On the structure of A-bar constructions in Dagbani: Perspectives of «wh»-questions and fragment answers

by Samuel Alhassan Issah (Author)
Thesis 238 Pages

Table Of Content

  • Cover
  • Title
  • Copyright
  • About the author
  • About the book
  • This eBook can be cited
  • Acknowledgments
  • Contents
  • Abstract
  • List of abbreviations
  • Chapter one General introduction
  • 1.0 Introduction
  • 1.1 The language and its speakers
  • 1.2 The objectives of the research
  • 1.3 Sources of data and orthography
  • 1.4 Theoretical insights of the analysis
  • 1.5 Word order and clause structure of Dagbani
  • 1.5.1 The aspect and tense system of Dagbani
  • 1.5.2 On the syntactic status of kà and ń in Dagbani
  • 1.5.3 A theory of focus and pragmatic uses of focus constructions
  • 1.5.4 Predicate focus in Dagbani
  • 1.6 The structure of the dissertation
  • 1.7 Interim summary
  • Chapter two The grammar of the Dagbani interrogative DP
  • 2.0 Introduction
  • 2.1 Wh-phrases as question operators in wh-questions
  • 2.2 Previous studies of Dagbani wh-phrases
  • 2.2.1 A novel account on the inventory of Dagbani wh-phrases
  • 2.2.2 On wh-pronouns and wh-determiners in Dagbani wh-phrases
  • 2.3 The grammatical characterization of Dagbani wh-phrases
  • 2.3.1 Distinction between human and non-human wh-phrases
  • 2.3.2 Number as a functional category of Dagbani wh-phrases
  • 2.3.3 Accounting for why -nìmá heads the functional NumP in the interrogative DP
  • 2.3.4 Ambiguities in Dagbani wh-phrases
  • 2.4 Interim summary
  • Chapter three The syntax of Dagbani ex-situ wh-questions
  • 3.0 Introduction
  • 3.1 Review of previous studies on Dagbani wh-questions
  • 3.2 The movement of wh-phrases within matrix/root clauses
  • 3.2.1 Syntactic evidence motivating focus movement of wh-phrases
  • 3.2.2 Pied-piping and feature checking in syntactic movement
  • 3.3 Extraction out of wh-phrases in embedded clauses in ex-situ wh-questions
  • 3.3.1 Extraction of subject wh-phrases out of embedded clauses
  • 3.3.2 Extraction of non-subject wh-phrases from embedded clauses
  • 3.3.3 Accounting for the reflexes of movement in ex-situ wh-questions
  • 3.3.4 Linguistic properties of resumptive pronouns
  • 3.3.5 Novel account on the spell-out of the focus heads in ex-situ wh-questions
  • 3.4 Typology of focus heads inventory in the Mabia languages
  • 3.5 Intermediate summary
  • Chapter four A syntactic analysis of Dagbani wh-in-situ questions
  • 4.0 Introduction
  • 4.1 Review of previous studies on Dagbani in-situ wh-questions
  • 4.2 Counterevidence for analysis of lá as a contrastive focus marker
  • 4.3 Dagbani wh-in-situ questions
  • 4.3.1 The descriptive facts on Dagbani in-situ wh-questions
  • 4.3.2 A minimalist analysis of Dagbani in-situ wh-questions
  • 4.3.3 An account for the ban on in-situ subject wh-phrases
  • 4.4 On the syntax of Dagbani wh-questions: Some typological insights
  • 4.5 Interim summary
  • Chapter five Constraints on wh-movement in Dagbani
  • 5.0 Introduction
  • 5.1 Island effects in Dagbani wh-constructions
  • 5.1.1 The complex NP constraint (CNPC)
  • 5.1.2 The syntactic properties of the Dagbani relative clause
  • 5.1.3 Extraction of wh-phrases from Dagbani relative clauses
  • 5.1.4 Extraction of wh-phrases out of the complex NP
  • 5.2 The coordinate structure constraint
  • 5.3 Interim summary
  • Chapter six On the syntax of answers to wh-questions and the derivation of fragment answers
  • 6.0 Introduction
  • 6.1 On the structural correlation between wh-questions and their answers
  • 6.1.1 Context-induced focus and question/answer correlation in Dagbani
  • 6.1.2 The projection of focus heads in ex-situ question/answer pairs
  • 6.1.3 Congruence of question/answer pairs
  • 6.2 Introduction to fragment answers
  • 6.2.1 An overview of the notion of fragment answers
  • 6.2.2 The syntactic derivation of Dagbani fragment answers
  • 6.3 Arguments in support of movement plus ellipsis account of Dagbani fragments
  • 6.3.1 The sensitivity of fragments to locality constraints
  • 6.3.2 Evidence in favor of analyzing fragment answers as an instance of elliptical process
  • 6.4 Semantic recoverability of fragment answers: the PF theory of fragment answers
  • 6.4.1 Semantic recoverability of fragment answers
  • 6.4.2 The PF theory of fragment answers and the [E];-feature
  • 6.4.3 Dagbani fragment answers and the [E];-feature
  • 6.5 Interim summary
  • Chapter seven Summary and conclusions
  • 7.0 Introduction
  • 7.1 Summary of the findings of the dissertation
  • 7.2 Summary of the contributions/significance of the dissertation
  • 7.3 Open issues for future research
  • Appendix
  • List of tables
  • References
  • Series index


