been only recently that studies of negation have begun to expand outside of the limited scope of Indo-European languages. Linguists are finding that certain patterns thought to be cross-linguistic occur mainly in this most heavily studied language family. The intent of this thesis is to survey the negation strategies in a collection of Kwa (Niger-Congo) languages in order to contribute to the literature on negation and tip the scales ever so slightly away from Indo-European. Commonly cited patterns such as Jespersen’s cycle (Jespersen 1917) are almost entirely unattested in this language group. There is a consistent pattern of marking negation in Akan, Ewe, and the North Guang languages involving the use of a preverbal nasal morpheme. Interestingly three South Guang languages utilize instead a verbal prefix bÉ -. The Ga-Dangme languages stand out from other Kwa languages in their use of verbal suffixes rather than prefixes. The Ghana-Togo Mountain branch of the Kwa language group also does not rely on preverbal nasal negation marking. 1. Introduction There is extensive literature written on the topic of negation but it has been only recently that studies of negation have begun to expand outside of the limited scope of Indo-European languages. Linguists are finding that certain patterns thought to be cross-linguistic occur mainly in this most heavily studied language family. The intent of this thesis is to survey the negation strategies in a collection of Kwa languages in order to contribute to the literature on negation and tip the scales ever so slightly away from Indo-European. Section 2 of the paper contains background information, first on previous work on negation in general and second on Kwa languages. Section 3 presents the data that has been collected on negation in Kwa languages. These languages include Akan, Ewe and a few other Gbe languages, ten Guang languages, seven 1 Lauren Schneider MA Thesis Ghana-Togo Mountain languages, and the two Ga-Dangme languages. Lastly, Section 4 contains a summary of the patterns and generalizations about negation in the Kwa language group. 2. Background 2.1 Previous work on negation As was mentioned above, the vast majority of the research on negation has focused on IndoEuropean languages. A few notable exceptions are discussed in Kahrels & van den Berg (1994) and Miestamo (2007). For example, two of the most talked about negation processes are negative concord (discussed in Section 2.1.3) and Jespersen’s cycle (discussed in Section 2.1.4). Both processes are very common in the languages of Indo-European decent or languages that have had extensive contact with this language family. This section will discuss these phenomena as well as general distinctions made in terms of the scope and symmetry of negation. 2.1.1 Negation scope When talking about the scope of negation, there are a number of different ways to classify negation. One distinction that should be acknowledged is the long philosophical tradition that holds that negation is ambiguous between internal and external readings (Miestamo 2009: 208). Below is the well-known “King of France” example: (1) The King of France is bald The King of France is not bald With internal negation (2a), the subject (the King of France) is interpreted as being outside the scope of negation while external negation (2b) would interpret the subject as included in the scope: 2 Lauren Schneider MA Thesis (2) a. b. Internal: The King of France is NOT[bald] External: NOT[The King of France is bald] Internal negation preserves the presuppositions of the corresponding affirmative while external negation denies the presuppositions (Miestamo 2009: 209). In natural language, negation typically does not affect the presuppositions of a sentence except in certain cases of metalinguistic negation: (3) The King of France is bald The King of France is not bald because there is no King of France (Miestamo 2009: 222) In this example, the fact that there is no King of France cancels the presupposition of the corresponding affirmative. The philosophical tradition that assumes negation is ambiguous between an internal and an external interpretation does not describe actual human language. There is no ambiguity in The King of France is not bald as to what is being negated in most contexts. Because this is an artificial distinction it will not be used to describe the negation patterns in the Kwa languages described in this paper. Other distinctions that have been made when talking about the scope of negation are sentential versus constituent negation and nexal versus special negation. Sentential negation takes the whole sentence in its scope whereas constituent negation only applies to a particular constituent (Miestamo 2009: 209). The example below demonstrates the basic distinction: (4) a. b. Constituent negation He is unhappy Sentential negation He is not happy There is a noticeable semantic difference between examples (a) and (b) above. Example (4a) involves the constituent negation of an adjective while (4b) involves negation of the sentence He is happy. While also dependent on context, these two clauses do not encode identical information and can be used to communicate different implications. For example, He is unhappy can be used 3 Lauren Schneider MA Thesis to entail a state of being sad or discontent while He is not happy is typically used to entail anger by emphasizing that the subject is the opposite of happy. Jespersen (1917) groups together sentential and constituent negation, instead distinguishing between nexal and special negation. Miestamo defines nexal negation as negation where the negative operator negates two ideas while special negation is where the negative operator negates one idea (2009: 209). Special negation may be expressed by some modification of the word (5a) or by the addition of not (5b). (5) Special negation Nexal negation a. b. unhappy not happy c. d. he doesn’t come he doesn’t come today (Jespersen 1917: 42-3) According to Jespersen, in (5c) the two positive ideas he and coming are negated. In (5d) the combination of the positive ideas he and coming today are negated. This paper will be mostly concerned with what is called standard negation. Standard negation is understood as the basic strategies languages use for negating declarative main clauses. In general, standard negation can typically be expressed by negative particles, negative verbs, or in the morphology of the verb (Payne 1985 cited by Miestamo 2009: 214). Negation encoded by negative particles and auxiliary verbs is referred to as syntactic negation while negation encoded through affixes, prosody, and reduplication is referred to as morphological negation. We will find examples of both morphological and syntactic negation in Kwa. 2.1.2 Symmetric vs. asymmetric negation Miestamo distinguishes two basic types of negative structures: symmetric and asymmetric negation (2009: 215). Symmetric negative constructions only differ from the corresponding 4 Lauren Schneider MA Thesis affirmative in that there are one or more negative markers. This type of system is demonstrated in the Spanish set below: Spanish: cantar ‘to sing’1 (6) AFFIRMATIVE 1SG 2SG 3SG 1PL 2PL 3PL canto cantas canta cantamos cantáis cantan NEGATIVE ‘I sing’ ‘you sing’ ‘s/he sings’ ‘we sing’ ‘you(pl) sing’ ‘they sing’ no canto no cantas no canta no cantamos no cantáis no cantan ‘I don’t sing’ ‘you don’t sing’ ‘s/he doesn’t sing’ ‘we don’t sing’ ‘you(pl) don’t sing’ ‘they don’t sing’ In asymmetric negation, the structure of the negative construction differs from that of the corresponding positive construction in other ways in addition to the presence of the negation marker. Asymmetric negation can be divided into three subtypes: A/Fin, A/NonReal, and A/Cat. In asymmetric subtype A/Fin, the finiteness of the lexical verb is reduced or lost in the negative: (7) Finnish (Miestamo 2009: 216) a. laula-n sing-1SG2 ‘I sing.’ b. e-n laula NEG-1SG sing.CNG ‘I do not sing.’ In subtype A/NonReal negatives, negatives are obligatorily marked for a category indicating a non-real state of affairs. In the example below, the hypothetical and negative constructions are marked with irrealis: 1 In this example, accent marks correspond to the stress marks used in Spanish orthography. In the rest of the paper, accent marks correspond to tone. 2 See page 71 for abbreviations. 5 Lauren Schneider MA Thesis (8) Maung (Capell & Hinch 1970: 67 cited by Miestamo 2009: 216) a. 3 ŋi-udba 1SG.34-put ‘I put’ b. ni-udba-ji 1SG.3-put-IRR.NPST ‘I can put’ c. marig ni-udba-ji 1SG.3-put-IRR.NPST NEG ‘I do/shall not put’ In the last subtype, A/Cat negatives, the grammatical category is marked differently in negative constructions than in affirmative ones: (9) Burmese (Cornyn 1944: 12-13 cited by Miestamo 2009: 216) a. θwâ-dé b. go-ACT ‘goes, went’ θwâ-mé go-POT ‘will go’ c. θwâ-bı́ go-PRF ‘has gone’ d. ma-θwâ-bû NEG-go-NEG ‘does/did/will not go, has not gone’ Examples (9a)-(9c) demonstrate a few of the grammatical categories available to the Burmese verb phrase while example (9d) illustrates how the negative clause has a distinct category marking. The categories most often affected by A/Cat negatives are TAM and person-numbergender (Miestamo 2009: 217). Symmetric vs. asymmetric negation is a useful distinction for classifying negation strategies but it is not a perfect dichotomy in actual language. Kwa languages exhibit both patterns and some Kwa languages may provide reasons to expand the criteria. 2.1.3 Negative concord Negative concord is defined by van der Auwera & Van Alsenoy as “a semantically single negation expressed both by a clause level negator and by a negative adverb, pronoun or determiner” (2016: 2). They provide the following non-standard English examples: 3 In the data provided, the morpheme ‘1SG.3’ is realized [ŋi] in (9a) and [ni] in (9b) and (9c) but no explanation is provided for this difference. 4 This is mostly like a reference to the object of ‘put’, but there is no reference to this object in the free translation provided. 6 Lauren Schneider MA Thesis (10) a. b. c. I ain’t never been to jail. You ain’t seen nothing yet. I can’t get no satisfaction. The clause level negator is n’t and is attached to the verb. Example (10a) contains a negative adverb never, (10b) a negative pronoun nothing, and (10c) a negative determiner no. Both of the negative words in each sentence take part in a single negation. Negative concord is well attested in a large variety of languages. In the Indo-European language family, a majority of the languages exhibit negative concord, with Dutch, German, Swedish, and Norwegian, being a few notable exceptions (Zeijlstra 2007: 504). There are different types of negative concord. Negative concord is often classified as either strict or non-strict. Example (11) below demonstrates a case of strict negative concord while example (12) demonstrates non-strict negative concord. In the Polish example below, in both (a) and (b), there is both a negative marker and a negative pronoun: (11) Strict Negative Concord Polish (Haspelmath 1997: 201 cited by van der Auwera & Van Alsenoy 2016: 16) a. Nikt nie przyszedł nobody NEG came ‘Nobody came.’ b. Nie widziałam saw NEG ‘I saw nobody.’ nikogo nobody In Polish, negative concord is obligatory in that negative clauses must have both a clausal negative marker and the negative indefinite. In Spanish, the presence of negative concord depends on the word order: 7 Lauren Schneider MA Thesis (12) Non-strict Negative Concord Spanish (Haspelmath 1997: 201 cited by van der Auwera & Van Alsenoy 2016: 16) a. Nadie vino nobody came ‘Nobody came’ b. No vi a nadie NEG saw to nobody ‘I didn’t see anybody’ If the negative indefinite, in this case nadie ‘nobody’, is preverbal, no negative marker is necessary. If the negative indefinite occurs after the verb, a negative marker is required preverbally. In languages that exhibit negative concord, strict negative concord is generally more common that non-strict negative concord. In their survey of 179 languages, van der Auwera & Van Alsenoy (2013) found that negative concord is most frequent in Eurasian languages, occurring in just over half of the sample languages. Negative concord occurred in only a fifth to a quarter of the South-East Asian & Oceanic, African, and South American languages surveyed. Only about a tenth of North American languages exhibited negative concord and only about 5% of the Australian and Papua New Guinea languages had this feature. Additionally, of the 26 non-Eurasian languages that exhibited negative concord, they found that six were directly influenced by the negative concord systems of a European language. Outside the reach of Indo-European languages, negative concord is much less common. As we would expect based on this fact, we will find that negative concord is uncommon in Kwa languages. 2.1.4 Jespersen’s cycle You cannot have a discussion about negation without talking about Jespersen’s cycle. This process, first described by Jespersen (1917), involves elements which were introduced into 8 Lauren Schneider MA Thesis negative clauses in order to strengthen the negation being reanalyzed as the primary markers of negation. According to Jespersen, there is a tendency to place the negative first or as soon as possible, often immediately before the word being negated (1917: 5). He goes on to claim that its position at the beginning of the sentence makes it more likely than elsewhere to weaken phonologically (Jespersen 1917: 6). The process is most commonly cited using the development of negative markers in French. Miestamo (2007: 566) notes that negators are often ancient elements whose non-negative origin cannot be traced by historical and comparative means. Jespersen’s cycle is the best-known process of negation marker development. The history of the negator *ne can be traced as far back as Indo-European. Jespersen’s cycle traces the development of the negative marker from Latin to modern French. This process is outlined in examples (17) - (21) below: (17) Latin (Jespersen 1917:7) ne dic-o say-1SG NEG ‘I do not say.’ The ne form was felt to be too weak and was strengthened by the addition of oenum ‘one thing’, resulting in ne oenum ‘not one thing’. These forms merged to form non: (18) Latin (Jespersen 1917:7) non dic-o say-1SG NEG ‘I do not say.’ In Old French, non becomes nen and then further weakens to ne: (19) Old French (Jespersen 1917:7) jeo ne di 1SG NEG say.1SG ‘I do not say.’ 9 Lauren Schneider MA Thesis This version of ne was also analyzed as too weak and it is strengthened by the addition of words such as pas ‘a step’ or mie ‘a crumb’. After this, pas began to grammaticalize and lose the semantic content that restricted it to clauses having to do with walking. It has become an obligatory marker of negation in French: (20) Written French (Jespersen 1917:7) je ne dis pas 1SG NEG say.1SG NEG ‘I do not say.’ The cycle has continued in modern spoken French; the word ne is too weak and pas has been reanalyzed as the obligatory negative component: (21) Spoken French (Jespersen 1917:7) je dis pas 1SG say.1SG NEG ‘I don’t say’ There is disagreement as to how to best break Jespersen’s cycle into stages. Some proposals involve three stages: (22) stage I stage II stage III NEG VERB NEG VERB NEG VERB NEG (Willis et. al. 2013: 16) This is the simplest and likely most idealized representation. The stages refer to particular construction types but this version doesn’t represent the instability of the stages nor what happens between them as they transition from one to the next. The following breakdown better describes the intermediate stages: 10 Lauren Schneider MA Thesis (23) stage 1 stage 2 stage 3 stage 4 stage 5 NEG VERB NEG VERB (NEG.EMPH) NEG VERB NEG (NEG) VERB NEG (Willis et. al. 2013: 18) VERB NEG These representations are to an extent a matter of preference and do not reflect different conceptualizations of the processes that are taking place. Languages in which the cycle occurs vary significantly in the rate at which they progress through it. Dixon (2012) suggests that there is potential for yet another stage in the cycle that returns negative marking to its original position before the verb. In a number of French-based creoles, the negative marker pa (from pas) has moved to the position before the verb: (24) Mauritius Creole (Bernini & Ramat 1996: 35 cited by Dixon 2012: 104) me mōte pa pe my watch NEG PROG ‘My watch doesn’t work.’ travaj work The potential developments in Jespersen’s cycle could be represented as follows: (25) stage i stage ii stage iii stage iv (i) NEG VERB NEG VERB NEG VERB NEG NEG VERB Unlike the other stages discussed above, there is not at this time data to support an intervening stage NEG VERB (NEG), as the negator shifts back to the preverbal position. This is a logical next stage based on Jespersen’s observation, mentioned earlier, of the tendency to place the negative marker first or as soon as possible, often immediately before the word being negated. It is important to consider what motivation a language would have for undergoing a cycle such as this. Why does a marker of emphatic negation come to replace the original marker? Jespersen’s presentation of the cycle as described above implies a pull-chain scenario in which the change is motivated by the phonological weakness of the preverbal negation (Willis et. al. 11 Lauren Schneider MA Thesis 2013: 15). More recently the focus has been on the overuse of the new negator over time reducing its emphatic nature. In this push-chain scenario, the preverbal marker is lost when the new marker takes over the function of expressing negation (Willis et. al. 2013: 16). More mixed approaches argue that a morphological weakening of the preverbal element, the establishment of a new negator, and the eventual elimination of the preverbal element, work together to motivate the process. According to Willis et. al, Jespersen’s cycle has occurred throughout the languages of western Europe as well as in a number of Afro-Asiatic and north African languages, such as Berber and Coptic (2013: 8). While traces of it have been found sporadically in Niger-Congo languages in central and west Africa, outside of Europe, Jespersen’s cycle has been identified less frequently and may be much less common (Willis et. al. 2013: 9). For this paper, the Kwa languages under study will be surveyed to determine whether or not Jespersen’s cycle is a feature that they have in common. 2.2 Kwa language background 2.2.1 Classification and location The Kwa language group (kwav1236)5 is a member of the Niger-Congo language family. The following chart is an abridged representation of the Niger-Congo language family: (26) Abbreviated Niger-Congo family tree: Kwa Mande Niger-Congo Atlantic-Congo Kordofanian 5 Volta-Congo Kru Volta-Niger Kwa Benue-Congo Gur Kwa Glottocode (Hammarström et. al 2016) 12 Lauren Schneider MA Thesis The term ‘Kwa’ was adopted by linguists near the end of the 19th century in order to group Akan, Ga, and Ewe together (Dakubu 2009: 630). There is some debate as to how to best classify these languages and there is even doubt as to whether these languages can actually be considered a genetic group.6 The intent of the diagram above (and other diagrams to follow) is not to propose a particular classification but simply to provide the reader with a basic idea of the group of languages under study. The data collected for this study, and the groundwork that it begins to lay, may lead to more clarity on the topic of classification in the future. The Kwa languages can be found in the Ivory Coast, Ghana, Togo, Benin and on the western border of Nigeria. The group is bordered to the west by the Kru language group, to the North by the Gur language group, and to the east by the Benue-Congo language group. More specific background on each language and language subgroup being surveyed will be provided in the section devoted to the language or language subgroup. 2.2.2 Phonology This section provides a brief summary of some of the phonological patterns typically found in Kwa languages. The data gathered for this paper is from published sources and is presented phonetically by some authors and orthographically by others. Due to the nature of the data collected the treatment of phonology in this paper is necessarily limited.7 Kwa languages tend to include the plosives: /p/, /b/, /t/, /d/, /k/, and /g/; the nasals: /m/, /n/, /ɲ/, /ŋ/; the fricatives: /f/, /s/; approximants: /j/, /l/ and/or /r/; and semivowels: /y/ and /w/. Some of languages contain additional fricatives such as /ф/, /β/, /v/, /z/, /ʃ/, /x/, and /ɣ/. It can be noted here that voiceless fricatives are less common that voiced ones. Other common phonemes include labiovelar sounds such as /kp/, /gb/, and /ŋm/ as well as the voiceless affricate /tʃ/. As 6 7 See Stewart (1989) for more on this. For a thorough treatment of phonology in the Kwa subgroup, Guang, see Snider (1990). 13 Lauren Schneider MA Thesis with the fricatives, the voiced affricate /dʒ/ is less common. Ewe contains a retroflex /ɖ/ but retroflex consonants are not common in this language group. A common feature in the Kwa consonant systems is labialization. The consonants /pʷ/, /bʷ/, /fʷ/, /sʷ/ and /kʷ/ are common. Vowel systems in Kwa typically consist of either a nine or seven vowel system. The ninevowel systems consist of the [+ATR] vowels /i/, /e/, /u/, and /o/ and the [-ATR] vowels /ɪ/, /ɛ/, /a/, /ʊ/, and /ɔ/. The seven vowel systems consist of the same [+ATR] vowels /i/, /e/, /u/, and /o/ but only the [-ATR] vowels /ɛ/, /a/, and /ɔ/. As can be gathered from the description of the vowel system, [ATR] is a relevant feature. [ATR] harmony is very common throughout the Kwa languages. While a slight over-generalization, in can be stated that affixes typically harmonize with stems. Some of these languages have phonemic nasal vowels. Additionally, it is important to note that in some of the Kwa languages vowel length is contrastive. Long vowels do not typically occur in closed syllables. The predominant syllable structure in Kwa is CV(C). Coda consonants are often limited to nasals and word-final glottal stops. Onsets with CC consonant clusters occurs in some languages but this is usually highly limited if it is allowed at all. Another feature typical of most Kwa languages is word-initial syllabic nasals. This feature will occur frequently in the data presented in the next section as it is common in negation marking. The last important feature of Kwa phonology that needs to be mentioned is tone. All of the Kwa languages are tonal. That being said, tone is often inconsistently marked in the grammatical descriptions utilized for this thesis. Tone will be addressed as it is relevant to the negative markers as much as possible. 14 Lauren Schneider MA Thesis 3. Presentation of data and analysis The negative morphemes presented here have largely been collected from the grammatical descriptions written by other linguists. This is important to note because of the way the negative morphemes are represented as either words or affixes. The treatment here relies somewhat on the conventions used by the previous presentations of the data. That being said, some cases are clearer than others and any disagreements with the original analysis will be noted. 3.1 Akan Akan [aka]8 is a Central Tano language spoken in Ghana. Akan can be placed within the Kwa language group as follows: (27) Abbreviated Kwa family tree: Akan9 Ga-Dangme Gbe Kwa Ka-Togo Na-Togo Nyo Agneby Attié Avikam-Alladian Potou-Tano Central Tano Potou Guang Tano Krobu Akan Western Tano This Kwa language has a number of well-known varieties. The first variety under study in this section will be Kwawu Akan. Like most other Kwa languages, Akan has a basic SVO constituent order (illustrated in (28) below). 8 When a language is introduced, the ISO 639-3 code for the language is provided in square brackets (e.g. Akan [aka], English [eng]). 9 Based on Hammarström et. al (2016) 15 Lauren Schneider MA Thesis (28) Akan (Campbell 1988: 214) yaw hu-u Yaw see-PAST 'Yaw saw him' no him All verbs inflect for negation in the same way, using a low tone nasal prefix, Ǹ-: (29) Akan (Campbell 1988: 214) yaw n-hú-ù no Yaw NEG-see-PAST 'Yaw hasn't seen him' him 10 The negative prefix is a homorganic nasal that assimilates to the place of articulation of the following consonant. This appears to be a clear case of morphological negation with the preverbal negative marker acting as an affix rather than a separate word. Like Kwawu Akan, other varieties of Akan, including Fante, Akuapem, and Asante, also employ a homorganic verbal prefix. Abakah (2005) describes the negative morpheme in Akan as a toneless11 syllabic nasal (N). In the example below the prefix is realized as [ǹ-] when used with the verb ‘sell’: (30) Negation in other varieties of Akan (Abakah 2005: 124) FANTE AKUAPEM ASANTE Gloss a. mʊ̀-ǹ-tɔ́ń mɪ-̀ ǹ-tɔ́ŋ́ mɪ-̀ ǹ-tɔ́ ʊ ̃ ́̃ I do not sell b. ɪ̀-ǹ-tɔ́ń wʊ́-ǹ-tɔ́ŋ́ wʊ́-ǹ-tɔ́ ʊ ̃ ́̃ you do not sell c. ɔ̀-ǹ-tɔ́ń ɔ̀-ǹ-tɔ́ŋ́ ɔ̀-ǹ-tɔ́ ʊ ̃ ́̃ s/he does not sell d. jɛ̀-ǹ-tɔ́ń jɛ̀-ǹ-tɔ́ŋ́ jɛ̀-ǹ-tɔ́ ʊ ̃ ́̃ we do not sell 10 Tone appears to be inconsistently marked on the examples in Campbell (1988). In (28), no tone has been marked and in (29) tone is marked on the verb but not elsewhere. 11 Abakah proposes that the N- prefix may have an underlying high tone in all dialects, see Abakah (2005) for details. 16 Lauren Schneider MA Thesis All varieties of Akan employ morphological negation through the use of verbal prefixes. This morpheme is clearly a prefix, rather than a particle, because it occurs prefixed immediately to the stem while the tense/aspect marking is prefixed to the negative marker, further from the stem. Negation in simple past tense clauses is symmetric in that the only thing that changes is the addition of the negative prefix to the verb. That being said, due to its low tone, the nasal prefix does affect the tone pattern of the entire phrase.12 Tone is not a criterion Miestamo (2009) utilizes to distinguish types of asymmetry and in these examples the tone shift appears to be largely a phonetic rather than grammatical effect. While negative asymmetry doesn’t occur in the negation of simple past tense clauses, it does occur in other tenses and aspects in Akan. In negative clauses, the distinction between the future tense and progressive aspect is neutralized. In non-negative sentences, future tense is marked on the verb by a prefix bɛ́-, and progressive aspect by the verbal prefix re-. In negative contexts, only re- is used: (31) Akan (Campbell 1988: 210) kofı́ re-ǹ-kɔ́ kùmáse Kofi FUT/PRES-NEG-go Kumase 'Kofi will not go to Kumase.' or 'Kofi is not going to Kumase.' While this is clearly a case of negative asymmetry, it does not fit nicely into one of Miestamo’s categories. It most closely resembles the subtype A/Cat, where marking of grammatical categories differs from their marking in affirmatives, usually affecting TAM and person-numbergender distinctions. Perhaps a more specific category is warranted, such as A/Neut, which would indicate that certain verbal tense/aspect contrasts are neutralized in negative contexts. This is supported by Dixon’s generalization that there are often fewer tense/aspect choices in a negative 12 For a full treatment of the tone rules in Akan, see Campbell (1988). 17 Lauren Schneider MA Thesis clause than in a positive one (2012: 129). For now, this will be treated as a case of A/Cat negation. Another feature that Akan has in common with other Kwa languages is its use of serial verb constructions (SVCs). A serial verb construction is a series of verbs acting together as a single predicate and which is conceptualized as a single event (Aikhenvald 2006). In Akan, the first verb in a non-negative SVC is marked for tense or aspect and the second verb in the SVC is marked with a consecutive prefix: (32) Akan (Campbell 1988: 211) kofı́ bɛ-yɛ àdwúma a-má yàw NAME FUT-do CONS-give Yaw work 'Kofi will work for Yaw.' Like in basic negative clauses, when a SVC with future tense or progressive aspect is negated, the bɛ́-/re- distinction is neutralized. Each verb in the SVC is marked for negation. The consecutive prefix is replaced by the negative prefix as these prefixes do not co-occur. (33) kofı́ re-ǹ-yɛ́ àdwúma work NAME FUT/PRES-NEG-do 'Kofi will not work for Yaw.’ or ‘Kofi is not working for Yaw.’ m̀-má NEG-give yàw Yaw Based on these examples, there is negative asymmetry occurring in future and progressive clauses. The closest asymmetry category would be A/Cat in that the grammatical category of the verbs is changing. In single verb clauses, the distinction between future tense and progressive aspect is neutralized and in serial verb constructions the same occurs with the additional change of the consecutive prefix being replaced by the negative prefix. 