129Issah Reflexive Pronouns in Dagbani Samuel Alhassan Issah1 Abstract This paper examines reflexivization in Dagbani, a Gur language spoken in Northern Ghana. The data are examined within the parameter of co-referentiality and antecedence. I argue that reflexives are formed by suffixing the reflexivizer, maŋa, meaning ‘self’, to the possessive pronominal. The pronominal varies in form depending on the singularity or plurality of the antecedent. I minimally compare reflexive pronouns with pronouns within the Binding Theory and observe that whilst pronouns are free within their domains, reflexives are governed within the clausal domain. I conclude that anaphors and pronouns of Dagbani match the principles of the Government and Binding theory. 1: Introduction The main focus of this paper is to examine the morpho-syntactic features of Dagbani reflexive pronouns. Genetically, Dagbani belongs to the South Western Oti-Volta central Gur languages spoken in Northern Ghana (Bendor-Samuel, 1989; Naden, 1988). Though Dagbani has a continuum of geographical/regional dialects, three major dialects stand out: Tomosili (the Western dialect) which is spoken in Tamale, the Northern Regional capital and its environs, and Nayahali (the eastern dialect), spoken in and around Yendi, the seat of the political head of Dagboŋ, that is the land that is occupied by the Dagbamba, and Nanuni spoken around Bimbilla, that is the Nanumba districts and south of Yendi. Hudu (2010) and Abdul- Rahman (2005) both argue that Nanuni be seen as a dialect of Dagbani rather than a distinct language. Dialectal differences between these two major dialects are mainly at the phonological and lexical levels. The canonical word order of Dagbani is SVO. The data used in this paper is based on the Tomosili dialect of the Dagbani language. Though Dagbani is a tonal language, for purposes of this paper 1 Samuel Alhassan Issah teaches at the University of Education, Winneba. 130 Legon Journal of the HUMANITIES Volume 22 (2011) I do not mark tone because it does not make a difference to the problem addressed. Reflexives are typologically grouped into two: those that are referred to as long distance anaphors, as with the Norwegian ‘seg’, and the Dutch ‘zich’. There are also those reflexives which are standardly viewed as local such as the English ‘himself. The “local” anaphors are anaphors that have strictly local antecedents, in that they receive their reference from strictly local syntactic arguments while “long distance” anaphors are anaphors which have non-local antecedents. They are said to have non- local antecedents because they can take an antecedent outside the local domain. It is argued by Faltz (1977) and Pica (1985, 1987), that when reflexives are complex expressions, then, they are universally assumed to be local, whereas the simplex reflexives as with the Norwegian ‘seg’, are generally the long-distance type of reflexives. Different languages employ different strategies to form reflexive pronouns. For example, in a language like English, reflexives are a combination of possessive pronouns and the reflexivizer “self” as in “myself” “herself” or “ourselves”, or a combination of the accusative pronoun and the reflexivizer as in “himself ”. I shall show that Dagbani has bimorphemic reflexives. Based on the observation that the accusative form of the pronominal does not participate in the formation of reflexive pronouns, I make a tentative proposal that accusative forms of reflexive pronouns might be simply absent in Dagbani. Dagbani thus shares the same features with a language like Buli where the reflexive pronouns are also made up of a suffix reflexivizer and the possessive pronoun as in wa- dek, “myself” and ba-dek “themselves” (Agbedor 2002). Dagbani also differs from other languages such as Norwegian and Dutch which have simple reflexives in that they are made up of monomorphemic items. For instance Norwegian has ‘seg’ as its reflexive pronoun whilst Dutch has ‘zich’ 2: The Binding Theory. Binding Theory deals with the distribution of pronominal and reflexive pronouns in languages. The canonical Binding Theory of Chomsky (1981, 131Issah 1986) groups nominal expressions into three basic categories: (i) anaphors (reflexives), (ii) pronominals, and (iii) R-expressions. Anaphors (also called reflexive pronouns) are characterized as expressions that have no inherent capacity for reference. Anaphors also refer to reciprocals. According to Haegeman (1994:228), the three principles that govern the interpretation of the established nominal expressions are referred to as the binding theory. Hence, anaphors must invariably depend on some other expression within a sentence for their interpretation. The expression on which the anaphor depends for its meaning is called the antecedent. The structural relation between a reflexive and its antecedent is accounted for using c-command. Haegeman (1994:212) claims that a node A c-commands a node B if (1) A does not dominate B; (2) B does not dominate A; and (3) the last branching node dominating A also dominates B. On the other hand, he claims that the pronominal is an abstract feature representation of the NP that may be referentially dependent but must always be free within a given syntactic domain. It could be deduced from these definitions that an anaphor (reflexive pronoun) must obligatorily have a local or a “nearby” antecedent within a given syntactic unit to which it will refer, whilst a pronominal may, but need not necessarily have its antecedent within the same syntactic domain. Adger (2004:54), on what he calls the co-referentiality hypothesis argues that for “two expressions to be co-referential, they must bear the same phi-features”. According to Adger (2004), “phi-features” is a linguistic term used to describe the semantic features of person, number and gender encoded in such lexical categories as nouns and pronouns. This, he further argues, is a “kind of general interface rule that relates syntactic features to semantic interpretation”. Compare the English sentences in (1) and (2): 1. Mandeeyaj admires himselfj 2. Suhuyinii destroyed himj These examples illustrate the (syntactic) distributional difference between an anaphor (a reflexive) and a pronominal. In sentence (1) for instance, himself can only refer to its antecedent, Mandeeya, which is found in the same local domain of the clause. In sentence (2) however, the pronominal 132 Legon Journal of the HUMANITIES Volume 22 (2011) him is free within the clausal domain as it cannot refer to Suhuyini. It could therefore only have some element that is not within the clause as its antecedent, and not Suhuyini since pronouns are free within the clausal domain in which they are found. The fact that himself can only refer to an entity already mentioned in the discourse, and him can refer to an entity outside the clausal domain, means that whilst reflexives are referentially dependent, pronouns are not referentially dependent. The abstract features of reflexives and pronominals make four major distinctions of NP, three of which are overt and the other non-overt. The three NP types, which include anaphors, pronouns, and R-expressions, are not syntactic primitives since they can further be broken down into small components as shown below. Lexical reflexives [+reflexives, -pronominal]: these are reflexives and reciprocals, e.g., himself, each other. Pronouns [-anaphor, + pronominal]: these are basically pronouns. e.g., he, she Name (full NP) [-anaphor, -pronominal]: names e.g., Samuel, Bonyeli, Jeremy. PRO [+anaphor, -pronominal] These three overt NP types are accounted for using principles called Binding Principles. Principle A of these principles is concerned with reflexives and reciprocals, Principle B deals with pronominals. Principle C on the other hand concerns itself with names or what have been called full NPs. In Haegeman (1994:228-229), the binding principles which govern the syntactic distribution of overt NP types are stated as follows: Principle A An anaphor must be bound in its governing category. Principle B A pronoun must be free in its governing category. 133Issah Principle C An R-expression must be free everywhere. These three principles govern the distributional properties of pronominals and reflexive pronouns in languages. 