The Syntax of Anaphors in Ki-Imenti: A Bantu Language Spoken in Kenya

This paper is an investigation of the distribution of anaphors in Ki-Imenti. KiImenti is a Bantu language spoken in Meru Central, North Imenti and Buuri sub-counties, Meru County. It is one of the dialects of Kimeru. The objective of this paper is to determine the syntax of anaphors in SMART MOVES JOURNAL IJELLH e-ISSN: 2582-3574 p-ISSN: 2582-4406 VOL. 8, ISSUE 4, APRIL 2020 www.ijellh.com 224 KiImenti. The study is guided by Chomsky’s Binding Theory. The binding theory divides noun phrases into three basic categories anaphors, pronouns and R-expressions. This theory develops three binding principles to explain the distribution of these noun phrases. These are the binding principle A, binding principle B and binding principle C. This study is only limited to the distribution of anaphors. The study adopts a qualitative study research design as it gives detailed descriptions and explanations of the phenomena studied. The researcher generated the data for the study herself using self-introspection and the data was corroborated by ten native speakers who were purposively sampled. The paper has established that Ki-Imenti conforms to the binding principles. Anaphors are bound in their binding domain, whereby the binding domain is the inflectional phrase or the noun phrase containing the anaphor. This paper will make a contribution to the knowledge of the syntax of anaphors in KiImenti and the description of the syntax of Bantu linguistics in general.


Introduction
The distribution of noun phrases in all languages is governed by the binding principles.
These binding principles determine the distribution of anaphors, pronouns and R-expressions.
The binding theory divides noun phrases into three basic categories anaphors, pronouns and R-expressions. Chomsky (1981) in his Binding Theory develops three binding principles to explain the distribution of these noun phrases. These are the binding principle A, binding principle B and binding principle C. This paper was, however, limited to the syntax of anaphors in Ki-Imenti. Carnie (2006) defines anaphors as noun phrases that obligatorily get their meaning SMART MOVES JOURNAL IJELLH e-ISSN: 2582-3574 p-ISSN: 2582-4406 VOL. 8, ISSUE 4, APRIL 2020 www.ijellh.com 225 from another noun phrase. They include reflexives (such as themselves, herself, and itself) and reciprocals (each other, one another). The binding principle that governs the distribution of anaphors is called binding principle A. It states that an anaphor must be bound in its binding domain. Carnie (2013) defines a domain as a syntactic unit that is clause-like. Binding requires two conditions to be met. One, the anaphor and the antecedent (a noun phrase that gives meaning to it) must be co-indexed and two, the antecedent must c-command the anaphor. The antecedent c-commands the anaphor if every node dominating the antecedent also dominates the anaphor and neither the antecedent nor the anaphor dominate the other. The way two noun phrases are indicated to be co-referential is by means of an index, usually a subscripted letter.

Theoretical Framework
This paper is guided by Binding Theory by Chomsky (1981). This is the part of the syntactic theory that shows how the interpretation of noun phrases is constrained by syntactic considerations. According to Chomsky (1981), the distribution of anaphors is determined by Binding principle A of this theory which states that: Anaphors are bound in their binding domain. For this to happen, there are two conditions that must be met.
The first one is that an anaphor must have a local (nearby) antecedent. Thus, Timothyi washed himselfi obeys Condition A: the antecedent of himself, which is Timothy, is in the same clause with the anaphor. In contrast, * Timothy i asked Sylvia to wash himselfi is unacceptable, because the reflexive and its antecedent are in different clauses. This is one of the conditions for anaphors, that the anaphor must find an antecedent in the same clause. Another condition indicated in the Binding principle A for anaphors is that anaphors must be ccommanded by their antecedent. Radford (2009) elaborates this condition. He explains that binding is a means of assigning reference and when one DP C-commands and is coindexed with another DP, the first is said to bind the other. A binds B iff (if and only if) A c-commands B iff every node dominating A also dominates B. In a nutshell this means a node c-commands its sister and everything its sister dominates. Therefore it follows that the antecedent must be in the subject position to c-command the anaphor. This is illustrated in sentence (a) and the syntactic tree in (b) (a) Mosesi loves himselfi In example (a) Moses is the antecedent of the reflexive himself. The antecedent is in the same clause with the reflexive and it is in a structural position where it c-commands the reflexive, that is every node dominating the antecedent also dominates the reflexive and this is the TP node as shown in (b). The antecedent and the reflexive corefer and they are appropriately coindexed. The Binding theory was helpful in establishing the binding relationship that holds between the anaphors in Ki-Imenti and the antecedents. This was done by analyzing the syntactic space in which anaphors must find their antecedents. Further the arguments that anaphors can refer to was investigated.