18 Lauren Schneider MA Thesis 3.2 Ewe Ewe [ewe] is a member of the Gbe subgroup and is spoken in south-eastern Ghana and southern Togo. The chart below illustrates Ewe’s genetic relationship to the rest of Kwa but is not meant to be an exhaustive representation of the Gbe language subgroup: (34) Abbreviated Kwa family tree: Ewe13 Aja (Benin) Ga-Dangme Fon Eastern Gbe Gbe Gun Ka-Togo Kwa Na-Togo Western Gbe Ewe Gen Nyo In Ewe, the basic word order is SVO which is demonstrated by the examples below. (35) Ewe (Nurse n.d.:1) a. nyɔ́nu=a ɸle dze woman=DEF buy.FAC salt ‘The woman bought salt.’ b. Kofí tási ná akutsá m etsɔ Kofi aunt give.FAC sponge me yesterday ‘Kofi’s aunt gave me a sponge yesterday.’ If an auxiliary verb is present, it immediately precedes the main verb. In the progressive and prospective aspects, instead of a S(Aux)VO constituent order, the ordering is instead S(Aux)OV and the verb appears to be a verbal noun (Nurse n.d): 13 Based on Hammarström et. al (2016) 19 Lauren Schneider MA Thesis (36) Ewe (Nurse n.d.: 2) a. me=le dɔ́ 1SG=BE.IPFV work ‘I’m working.’ wɔ-ḿ do-PRG b. me=le dɔ́ wɔ-gé 1SG=BE.IPFV work do-PROS ‘I am going to/intend to work.’ Ewe, like Akan, allows serial verb constructions: (37) Ewe (Nurse n.d.: 3) míí ɖa fufu ɖu we cook fufu eat ‘We cooked fufu and ate it.’ All negation in Ewe is encoded by the preverbal particle mé and a clause-final negative particle o: (38) Ewe (Nurse n.d.: 8) a. b. c. atí lá mé kɔ́ tree DEF NEG tall ‘The tree is not tall.’ o NEG nye mé ga le 1s NEG ITR be.IPFV ‘I’m no longer growing.’ mé ga NEG ITR yi go tsitsi-ḿ grow.grow-PRG o NEG o NEG ‘Don’t go!’ d. 14 wó má14 ga NEG.POT ITR 3p ‘They shouldn’t go.’ yi go o NEG ma is the combination of me + a (Nurse n.d.: 8). These should both be analyzed as clitics. 20 Lauren Schneider MA Thesis In Ewe, negation is expressed by the bipartite morpheme demonstrated in (38) above. The morpheme me is cliticized15 onto the first element of the verb phrase while o occurs at the end of the clause but before sentence final particles (Adjei 2014: 34). Unlike Akan, in a negated SVC, the negative is only stated once before the first verb: (39) Ewe (Nurse n.d.: 7) Kofi mé=mli kpé=á Kofi NEG=roll.FAC stone=DEF ‘Kofi didn’t roll the stone here.’ vɛ́ come o NEG This system of double negation closely resembles the bipartite negation seen in stage 3 of Jespersen’s cycle. In order to determine whether or not this is a case of Jespersen’s cycle additional historical and comparative data is needed. Adjei (2014) provides an interesting account of children’s speech. If either of the Ewe negation markers were being eroded, children’s speech would be a good place to look for variation if there was any. Adjei’s (2014) data contains only negative clauses with both negative particles. As far as current data shows, the bipartite system of negation in Ewe is in no danger of disappearing. Possible comparative evidence for the existence of Jespersen’s cycle in Gbe languages can be seen by comparing Ewe with a couple of its fellow Gbe languages. Like Ewe, Gen (or Gengbe) [gej], has a bipartite negation marking system. 15 Adjei (2014) describes the mé negation marker as a clitic but Nurse (n.d.) does not mark the morpheme as a clitic in the examples in 38. 21 Lauren Schneider MA Thesis (40) Gen (Samson Lotven, personal communication, April 7, 2017) a. b. ɲɛ̃̀ mú ̃ 1SG NEG ‘I don’t go.’ jı̀ go ɲɛ̃̀ má ̃ 1SG NEG.POT ‘I won’t go.’ nà ̃ ò HAB PART jı̀ go ò PART In (40b), the negative marker mú ̃ combines with the potential marker to form má ̃ , encoding both negation and potentiality. The phrase-final marker, glossed PART (PRT elsewhere in this paper) by Lotven, must occur with the preverbal negative marker and cannot be left out. The firstperson-singular subject pronoun is segmentally identical to the negative marker, distinguished by having a low tone: mù ̃. In negative clauses, like those in (40) above, a different version of the pronoun is used. This type of negative asymmetry can be classified as A/Cat, which incorporates person-number-gender distinctions in negative clauses. Fon (or Fongbe) [fon] is another Gbe language and is spoken in southwestern Benin. This language, as is expected in a Kwa language, exhibits an SVO constituent order: (41) Fon (Lefebvre 1991: 23) Kɔ̀kú gbà mɔ́tò NAME destroy car ‘Koku destroyed cars.’ Unlike Ewe, Fon does not exhibit bipartite negation. Instead, this language only has a post-verbal negative particle: (42) Fon (Lefebvre 1991: 27) wá ǎ NAME arrive NEG ‘Koku has not arrived.’ Kɔ̀kú 22 Lauren Schneider MA Thesis This particle differs from Akan and many other Kwa languages first in that is it post-verbal and second in that it is not a nasal morpheme. As more data is presented we will see that a preverbal negative morpheme is the most common marker of negation in Kwa languages. For example, Gun [guw], another Gbe language which is spoken near the southeast border of Benin and Nigeria, exhibits this pattern: (43) Gun (Aboh 2009: 5) Sɛ́sın má ná sɔ̌ kùn mótò cè sɔ́ ́ ú! À Sesinou 2SG NEG FUT again drive car 1SG.POSS hit ‘Sesinou! You will not again drive my car hit (i.e. into) the wall today!’ àdó wall égbè today This example indicates that the negative marker is a particle rather than a prefix because additional words can occur between the negative marker and the verb stem. That being said, it is also possible that the negative marker could be behaving as a clitic, as was seen in Ewe (see example 39). Additional data would be required to determine whether the negative marker cliticizes onto the first element of the verb phrase or is a distinct phonological word. Both Fon and Ewe have a vowel sound as the negation marker after the verb while both Ewe and Gun have mV preverbal negation marking. When comparing Fon, Gun, and Gen to Ewe, we see negation patterns that resemble the stages in Jespersen’s cycle: (44) Gun Ewe/Gen Fon NEG VERB NEG VERB NEG VERB NEG STAGE I STAGE II STAGE III While Ewe’s preverbal negative marker is not currently showing any signs of eroding, it is possible that Fon did have one. Without historical data, it is difficult to say for sure but this comparative data does seem to indicate that Gun may be in the first stage of Jespersen’s cycle, Ewe in the middle stage, and Fon in the final stage. Another proposal could be that Ewe represents an older stage in the development of negative markers while Gun and Fon represent alternative methods that have developed for reducing the negative marking. Double negation is 23 Lauren Schneider MA Thesis not common in Kwa languages and frequent contact with other languages of the group may wear away one marker or the other. 3.3 Guang languages The Guang languages are spoken largely in Ghana, concentrated in the southeast around the Volta lake region. As can be seen in the map below, Guang languages can also be found in Togo and Benin: (45) Distribution of Guang languages in Ghana, Togo, and Benin16 Benin Togo Ghana The Guang languages are classified under the Tano branch of the Kwa language group. The following chart breaks down the subgroup: 16 All of the maps used in this paper have been created using QGIS software. The coordinates for the languages were obtained from Glottolog 2.7 (Hammarström et. al 2016). 24 Lauren Schneider MA Thesis (46) Abbreviated Tano family tree: Guang17 Foodo Gonja Central Tano North Guang Krobu Western Tano Mountain Nkonya (Nkami) Nterato Oti Guang Tano Gikyode-Ginyanga Dompo Awutu South Guang Hill Efutu Awutu Senya River Gua Cherepon Larteh (Leteh) Nawuri Chumburung Dwang Kplang Krache Nchumbulu Tchumbuli Like the other Kwa languages surveyed so far, Guang languages exhibit an SVO basic constituent order. The examples in (47) have been provided to illustrate the consistent word order across the branches of Guang: (47) a. Nkami (Akanlig-Pare & Asante 2016: 38) Otu lɛ̀-dʊ oyi Otu tree PRF-climb ‘Otu has climbed a tree.’ b. Krache (Abunya 2010: 110) Kofi kɛ́-gyɪ̀ Ama Kofi FUT-know Ama ‘Kofi will know Ama.’ c. Leteh (Ansah 2015: 35) Kofi bè-sú Ama Kofi FUT-send Ama ‘Kofi will send Ama.’ 17 Based on Hammarström et. al (2016). 25 Lauren Schneider MA Thesis According to Dolphyne & Dakubu (1988: 84) the Guang languages consistently mark negation “by means of a prefix, almost invariably consisting of m plus a vowel, that usually precedes the tense/aspect markers.” This is generally true of the North Guang languages but is less consistently true of the South Guang languages. The Guang languages vary in whether they employ morphological or syntactic negation to encode sentential negation. The discussion of Guang languages has been broken into two sections. Section 3.3.1 will discuss negation patterns in the North Guang languages Nkami, Foodo, Krache, Gonja, Chumburung, and Nawuri. Section 3.3.2 will cover the South Guang languages Efutu, Leteh, Cherepon and Gua. 3.3.1 North Guang The North Guang languages generally follow Dolphyne & Dakubu’s (1988) generalization, marking negation with an /m-/ prefix, often with a vowel. The North Guang languages seem to employ mostly morphological negation but this is not always clear from the data provided. As was mentioned earlier in the paper, the negative morphemes presented here have been collected from the grammatical descriptions written by other linguists. It can be difficult to determine solely based on these analyses whether a particular morpheme in a given language should be considered as an affix or a separate word. The representations of the data are often presented orthographically rather than phonetically. There is an orthographic tendency to represent certain preverbal morphemes as separate words when they would be better treated as affixes or clitics. The treatment here relies somewhat on the conventions used by the previous presentations of the data. Nkami (or Nkonya) [nko] is a North Guang language that exhibits a clear case of morphological negation in the form of a negative verbal prefix: 26 Lauren Schneider MA Thesis (48) Nkami negative verbal prefixes mʊnPast mʊnɛProgressive mʊntɪPerfect màFuture máHabitual Nkami differs from Akan in that it has different forms of the negative verbal prefix for different tenses and aspects. The future and habitual negatives in Nkami differ only in terms of tone: (49) Nkami (Akanlig-Pare & Asante 2016: 42) a. Kofi mà yɪrɪ stand NAME NEG.FUT ‘Kofi will not stand there.’ mʊ there b. Kofi má mʊ there yɪrɪ NAME NEG.HAB stand ‘Kofi does not stand there.’ The future and habitual negatives are expressed by the same segment /ma/ and the future versus habitual distinction is marked by a tonal morpheme. Nkami also exhibits symmetrical negation: (50) Nkami (Akanlig-Pare & Asante 2016: 33) 18 a. be-be-tʃu mʊ. 3PL-FUT-lift 3OBJ ‘They will lift her up.’ b. bɛ-ma-tʃu mʊ. 3PL-FUT.NEG-lift 3OBJ ‘They will not lift her up.’ The only difference in this set is that the third person plural pronoun is realized as [be] in (50a) and [bɛ] in (50b). This occurs because the /a/ in the future negative prefix in (50b) blocks the [+ATR] spreading that occurs in (50a). 18 The tone is not provided for these examples. The negative future has a low tone, as demonstrated by example (48). The data provided does not illustrate how the tones interact with their environment. 27 Lauren Schneider MA Thesis Foodo [fod] is another North Guang language that has been analyzed as having negative verbal prefixes. (51) Foodo negative verbal prefixes maNPerfective mÈ -nÉ Imperfective má Future In Foodo, the future negative prefix is consistently realized as [máá-] (example (52a) below) while the realizations of the perfective and imperfective negative prefixes are determined by the phonetic context: (52) Foodo (adapted from Plunkett 2009: 129) 19 a. b. c. FUTURE NEGATIVE ɔ̀-ɲɪ-́ ḿ máá-↓náánáá CP-man-NSF NEG.FUT-go ‘The man will not go.’ ↓ ɔ̀-ɲɪ́-ḿ máá-wù CP-man-NSF NEG.FUT-see ‘The man will not see.’ PERFECTIVE NEGATIVE ɔ̀-ɲɪ-́ ḿ màǹ-náá CP-man-NSF NEG.PFV-go ‘The man did not go.’ ɔ̀-ɲɪ́-ḿ màŋ̀-wù CP-man-NSF NEG.PFV-see ‘The man did not see.’ IMPERFECTIVE NEGATIVE ɔ̀-ɲɪ-́ ḿ mɛ̀-nɛ́ɛ́-↓náánáá CP-man-NSF NEG-IPFV-go ‘The man is not going.’ ɔ̀-ɲɪ́-ḿ mè-néé-wù CP-man-NSF NEG-IPFV-see ‘The man is not seeing.’ In (52b) above, the perfective negative prefix maN- contains a homorganic nasal and is realized as [màǹ-] before /n/ and [màŋ̀-] before /w/. In (52c), the imperfective prefix mÈ -nÉ - is realized first as [mɛ̀-nɛ́ɛ́-] and second as [mè-néé-] agreeing with the [ATR] value of the stem. 19 Class prefix glossing in Plunkett (2009) is complex and confusing without sufficient contextualization and so has been simplified in the examples used in this paper. 28 Lauren Schneider MA Thesis Krache [kye] (also known as Krachi or Kaakyi) is another North Guang language that exhibits negative nasal prefixes. Abunya (2010) defines the negative prefixes as follows: (53) Krache negative verbal prefixes (Abunya 2010) mFuture ḿPast ḿProgressive ḿpɛ̀ Perfect anImperative/optative àáHabitual The future tense marker is /kÉ -/ and the negative future tense marker is /mÉ -/. The examples below demonstrate the future and negative future: (54) Krache (Abunya 2010: 110) FUTURE NEGATIVE FUTURE a. àlı-́ ké-yɔ̀ obuase 1PL.SUBJ-FUT-go house20 ‘We will go home.’ b. àlı́-mé-yɔ̀ obuase 1PL.SUBJ-NEG.FUT-go home ‘We will not go home.’ c. bɛ̀-kɛ́-tɛnsu 3PL.SUBJ-FUT-forget ‘They will forget.’ d. bɛ̀-mɛ́-tɛnsu 3PL.SUBJ-NEG.FUT-forget ‘They will not forget.’ In this example, the tone pattern of the phrase as a whole is unchanged by negation in this case but we will find that this is not always the case. In Krache, the past tense is typically encoded by a high tone /É -/ prefix. In the negative past, the past tense prefix is replaced by a high tone syllabic homorganic nasal which is prefixed to the verb stem. 20 Abunya (2010) glosses this ‘house’ and ‘home’ in the free translation and in other examples (see 54b). 29 Lauren Schneider MA Thesis (55) Krache (Abunya 2010: 113, 115) PAST NEGATIVE PAST a. Ama ɛ́-kpɪ́sáŋ̀ Ama PST-sneezed ‘Ama sneezed.’ b. Ama ŋ́-kpɪ́sáŋ̀ Ama NEG.PST-sneeze ‘Ama did not sneeze.’ c. gyoro wʊ́ é-wú DET PST-die dog ‘The dog died.’ d. gyoro wʊ ń-wú DET NEG.PST-die dog ‘The dog did not die.’ e. ó-múŋà 3SG.SUBJ.PST-kneel ‘S/he knelt down.’ f. ò-ḿ-mùŋá 3SG.SUBJ.PST-NEG21-kneel ‘S/he did not kneel down.’ akontɪ down akontɪ down Like the negative future, the negative past construction is straightforward, with the only change being the replacement of the past marker with the negative past. One noteworthy difference is that the negative past marker carries the same high tone as the affirmative past marker and so does affect the tone pattern of the negated phrase as can be seen in (55f). In the affirmative, the progressive aspect is encoded by a similar prefix to the past tense and is distinguished only by having a low tone instead of a high tone. In the negative progressive, a high-tone bilabial nasal /ḿ-/ is prefixed to the low-tone /È -/ progressive prefix. (56) Krache (Abunya 2010: 117-8) PROGRESSIVE 21 NEGATIVE PROGRESSIVE a. Yaw è-fùkı́ Yaw PROG-jump ‘Yaw is jumping.’ b. Yaw m-é-fùkı́ Yaw NEG-PROG-jump ‘Yaw is not jumping.’ c. kegyifɔrɪ wʊ́ ɛ̀-dɪdɪ DET PROG-sleep child ‘The child is sleeping.’ d. kegyifɔrɪ wʊ́ m-ɛ́-dɪɪ́ dɪ DET NEG-PROG-sleep child ‘The child is not sleeping.’ Gloss for the negative marker was left out Abunya (2010). 30 Lauren Schneider MA Thesis In this example, when the negative marker is prefixed to the progressive prefix, the high tone of the negative replaces the low tone of the progressive marker. Other than the effect on the tone pattern, the change from affirmative to negative is largely symmetric. The perfect aspect is marked by the prefix /ɛ́kà-/ and the negative perfect is marked by the prefix /ḿpɛ̀-/, both with a high low tone pattern: (57) Krache (Abunya 2010: 115) PERFECT a. NEGATIVE PERFECT ɔdɔɔpʊ wʊ́ ɛ́kà-bà farmer DET PERF-come ‘The farmer has come.’ b. ɔdɔɔpʊ wʊ́ ḿpɛ̀-bà farmer DET NEG.PERF22-come ‘The farmer has not come.’ Like the previous examples, the negative of the perfect aspect is symmetric. The negative habitual is the only negative prefix in Krache that does not typically involve a nasal phoneme. The habitual is marked by a /ɪ̌ -/ morpheme with a rising tone and the negative habitual is typically encoded by a long /àá-/ morpheme as is demonstrated by (58b and d) below: (58) Krache (Abunya 2010: 112) HABITUAL NEGATIVE HABITUAL a. Kofi ɪ́-kyá HAB-dance Kofi ‘Kofi dances.’ c. Kofi ɪ́-nyɪ́ŋɪ́ ɔtɛ Kofi HAB.remember story ‘Kofi remembers the story.’23 e. Kwaakru ı́-nyı́ Kwaakru HAB-know ‘Kwaakru knows Ama.’ Ama Ama wʊ́ b. Kofi àá-kyà NEG.HAB-dance Kofi ‘Kofi does not dance.’ d. Kofi àá-nyɪ́ŋɪ́ ɔtɛ wʊ́ Kofi HAB.remember story DET ‘Kofi does not remember the story.’ f. Kwaakru ǹ-nyı̀ Ama Kwaakru NEG.HAB-know Ama ‘Kwaakru does not know Ama.’ DET 22 This is glossed only ‘NEG’ in the data. There are errors in glossing and free translation in original paper. (58c) is missing a gloss for ɔtɛ ‘story’. The free translation provided for (58c) is ‘I remember the story’ and (58d) is ‘Kofi remembers the story.’ 23 31 Lauren Schneider MA Thesis There is a single exception to the use of the long /àá-/ as the negative habitual prefix as shown in (f) above. In this example, the verb nyi ‘know’ takes a nasal /n-/ prefix. No additional data is provided on what types of verb take this prefix or what grammatical conditions require it. The negative imperative and the negative optative utilize the marker /aN-/. According to Abunya’s (2010) analysis, it is prefixed to the second person singular subject /f/ to create the compound prefix /faN-/. Typologically, few Guang languages have prefixes consisting of a single non-nasal consonant, so this analysis may require additional review. The examples below demonstrate the use of the negative imperative: (59) Krache (Abunya 2010: 118-9) IMPERATIVE a. kyà! IMP.dance ‘Dance!’ c. yɔ̀! e. mɔ̀sɪ!́ IMP.laugh/smile ‘Laugh/smile!’ f. fam-mɔ̀sɪ́ 2SG.SUBJ.NEG.laugh/smile ‘Laugh/smile!’ IMP.go ‘Go!’ NEGATIVE IMPERATIVE b. faŋ-kyà! 2SG.SUBJ.NEG-dance ‘Do not dance!’ d. fan-yɔ 2SG.SUBJ.NEG.go ‘Do not go!’ Lastly, the optative is demonstrated by example (60) below: (60) Krache (Abunya 2010: 119-120) OPTATIVE a. a-kyà sɛ let 2PL.SUBJ.OPT-dance ‘Let us dance; I wish that we dance.’24 NEGATIVE OPTATIVE b. fan sɛ-a-kya 2SG.SUBJ.NEG let-2PL-dance ‘Let us not dance.’ Negation in Nkami, Foodo, and Krache is asymmetric. These languages fall under the subtype A/Cat because the marking of the grammatical categories differs from their marking in 24 The example is glossed second person plural but the free translation indicates first person plural. It might be better translated ‘Let you all dance’. 32 Lauren Schneider MA Thesis affirmatives. In each of these languages, the negative construction has asymmetry in that the marking of tense and aspect forms a portmanteau negative marker. In the Kwa languages surveyed so far, a few criteria for determining whether a negator is a prefix or a particle have been established. Clear cases of particles include those in which intervening words may occur between the negator and the verb (see (43) Gun). Evidence for a negator’s status as a prefix includes occurring closer to the stem than other verbal prefixes (see (50) Nkami) as well as extensive assimilation with the stem ((52) Foodo). The next set of North Guang languages have been described as using preverbal negative particles in existing analyses: (61) North Guang languages with two or fewer distinct negative morphemes a. Gonja25 b. Chumburung maa mʊŋ Negative imperfective Negative perfective c. Nawuri Negative Negative incompletive máN- maŋ mEE Gonja [gjn], spoken in the Northern and Brong-Ahafo regions of Ghana, has a single negative morpheme. The Gonja data is written orthographically. There are three forms found in the data: maŋ, maa, and maaŋ. The form maaŋ is shown in example (62) below, maa in (63), and maŋ in (64). Gonja, like Akan and Ewe, allows serial verb constructions but it differs from them in that Gonja requires a homorganic N as a serializing conjunction (underlined in example (62) below). 25 All Gonja data is selected from data collected in Ghana by Roderic Casali in 2014 and 2015. The 2014 data is from audio recordings of an adult male speaker named Francis Zakaria. The 2015 data is from audio recordings of an adult male speaker named Amidu Changa. The research and recording for both sets of data was carried out at the Kanvili office complex of the Ghana Institute of Linguistics, Literacy, and Bible Translation (GILLBT) near Tamale, the capital city of the Northern Region. The GILLBT staff also provided logistical support for the project. The texts collected have preliminary glosses compiled by previous students at the Canada Institute of Linguistics at Trinity Western University and may be in need of further clarification. 33 Lauren Schneider MA Thesis (62) Gonja nɛ b yɔ ndoŋ maaŋ bɛta 3PL go place return CNJ NEG ‘they who went there would not return at all' m ba come CNJ kikɛ at.all As can be seen in (62) above, in Gonja, the negation is marked only before the first verb of the SVC and the serializing conjunction occurs in front of the second verb. (63) Gonja bɔkwe lelemu nɛ be maa CNJ week entire 3PL.SUBJ NEG ‘For a whole week, they did not eat anything.’ (64) ji eat sheŋ anything Gonja ndoŋ nɛ Ewura yɛ men maŋ wu There CNJ chief say 2PL.CONT NEG see ‘Then the chief said, “Didn’t you see?”’ a EMPH The differences between maŋ, maa, and maaŋ do not appear to have phonetic basis. For now, these forms have been analyzed as allomorphs of the morpheme maN. If there is a difference in meaning between these forms, more data would be required to determine this. While this marker is written as distinct word in the orthography, there is evidence that the negative marker is subject to assimilation with the verb stem. The morpheme-final nasal assimilates to the point of articulation of the following consonant and the /a/ vowel is subject to [ATR] harmony. Based on this and the group’s preference for affixes so far, it seems likely that the negative marker in Gonja is an affix or clitic. Chumburung [ncu] is analyzed as having two negative particles, the first is mʊŋ which encodes negation and the perfective aspect. Examples (65a and b) demonstrate the use of this marker: 34 Lauren Schneider MA Thesis (65) Chumburung (Hansford 2010: 128; 2011:54)26 a. ɔ mo̱ŋ yɔ Aburikyire 27 he NEG.PFV go.to Europe ‘He/she has never been to Europe.’ daa ever b. Fo̱ ya de̱e̱re̱, fo̱ mo̱ŋ You COND look, you NEG.PFV ‘If you look at it, you will not see its legs.’ ke̱e see mò̱ its ayaa. legs In (66) below mʊŋ functions in the clause expressing the condition of a conditional construction which is expressing a non-existent state: (66) Chumburung (Hansford 2011: 52) Ɔbɛɛ Person ko̱ŋko̱ one e̱ COND mo̱ŋ bo̱-rɔ, NEG.PFV be.there, bo̱ maa taare̱ a bo̱ so̱rɔ tɔso̱rɔ. 28 they able that they carry.SBJV headload. NEG.IPFV ‘If one is not there, they can’t carry a headload.’ This example also provides a case of maa, which is the negative particle used to encode negation and imperfective aspect. Another example is provided by (67) below: (67) Chumburung (Hansford 2011:53) Mo̱ e̱ kyaa na bo̱ maa laŋ PRES dance I and they NEG.IPFV beat ‘I dance but they are not drumming.’ akɔye̱. drum. There is one other case of maa in Hansford’s (2011) paper: 26 Hansford (2010) & (2011) uses the Chumburung orthography rather than IPA. Additionally, the representations of the data do not account for tone. Correspondences between the orthography and IPA are shown below: Chumburung orthography IPA “gy” /ʤ/ “ky” /ʧ/ “ny” /ɲ/ “e” /ɪ/ “o” /ʊ/ 27 This morpheme is glossed ‘NEG.PAST’ by Hansford (2010), and ‘STAT.NEG’ and ‘NEG’ in Hansford (2011). According to a conversation with Keith Snider, Chumburung negative markers contain no tense marking, only aspect. 28 This morpheme is glossed ‘NEG.FUT’ and ‘PRES.NEG’ by Hansford (2011). 35 Lauren Schneider MA Thesis (68) Chumburung (Hansford 2011:53) taare̱ able NEG.IPFV ‘I can’t roll it up.’ Ma 29 a that m bɔɔ. 1SG roll.up.SBJV In this example, the vowel is shorter and it encodes first person singular as well as negation. This is the only case and it will be treated as an allomorph. It should also be noted that a phrase-final glottal stop occurs in all negative utterances. This is not reflected in the Hansford (2010) & (2011) data, which uses the Chumburung orthography rather than IPA. The glottal stop is not a contrastive phoneme in Chumburung. Like Chumburung, Nawuri [naw] also is analyzed as having two preverbal negative particles which always occupy the initial position in the verb phrase: (69) Nawuri (Casali 1995: 74, 75) a. ɔ maŋ 3SG NEG ‘She isn’t there.’ bʊ be b. ɔ mɛɛ tɔwɪ 3SG NEG.INCOMP speak ‘He doesn’t speak.’ tɔ there Casali (1995) proposes that maŋ is used to indicate negation alone while mEE is used to mark both negation and incompletive aspect. mEE has two allomorphs, [mee] and [mɛɛ], the latter of which is shown in example (69b) above. The negative marker in Nawuri, unlike in Chumburung and Gonja, exhibits assimilation with the verb stem. Gonja is the clearest in its use of particles rather than prefixes and at this point in its development it is safe to analyze Chumburung as having particles as well. Nawuri is 29 This morpheme is glossed ‘1SG.PRES.NEG’ by Hansford (2011). This is the only case where a negative morpheme is glossed as encoding person and number. This is not typologically expected in Kwa. The forms [ma] and [maa] will be treated as allomorphs in this paper. 36 Lauren Schneider MA Thesis likely undergoing a transition from particle to prefix as the negative marker is weakening, taking on features of the verb it modifies. Negation in these languages is largely symmetric. North Guang languages are consistent in that they all use preverbal negative nasal morphemes to encode negation. Nkami and Foodo use negative verbal prefixes and have three or more options for marking tense and aspect. Gonja and Chumburung are clear cases of the use of preverbal negative particles and each of these languages has two options for different tense and aspect contexts. Since there is a consistent nasal morpheme that appears to be a split between languages with preverbal particles and languages with prefixes, it seems likely that a protolanguage for this group likely contained nasal preverbal particles which in some languages have become verbal prefixes. The Nawuri data provides a potential example of what a stage in that process looks like. In the next section, we will look at South Guang languages and find that this subgroup of Kwa tends to be less consistent. 3.3.2 South Guang Dolphyne & Dakubu (1988) make the generalization that the Guang languages consistently mark negation by means of a /m-/ prefix. The North Guang languages do exhibit this type of behavior, most often utilizing an mV(C) pattern. Unlike the consistency found in North Guang the South Guang languages exhibit greater variation. Efutu [afu] is spoken in southern Ghana and is a member of the Awutu sub-branch of South Guang. Like the North Guang language group, Efutu marks negation using preverbal negation marking: 37 Lauren Schneider MA Thesis (70) Efutu (Obeng 2008: 29, 84-5) a. mɪ 1SG mɑ-ɑ NEG-have apʲɪɛkʊ̃ money pii much ‘I don’t have much money.’ b. mʊ̀ wó 3SG RFLX mɑ́30 NEG ɑbɔ́ be fʲɪ̀ɛ̀w nice ‘S/he is not nice.’ The first thing of note is that Efutu appears to be following the same pattern as the North Guang languages, using a preverbal negative nasal morpheme. The negative prefix is realized either as /mɑ/ or /mɔ/ depending on the vowel of the verb it immediately precedes: (71) Efutu (Obeng 2008: 84) a. ɔ̀sé31 nʊ́ mʊ̀bɔ́ woman the she ‘The woman is good.’ ɔ̀ dɛ̀ be good b.32 ɔ̀sé n̄ mʊ̀ḿ woman the she ‘The woman is bad.’ mɔ́ NEG ɔ̀ be dɛ̀ good In this example, is appears that [mɔ́] is the preverbal negative prefix but this is not the only change that takes place in the sentence. The definite article loses its vowel and the form of the 3rd person pronoun changes. According to Obeng (2008: 16), n is the definite article in Efutu and [nʊ́] appears to be one of its allophonic variations. A similar example to (71b) above is (72) below: 30 The prefix vs. particle distinction is inconsistently marked in Obeng (2008). Based on the data provided, the negative marker is analyzed here as a prefix rather than as a particle. 31 This is a sequence where one would expect to see vowel harmony. The Efutu data provided by Obeng is largely orthographic and so does not always reflect cases of vowel harmony. 32 The morpheme-by-morpheme gloss for this example was not provided by Obeng (2008). The gloss has been constructed based on previous glosses provided in the data. 38 Lauren Schneider MA Thesis (72) Efutu (Obeng 2008: 26) ɔ̀sé ḿ òmɔ́-ɔ́33 dɛ̀ woman this NEG-be good ‘This woman is not good.’ Based on examples (70) - (72), I propose that the negative prefix is underlyingly mV- and that it assimilates with the initial vowel of the verb. The negative prefix can be further reduced, as is demonstrated by the next example: (73) Efutu (Obeng 2008: 26) a. ɔ̀mʊ́ èdǽ this it.be ‘This is a stick.’ jı̀bı́ stick 35 b.34 ɔ̀mʊ́ ̃ idæ jibi36 this it.be stick ‘This thing is a stick.’ c. ɔ́mʊ́ ̃ m-ıd ́ æ̋ this NEG-it.be37 ‘This is not a stick.’ jı̀bı́ stick This example shows how the prefix [m-] is attached to the verb in order to negate the clause. The only time the vowel is elided is in negative copula clauses, like the one demonstrated in example (73c) above. The Efutu negative marker has been analyzed here as a prefix. The inconsistency of 33 The morpheme-by-morpheme gloss for this example was not provided by Obeng (2008). The morpheme òmɔ́ is the only case where a vowel occurs before the negative morpheme. The gloss has been constructed based on previous glosses provided in the data but the /o/ has not been accounted for. 34 As indicated with certain previous examples, the morpheme-by-morpheme gloss for this example was not provided by Obeng (2008) and the gloss shown was constructed based on previous glosses. 35 While not glossed as such, it is possible that the change in vowel from e in (73a) to i in (73b) is meant to differentiate between the use of ɔ̀mʊ́ as a demonstrative in (a) and as a determiner in (b). 36 The data for these examples are presented here as they were presented in Obeng (2008). Tone has been included where provided but it has been inconsistently marked in the data. Due to the inconsistency of the tone marking it is difficult to tell how the negation affects the tone pattern of the utterance. 37 This morpheme is written mı́dæ̋ with no hyphens and glossed ‘it-NEG-be’ in the original data. Based on other glosses, it has been revised as shown in (73c). 39 Lauren Schneider MA Thesis the marking may be caused by either the fact that the marker had particle status previously but has since lost its status as a distinct word or by the orthographic tradition. Leteh (or Larteh) [lar], spoken in south-eastern Ghana, is a member of the Hill subbranch of South Guang. Like many other Kwa languages, Leteh exhibits negative verbal prefixes: (74) Leteh (Ansah 2015: 34, 37) a. ɛ̀né bɛ́-dɛ́-sɔ̀ oburodwo 1PL NEG-PRG-buy plantain. ‘We are not buying plantain.’ b. Kofi bé-gyí dɔkɩta Kofi NEG-be doctor ‘Kofi is not a doctor.’ The high tone morpheme bÉ - can be realized as either [bɛ́-], as in example (74a) above, or [bé-], as in example (b), depending on the [ATR] value of the verb it modifies. Example (a) also shows that when the negative marker co-occurs with the tense/aspect marker, the negative morpheme precedes the tense/aspect marker. This language stands out from other Kwa languages in that the marker of negation is not a nasal phoneme. This is especially interesting because of the fact that Efutu, another South Guang language spoken in the same region, follows the pattern of the other Kwa languages in using preverbal nasal negation marking. Two other South Guang languages share the bilabial plosive rather than nasal negative morpheme. Cherepon (or Ɔkere38) [cpn], another member of the Hill sub-branch, is located in the same region of South-East Ghana as Leteh. In Cherepon, 38 The Ɔkere people use ‘Ɔkere’ to refer to themselves and their language. The name was changed to ‘Kyerepong’ (or ‘Cherepon’) by the Akwamu people and is used by non-natives to refer to the language and the people (Animah 2015: 2). 40 Lauren Schneider MA Thesis negation is also marked with a verbal prefix bÉ - which can be realized as [bé-] as in (75a) below or [bɛ́-] in (b): (75) Cherepon (Animah 2015: 55, 94) a. Kofi Kofi bé-wè NEG-go.HAB sukuu. school ‘Kofi does not go to school.’ b. Papa Larbi Papa Larbi bɛ́-bͻ̀ɛ̀ NEG-do.HAB ͻtͻkoa, nothing a-bɛ́-dı̀ 3SG-NEG-sleep.HAB nso. too ‘Papa Larbi does nothing, he doesn’t sleep either.’ If there is a verbal aspect prefix, the negation marker is attached to that prefix which is attached to the verb stem: (76) Cherepon (Animah 2015: 55, 94) Kofi bé-né-wè sukuu. Kofi NEG-PFV-go school ‘Kofi has not gone to school.’ In this example, the perfective marker is attached to the verb stem and then the negation marker is attached to the perfective marker. The other South Guang language that shares the bilabial plosive /b/ feature is Gua [gwx]. Gua is located to the north of Leteh and Cherepon in the southern part of the Volta Lake region. Like the other two Hill Guang languages, Gua utilizes the high-tone prefix /bÉ -/ to mark negation: 41 Lauren Schneider MA Thesis (77) Gua39 AFFIRMATIVE a. NEGATIVE kòfı́ hù yâw Kofi see.HAB/PRES Yaw ‘Kofi sees Yaw.’ b. yâw Yaw kòfı́ bé-hù NEG-see.HAB/PRES Kofi ‘Kofi does not see Yaw.’ This example is a clear case of symmetrical negation. Like the other Hill Guang languages, the vowel in the prefix is subject vowel harmony with the stem. (78) Gua a. kòfı́ bè-hú yâw áʨɪ̃̀ FUT-see Yaw Kofi tomorrow ‘Kofi will see Yaw tomorrow.’ b. kòfı́ bèé-hú yâw áʨɪ̃̀ FUT.NEG-see Kofi Yaw tomorrow ‘Kofi will not see Yaw tomorrow.’ c. kòfı́ ɛ̀ɛ́-bwɛ̀ PROG-do Kofi ‘Kofi is working.’ d. kòfı́ bɛ̀ɛ́-bwɛ̀ NEG.PROG-do Kofi ‘Kofi is not working.’ ésım ́ ı̀ work/job ésım ́ ı̀ work/job In examples (78b and d) above, the negative marker combines with the tense or aspect marker. In (b) the low-tone future marker bè- combines with the high-tone negative marker to form the compound negative future prefix bèé-. In (d), the progressive marker combines with the negative marker to form the negative progressive prefix bɛ̀ɛ́-. The South Guang languages are largely symmetric. Gua exhibits a potential exception in example (78) above. This is another example of A/Cat asymmetry. The three Hill Guang languages stand out from the rest of the Guang languages in their use of bÉ - as their negative marker rather than a nasal phoneme. Possible reasons for these typological outliers will be discussed in the next section. 39 Gua data was provided by Michael Obiri-Yeboah, a native speaker of the language, by means of email correspondence. The data collected is featured in the Appendix. 42 Lauren Schneider MA Thesis 3.3.3 A solution for the Guang typological outliers Leteh, Cherepon, and Gua stand out from the other Kwa languages surveyed so far in that these languages do not utilize a nasal negative morpheme. This is especially interesting because of the fact that Efutu, another South Guang language, follows the same pattern as the majority of the other Kwa languages in using preverbal nasal negation marking. Leteh, Cherepon, and Gua make up the Hill sub-branch of the South branch. As can be seen in the map below, Leteh and Cherepon are geographically near the Ga-Dangme languages, which are also spoken in SouthEast Ghana: (79) Location of typological outliers Ghana Dangme Ga Ga-Dangme is a small branch of the Kwa language family that consists only of Ga [gaa] and Dangme (or Adangme) [ada]. Because of the small size of the subgroup and their relevance to the discussion of Leteh and Cherepon, they will be treated here. Both Ga and Dangme contain negative verbs with the phoneme /b/. Ga, unlike other Kwa languages, uses mostly negative suffixes (demonstrated in (80a) below) but it has a negative locative copula verb bɛ́ (shown in (b) below): 43 Lauren Schneider MA Thesis (80) Ga (Campbell 2014: 90) a. gbèé-ɛ̀ dog-DEF kɛ̀ take è-yı́tsò 3SG-head hɔ̀ put jar jar ml̀ı̀ fɛ́ɛ́ inside all àmɛ́ -̃ !ná ̃á ̃á ̃ 3PL-see.NEG lɛ̀ him ‘The dog even put his head into the jar but they still couldn’t find him.’ b. kɔ̀kɔ̀dé!né-!ɛ́ frog-DEF bɛ́ be.located.NEG è-tɔ́-ɔ̀ 3SG-bottle-DEF mlı̀. inside ‘The frog was not in his bottle.’ The negative suffix in (80a) should not be mistaken for a nasal negative morpheme. The nasal phoneme in this example comes from the verb ‘see’. The negative suffix and negative copula verb in Ga are illustrated below: (81) Ga (Campbell 2014) VERB kwɛ̂ ‘look’ INFLECTED yà-kwɛ́ ITIV-look NEGATIVE àmɛ̀ -̃ kwɛ́ɛ́ɛ́ 3PL-look.NEG nà ̃ ‘see’ àmɛ̀-yà-nà ̃ 3PL-ITIV-see àmɛ́ -̃ !ná ̃á ̃á ̃ 3PL-see.NEG yɛ̀ ‘be located’ bɛ̃ be.located.NEG In Ga, the negative suffix seems to be a lengthening of the vowel of the verb. In Dangme, there are several different negation strategies. The first involves a post-verb particle we and seems to resemble Ewe’s clause-final negator, the particle o. A difference from negation in Ewe is that negation in Dangme always involves a high tone on the verb stem (Ameka & Dakubu 2008: 217). A Dangme verb paradigm is shown below: 44 Lauren Schneider MA Thesis (82) Dangme verb paradigm (Ameka & Dakubu 2008: 251) Aspect Perfective Habitual Negative Subjunctive Absolute Form Nà lá Nà lá-á Nà lá wē Nà á lá lá Gloss ‘Na sang’ ‘Na sings’ ‘Na did not/ does not/ is not singing’ ‘Na is to sing, would sing’ ‘sing!’ This negation pattern fits roughly into Miestamo’s asymmetry category A/Cat, in which the marking of grammatical categories in negative clauses differs from their marking in affirmative clauses. In the negative, the aspect distinction found in the affirmative is neutralized. The negative particle cannot co-occur with marking for habitual or subjunctive (Ameka & Dakubu 2008: 253). This first negation strategy neutralizes the contrast between the realis forms. The phonological form of the realis negative depends on the tone class of the verb and varies slightly from one dialect to another: (83) Dangme negation of realis (Ameka & Dakubu 2008: 255) Tone class High Low Mid Verb lá dò dū ‘sing’ ‘dance’ ‘bathe’ Negated verb lá wē dú-ı́ dú wē (Krobo dialect) dú-ı́ (Ada dialect) If the verb stem has a final high tone, the tone does not change and the particle, we follows the verb. If the verb has a final low tone, the stem vowel is raised, there is a suffix consisting of a high vowel, and the whole construction has high tone. Final mid tone verbs follow the patterns of high tone verbs in the Krobo dialect and follow the pattern of low tone verbs in the Ada dialect (Ameka & Dakubu 2008: 255). In their analysis, Ameka and Dakubu (2008) describe we as a particle with phonological realizations as suffix. It seems likely that we could be analyzed as a suffix (-we) with different allomorphs, depending on the tone of the verb stem. 45 Lauren Schneider MA Thesis Irrealis in Dangme undergoes a different set of negation strategies. Irrealis forms are negated by means of negative verbs. The future is negated by means of bé, which is a negative locative verb that has been glossed as ‘absent’ (Ameka & Dakubu 2008: 266). Imperatives are typically negated though the use of the counter-factual verb ko, which indicates a situation to the contrary. Examples of both bé and ko are demonstrated in (84) below: (84) Dangme negation of irrealis (adapted from Ameka & Dakubu 2008: 270) Aspect/Mood Positive Negative a. Future è mā ̃ á ̃ tró kɛtɛ ‘Sh/e will carry the basket’ è bé kɛtɛ tró-ē ‘S/he will not carry the basket’ b. Indir. Imperative è tró kɛtɛ ‘carry the basket’ é kó tró kɛtɛ ‘don’t carry the basket’ c. Dir. Imperative tró ‘carry!’ kò ó tró ‘don’t carry!’ When no subject is expressed, an expression using ko is interpreted as a singular negative direct imperative (Ameka & Dakubu 2008: 259), as in (84c) above. Both Ga and Dangme contain negative bÉ verbs which might explain where Leteh and Cherepon acquired a /b/-initial negation marking. Ga and Dangme also differ from the rest of Kwa in that they can place the negative marker after the verb. Ga marks negation through a negative suffix while Dangme uses a split strategy of a negative suffix -we for realis and premain-verb negative verbs for irrealis. The use of /b/-initial negators as well as negators containing /w/ is more common in other language groups in and around eastern Nigeria, as is demonstrated by (85) below. The map shows the distribution of a handful of languages with /b/initial negation morphemes and /w/ negation morphemes. Language family has been largely disregarded for this illustration but the cluster of Kwa languages has been circled. 46 Lauren Schneider MA Thesis (85) Distribution of /b/-initial and /w/ negation morphemes40 Burkina Faso Benin Ghana Togo Nigeria Cameroon /b/-initial morphemes /w/ morphemes As can be clearly seen by this graphic, /b/-initial and /w/ negation markers are much more common on the east side of Nigeria.41 They do not occur in Kwa other than in the small area that has been circled on the map. It seems most likely that these negation features were acquired by a small group of Kwa languages, possibly from the Ga-Dangme subgroup, and have since then spread to a small selection of South Guang languages. Without additional historical data, this cannot be confirmed. The correlation of negative words containing the phoneme /b/ in southeastern Ghana is interesting but additional data would be required in order to confirm whether or 40 All of the maps used in this paper have been created using QGIS software. The languages included are based on data provided by Dryer (2009). The coordinates for the languages were obtained from Glottolog 2.7 (Hammarström et. al 2016). 