3: The morpho-syntax of Dagbani reflexive pronouns Once it is observed that Dagbani reflexive pronouns are a combination of pronouns and the –maŋa morpheme, there may be justification for one to assume that in the language, the pronoun prefixes of reflexives are lexically pre-specified as possessive. This explains why it is not possible to form or generate reflexives using the accusative forms of the pronominals. Considering the fact that the nominative and possessive pronominal forms are morphologically the same (as will be seen in table 1), it may seem rather tasking to tell which form of the pronominal is suffixed to the reflexivizer. My proposal that the attached pronominal is the possessive and not the nominative form is based on the fact that cross-linguistically, nominative forms of pronominals do not seem to be specified for reflexive pronouns (cf. English: *Iself, *Heself, *youself, *sheself etc). I argue that Dagbani contrasts with English in this light since it is not possible to form reflexives in Dagbani via the suffixation of the accusative pronoun to the reflexivizer forms as in English- “himself”, which is made up of an accusative pronominal “him” and the reflexivizer, ‘self’. An insight into the pronominal system of Dagbani will help give a better insight into the reflexive pronominal system. Table 1: Dagbani pronominal system Nominative case Accusative case Genitive case Person singular plural singular plural Singular plural 1st m,n,ŋ Ti Ma ti m,n,ŋ ti 2nd a Yi A ya A yi 3rd o bɛ O ba O bɛ Critical readers will observe from table 1 that, the first person plural and the third person singular pronominals do not make a morphological 134 Legon Journal of the HUMANITIES Volume 22 (2011) distinction between the nominative and the accusative plural forms. The third person plural pronominal and the second person singular and plural pronominal, however, do make a morphological distinction between the nominative and accusative plural forms. As already mentioned, there is a tight relationship between the pronominal and the reflexive system of Dagbani. This is because the reflexives of Dagbani are complex (bimorphemic) expressions which are formed via the suffixation of a reflexivizer -maŋa meaning ‘self’ to the possessive or genitive form of the pronominal as shown in table 2. We also observe in this table that attempts to suffix the – maŋa morpheme to the accusative form of the pronominal yields an ungrammatical form. Table 2: Reflexive pronouns in Dagbani Possessive pronouns suffix reflexiziver reflexive pronoun n/m/ŋ- “my” -maŋa m -maŋa “myself” *ma-maŋa o-“his/her” - maŋa o- maŋa “himself/herself” a-“your” (singular) - maŋa a- maŋa “yourself” ti- “our” - maŋa ti- maŋa “ourselves” b-“their” -maŋa b-maŋa “themselves” *ba-maŋa yi-“your” (plural) - maŋa yi-maŋa “yourselves” *ya-maŋa I argue that Dagbani reflexives manifest the cross-linguistic syntactic distribution associated with reflexive pronouns as they do not occur as semantic subjects of sentences. Based on this syntactic property, when they occur as the semantic subjects of sentences, the resulting structure will be ungrammatical. In the data that follow, I illustrate the distributional properties of the reflexive pronouns in Dagbani. 135Issah 3. *m-maŋa ku-ri bi-hi maa pam 1stsg-self kill-imperf child-pl def inten 4. Abuj sa ku-Ø o-maŋaj Abu tdp kill-perf 3sg-self “Abu killed himself yesterday” 5. *Abuj sa tu-Ø Abuj Abu tdp insult-perf Abu 6. *Bɛ-maŋa tu-ri bɛ-maŋa 3pl-self insult-imperf 3pl-self From examples (3) to (6), we observe that (3), (5) and (6) are ungrammatical. The ungrammaticality of (3) and (6) is based on the fact that reflexive pronouns have been used as semantic subjects of those two sentences. However, reflexives are known, cross linguistically, to pick up their reference from an old syntactic element as their antecedents. They however, defy this syntactic expectation, hence their ungrammaticality. Note also that the ungrammaticality of (5) is based on the fact that Abu has Abu as its antecedent as seen in the co-indexization, which is not allowed by the binding principles. However, the data in (4) is grammatical because o-maŋa has Abu as its antecedent with which it agrees in number and person. Consider (7) and (8) below for further explication on the distribution of Dagbani reflexives: 7. Bɛj sa tu-Ø bɛ-maŋaj sɔhala 3pl tdp insult-perf 3pl-self yesterday “They insulted themselves yesterday” 8. Bɛ- maŋa ka b sa tu-Ø sɔhala 3pl-self foc 3pl tdp insult-perf