This dissertation provides a description and analysis of the structure of A-bar constructions, focusing on the derivation of wh-questions and fragment answers in Dagbani, a Mabia (Gur) language spoken in Northern Ghana. I show that wh-phrases are crucial syntactic elements in the derivation of wh-questions and further examine the inventory and grammatical characteristics of these wh-phrases. I demonstrate that in the formation of wh-questions, wh-phrases occur in two distinct positions, ex-situ and in-situ, except for subject wh-phrases, which only occur in the former position. I provide a theoretical analysis of the distribution of wh-phrases couched within Minimalism (Chomsky 1995). I assume that Dagbani has two focus feature specifications in its lexicon: a strong and a weak focus feature. Whereas the former triggers overt syntactic movement of the wh-phrase from its base position to the clausal left periphery, the latter licenses covert movement at LF. I further show that while the ex-situ wh-questions mandatorily require the overt morphological presence of the particles and ń which must be c-commanded by the extracted wh-operator, their in-situ counterparts do not have these syntactic items overtly expressed. I account for this absence of phonologically visible focus heads in in-situ wh-questions by referring to the fact that LF movement is a post-syntactic phenomenon, and therefore, although the focus heads are present in the in-situ wh-questions, they are not phonologically visible to the syntax. Consequently, I conclude that and ń are spell-outs of this strong focus feature which establish the needed Spec-Head configuration for feature checking. The fact that both ex-situ and in-situ wh-questions exhibit sensitivity to island effects (Ross 1967), motivates the proposal that they are derived via a movement operation. The syntactic incompatibility between ex-situ wh-questions and fronted focus constituents is interpreted to mean that they compete for the same syntactic position, supporting an analysis according to which wh-fronting is analyzable as an instance of focus movement. Contrary to reports of previous studies that the spell-out of focus heads in ex-situ wh-questions is regulated by the grammatical category of the extracted constituent (cf. Fiedler 2007, Fiedler & Schwarz 2005, Fiedler et al. 2010, Hudu 2012, Issah 2012, Schwarz & Fiedler 2007), I show that there is empirical evidence to suggest that this generalization is not tenable. Building on a proposal made in Issah (2008), I thus, develop a novel account according to which the spell-out of a focus head arises from clausehood. I further offer formal explanations for some asymmetric patterns in the formation of wh-questions including the ban on ←17 | 18→in-situ subject wh-questions in contrast to their non-subject counterparts and the licensing of traces and resumption pronouns in the extraction of matrix and embedded subjects respectively. I conclude that subject in-situ wh-questions are barred because Spec, TP is the unmarked topic position. I propose that the complementarity of traces and resumptive pronouns can be accounted for by employing the fact that (i) there is a strong EPP requirement in Dagbani, the reason for which the Spec, TP position should always be filled with a DP material and that there is a blocking effect of this for local wh-subject extraction due to the Highest Subject Restriction (McCloskey 1990, 2002) or (ii) Dagbani is sensitive to the that-trace effect for which reason the overt complementizer head cannot be c-commanded by a trace (Perlmutter 1968, 1971). I further show that although wh-questions and their answers may share syntactic parallelisms, this is not a requirement for congruency in question/answer pairs. Finally, I give an account of the syntactic derivation of fragment answers, analyzing them as instances of ellipsis, employing the PF theory of ellipsis by Brunetti (2003) and Merchant (2001, 2004, 2008) among others. The sensitivity of fragment answers to island constraints (Ross 1967), the ban on certain categories in fragment answers as well as the connectivity effects, serve as syntactic evidence supporting an analysis according to which the derivation of elliptical answers involves A-bar movement together with PF deletion of the remnant TP. The elided component is argued to be retrievable from a property that licenses its semantic recoverability. I conclude that within the Minimalist assumption (Chomsky 1998), the processing of fragments is more economical than their non-elliptical counterparts.

