R-Expressions and the Interrogative Use of Anaphora in Najdi Arabic

This paper endeavors to explain how Najdi Arabic (NA), one of the dialects spoken in the central region of the Arabian Peninsula, diverges from Modern Standard Arabic (MSA) in its anaphoric treatment of R-expressions and pronominals. Data from a native Najdi Arabic informant suggest that only a subset of NA verbs allow proper names to be referentially bound by their antecedent pronouns in interrogative structures. Although this property is characteristic of Najdi Arabic not MSA, it yields certain challenges to the basic tenets of the Binding Theory. While Principle C of the Binding Theory requires R-expressions to be free, a referential reading of the NA data, which syntactically binds proper names with their pronominal referents, violates such principle.


Introduction
While most of the literature on the semantic nature of anaphors in Arabic is devoted to Classical Arabic (Abdul-Ghany, 1981), or Modern Standard Arabic (MSA), many grammatical aspects of Arabic dialects remain unstudied. This paper examines the use of anaphoric pronominals as interrogatives in Najdi, one of the widely-spoken dialects of Arabic. Arabic is the native language for over 186 million speakers and the liturgical one of almost 250 million people, ranking the sixth among the widely spoken languages of the world. It belongs to the Semitic group of languages and is closely related to Hebrew, Aramaic, Amharic (spoken in Ethiopia), and Tigrigna (spoken in Eritrea) (Versteegh, 2014). Najdi Arabic (NA), however, is far more limited in its use and is the native dialect of the people in the central region of the Arabian Peninsula, nowadays Saudi Arabia.