yesterday “It was they themselves (that) they insulted yesterday”. Example (7) is also grammatical since the anaphoric expression bɛ-maŋa “themselves” has bɛ “they” as its antecedent and the two share phi-features 136 Legon Journal of the HUMANITIES Volume 22 (2011) in terms of number. The grammaticality of sentence (8) indicates that, the prohibition on the syntactic occurrence of reflexive pronouns is not on their being in subject positions, but on their being semantic subjects of the sentences as in (6) and (3). For instance, in (8), although bɛ-maŋa is in the subject position, it is not a subject; rather, it is the object that has been moved from the in-situ position to the sentence initial position or left periphery position. Perhaps, the reflexive pronoun being in that syntactic slot is just to fulfil the Extended Projection Principle (EPP) which has it that every sentence must have a subject. By this observation, I make the hypothesis that Dagbani reflexives can occur in subject position, once they are syntactic, but not semantic subjects. If this were not possible in the language, then example (8) above should have been ungrammatical. I further submit based on evidence from the data that, it is possible to have non-subject oriented reflexive pronouns in Dagbani. When a reflexive pronoun is a non-subject oriented reflexive, such a pronoun will have another syntactic element as its reference rather than the subject of the sentence. This observation seems to tally with the generalization that one usually finds in the generative literature claiming that monomorphemic reflexive pronouns are subject oriented (and can be long-distance bound) whereas bimorphemic ones are strictly clause-bound and not subject oriented (Reinhart & Reuland 1993). Based on the current claim that Dagbani has bi-morphemic reflexives, it is no surprise that one finds non- subject oriented reflexive pronouns in Dagbani. The same is observed in Ewe and Buli (Agbedor 2002). The Dagbani example in (9) below explicates the claim about the non-subject oriented nature of reflexive pronouns in Dagbani. 9. Ni yɛli-Ø Bonyelij o-maŋaj yɛla 1stsg talk-perf Bonyeli 3sg-self matters “I have spoken to Bonyeli about himself/herself ”. We see in (9) that though the subject of the sentence is N, meaning “I”, it is not the antecedent of the reflexive pronoun, o-maŋa [“himself/hersef”]. It does have a no-subject reference which is Bonyeli. This situation also pertains in English. When the reflexive pronoun in English has a non- subject reference or antecedent, there is usually an oblique subject in 137Issah cases where a reflexive could be assumed to be non-subject oriented, as in (10) below: 10. Wej spoke to Theresak about herself k. Though we is the subject of the sentence, the reflexive “herself” definitely does not refer to we, but to Theresa, which is not the subject of the sentence. 4. Dagbani pronouns, reflexive pronouns and Binding principles In this sub-section, I investigate how the binding principles A and B capture the distribution of pronouns and reflexives in Dagbani. It will be recalled that principle A deals with the distribution of reflexive pronouns whilst principle B deals with the distribution of pronouns. 11. Bonayoi tu-ri o-maŋa Bonayo insult- imperf 3sg-self “Bonayo is insulting/insults himself ” 12. Bonayoi tu-ri oj Bonayo insult-imperf 3sg “Bonayo is insulting/insults him/her” 13. Bi-hij maa tu-Ø b-maŋaj Child-pl def insult-perf 3pl-self “The children have insulted themselves” 14. Paɣabai maa sa ku-Ø baj/*i sɔhala Woman-pl def tdp kill-perf 3pl adjun “The women killed them yesterday” In sentence (11), it is clear that the sentence is grammatical as the reflexive has its antecedent within the same clause (which, as observed above, is the subject of the clause). In (12) and (14) however, it is seen that the pronominal o meaning ‘him/her’ and ba meaning ‘them’ can only refer to entities that are not within the same clause. It is seen from the co-indexization in (14) that if ba, meaning ‘them’, should have paɣaba maa, meaning ‘the women’, as its antecedent, the resulting structure will 138 Legon Journal of the HUMANITIES Volume 22 (2011) be ungrammatical. I assume that the entity that the pronominal refers to within the syntactic system