41 It can also be seen from the map that there are additional cases in and around Cameroon as well as a single data point in Burkina Faso. 47 Lauren Schneider MA Thesis not Leteh replaced a negative nasal morpheme, as is more common in the Guang language group, with the one it has now. Unlike the North Guang languages, the South Guang languages do not encode tense or aspect in their negative particles or prefixes. These languages show less consistency as a subgroup than the North Guang languages and were likely influenced by the Ga-Dangme languages. 3.4 Ghana-Togo Mountain languages Most of the languages known as Ghana-Togo Mountain (GTM) languages are spoken in the mountain region near the borders of Ghana and Togo. The map below shows the geographic distribution of these languages: (86) Ghana-Togo Mountain languages Togo Ghana Originally the label “Ghana-Togo Mountain” was a daughter node under Kwa but there has since been debate as to whether these languages are any more closely related to one another than they are to the rest of the Kwa languages. More recent classifications accept the Na-Togo and KaTogo groupings shown below and put these directly under Kwa (Dakubu 2009: 631). 48 Lauren Schneider MA Thesis The label, “Ghana-Togo Mountain” will be used in this paper but refers to a group of languages that are geographically close but does not really constitute a cohesive genetic subgroup42. (87) Abbreviated Kwa family tree: Ghana-Togo Mountain languages AvatimeNyangbo Ka-Togo KebuAnimere Kposo-AhloBowili Kwa Basila-Adele Boro Na-Togo Nyangbo Tafi Igo Ikposo Tuwuli LelemiAkpafu Lelemi Siwu Lelemic Logba LikpeSantrokofi Sekpele (Likpe) Selee Like the other Kwa language subgroups, these languages tend to be SVO. Unlike the majority of the Kwa languages surveyed so far, languages found in the GTM area exhibit more diversity in negation patterns, including a few significantly more complicated systems of negation. 3.4.1 Ka-Togo language data The first pair of Ka-Togo languages that will be covered here are Tafi (or Tɪgbɔ) [tcd] and Nyangbo (or Nyagbo or Tutrugbu) [nyb]. Tafi is spoken in the Volta Region of South-eastern Ghana. Like many other Kwa languages, Tafi is a serializing language and in SVCs the verbs cannot be independently negated (Bobuafor 2013: 10). Like Chumburung, there is an utterancefinal glottal stop /ʔ/ in Tafi which is used to mark negative utterances. It is considered as a prosodic clause marker and is not a contrastive sound in Tafi (Bobuafor 2013: 23). 42 Classification based on Hammarström et. al (2016). 49 Lauren Schneider MA Thesis Negation in Tafi is expressed by a verbal prefix tÍ- in the present and dI- in the nonpresent. These prefixes exhibit ATR vowel harmony which means that their realization depends on the ATR value of the initial vowel in the verb: (88) Tafi tÍ - prefix with verb paradigms (Bobuafor 2013: 215) a. ɪ́-tɪ́-bhɪtɪ̄ʔ ɔ́-tɪ́-bhɪtɪ̄ʔ á-tɪ́-bhɪtɪ̄ʔ bʊ́-tɪ́-bhɪtɪ̄ʔ nɔ-tɪ́-bhɪtɪ̄ʔ bá-tɪ́-bhɪtɪ̄ʔ ‘I don’t do’ ‘You don’t do’ ‘S/he doesn’t do’ ‘We don’t do’ ‘You don’t do’ ‘They don’t do’ b. í-tí-huʔ ó-tí-huʔ é-tí-huʔ bú-tí-huʔ no-tí-huʔ bé-tí-huʔ ‘I don’t hit’ ‘You don’t hit ‘S/he doesn’t hit’ ‘We don’t hit’ ‘You don’t hit’ ‘They don’t hit’ The negative morphemes tÍ - and dI- are selected on the basis of tense and aspect features of the predicate they negate. For example, with active and stative predicates, tÍ - is used to indicate present negation while dI- is used to indicate past negation: (89) Tafi negative prefixes (Bobuafor 2013: 215) ɪ́-tɪ́-báʔ ɪ́-tɪ́-waʔ ɪ́-tɪ́-shɪʔ̃ í-tí-sǐʔ ‘I do not come’ ‘I do not do’ ‘I do not leave’ ‘I do not run’ ɪ-dɪ-báʔ ɪ-dɪ-waʔ ɪ-dɪ-shɪʔ̃ i-di-síʔ ‘I did not come’ ‘I did not do’ ‘I did not leave’ ‘I did not run’ There is a third marker of negation in Tafi, gɪ-, which has a much more restricted use. In the corpus used by Bobuafor (2013), it occurs with the verb sɪ́ ‘say’ and expresses ‘did not intend to do something’: (90) Tafi (Bobuafor 2013: 216-7) e-yí nɔ̂ nɪ́ gɪ a-sɪ́ yɪ́ é-nī CP-child WH.ever DEF REL 3SG.DEP-say 3SG.IND CP-mother á-tɪ́-dɔ tɪ-rá e-lishí yɪ́ tsyɪ́ a-gɪ-sɪ́ 3SG-NEG3-say SM-NEG1-contact CP-sleep CP-night 3SG.IND too ‘The child who would not let its mother sleep will also not sleep.’ á-ba-dɔ SM-FUT-contact Negation in Tafi also interacts with other aspectual categories. For example, when the negative and present progressive markers co-occur, the vowels of the resulting form are the same as the 50 Lauren Schneider MA Thesis vowel of the pronominal form they occur with, with the exception of the first-person plural (Bobuafor 2013: 217). In the examples (91a-c, e and f), the vowel of the negative morpheme matches that of the pronominal prefix while in (91d), the first-person plural, the vowel does not change: (91) Tafi negative present progressive (Bobuafor 2013: 217) a. b. c. d. e. f. ɪ-́ tɪ́ɪ́-gāʔ ɔ́-tɔ́ɔ́-gāʔ á-táá-gāʔ bʊ́-tɪ́ɪ́-gāʔ nɔ-tɔ́ɔ́-gāʔ bá-táá-gāʔ ‘I am not walking’ ‘You are not walking’ ‘S/he is not walking’ ‘We are not walking’ ‘You are not walking’ ‘They are not walking’ Nyangbo is located in the same region of South-eastern Ghana as Tafi and has a similar, but simpler mode of negation. There are two markers in Nyangbo, tV- and gɛ-. The prefix tV- can be realized as [ti-], as in (92a), or as [tɛ-], as in (b), depending on the ATR value of vowel in the verb stem: (92) Nyangbo (Essegbey 2009; 2012) a. i-ti-nyı́ m’ élı ̃ 1SG-NEG-know 1SG:POSS mother I don’t know my mother’s name. b. Kofi a-tɛ́-ka-á-tsɛ̅ AM-NEG-still-PROG-pick Kofi ‘Kofi is no longer taking biscuit.’ c. nyé name biskwit biscuit o-dzinidzini a-pɛ a-nyɛ-ɛ petee lɔ-gɛ-mɔ-ɛ́ AM-search CM-man-DEF all 3SG-NEG-see-3SG ‘Earthworm looked everywhere for the man but did not see him.’ CM-earthworm Based on this data, it appears that, similar to the negative prefixes in Tafi, tI- may have a present meaning and gɛ- a past meaning. Additional data would be required to establish any further similarities with the Tafi negation system. 51 Lauren Schneider MA Thesis The /t/, /d/, and /g/ initial negators are features of these two GTM languages that set them apart from the rest of Kwa. We will see as we examine other members of Ka-Togo and Na-Togo that this feature appears in both of these branches. As these are not features found elsewhere in Kwa, there are two possible solutions. First, that it was acquired from another Niger-Congo language group, where these types of features are more common,43 or second, that it is an innovation of a proto-Ghana-Togo-Mountain language that has been inherited by these languages. At this point, the first solution is preferred but would require data that indicates a relatively long-distance migration in order to confirm. Additionally, the most recent analyses of Kwa argue that the Ka-Togo and Na-Togo branches are not necessarily more closely related to one another than they are to the rest of Kwa (Dakubu 2009: 631), which might indicate that these languages might be better categorized elsewhere. The similarities between these branches are likely due to contact with one another. This next pair of languages are Ikposo and Tuwuli and these fall under the Kposo-AhloBowili branch of Ka-Togo. Ikposo [kpo], spoken in Southwest Togo, exhibits a negation pattern much like that of the rest of Kwa. It is described as having a rigid SVO constituent order (Soubrier 2009: 189). This language utilizes a much simpler pattern than most other GTM languages, encoding negation with a verbal prefix na-: (93) Ikposo (adapted from Soubrier 2009: 195) ɛ̀dɪ̄-ɛ̌ bwakʊ́ ɔ́-na-fʊ́ that-DEF REL 3SG-NEG-like ‘The one he does not like ...’ yɪ̏ OBJ3SG This is clearly a case of morphological negation because the marker of negation occurs closer to the verb than other verbal prefixes encoding person and number. 43 Based on data obtained from Dryer (2009). 52 Lauren Schneider MA Thesis The next language in this subgroup has a more complicated system of negation. Tuwuli [bov] is another Ka-Togo language and is spoken in the Central Volta Region of Ghana. Like the rest of Kwa, Tuwuli exhibits a basic SVO constituent ordering: (94) Tuwuli (Harley 2009: 77) Kòfı́ lı-̀ yé fɔ́fɛ̀ Kofi rice NPS-steal ‘Kofi stole the rice.’ à ID Tuwuli contains several verbal prefixes for marking negation. The first is tá-, shown in example (95) below, which is used to negate simple declarative sentences: (95) Tuwuli (Harley 2009: 79) Kòfı́ tá-náà kùgbénı́ Kofi NEG-go hunting ‘Kofi didn’t go hunting.’ The second negative verbal prefix is l(V́ )-. All verbs with certain TAM and negative polarity require this prefix: (96) Tuwuli (adapted from Harley 2008; 2009) a. nɔ̀ óvólı̂ lɛ́-la̋-mlà Your letter NEG-be:MANNER-with ‘Your letter is not coming.’ b. ɛ́-la̋-mlà 3SG-NEG.be:MANNER-with ‘He isn’t reading (a book).’ c. ɔ́wɔ̂ l-a̋-vıɔ́ ̀ ̃ mɔ̀ river with NEG-PRES.IPFV-dry:up ‘A river doesn’t dry up with stones.’ d. Kofi lɛ́-l-a̋á-yá Kofi NEG-NEG.FUT-FUT-come ‘Kofi will not come.’ (óvólì) book kà-yá NOM-come kà-kã́ NOM-reading bóè stones 53 Lauren Schneider MA Thesis In (96d) above, the negative is marked twice. This is the only place in Tuwuli where this seems to occur. Harley (2009) glosses the construction as shown in (97a) below: (97) Tuwuli (Harley 2009:79) a. Kòfı́ kɔ́ lɛ́-l-a̋á-náà Kofi TOP NPS.NEG-NEG.FUT-FUT-go ‘As for Kofi, he won’t go.’ b. Kòfı́ lɛ̀-tá-náà NPS.FOC-NEG-go Kofi ‘Kofi didn’t go hunting.’ kùgbénı́ hunting In (97a) /lɛ-/ has a high tone and is glossed as the noun phrase subject and negative while in (b) it has a low tone and is glossed as the noun phrase subject and focus. It may be that the [l-] morpheme in (a) is the marker of negation and is accompanied by a high tone that affects the tone of the /lɛ-/ morpheme. This will not be treated as a case of double negation. The last negative prefix found in Tuwuli is a-. This prefix is shown in (98a) below alongside a construction with the same meaning using ta- in (b): (98) Tuwuli (Harley 2008: 318) a. b-ǎ-aka-kɛna adzuma 3PL-NEG.IPFV-SBJV.IPFV-do work ‘They should not be working.’ b. bɛ̀-ta-ka-kɛna adzuma 3PL.SBJV-NEG-IPFV-do work ‘They should not be working.’ The prefix /a-/ encodes both imperfective aspect and negation. As can be seen in (98a) above, when this prefix is used, imperfective aspect is marked by the negative morpheme and by another affix that also encodes the subjunctive. Alternatively, the same idea can be encoded using the original negative marker introduced, /ta-/. Example (98b) illustrates how the same meaning is conveyed by encoding the subjunctive on the affix that also encodes person and 54 Lauren Schneider MA Thesis number and imperfective is encode by another affix and is separate from the marker of negation. Additional data would be required to tease apart the nuances of these two examples and better understand the contexts in which one might be prioritized over the other. Tuwuli is interesting in that it is another language that stands out from the normal Kwa negation marking by failing to use preverbal nasal negation marking. Tuwuli’s final negation strategy is another that differentiates it from other Kwa languages. In Tuwuli there is an extrahigh tone level, which is strongly linked with the marking of negation. In the case of state verbs, this negative high tone is frequently the only indication of negation in the clause (Harley 2008: 294): (99) Tuwuli (Harley 2008: 323-4) a. fɔ́-nɛ́nɛ̀ 3SG.REF-be:good ‘It’s good.’ b. fɔ́-nɛ̋nɛ̀ 3SG.REF-NEG.be:good ‘It’s not good.’ The use of only tone to mark negation is a feature that is almost absent from the Kwa language group. The table below provides a summary of the negation features found in the Ka-Togo branch: 55 Lauren Schneider MA Thesis (100) Summary of negators in Ka-Togo GTM languages Ikposo naTuwuli taalV́ V̋ Nyangbo tVgɛTafi tÍ dI(gɪ-) Basic declarative clauses Negative imperfective Clauses with certain TAM and negative polarity Negation of certain state verbs Negative present Negative past Negative present Negative past In the next section, we will add Na-Togo languages to our analysis of GTM languages. 3.4.2 Na-Togo language data Sekpele (or Likpe) [lip] is spoken along the southern part of the border of Ghana and Togo. This language follows the SVO constituent ordering of the Kwa language group: (101) Sekpele (Delalorm 2016: 378) o-sani əmə a-mɪnɪ NCP-man NCP-the HAB-swallow ‘The man swallows the medicine’ ko-fa NCP-medicine On the surface, this language appears to have the simplest of the Na-Togo negation systems. The negative marker in Sekpele is represented by a homorganic nasal verbal prefix /N-/, aligning with the expected patterns of a Kwa language. This negation marker is a confirmed case of morphological negation because it can occur closer to the verb stem than other prefixes for TAM as well as occurring closer to the stem than person and number marking. For example, 56 Lauren Schneider MA Thesis (102) Sekpele (Delalorm 2016: 352) wé=m-fi kólèsá 3SG.NOM.PST=NEG-take anything ‘He didn't take anything’ In (102) above, 3rd person singular and past tense are encoded by a clitic, which occurs further from the verb stem than the marker of negation. This example also shows how the underlying /N-/ prefix is realized as [m-] through assimilation with the place feature of the stem. While in most contexts, a single negation marker is used, Sekpele differs from most other Kwa languages in that there are specific contexts in which negation is marked twice. The negative prefix occurs twice on the same verb when the perfective aspect and/or 2SG pronominal is present. In the example below, (103a) demonstrates single negation with the 1SG pronominal and the past tense while (b) contains both the perfective aspect and the 2SG pronoun: (103) Sekpele (Delalorm 2016: 354-5) a. mé=n-kè-sù àsɔ̀lé ə́-mə́ 1SG.PST=NEG-PROG-go church CP-the ‘I wasn't going to the church’ b. á=m-è-n-kè-sù 2SG.PST= NEG-PFV-NEG-PROG-go ‘You weren't going to the church’ àsɔ̀lé əmə́ church CP-the The negative prefix occurs once in (103a) and twice in (b). The following example demonstrates double marking of negation with the perfective aspect and no pronominal forms: (104) Sekpele (Delalorm 2016: 229) di-yìbíbí nə́-mə́ NCP-fruit CP-the ‘The fruit is unripe’ m-à-n-sɛ́ NEG-PFV-NEG-ripe As the double marking of negation occurs both times as a prefix, this is not a typical case of double negation. In most cases of bipartite negation, one negative marker occurs immediately 57 Lauren Schneider MA Thesis before the verb or verb-phrase being negated and the second marker occurs after the verb, either immediately, as in French, or phrase finally as in Ewe. Lelemi [lef] is another Na-Togo language and is spoken in the Volta Region of Ghana. Like Sekpele and the rest of Kwa, Lelemi exhibits SVO word order: (105) Lelemi (Fiedler & Schwarz 2009: 61) bé‐yè èjı̄bı.