Statement of the Problem
A 29-year-old native speaker of the Najdi Arabic dialect provided the data for this research paper. The informant has lived in the central region of Najd, in the city of Riyadh for most of his life. His knowledge of both standard and dialectal Arabic is exemplary as his college major was Arabic. The informant was queried by the researcher himself on various constructions and possible utterances of standard as well as Najdi Arabic and his responses were recorded and analyzed.
According to the data provided by the informant, the following example in (1) best describes a question on whether someone has arrived or not in MSA: (1) hal ʤaʔ-a Ali?
"Did Ali arrive?" The question word hal in (1) is used to construct the interrogative and can be interpreted to mean "did" here. The verb ʤaʔ "arrived" is in the simple past tense, and is suffixed by the masculine clitic -a through subject-verb agreement in gender. Verbs in Arabic are strong so there is no need for an auxiliary verb in question formation. The construction in (1) is uttered with a rising intonation, as is the case with all questions set off by the question word hal. The R(eferring)-expression Ali (all proper names are R-expressions) is used referentially and its antecedent can be an individual determined by the context of speech or conversation. Similar to (1) is the construction in (2) below: (2) hal ʤaʔa-t Nurah?
"Did Norah arrive?" The verb in (2) carries the feminine clitic -t in agreement with the R-expression Norah, which is also used referentially here.
The data also show that it is possible for the R-expressions in (1) and (2) to be replaced, although less commonly, by pronouns in MSA: (3) hal ʤaʔ-a hʊwa?
Q-word arrived-mas he?
Q-word arrived-fem she?
"Did she arrive?" The pronominal pronouns hʊwa "he" and hɪja "she" in (3) and (4), respectively, make reference to an antecedent whose mention is determinable by context or previous conversation. In other words, (3) and (4) differ from (1) and (2) only in the substitution of R-expressions for pronouns.
The questions in (1) and (2), although construed in a similar fashion, can be formed differently in Najdi Arabic. According to the informant in this study, the following constructions in (5) and (6)  Note that in both (5) and (6) the question word hal is optional, albeit obligatory in all of the MSA examples (1-4) above. Interestingly, Najdi Arabic utilizes the masculine pronoun hu "he" in (5) and the feminine pronoun hi "she" in (6) to form the interrogative (Note 1). In other words, the NA pronominal pronouns hu "he" and hi "she" can function as the question word hal. In (5) the masculine pronoun hu "he" refers to the R-expression, Ali. This referential relationship is indicated by the coindexation of the proper name and the pronominal pronoun (i.e. Ali bears the same index i as the pronoun hu "he"). Similarly, the proper name Norah in (6) is referential and coindexed with the feminine pronominal hi "she". Note also that the verbs in (5) and (6), compared to (3) and (4), undergo glottal plosive /ʔ/ deletion with the vowel /a/ lengthened (Note 2).
The NA treatment of R-expressions referenced with pronominal pronouns, which function as interrogatives, is unprecedented; and no evidence exists, according to my knowledge, of other languages that behave similarly. Data from the informant confirmed that such usage is equally common in embedded structures as well.
Consider the examples (7) and (8) "Did you ask her if Norah arrived?" The structures above are comprised of the main clauses ʔant-ø saʔalt-uh and ʔant-ø saʔalt-ah "you asked him/her", and the subordinate null (if) complementize phrases, namely hu ʤaa-a Ali and hi ʤaa-t Nurah, which are identical to the ones in (5) and (6). Both (7) and (8) are statements uttered with rising intonation to form interrogatives in the main clause and the subordinate one as well.
"Did she arrive?" The constructions in (9) and (10) are judged ungrammatical by native speakers of Najdi Arabic. It appears that their ungrammaticality stems from the fact that the pronominal pronouns are referenced with each other within the same clausal context (i.e. governing domain); in (9) the pronoun hu "he" is tied to the other pronoun hu "he" within the same question and so is the pronoun hi "she" in (10). In other words, the antecedent of the pronominal pronoun hu "he" in (9) is the same male person, and the antecedent of the pronominal pronoun hi "she" in (10) is the same female person. It seems that the illicitness in these structures is due to both pronominal pronouns being not free within their governing domain (more on this in section 3).
To sum up, it is evident from the examples above in (5) and (6) that NA presents an intriguing case in which R-expressions and proper names are used referentially with pronominal pronouns within the same syntactic domain to form the interrogative. Not only is this usage deemed novel in language, it appears to be idiosyncratic of Najdi Arabic. In fact, such structures where the proper name is linked to its antecedent are judged ungrammatical in Modern Standard Arabic: "Did Norah arrive?" The treatment of R-expressions referentially with the pronominals presents a serious dilemma to the Binding Theory, which predicts, erroneously in case of NA, the ungrammaticality of (5) and (6). In what follows a discussion of how the Binding Theory fails to exclude such structures is presented; and what alternative analysis is possible to account for the emergence of anaphoric pronominals along with their proper name referents as a device for question formation in Najdi Arabic.

Discussion
It is important to understand before contemplating a possible explanation to the newly found syntactic configuration of NA question formation how the Binding Theory operates in anaphoric relations, and why it fails to rule out the occurrence of such structure in NA (see Hicks, 2009 for a critique of the BT).

The Binding Theory
The Binding Theory (BT), also known as the Government and Binding Theory (GBT), is part of a larger theoretical framework on Transformational Grammar (TG) advanced by Chomsky (1982), and later revisited in his well-acclaimed book the Minimalist Program (1995). While an extensive overview of the Binding Theory is beyond the scope of this paper, focus will be on the relevant principles and concepts of the BT germane to the issue at hand (see Büring, 2005;Haegman, 1994;Webelhouth, 1995 for in-depth discussion of the BT).
According to the Binding Theory, there are basically three conditions or states which define the binding relationship between the noun or the pronoun and its antecedent or binder, and can be formalized in the following three principles (Grohmann, Hornstein, and Nunes, 2005, p. 236): (13) Principles of the Binding Theory I. Principle A: An anaphor (e.g. a reflexive or reciprocal) must be bound in its domain or governing category.
II. Principle B: A pronoun must be free, not bound, in its domain or governing category.