could be an entity that has already been mentioned in the discourse, but does not necessarily have to be in the same local or clausal domain. This is what calls for the cross-linguistic generalisation that pronouns are free within their syntactic domain. The data given so far seems to indicate that reflexive pronouns and pronouns in Dagbani perfectly match the behavior predicted by principles A and B of the Government and Binding theory. All the data I have dealt with are simple sentences and thus may not provide enough basis for making a generalization on the behavior of reflexive pronouns and pronouns. I will therefore look at complex sentences as well to help ascertain the behavior of reflexives and pronouns in relation to binding principles: 15. [Bɛnitichei mi [ni Fatij tu-ri oi/k] Bɛnitiche know that Fati insult-imperf 3sg “Bɛnitiche knows that Fati insults/is insulting him/her” 16. [Bɛnitichei yɛli-ya [ni oi/j bi bɔ-ri Fati] Bɛnitiche say-perf that s/he neg want-imperf Fati “Bɛnitiche has said that s/he does not want Fati” 17. Babai yɛli-ya [ni Mikashinij tu-Ø oi/k/*j] Baba say-perf that Mikashini insult-perf 3sg “Baba said that Mikashini has insulted him/her”. In sentence (15), it is observed that the pronominal item, o is free in its minimal domain. Accordingly, it is observed from the co-indexization that it is possible for the pronominal to refer to the subject of the independent clause, which is Bɛnitiche, or to an item labeled k which is an item not mentioned within the clausal structure or domain. The same explanation holds for sentence (16) where the pronominal o, meaning “s/he”, could refer to the subject of the independent clause, Bɛnitiche, or to any item that is outside the clausal structure. These sentences are nevertheless grammatical as they do not defy the binding principle B, which asserts that a pronominal and its potential antecedent may be found within the same clausal structure and that it is also possible for the 139Issah pronoun to refer to an entity outside the sentence. The example in (17) also suggests that the object pronoun o is free within its domain. We observe from the co-indexation that the pronoun o could have the subject of the independent clause, Baba, as its antecedent. It could however also have as its antecedent an item not mentioned within the sentence which is labeled as k. However, it will be ungrammatical for the pronoun o to have the subject of the embedded clause, Mikashini as its antecedent. This observation leads me to tentatively postulate that the pronominal in Dagbani o of the complex sentence is ambiguous since it can refer to the subject of the independent clause within which it is found or to an unmentioned entity, that is, an entity that is found within the clausal structure. Discussing the distributional properties of the pronominal in Ewe, Agbedor (2002:151) argues that Ewe is able to deal with this kind of ambiguity in the pronominal by using a special type of pronoun called a logophoric pronoun. Clements (1975:142) is of the view that the logophoric pronoun is a special pronoun form that is used to indicate reference to the person whose speech, thoughts and perceptions are reported. This pronoun thus invariably has the subject of the main clause as its reference. Consider these data from Agbedor (2002:150) which clearly show the distributional pattern of the logophoric pronoun and the pronoun in Ewe. 18. John nya be Ama lɔ ye John know that Ama loves LOG “Johni knows that Ama loves himi”. 19. John nya be ye- lɔ Ama John know that LOG-love Ama “Johni knows that hei loves Ama”. 20. John nya be Ama lɔ -e John know that Ama loves-3sg “Johni knows that Ama loves himj”. 21. John nya be e-lɔ Ama John know that 3sg-love Ama “John knows that he loves Ama”. 