Chapter one General introduction

1.0 Introduction

The goal of this dissertation is to provide a syntactic account of A-bar constructions in Dagbani, focusing on the structure of wh-questions and their answers. In this chapter, I present some background information on the genetic affiliation and basic clause structure of Dagbani, the objectives of the study, the theoretical frameworks within which the analyses are couched, a theory of focus, given the fact that the idea of focus is relevant to later discussions in the dissertation, the sources of the data used, orthography used in the presentation of the Dagbani data used in this current research, and the structure of the dissertation. This chapter shall proceed as follows. In section 1.1, I present a brief overview of the language and its speakers focusing on the genetic affiliation, the population of speakers and their geographical location. Section 1.2 outlines the objectives that underline the present study, while section 1.3 briefly comments on the sources of the data used in the current study as well as the orthography used in presenting the Dagbani data. Though I show that the writing system is based on the current approved orthography for Dagbani, I also point out the fact that where there are apparent unscientific spelling rules that would pose a threat to syntactic analysis, or are empirically unmotivated, I defy such rules. Section 1.4 is devoted to a discussion of the theoretical frameworks within which the analyses are cast. I show that the analyses are couched within the syntactic views of Chomsky (1995) Minimalist syntax as well as the ellipsis as PF-deletion Theory of Merchant (2001, 2004, 2008), while section 1.5 outlines the basic word order and clause structure of Dagbani. Some basic grammatical properties of Dagbani, including tense, aspect, mood and negation, are discussed in this section. In addition to these, I also discuss a theory of focus and offer an overview of term and predicate foci of Dagbani in this section. In section 1.6, I outline the structure of the dissertation, while section 1.7 offers a summary to the chapter.

1.1 The language and its speakers

Genetically, Dagbani belongs to the South-Western Oti-Volta subgroup of the Mabia (Gur), group of languages (Bendor-Samuel 1971 and Naden 1988, 1989). Approximately, Dagbani is spoken by around two million people in Ghana (Abukari 2018). Bodomo (1993), suggests the indigenous term Mabia as a classificatory term for these languages which include Dagbani, Dagaare, Mampruli ←21 | 22→and Kusaal. The use of the term Gur to refer to these languages was principally motivated by the assumption that the morpheme Gur was prevalent in the syllable structure of the languages. Contrary to this claim is the fact that it is not as common as assumed. It is for instance not motivated empirically for Dagbani since it is not a common syllable. In the light of this, there has been the need to reconsider nomenclature primarily based on the socio-cultural aspirations of the speakers of these languages. Bodomo’s proposal of the term Mabia is a reflection of the socio-cultural affinities that are shared by speakers of these languages. Following Bodomo (1993 et seq.), I adopt the term Mabia for this cluster of languages to which Dagbani belongs and which had once been referred to as Gur languages. The term Mabia as explained by Bodomo (1993 et seq.) is made up of two independent morphemes ma ‘mother’ and bia ‘child’. These two morphemes are predominant in these languages and has often resulted in speakers of these languages referring to one another as Mabia. The term is, therefore, assumed to be a reflection of the socio-cultural similarities among these languages, a reason for which it is deemed more appropriate.