International Journal of Linguistics
The Governing Category means the minimal or local domain that contains the governor, governee and the subject, and is often used synonymously with the term domain. Armed with these definitions, the notion of Syntactic Binding in the Binding Theory is stated as follows: (16) Syntactic Binding α syntactically binds β if and only if α c-commands β; α and β are co-indexed; and α is an A-position.

R-Expressions
In order to appreciate the discrepancies in the behavior of R-expressions in MSA and NA interrogative structures, a reexamination of the data vis-à -vis Principle C of the Binding Theory is necessary. Consider the following examples from MSA, discussed earlier in section 2 and repeated here for convenience sake: "Did Norah arrive?" Principle C of the Binding Theory, which states that R-expressions must be free everywhere, seems to be obeyed in (17) and (18) as the proper names Ali and Norah are not coindexed with nor are they bound by any element in the clause. However, the same is not true for (19) and (20); the binding of the R-expressions Ali and Norah by the pronouns hʊwa "he" and hɪja "she" violates Principle C and renders the construction ungrammatical.
In contrast to MSA, NA allows R-expressions to be syntactically bound to the pronouns; that is to say, they are not free: To illustrate the syntactic binding relationship between the R-expressions and the pronouns in (21) and (22), consider the following tree structure diagrams: Figure 1. Binding between R-expression Ali and Pronoun hu "he" International Journal of Linguistics ISSN 1948-5425 2020 www.macrothink.org/ijl 8 Figure 2. Binding between R-expression Nurah and Pronoun hi "she" All conditions for syntactic binding are met in Figures 1 and 2. Assuming that the Binding Theory accounts only for bound variables, the pronouns hu "he" in Figure 1 and hi "she" in Figure 2 must raise to the quantifier position in the tree structure, the NP head position, leaving a trace behind (for further discussion on Quantifier Raising, see Heim and Kratzer, 1998). The A-position pronominals hu "he" and hi "she" c-command the R-expressions Ali and Norah respectively since neither one dominates the other. The R-expressions are referentially coindexed with the pronouns. More specifically, the syntactic binding here occurs between the trace of the pronominal and the proper name, both assigned the same indexation i. Regardless of whether the binding is local and operates within the governing category or not, examples (21) and (22) constitute a clear violation of Principle C, and should not, therefore, surface in language. In other words, the Najdi Arabic structures (21) and (22), contrary to the MSA examples (17) and (18), should be deemed ill-formed, yet they are reported by the informant to be fully acceptable by native speakers of Najdi Arabic.
To sum up, it seems that while MSA systematically obeys Principle C of the Binding Theory as examples (17-20) demonstrate, NA allows structures that are otherwise prohibited by Principle C; hence, a stark discrepancy between MSA and NA exists in the treatment of R-expressions.
Q-word arrived-mas he?
Q-word arrived-fem she?
"Did she arrive?" ISSN 1948-5425 2020 Note how the pronouns hʊwa "he" and hɪja "she", as evidenced by lack of coindexation, are not bound in their governing category, in accordance with Principle B of the Binding Theory; that is, they do not refer to an entity within the clause. Likewise, the examples (9) and (10), repeated here as (25) and (26) "Did she arrive?"

International Journal of Linguistics
The following diagrams demonstrate the binding relationships between the pronouns in (25) and (26):  ISSN 1948-5425 2020 (25) and (26) are ungrammatical because the pronouns are not free as they are used referentially. Figures 3 and 4 show how the post verbal pronouns hu "he" and hi "she" are coindexed and c-commanded by the preverbal ones; i.e. both pronouns refer to each other and both refer to the same person.

R-expressions and Verb Transitivity (Note 3)
Further examination of the data provided by the informant reveals some interesting distribution restrictions on the referential use of R-expressions in Najdi Arabic. Recall that contrary to Principle C of the BT, Najdi Arabic permits R-expressions to be bound as argued in 3.2.1. Nonetheless, data elicited from the informant suggest that this referential treatment is only possible with intransitive verbs: (27) *hu i tarab Ali i ? he hit-mas Ali?
"Did Norah hit (someone or something)?" In (27) and (28) the R-expressions Ali and Norah refer to hu "he" and hi "she", respectively, but because the verb tarab "hit" is transitive, the structures are impossible in NA. This is reasonable since the verb tarab "hit" requires an object different from the subject hu "he"/hi "she"; i.e. the subject and object cannot refer to the same person.