140 Legon Journal of the HUMANITIES Volume 22 (2011) In the Ewe data taken from Agbedor (2002:150), we see that there is no ambiguity in the sentences unlike their English or Dagbani counterparts which will invariably be ambiguous. In English and Dagbani, it is possible for him to refer to either the subject of the sentence or to an entity which is not within the clausal structure (an oblique object). In Ewe, however, such ambiguity is avoided as Agbedor (2002) argues using the logophoric pronoun ye which he claims could only have one reference, and that is the subject of the main clause. He further postulates that unlike the logophoric pronoun, the Ewe pronoun can only pick up its reference from outside the entire sentence. He posits, based on this observation that the Ewe pronoun and logophoric pronouns are in complementary distribution. Data used in this paper suggest that Dagbani does not have a technique of dealing with this kind of ambiguity. I therefore conclude that Dagbani, like most other Ghanaian languages, does not have a logophoric pronoun. Cross-linguistic studies of reflexives seem to suggest that a reflexive picks up its reference from a local subject NP. According to Haegeman (1994: 192) “the NP on which a reflexive is dependent for its interpretation is the antecedent of the reflexive”. Haegeman (ibid: 207) further argues that we use co-indexation to indicate that a reflexive and its antecedent have the same referent and that the reflexive and its antecedent must agree with respect to the nominal features of person, number and gender. These prescriptions on reflexives give rise to issues of agreement and locality constraints on reflexives. However, the typological generalization that one finds in generative literature, including the work of Cole and Hermon (1998), Yang (1983) and Vikner (1985), is that monomorphemic reflexives are subject-oriented and can be long-distance bound, whereas bimorphemic ones are strictly clause-bound and are not subject-oriented. I thus argue based on evidence from the data presented in this paper, that Dagbani reflexives can be used as genuine evidence in favour of this typological claim. The data in (22-25) will further indicate that Dagbani pronouns may also function as antecedents of reflexives. 22. [*Bɛi tɛhi-ya [ni Maryj je bɛ-maŋai] 3pl think-perf that Mary like-not 3pl-self 141Issah 23. [Oi yɛli-ya [ni Abuj ku-Ø o-maŋaj] 3sg say-perf that Abu kill-perf 3sg-self. “S/he has said that Abu has killed himself”. 24. [*Oi tɛhi-ya [ni Abuj ku-Ø o-maŋai] 3sg think-perf that Abu kill 3sg-self 25. [Oi yɛli-ya [ni bi-hij maa tu-Ø b -maŋaj] 3sg say-perf that child-plu def insult-perf 3pl-self “S/he has said that the children have insulted themselves”. We observe from these data that sentence (22) is ungrammatical. The ungrammaticality of this sentence is caused by the violation of the locality constraint imposed on reflexives. If the antecedent and its referent were found in the same local domain, then the constraint would not have been violated, thereby resulting in ungrammaticality. We observe however that sentence (23) is grammatical. The grammaticality of this sentence is borne out of the fact that the locality constraint is not defied. This is because the reflexive o-maŋa “himself” has the subject of the embedded clause Abu as its antecedent. These two arguments, Abu and o-maŋa are however found in the same local domain. When the reflexive and its antecedent are in the same local domain, the reflexive is said to be clause bound. The sentence in (24) on the other hand is ungrammatical. The ungrammaticality of this sentence is caused by the fact that the reflexive pronoun o-maŋa “himself or herself and its antecedent o “s/he” are too far away from each other in the sentence structure; whilst o “s/he” is found within the independent clause, o-maŋa “himself or herself” is found within the dependent clause. A tentative conclusion could then be drawn, based on these data provided, that the locality constraint imposed on bimorphemic reflexives is valid for Dagbani reflexives since defying that condition results in the formation/ generation of ungrammatical sentences. It is then shown that Dagbani is like English in that the reflexives are clause bound. Thus, the syntactic behaviour of Dagbani reflexives patterns 142 Legon Journal of the HUMANITIES Volume 22 (2011) with the cross linguistic generalization that languages with complex or bimorphemic reflexives are strictly clause-bound. In the data that follow, adapted from Cole and Hermon (1998:57), we see the contrast between the syntactic properties of mononmorphemic reflexives and bimorphemic reflexives as typologically argued. 