Research Methodology
The researcher generated the data herself using self-introspection. This is because the researcher is a Ki-Imenti native speaker. She is therefore competent and qualified to generate some of the data needed for analysis. Radford (1988) notes that for practical purposes, most linguists describing a language of which they are native speakers rely on their own intuitions and thus the grammar they devise is essentially the grammar of their own idiolect, which they assume is a representative of the language as a whole. The choice of this method was motivated by the fact that the researcher is a native speaker of the language under study and knows what she is looking for. This saves a lot of time and prevents the possibility of unwanted data. For further verification, the researcher counter checked with ten other Ki-Imenti native speakers.

Findings
This section presents the environment in which an anaphor must find an antecedent and the kind of relationship that holds between anaphors and their antecedents. To illustrate this relationship, syntactic trees have been drawn. Also discussed are the arguments that anaphors can refer to.

The Binding of Anaphors in Ki-Imenti
The reflexive in Ki-Imenti is bound in its binding domain. This is in accordance with the binding principle A (Chomsky 1981). The reflexive in Ki-Imenti must find an antecedent in its local domain as shown in sentence (1).

Mwitimii bought himselfi a car
In sentence (1) the reflexive morpheme -ciderives its interpretation from the antecedent Mwitimi in the same clause. The reflexive in Ki-Imenti cannot be co-indexed with an antecedent outside its local domain as illustrated in (2).
Sentence (2) is wrongly coindexed because the reflexive morpheme -cicannot have Arimu as its antecedent because they are not in the same clause. This conforms to the binding principle A which requires anaphors to be bound in their own domain.
Further the reflexive in Ki-Imenti cannot be coindexed with a noun phrase embedded inside the subject noun phrase as shown in (3a). To bind the reflexive, the antecedent has to be in the subject position as shown in (4a) and if it is not the sentence becomes ungrammatical as shown in (4b). In Ki-Imenti reciprocals must be bound locally as shown in (5) (5) Aanai i-ba-kũ-rum-ani-a Children2 f-sm-pres-insult-rec-fv The childreni are insulting each otheri.
In sentence (5) the reciprocal -anis bound by the antecedent aana in the same clause through the pronominal affix ba. If the reciprocal does not find an antecedent in the same clause the sentence becomes ungrammatical as illustrated in (6).
Sentence (6) is ungrammatical because the reciprocal -ancannot get its interpretation from the subject Gatwiri and Koome given that this antecedent is very far from the reciprocal because they are in different clauses.
Further the reciprocal morpheme cannot have a noun phrase embedded in the subject noun phrase as its antecedent as indicated in sentence (7b). The right coindexing is as indicated in (7a) where the whole subject noun phrase is coindexed with the reciprocal.
(7a) {Athingatĩri ba arimũ}i i-ba-men-eni-e Supporters of teachers2 f-sm-hate-rec-fv Muriungi (2010) notes two types of causative morphemes in his study of Kiitharaka causation. These are the transitive -iwhich he calls an internal cause and the -ithi referred to as the synthetic causer. He observes that when the causative -iis suffixed to the verb, it introduces a causer argument and makes the verb take an object hence transitivizing it. He notes the synthetic causer -ithis suffixed to transitives and agentive intransitives thereby introducing a causer.
Mwangi (2001) indicates that there are two causative affixes, the -iand the -ithin Gikuyu, Kiembu and Kikamba. In Gikuyu, and Kiembu they are realized in these forms regardless of the phonological structure of the verbs to which they are affixed, but in Kikamba phonological processes affect the realization thus -iis also realized as -y-, -e-, and -sywhileith-is also realized as: -ethy-or -ithy-. In many languages, causation is expressed morphologically by use of affixes in the verb (Mwangi, 2001).
The causative is marked by the morpheme -ithor -iin Ki-imenti.
The applicative licenses a wide range of object roles that include the beneficiary, recipient, maleficiary, goal, instrument, reason and location (Ngonyani, 1998).
The applicative is marked by the morpheme -irin Ki-Imenti. In Kiswahili the applicative is marked by the morpheme -i- (Marten, 1999). In Chichewa, the applicative affix has two allomorphs: -il-and -el- (Mchombo, 2004). The allomorphs that is selected and affixed to the verb is constrained by rules of vowel harmony.
Just like the causative the applicative adds an extra argument to the sentence. The prototypical thematic role of the argument added by applicative is the beneficiary  (Mchombo, 2004). The beneficiary is a participant for whose benefit the action expressed by the verb is performed.