̀ 3PL.PFV-buy fruits ‘They bought fruits.’ Lelemi has a relatively complex verbal prefix system and has a system of negation that resembles that of Tuwuli. The tense and aspect, person and number, and negation prefixes sometimes exhibit portmanteau and sometimes are distinct affixes. In the perfective aspect, negation is marked by a distinct affix: (106) Lelemi (Fiedler & Schwarz 2009: 63) ń‐tá‐nū 1SG.PFV-NEG-see ‘I didn’t see her.’ ŋ̀. 3SG Unlike the perfective aspect, negation in the stative aspect is encoded by a portmanteau affix, as demonstrated by (107b) below: (107) Lelemi (Fiedler & Schwarz 2009: 62-3) a. b. n̄‐jı̄ ká sūè 1SG.STAT-know COMPL Sue ‘I know that Sue and Agnes…’ lɛ̀ɛ̀‐jı̄ kàbá 1SG.STAT.NEG-know family I do not know this family. kū CNJ ágnès … Agnes kāmɛ́nı̄. DEM Depending on the person and number marking, the negative stative marker can take a different form: 58 Lauren Schneider MA Thesis (108) Lelemi (Fiedler & Schwarz 2009: 62-3) ùlòkú ɔ̄ mɛ ù-bō ̃ ́ nı̄ woman DEM 3SG.STAT-have CNJ ‘… and this woman had three children.’ a. … nà b. óòwò, ɔ̀ɔ̀‐dı̄ ùbıd ú‐yè èjibı̀ ̀ ı̄ nà no, 3SG.STAT.NEG-be.qual child CNJ 3SG.PFV-buy fruits ‘It is not her child that she bought the fruits for.’ ú‐tè. 3SG.PFV-give Both of these negative stative markers are characterized by a low tone. The shape of these negation markers is determined by the person and number marking. Fielder & Schwarz (2009:66) generalize the marking as follows: (109) Person-number marking in the Stative aspect AFFIRMATIVE N 1SG U 3SG NEGATIVE lÌ + Ì 1SG + NEG Ù + Ù 3SG + NEG Negation in the stative aspect in Lelemi will be generalized to a lengthening of the vowel of the affix marking person and number. Like the stative aspect, the negative imperfective aspect has two realizations depending on the person and number. The first, shown in (110a) is a distinct prefix while the second, shown in (110b), is a portmanteau affix: (110) Lelemi (Fiedler & Schwarz 2009: 63) a. óòwò, bèlòkúbı̀ ı́nyɔ̀ bà‐là‐kā no, girls two 3PL.IPFV-NEG-read ‘No, the two girls are not reading a book.’ b. nàà‐kālı̀ ká ... 1SG.IPFV.NEG-think COMPL ‘I don’t think that …’ ɔ̀kū. book The following generalization is adapted from Fielder & Schwarz (2009: 66): 59 Lauren Schneider MA Thesis (111) Person-number marking in Imperfective aspect AFFIRMATIVE mÙ 1SG Ù 3SG NEGATIVE Ǹ + (l)A 1SG + NEG Ù + (l)A 3SG + NEG Negation in the imperfective will be summarized as the addition of the affix (l)a- with the application of additional phonological processes. The tone of this marker is opposite to the tone of the lexical verb (Fielder and Schwarz 2009: 66). Lastly, we will look at the negative future tense: (112) Lelemi (Fiedler & Schwarz 2009: 63-4) a. ù‐là‐dı̄. 3SG.FUT-NEG-eat ‘He will not eat.’ b. è‐lé‐bō kùdı́kùdı́ 2SG.FUT-NEG-come never ‘You will never come again.’ lɛ́nɔ̀. also And the future markers are generalized as follows: (113) Person-number marking in Future tense AFFIRMATIVE Ǹ 1SG Ù 3SG NEGATIVE Ǹ + (l)A 1SG + NEG Ù + (l)A 3SG + NEG The tone of this marker is identical to the tone of the first syllable of the verb. Based on the data and Fielder and Schwarz’s (2009) generalizations, the negative prefixes in Lelemi will be summarized as follows: 60 Lauren Schneider MA Thesis (114) Summary of negative prefixes in Lelemi PREFIX táVV(l)A(l)A- TENSE/ASPECT Perfective Stative Imperfective Future NOTES long vowel on person & number prefix polar tone same tone as first syllable of verb Lelemi differs from Akan and most Guang languages in the lack of nasal phonemes used for negation marking. Sɛlɛɛ (or Selee) [snw] is another Na-Togo language spoken in the Volta Region of Ghana. Like the others, it is SVO: (115) Sɛlɛɛ (Agbetsoamedo 2014: 103)44 ba-a-tɔ̀ɔ̀-pɛ si-si 3PL-DP-PROG-plant CP-yam ‘They were growing yams.’ Like many of the previous Kwa languages discussed so far, Selee employs negative prefixes. The first prefix found in the language is nA- which follows the pattern we would expect to find in a Kwa language. The nA- prefix is realized differently in certain contexts: (116) Sɛlɛɛ (adapted from Agbetsoamedo 2014: 104, 105) a. nkpo oso sé fa-le ɔ-lɛɛ o-mu Then so if 2SG-COP CP-native of Santrokofi CP-full ‘Then if you are a native of Santrokofi and you don't know…’ b. a-pípì ni-kpɛ alɛ di-fila CP-sweat 3SG.NEG-work that CP-be.hot ‘There is no sweat without heat or warmth.’ ee or fa-naa-ye… 2SG-NEG-know le-sansa CP-warm n-naa 3-not.exist In (116a) above, [naa] is used as a negative prefix on the verb while in (b) it performs the function of a negative existential verb. [ni] also appears in (b) as a third person singular negative verbal prefix. Based on this data the prefix naa- appears to have a more generic negative 44 Class prefix glossing in Agbetsoamedo (2014) is confusing without sufficient contextualization and so has been simplified in the examples used in this paper. 61 Lauren Schneider MA Thesis function as it was derived from the negative existential verb (as suggested by the gloss in (116b) above) while ni- encodes additional information about person and number. The negative prefix in the following example is similar: (117) Sɛlɛɛ (Agbetsoamedo 2014: 115) sɛ-fa CP-grass se-wo CP-some nín-sin-wɛ LSM.NEG-NEG-have bia all nin-yɛ LSM-stand n-tù CP-water ni PRT si-a-nyi CP-DP-drink kaa-sɔ CP.PP-ground ni then oso, so sɛ-fa CP-grass n-kpi LSM-die ‘even the grass that is on the ground does not get water, so they die.’ Selee contains what Agbetsoamedo (2014) analyzes as Lexical Subject Markers (LSM). In (117) above, [nı́n-] is analyzed as an LSM and as negative, [sin-] is analyzed as negative, and [nin-] and [n-] are analyzed as LSMs. The Selee data from Agbetsoamedo’s (2014) paper does not contain any additional examples of the nı́n-sin- negative prefix combination. The next negative prefix found in Selee loosely resembles the lV- prefix seen in Tuwuli and Lelemi: (118) Sɛlɛɛ (Agbetsoamedo 2014: 105) Sé If lɛ-nyɛ̀nɛ̀nɛ̀-ɛ CP-cold-INT le sé if fe-e-kutu 2SG-RP-uproot si-sí CP-yam o-kóso PRG-dry si-ba-kóso CP-FUT-dry ni-ɛ-wa REL-PRES-come CP ni then ni PRT si-lɔ́ɔ́ba-bùù CP-NEG.FUT-rot ni PRT si-lɔ́ɔ́ba-bùù CP-NEG.FUT-rot ‘Yams don’t get rotten when uprooted in cold, they only get dry.’ The negative verbal prefix [lɔ́ɔ́ba-] is made up of lɔɔ- and the future prefix ba-. Agbetsoamedo’s (2014) data contains no examples of lɔɔ- occurring on its own so more data would be required to determine whether or not this is a productive negative prefix or is only used in this context. The 62 Lauren Schneider MA Thesis compound prefix [lɔ́ɔ́ba-] encodes both negation and future tense like the (l)A- prefix in Lelemi (see 114). 3.4.3 Summary of GTM languages All of the GTM languages use verbal prefixes to mark negation. Ikposo, Sekpele, and Selee contain negative markers that resemble those in the rest of the Kwa language group. Additionally, Selee shares the lV- prefix with Tuwuli and Lelemi, which have relatively complex negation systems. Example (119) below summarizes the patterns found in these languages: (119) KA-TOGO NA-TOGO Ikposo na- Sekpele N- Nyangbo tVgɛ- Selee nAlV- Tafi tÍ dI(gɪ-) Lelemi taVV (l)A- Tuwuli taalV́ V̋ The Ghana-Togo Mountain languages tend to have more complex negation strategies than other Kwa languages. It is often not just a matter of adding a negative particle or prefix. The prefixes can be conditioned by the tense, aspect, or person and number. Unlike Akan, Ewe, and the majority of the Guang languages, the Ghana-Togo languages can but do not consistently use a preverbal nasal morpheme to mark negation. The /t/, /d/, /g/, and /l/-initial negators are features of these GTM languages that set them apart from the rest of the Kwa languages. These features are not found elsewhere in Kwa, so the question arises of where these prefixes came from. It is most likely that, through processes of 63 Lauren Schneider MA Thesis migration and language contact, they were acquired from another language group. Example (120) below contains examples of negative morphemes containing these phonemes: (120) Various (cited by Dryer 2009: 8-9) a. Izi (Igboid, Niger-Congo; Meier, Meier & Bendor-Samuel 1975: 217) nwó!ké té è-pfú-du man 3SG-speak-NEG NEG ‘the man is not speaking it.’ b. í!yá 3SG Tiv (Bantoid, Niger-Congo; Abraham 1940: 22) a kạha̱ ga ̣ surə hoe farm NC1 NEG ‘he did not hoe the farm.’ c. Bagirmi (Bongo-Bagirmi, Nilo-Saharan; Stevenson 1969: 92) deb-ge tol tobio person-PL kill lion ‘the people did not kill the lion.’ li NEG (a-c) above are intended not as proposals for the origin of the GTM negative markers but simply as examples of similar negative markers found in other language groups. In conclusion, it is safe to say that additional historical and comparative data would be required to determine the origin of the GTM negative markers. 4. Conclusions 4.1 Summary The following table summarizes the distribution of particles, prefixes, and, in a few cases, suffixes in Kwa languages: 64 Lauren Schneider MA Thesis (121) Particle-affix distribution in the Kwa languages surveyed SUBGROUP LANGUAGE FORM MORPHEME Akanic Akan Prefix N- Gbe Ewe Preverbal clitic & Post-verbal particle mé=…o Gen Preverbal particle & Post-verbal particle mú ̃… ò Fon Particle (post-verb) ǎ Gun Particle má Nkami Prefix Foodo Prefix mʊnmʊnɛmʊntɪmàmámaNmÈ -nÉ má- Krache Prefix màáḿpɛ̀aN- Gonja Particle Chumburung Particle máŋ máaŋ maa mʊŋ Nawuri Particle/Prefix maŋ mEE Efutu Prefix mV- Leteh Prefix bÉ - Cherepon Prefix bÉ - Gua Prefix bÉ Ga Dangme Suffix Suffix -VV -we Guang Ga-Dangme 65 Lauren Schneider MA Thesis SUBGROUP LANGUAGE FORM MORPHEME Ka-Togo Ikposo Prefix na- Nyangbo Prefix tVgɛ- Tafi Prefix tÍ dI(gɪ-) Tuwuli Prefix tAlV- Sekpele Prefix N- Selee Prefix nAlV- Lelemi Prefix taVV (l)A- Na-Togo All of the languages designated “particle” refer to preverbal particles unless otherwise stated. Languages in which the prefix-particle distinction is unclear have been designated “particle/prefix”. These are most likely cases in which a preverbal particle is becoming or has become a verbal prefix, but there are still remnants of its history as an independent particle in the orthography. Prefixes are generally preferred by this language group and preverbal negative particles have a tendency to move toward becoming prefixes as they assimilate to more of the phonetic features of the verb and take on more verbal features such as tense, aspect, person, and number. The Ga-Dangme group stands out from the rest of Kwa in its used of post-verb negation. There is a strong preference in Kwa for the marker of negation to precede the verb. The table below provides a summary of the word-order patterns found in Kwa: 66 Lauren Schneider MA Thesis (122) Neg-Verb ordering SUBGROUP ORDER Akanic NEG-V Gbe Ewe & Gen Fon Gun NEG V NEG V NEG NEG V Guang NEG V Ga-Dangme V-NEG GTM NEG-V The Gbe group is the only one that contains variation within the subgroup. All of the others exhibit consistent Neg-Verb or Verb-Neg ordering, regardless of whether the negative marker occurs before or after the verb. Fourteen of the Kwa languages have symmetric systems and ten have asymmetric or partially asymmetric systems. (123) Summary of negation symmetry SUBGROUP LANGUAGE (A)SYMMETRY Akanic Guang Akan Nkami Foodo Krache Gonja Chumburung Nawuri Efutu Leteh Cherepon Gua Symmetric; Asymmetric: A/Cat Asymmetric: A/Cat Asymmetric: A/Cat Asymmetric: A/Cat Symmetric Symmetric Symmetric Symmetric Symmetric Symmetric Symmetric; Asymmetric: A/Cat 67 Lauren Schneider MA Thesis SUBGROUP LANGUAGE (A)SYMMETRY Gbe Ewe Gen Fon Gun Symmetric Symmetric; Asymmetric: A/Cat Symmetric Symmetric Ga-Dangme Ga Dangme Symmetric Asymmetric: A/Cat Ka-Togo Tafi Nyangbo Ikposo Tuwuli Symmetric Symmetric Symmetric Asymmetric: A/Cat Na-Togo Sekpele Lelemi Selee Symmetric Asymmetric: A/Cat Asymmetric: A/Cat If a language is listed as having both symmetric and asymmetric negation, this indicates that the system is largely symmetric but exhibits asymmetry in a limited number of grammatical contexts. All cases of asymmetry in the Kwa language group fall under the subtype A/Cat. Nine of the cases of A/Cat asymmetry involve a change in marking of tense and aspect in the negative. Gen provides a single example of a change in the marking of person and number in the negative. 4.2 Jespersen’s cycle in Kwa The only language to exhibit anything resembling Jespersen’s cycle or negative concord is Ewe. (124) Ewe (Nurse n.d.: 8) a. atí lá mé kɔ́ DEF NEG tall tree ‘The tree is not tall.’ o NEG As was discussed earlier, in Ewe, clausal negation is expressed by the morpheme me, which is cliticized onto the first element of the verb phrase and o, which occurs at the end of the clause. 