International Journal of Linguistics
she hit-fem Norah.
"She hit Norah." Note that a referential reading where Ali and Norah are related to the pronouns hu "he" and hi "she" is not possible here regardless whether the structure is interrogative or declarative. The pronouns are indexed with the subject (a person specified by the context of speech). The proper names Ali and Norah serve as the object of the verb tarab "hit". In other words, the antecedents of the pronominals hu "he" and hi "she" must not be the same individuals to whom the R-expressions refer. The antecedents have to be other individuals (or things) that are salient in the conversation and are believed to be the agent of the action.
Aside from the fact that they violate Principle C of the BT, one reason for the unacceptability of (27) and (28) in the first place is the semantic ambiguity that arises with such structures since the pronouns hu "he" and hi "she" can have multiple readings: A. Be indexed and referential with the R-expressions as in (27) and (28). In this case, the construction has to be a question about the subject proper name mentioned in the sentence. The sentence then carries the meaning of confirmation and is deemed ungrammatical.
B. Be referential with a subject not mentioned in the sentence. In this case, the construction can either be a question asking about the subject as in (29) and (31), or a declarative statement about the subject, informing us that someone hit Ali or Norah as in (30) and (32).
In other words, when the R-expression is the subject, a referential reading with the pronoun is construed. But when the R-expression is the object, a reading of a declarative or a question about a subject not present in the sentence is obtained.
However, consider the following set of sentences: "Did she eat Norah?" Even though the verb ʔakal "ate" is transitive, the R-expressions Ali and Norah must be coindexed with the pronouns. According to (34) and Principle C of the BT, neither (35) nor (36) should exist in NA, but the informant in this study emphasized their grammaticality. Two readings for the aforementioned construction are possible: A. A referential reading where the pronouns refer to the agentive R-expressions as in (35) and (36); this is the most natural and acceptable reading obtainable.
B. The R-expressions are free and the pronouns refer to someone else as in (37) and (38); in this case, the questions inquire whether someone ate Ali and Norah. Such reading is fortunately not acceptable and can only be true in a world where humans are eatable! But how is it that some verbs, although transitive, can permit such constructions while others such as the ones in (34) do not? It is argued that the two verbs tarab "hit" and ʔakal "ate" behave differently when it comes to the choice of object they take: while in every day speech, the verb tarab "hit" can take both human and nonhuman objects, the verb ʔakal "ate" is restricted to nonhuman objects. Other verbs that fit into this category include: ʃarab "drank", saaq "drove (a car)", χabaz "baked", tabaχ "cooked", banaa "built". These verbs, among many others, would allow R-expressions to be referential with the pronouns to form interrogatives since no ambiguous reading can arise: such verbs must only take nonhuman objects, and the pronouns hu "he" and hi "she" have to refer to the R-expressions within the sentence.

Analysis
The discussion so far can be summarized as follows; MSA treats pronouns and R-expressions in accordance with Principles B and C of the Binding Theory while NA employs R-expressions referentially, in violation of Principle C. While no clear explanation for the referential behavior of R-expressions in NA exists, a number of possible arguments can be entertained.
One possible analysis is to consider examples (21) and (22), repeated here as (39) and (40), as instances of apposition: Note 2. Although irrelevant to the current discussion, this process of Compensatory Lengthening is quite common in Najdi Arabic and is found in a number of other languages as well (see for example Hualde, 1990;Poser, 1988;Sukač, 2011; among many others).

Copyrights
Copyright for this article is retained by the author(s), with first publication rights granted to the journal.
This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/)