26. Zhangsani renwei Lisij zhudao Wangwu xihuan zijii/j/k Zhangsan thinks Lisi knows Wangwu likes self ‘Zhangsan thinks Lisi knows Wangwu likes him/himself’ We see from the data in (26) that Ziji could have a reference that is non- local, that is the possibility that it can have Zhangsan or even Lisi as its antecedent. The same property of non-local reference is not available for Dagbani reflexives as it has so far been observed from the data examined. This seems to suggest that the reflexive pronouns of Mandarian Chinese should be non-local. The same property is reported of the reflexive pronouns of Russian which are also argued to be non-local as argued by Rappaport (1986). The same non-local syntactic property has been associated with monomorphemic reflexives in languages like Danish. The data is adapted from Wayne (1996:193). 27. At Peteri bad Anne om [PROk at ringe til sigi] That Peteri asked Annk (for) at ring to selfi Vikner (1985) From the data in (26) and (27), we see that in Mandarian Chinese and Danish, it is possible for the antecedent of the reflexives to be non-local. We see from these examples that the reflexive sig can have Peter as its antecedent, though the two are syntactically apart from each other. One common feature that runs across the data from Danish and Mandarian Chinese is the fact that the reflexives in both languages can have non-local antecedents. Both languages however have monomorphemic reflexives, as seen in their morphological composition. These data adapted from Cole and Hermon (1998:57) and Vikner (1985) cited in Wayne (1996:193), also provide a basis for a genuine argument in favour of the claim that Dagbani reflexives, which are morphologically complex (bimorphemic) reflexives, differ in their locality property. 143Issah Also, it is a cross linguistic expectation that reflexives agree with their antecedents in number. Number is a feature of nouns that distinguish between singular and plural nouns. Since reflexives are not independent items in languages, Haegeman (1994:207) states that “the reflexive and its antecedent must agree with respect to their nominal features of person, gender, and number” in languages that have number agreement. A close look at the distribution of Dagbani reflexives suggests that the assumption on number agreement between an antecedent and its reflexive is valid for Dagbani, as in (28), (29), (30) and (31). 28. *Zaapayim tu-ri bɛ-maŋa Zaapayim insult-IMPERF 3pl-self 29. *A zu-Ø yi-maŋa 2sg steal-PERF 2pl-self 30. Yi zu-Ø yi-maŋa 2pl steal-PERF 2pl -self “You have stolen yourselves”. 31. Abu ŋme-Ø o-maŋa Abu knock 3sg-self “Abu has knocked himself”. In (28), the ungrammaticality is borne out of the fact that there is a mismatch between the subject NP Zaapayim and the reflexive bɛ-maŋa “themselves” in terms of number. The reflexive bɛ-maŋa cannot have Zaapayim as its antecedent since the two do not agree in number. Zaapayim is a singular noun functioning as a subject of that sentence whilst bɛ-maŋa is a plural reflexive pronoun with its English equivalent as themselves. In example (29) too, we see that the second person singular pronoun a, “you” does not agree in number with yi-maŋa “yourselves” which is plural. The lack of number agreement between these two accounts for the ungrammaticality of example (29). Example (30) however is grammatical since the subject of the sentence, the second person plural pronominal yi [“they”] agrees in number with the reflexive yi-maŋa. In (31) too, we observe that Abu is the antecedent whilst o-maŋa is the reflexive. These two agree in terms 144 Legon Journal of the HUMANITIES Volume 22 (2011) of number since the antecedent of the reflexive Abu is singular and the reflexive pronoun o-maŋa [“himself” or “herself”] is also singular. The agreement in number between the antecedent and the reflexive results in the grammaticality of the sentence. 