Reflexives and Causatives
In reflexive causative constructions in Ki-Imenti, binding is only possible between the causer and the external argument (8a). The external argument and the object cannot be coindexed (8b).

(b) * Kimathi forced Karimi to kill herself
The observation in (8) means that the interpretation of a reflexive causative construction is that the agent forced himself to do a particular action as indicated in (8a). The reflexive marker cannot be interpreted as the object of the verb root as indicated in (8b).

Reflexives and Applicatives
In applicative constructions involving the reflexive, binding is possible between the subject and the applied argument as shown in (9a). Binding is also possible between the subject and the direct object (9b). However, binding is not possible between the applied and direct object (9c) In (9a) the interpretation of the reflexive applicative construction is that subject is the beneficiary of the action of the verb while in (9b) the subject initiated an action upon himself for another person's benefit. The interpretation in (9c) is not possible, that the subject initiated an action upon another person for that persons benefit.

Reciprocals and Causatives
In reciprocal causative constructions in Ki-Imenti, binding is possible between the external argument and the direct object as shown in (10a). The causee and the external argument cannot be coindexed as shown in (10b). Further the causer and the direct object cannot be coindexed (10c).

Reciprocals and Applicatives
In applicative constructions, that have reciprocals binding is possible between the subject and the applied argument (11a). It is also possible to coindex the subject and the basic object (11b). The basic object and the applied object cannot be coindexed (11c).
(c) *The parents tied the childreni for each otheri From the observation of anaphors and multiple argument involving applicatives and causatives, the binding generalization is that binding is between the closest arguments. Further in reflexive and reciprocal constructions in Ki-imenti, only one of the objects of the causative is involved in the binding process. This is the causee. In applicative constructions on the other hand any of the objects is involved in binding. That is either the basic or the applied verb. This aspect makes Ki-imenti both an asymmetrical and symmetrical object language. In an asymmetrical object language in the context of multiple object constructions like applicatives and causatives only one of the objects can behave like a real object. This is in contrast to symmetrical object languages whereby in multiple object constructions, both objects are given the same status such that any object can be involved in binding process, (Bresnan, 1990). SMART MOVES JOURNAL IJELLH e-ISSN: 2582-3574 p-ISSN: 2582-4406 VOL. 8, ISSUE 4, APRIL 2020 www.ijellh.com 235

Discussion of Findings
As it has been demonstrated, an anaphor in Ki-Imenti must be bound within its binding domain according to binding principle A (Chomsky, 1981). For this to happen, first the antecedent must c-command the anaphor. The antecedent c-commands the anaphor if every node dominating the antecedent also dominates the anaphor and neither the antecedent nor the anaphor dominates the other. This is demonstrated in sentence (12) with the syntactic representation in (13).
(12) Mucore wa John n-a-ci-end-et-e Friend2 of John f-sm-om-love-pfc-fv John's friend loves himself. In diagram (13) the antecedent of the reflexive morpheme -cican only be understood to be the whole NP1 phrase Mucore wa John. This is because the node that dominates NP1 also dominates the node that has the reflexive. This is the Top P node. This phrase qualifies to be an appropriate antecedent because it c-commands the reflexive according to the binding principle A. However the reflexive in diagram (13)  John to be an appropriate antecedent to the reflexive.
In the same way as relexives, a reciprocal must be c-commanded by an antecedent as shown in sentence (14) with the syntactic tree (15).
(14) {Athingatĩri ba arimũ}i i-ba-men-eni-e Supporters of teachers2 f-sm-hate-rec-fv The supporters of teachers hate each other. The right antecedent for the reciprocal in diagram (15) is the whole phrase athingatiri ba arimu in the NP1 node. There is only one node dominating NP1 and that is the TopP. The same node dominates the verbal complex with the reciprocal. Verbal complexes are derived via incorporation that is successive head to head movement (Baker, 1988). Therefore, the verb moves from head to head in the structure, picking up reciprocal and the final vowel affix as shown in (15). Thus the antecedent athingatiri ba arimu c-commands the verbal complex (VP) containing the reciprocal that is menene The word arimu in NP2 cannot be an antecedent to the reciprocal. There is an NP1 node dominating it that does not dominate the reciprocal therefore it