68 Lauren Schneider MA Thesis This system closely resembles the bipartite negation seen in stage 3 of Jespersen’s cycle, exemplified by Written Standard French: (125) Written Standard French je ne dis pas 1SG NEG say.1SG NEG ‘I do not say’ At this point in Ewe, as in French, neither of the negative particles have any meaning other than grammatical marking of negation. Adjei’s (2014) account of children’s speech provides evidence that unlike what has occurred in French, the preverbal negative marker is not in the process of eroding, at least as of when the 2014 data was collected. Based on the available data, the bipartite system of negation in Ewe is not yet in danger of disappearing. As was demonstrated by the French examples in Section 2.1.4, the bipartite system ne… pas persisted for centuries, aided by the writing system, and the verb-initial marker ne has only eroded from spoken language in the language’s more recent history (noted by Jespersen 1917). Potential evidence for the existence of Jespersen’s cycle can be seen when comparing Ewe with other Gbe languages. Both Fon and Ewe have a vowel sound as the negation marker after the verb while both Ewe and Gun have mV preverbal negation marking: (126) Gun Ewe/Gen Fon NEG VERB NEG VERB NEG VERB NEG STAGE I STAGE II STAGE III While Ewe’s preverbal negative marker is not currently showing any signs of eroding, it is possible that Fon did have one at some point. Without historical data, it is difficult to say for certain but this comparative data does seem to indicate that Gun may be in the first stage of Jespersen’s cycle, Ewe in the middle stage, and Fon in the final stage. 69 Lauren Schneider MA Thesis Other than in the Gbe branch of Kwa, there appears to be little to no evidence that Jespersen’s cycle is taking place in the Kwa language group. This is only one pattern of development for negation strategies and its attestation outside of Indo-European languages is limited and cannot be considered a universal cross-linguistic phenomenon. 4.3 Concluding remarks This survey of negation patterns in the Kwa language group intentionally provides only a surface level analysis of negation in each of the languages covered. Future work to be done could focus on each subgroup in turn. These studies would necessarily include prohibitives, negative verbs, and negative indefinites, to name a few important topics. As these are not topics that have been broached in most of these languages, topic-specific language data would need to be collected or elicited. For example, negative indefinites are important to study negative concord. In order to determine a language’s use of negative concord, many examples of negative indefinites would need to be collected. There is still much work to do in order to describe negation patterns in the world’s languages, especially outside of Indo-European. Markers of negation are often very old morphemes whose history is difficult to trace. Jespersen’s cycle provides a unique look into the development of negation from the IndoEuropean *ne to modern spoken French pas. By looking at records it is possible to trace the development of pas from a lexical to a grammatical word. Unlike many languages in the IndoEuropean language family, other language families, such as Niger-Congo, do not have the wealth of historical linguistic information preserved in centuries of texts. Without this historical and comparative information, it is difficult to trace the origin of many grammatical morphemes, including the negative marker, back to their lexical origins. For these language families with relatively younger writing systems, it will be important to document the changes in their 70 Lauren Schneider MA Thesis negation patterns as the languages change. By doing this, a new foundation can be laid for observing the development of negation patterns in a wider variety of languages. Abbreviations 1 2 3 ACC ACT ATR CNG COND CNJ CONS CONT COP (N)CP DEF DEM DIRV DP EMPH FAC FOC FUT GEN GTM HAB ID IMP INT IPFV INDF IRR First person Second person Third person Accusative Active Advanced tongue root Connegative Conditional Conjunction Consecutive Continuous Copula (Noun) Class prefix (& agreement marking) Definite Demonstrative Directional verb extension Distant past Emphatic Factative Focus Future Genitive Ghana-Togo mountain Habitual Identifiability marker Imperative Intensifier Imperfective Indefinite Irrealis ITR ITIV LOC LSM NEG NPS NPST NSF OBJ PFV PL POSS POT PP PRES PRF PRG PROS PRT REL REF RFLX RP SBJV SG STAT SUBJ SVC TAM TOP Iterative Itive Locative Lexical subject marker Negative Noun phrase subject focus Non-past Noun class suffix Object Perfective Plural Possessive Potential Tone marking V́ High tone V̀ Low tone v̂ Falling tone v̋ Extra-high tone Downstep V↓ Preposition Present Perfect Progressive Prospective Particle Relative Referential Reflexive pronoun Recent past Subjunctive Singular Stative Subject Serial verb construction Tense, aspect, modality Topic Capital letters within morphemes indicate that the realization of the sound is dependent on some assimilation process. 71 Lauren Schneider MA Thesis References Abakah, Emmanuel N. 2005. Tone rules in Akan. Journal of West African Languages 32(1-2). 109-134. Aboh, Enoch Oladé. 2009. Clause structure and verb series. Linguistic Inquiry 40(1). 1-33. Abunya, Levina Nyameye. 2010. Kaakyi verbal morphology. MA Thesis. University of Ghana. Legon, Ghana. Adjei, Francisca Adzo. 2014. Children’s use of nyá constructions in Ewe. Journal of West African Languages 41(1). 31-46. Agbetsoamedo, Yvonne. 2014. Noun classes in Sɛlɛɛ. Journal of West African Languages 41(1). 95-124. Aikhenvald, A. Y. 2006. Serial verb constructions in typological perspective. In Robert M. W. Dixon & A. Y. Aikhenvald (eds.), Serial verb constructions: A crosslinguistic typology. 1-61. Oxford: Oxford University Press. Akanlig-Pare, George & Asante, Rogers Krobea. 2016. Vowel harmony in Nkami. Journal of West African Languages 43(1). 21-44. Ameka, Felix K. & Dakubu, M.E. Kropp. 2008. Imperfective constructions: Progressive and prospective in Ewe and Dangme. In Felix K. Ameka & M.E. Kropp Dakubu (eds.). Aspect and Modality in Kwa languages. Amsterdam: John Benjamins Publishing Co. Amfo, Nana Aba Appiah. 2011. The Dangme clausal connective nɛ̄. Journal of West African Languages 38(1). 53-68. Animah, Diana Savala. 2015. Focus constructions in Ɔkere. University of Ghana Digital Collections. Ansah, Mercy Akrofi. 2015. Typology of Leteh (Larteh) restricted verbal clauses. Journal of West African Languages 42(1). 31-43. Bobuafor, Mercy. 2013. A grammar of Tafi. PhD dissertation. Leiden University. Campbell, Akua A. 2014. Motion expression in Ga narratives. Journal of West African Languages 41(2). 69-101. Campbell, Richard. 1988. Verbal inflection in Kwawu Akan. Studies in African Linguistics 19(2). 205-232. Casali, Roderic. 1995. An overview of the Nawuri verbal system. Journal of West African Languages 25(1). 63-86. Dakubu, M. E. Kropp. 2009. Kwa languages. In Keith Brown & Sarah Ogilvie (eds.), Concise encyclopedia of languages of the world. 630-633. Oxford: Elsevier Ltd. 72 Lauren Schneider MA Thesis Delalorm, Cephas. 2016. Documentation and description of Sɛkpɛlé: a Ghana-Togo mountain language of Ghana. PhD Thesis. SOAS, University of London. Dixon, R.M.W. 2012. Basic linguistic theory 3: Further grammatical topics. Oxford: Oxford University Press. Dolphyne, F.A. & Dakubu M.E. Kropp. 1988. The Volta-Comoé Languages. In M.E. Kropp Dakubu (ed.). The languages of Ghana. London: International African Institute. Dryer, Matthew S. 2009. Verb-object-negative order in central Africa. In Norbert Cyffer, Erwin Ebermann & Georg Ziegelmeyer (eds.). Negation patterns in West African languages and beyond. Amsterdam: John Benjamins Publishing Co. Essegbey, James. 2009. Noun classes in Tutrugbu. Journal of West African Languages 36(1-2). 37-56. Essegbey, James. 2012. Aspectual contrasts in Tutrugbu (Nyagbo). In Bruce Connell & Nicholas Rolle (eds.). Selected proceedings of the 41st Annual Conference on African linguistics, 40-49. Somerville, MA: Cascadilla Proceedings Project. Fiedler, Ines & Schwarz, Anne. 2009. Focal aspects in the Lelemi verb system. Journal of West African Languages 36(1-2). 57-73. Hammarström, Harald & Forkel, Robert & Haspelmath, Martin & Bank, Sebastian. 2016. Glottolog 2.7. Jena: Max Planck Institute for the Science of Human History. Hansford, Gillian F. 2010. Red is a verb: The grammar of color in Chumburung. Journal of West African Languages 37(2) 109-137. Hansford, Gillian F. 2011. I have a certain child… Riddles in Chumburung. Journal of West African Languages 38(2). 49-64. Harley, Matthew. 2008. Tense, Aspect and Mood in Tuwuli. In Felix K. Ameka and Mary Esther Kropp Dakubu (eds.). Aspect and Modality in Kwa Languages. Amsterdam: John Benjamins Publishing Co. Harley, Matthew. 2009. Focus constructions in Tuwuli. Journal of West African Languages 36(1-2). 75-90. Jespersen, Otto. 1917. Negation in English and other languages. Copenhagen: A. F. Høst. Kahrel, Peter & van den Berg, Ren. 1994. Typological studies in negation. Amsterdam: John Benjamins Publishing Co. Lefebvre, Claire. 1991.On the distribution of clausal wɛ̀ in Fongbe. Journal of West African Languages 21(2). 21-34. Miestamo, Matti. 2009. Negation. In Jef Verschueren & Jan-Ola Ostman & Frank Brisard (eds.). Grammar, meaning and pragmatics. 208-229. Amsterdam: John Benjamins Publishing Company. 73 Lauren Schneider MA Thesis Miestamo, Matti. 2013. Subtypes of Asymmetric Standard Negation. In Matthew S. Dryer & Martin Haspelmath (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Nurse, Derek. n.d. Ewe. In Derek Nurse & Sarah Rose & John Hewson (eds.) Verbal categories in Niger-Congo languages. Newfoundland: Memorial University. Plunkett, Gray C. 2009. An overview of Foodo, a linguistic island in Benin. Journal of West African Languages 36(1-2). 107-138. Snider, Keith L. 1990. Studies in Guang Phonology. Leiden: University of Leiden Ph.D. dissertation. Soubrier, Aude. 2009. Emergence d’une marque de causation en Ikposo? (Kwa, Togo). Journal of West African Languages 36(1-2). 188-201. van der Auwera, Johan & Van Alsenoy, Lauren. 2016. On the typology of negative concord. Studies in Language 40(3). 1-41. Willis, David, Anne Breitbarth & Christopher Lucas. 2013. Comparing diachronies of negation. In David Willis, Anne Breitbarth & Christopher Lucas (eds.), The history of negation in the languages of Europe and the Mediterranean, volume 1: Case studies, 1–50. Oxford: Oxford University Press. Zeijlstra. Hedde. 2007. Negation in natural language: On the form and meaning of negative elements. Language and Linguistics Compass 1. 498-518. 74 Lauren Schneider MA Thesis Appendix Overt markers of negation are usually easy to detect. Each marker’s form, placement, and scope should be investigated. In an investigation of negation in a given language, the following questions need to be answered (Dixon 2012:133-4):  If a negative particle is used in the language, does it apply to main clauses, subordinate clauses, clausal constituents, and/or internally within a noun phrase?  If morphological negation is used, what grammatical category of word does it apply to.  Any other idiosyncrasies found in negative constructions. Consultant Name: __Michael Obiri-Yeboah___ Date: __March 26, 2017 ___ Gua (South Guang) data: (1) kòfı́ Kofi hù see.HAB yâw Consultant notes: Yaw There is no distinction between habitual aspect and present tense in Gua. ‘Kofi sees Yaw.’ (2) kòfı́ bé-hù yâw Kofi NEG-see.HAB Yaw The negative marker is bɛ́- but it undergoes harmony to become bé- ‘Kofi does not see Yaw.’ (3) kòfı́ kofi hú see.PST yâw ɪ́ɲɪ́dɪ̀ Yaw yesterday ‘Kofi saw Yaw yesterday.’ (4) kòfı́ bé-ǹ-hú yâw ɪ́ɲɪ́dɪ̀ Kofi NEG-?-see.PST Yaw yesterday ‘Kofi did not see Yaw yesterday.’ (5) kòfı́ Kofi bè-hú FUT-see.FUT yâw Yaw Not sure what the nasal is but it could be separate so you may have an idea by considering other works on Kwa languages áʨɪ̃̀ tomorrow ‘Kofi will see Yaw tomorrow.’ 75 Lauren Schneider MA Thesis (6) yâw kòfı́ bè-é-hú Kofi FUT-NEG-see.FUT Yaw áʨɪ̃̀ Consultant notes: tomorrow ‘Kofi will not see Yaw tomorrow.’ (7) kòfı́ Kofi ɛ̀ɛ́-bwɛ̀ PROG-do.PROG ésım ́ ı̀ work/job ‘Kofi is working.’ (8) kòfı́ Kofi bɛ̀ɛ́-bwɛ̀ ésım ́ ı̀ NEG.PROG-do.PROG work/job bɛ́ + bɛ̀ɛ́ becomes bɛ̀ɛ́- ‘Kofi is not working.’ (9) kòfı́ Kofi bɛ̀ɛ́-bwɛ̀ ésı́mı̀ NEG.PROG-do.PROG work/job hɛ̀ again There is another way of saying it which could be described as perfective ‘Kofi is not working again.’ kòfı́ Kofi ɛ́ɛ̀-fá ɪ̃ ̀ ̃ PERFstop.PERF nɪ́ ésım ́ ı̀ work/job FOC ɔ̀-bwɛ́ ?-do.PST I’m not sure of what the ɔ- is. ‘Kofi has stopped working.’ (10) kòfı́ kofi ɛ̀ɛ́-ʥà ɪ̃ ́ ̃ PROG-search dɪ́ á-bwɛ̀ ésı́mı̀ COMP 3SG-do.PROG work/job ‘Kofi intends to work.’ (11) kòfı́ kofi bɛ̀ɛ́-ʥà ɪ̃ ́ ̃ NEG.PROG-search dɪ́ á-bwɛ̀ ésı́mı̀ COMP 3SG-do.PROG work/job ‘Kofi does not intend to work.’ (12) kòfı́ éè-hú yâw Kofi PERF-see.PERF Yaw ‘Kofi had seen Yaw.’ 76 Lauren Schneider MA Thesis (13) kòfı́ béè-ǹ-hú yâw Kofi NEG.PERF-?-see.PERF Yaw Consultant notes: bɛ́- + ɛ́ɛ̀- + nbecomes bɛ́ɛ̀n- and ends up becoming béèn- when they harmonize. I still don’t know what the nasal represents ‘Kofi had not seen Yaw.’ (14) kòfı́ Kofi bɛ̀tɪ́ never yâw hú see.PST Yaw ‘Kofi has never seen Yaw.’ (15) sɛ̀ kòfı́ é-ǹ-hú yâw à, á-bɛ̀ɛ́-bwɛ́ ésım ́ ı̀ if Kofi ?-?see.PST Yaw DET 3SG-NEG.PROGdo.FUT work Here too I can’t tell the appropriate gloss for both éand n- ‘If Kofi1 sees Yaw2, he1 will not work.’ (16) kòfı́ hù yâw ǹsó á-bɛ̀ɛ́-bwɛ́ ésı́mı̀ Kofi see.HAB Yaw but 3SG-NEG.PROGdo.PST work/job ‘Kofi1 sees Yaw2 but he1 is not working.’ (17) kòfı́ Kofi hù see.HAB yâw Yaw ásɛ̀ bɛ̀ɛ́-bwɛ́ REL FUT.NEG.PROGdo.FUT ésı́mı̀ work/job ‘Kofi1 sees Yaw2, who2 will not work.’ (18) kòfı́ Kofi bè-é-hú FUT-NEG-see.FUT yâw Yaw ‘Kofi will not see Yaw.’ 77 Lauren Schneider MA Thesis (19) Kòfı́ bé-hù yâw ǹdɪ́ Kofi NEG-see.HAB Yaw today Consultant notes: ‘Kofi has not seen Yaw today.’ (20) kòfı́ Kofi k͡pà tall/long ‘Kofi is tall.’ (21) kòfı́ Kofi bɛ́-ŋ͡m̀-k͡pà NEG-?-tall same n- as before /n-/ is becoming [ŋ͡m]/ k͡p ‘Kofi is not tall.’ (22) ɜ́jı̀ à bɛ́-ŋ͡m̀-k͡pà tree DET NEG-?-long/tall The nasals without specific glosses seem to occur before verbs marking past tense and perfective aspect. It can also occur before adjectives so you may need to figure them out. ‘The tree is not tall.’ (23) bwɛ̀ ésı́mı̀ do.IMP work ‘Work!’ (24) bɛ́-bwɛ̀ ésım ́ ı̀ NEG-do.IMP work ‘Don’t work!’ (25) kòfı́ bé-ǹ-hú yâw á Kofi NEG-?-see.PST Yaw Q ‘Didn’t Kofi see Yaw?’ 78