5. Summary and conclusion This paper has investigated the morpho-syntactic properties of Dagbani reflexive pronouns using the theoretical framework of co-referentiality and antecedence as an analytical tool. Morphologically, I argued that Dagbani has complex (bimorphemic) reflexives. I also investigated the syntactic distribution of Dagbani reflexive pronouns. It was discovered that Dagbani reflexives as cross-linguistically assumed, cannot work as semantic subjects. They could however work as syntactic subjects, particularly in ex-situ focus type constructions. Dagbani reflexive pronouns were also found to be bound within their local domain since they cannot have an antecedent outside their clausal domain. I minimally compared the distributional properties of reflexives with those of pronominals and concluded that the two differ syntactically: pronouns are free within their syntactic domain whilst reflexives are clause-bound. I therefore came to the conclusion that essentially, reflexive pronouns and pronominals in Dagbani match the behavior predicted by principles A and B of the GB binding theory. Based on evidence from the data analyzed, I conclude that Dagbani patterns with the typological assumption in the generative literature that monomorphemic reflexives are subject oriented (and can be long-distance bound) whereas bimorphemic ones are strictly clause-bound and not subject oriented. _________________________ iI am very much obliged to the anonymous referee for his or her comments and suggestions which have been duly integrated into the final version of this paper. iiAbbreviations used in this paper include: tdp= time depth particles, perf=perfective, imperf=imperfective, plu=plural, def=definite, sg=singular, inten=intensifier 1st, 2nd, 3rd for first, second, and third person respectively, foc=focus, adjun=adjunct, neg=negative. iii Though Dagbani does make a formal distinction between reflexives and reciprocals, I do not make a further discussion on reciprocals since the focus of this current paper is to discuss reflexive pronouns. ivThe first person singular pronominal n assimilates to the place of articulation of the following segment. v Note that the third person singular pronoun in Dagbani, o, “s/he” is not sensitive to gender since the language generally does not have gender as a prominent grammatical feature. 145Issah References Abdul-Rahman, F. 2005. A Spectrographic Analysis of Dagbani Vowels. Mphil Thesis, Department of Applied Linguistics, University of Education, Winneba. Adger, D. 2004. Core Syntax: A minimalist Approach. Oxford: Oxford University Press. Agbedor, P. 2002.Reflexivation in Buli: A Preliminary Observation. In M.E. Kropp Dakubu and E.K. Osam. Selected proceedings of the Annual Colloquium of the Legon-Trondheim Linguistic Project. 1: 147-152. Bendor-Samuel, J.T. 1989. The Niger-Congo languages:a classification and description of Africa’s largest language family. University press of America. Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris publications. Chomsky, N. 1986. Barriers. Linguistic Inquiry Monograph 13. MIT Press. Clements, G. N. 1975. The Logophoric Pronoun in Ewe: its Role in Discourse: Journal of West African languages 10:139-177. Cole, P. and Gabriella, H. 1998. Long-Distance Reflexives in Singapore Malayan: Apparent Typological Anomaly. Linguistic Typology. 2(1). Faltz, L. M. 1977. Reflexivization: A Study in Universal Syntax. Doctoral Dissertation, University of California at Berkeley. Haegeman, L. 1994. Introduction to Government and Binding Theory. Second Edition. Cambridge, Massachusetts: Blackwell. Oxford UK& Cambridge USA. Hudu, A.F. 2010. Dagbani Tongue-root Harmony: a Formal Account with Ultrasound Investigation. PhD Dissersation, University of British Columbia, Canada. 146 Legon Journal of the HUMANITIES Volume 22 (2011) Naden, A.J. 1988. Language, History and Legends in Northern Ghana, Ms GILLBT, Tamale. Pica, P. 1985. Subject, Tense and Truth: Towards a Modular Approach to Binding. In Gueron, Obenauer, and Pollock 1985: 259-292. Pica, P. 1987. On the Nature of the Reflexivization Cycle. In Proceedings of NELS 17: 483-499. GLSA, University of Massachusetts, Amherst. Rappaport, G. C. 1986. On Binding and Reflexives in Russian. Natural Language and Linguistic Theory 4: 97-120. Reinhart, T. and Reuland E. J. 1993. Reflexivity. Linguistics Inquiry 24. 4: 657- 720. Vikner, S. 1985. Parameters of Binder and Binding Category in Danish. Working Papers in Scandinavian Syntax. 23: 1-58. Wayne, H. 1996. Binding Theory, Control and Pro. In Gert Webelhuth (ed). Government and Binding and the Minimalist Program. (Place of publication/publisher?) 180-240. Yang, D.W. 1983. The Extended Binding Theory of Reflexives. Language Research 19:169-92