Dagbani has three major dialects, namely Tomosili (Western Dialect), Nanunli, and Nayahali (Eastern dialect) which are spoken in and around Tamale (the political capital of the Northern Region), Bimbilla and Yendi respectively (Olawsky 1999, Hudu 2010 et seq.). According to Hudu (2010) and Olawsky (1999), inter alia, one major noticeable disparity is at the level of segmental and suprasegmental phonology with no noticeable differences in the structural domains among these three major dialects of the language. As a speaker of the Tomosili (Western) dialect, and considering the fact that I want to make this piece more representative of all three dialects, I have consulted native speakers of the other two dialects (Nanunli, and Nayahali) so as to enable me account for possible dialectal discrepancies that might exist.

Here, I discuss only grammatical properties that are considered relevant to later discussions in the dissertation. However, before I discuss these syntactic phenomena, I must point out one crucial phonological issue which especially supports my motivation to mark tone on all the data used in this thesis. Dagbani is a tone language. As a phonological characterization of tonal languages, tone is phonemically contrastive since there could be two words that are segmentally same, but the difference in meaning is based on the suprasegmental feature of tone. Dagbani has two basic contrastive tones, High (H) marked with (ˊ) and Low (L) which is marked with (ˋ). The interaction of the level tones may lead to the application of tone rules such as down step (Hyman 1993, Hyman & Olawsky 2004, Olawsky 1999). In the data used throughout the dissertation, ←22 | 23→only phonemic tones are marked. The data in (1) illustrate the grammatical uses of tone in Dagbani.

Previous linguistic research on Dagbani (Hudu 2010, 2013, 2014a, 2014b) has focused on the phonology with relatively little investigation into the syntax of the language. With the exception of Olawsky (1999), there is no known work on Dagbani that has covered significant aspects of the syntax. Previous aspects of the information structure mainly issues of focus marking have been addressed in Fiedler (2007), Fiedler & Schwarz (2005), Fiedler et al. (2010), Hudu (2012), Issah (2008, 2012, 2013c) and Schwarz & Fiedler (2007). The aspect of the formation of wh-questions and other A-bar construction like the derivation of fragment answers however, remain largely uninvestigated. This research gap is filled by this dissertation.

1.2 The objectives of the research

The specific objectives that underpin this present study include the following:

Investigate the inventory, internal structure and grammatical properties of Dagbani wh-phrases

Give a theoretically informed description of the strategies that are employed in the derivation of wh-questions

Argue that the formation of ex-situ and in-situ wh-questions can be accounted for using a movement analysis since both derivations are constrained by island effects

Motivate a syntactic analysis according to which movement of wh-operators is an instance of focus movement

Show that although there are constraints that regulate wh-questions and their congruent answers, formal structural correlation between the two is not required

Offer a systematic study into the syntactic derivation of fragment answers according to which they are derived via A-bar movement and an elliptical process

←23 | 24→

1.3 Sources of data and orthography

The data used in this dissertation are from different kinds of sources including: (i) data based on my introspective judgments as a native speaker together with other native speakers and (ii) the data gathered through field trips covering the three major dialects of Dagbani. Being a native speaker of Dagbani, most of the data used in this dissertation are based on my native speaker’s intuitions. All the data that are constructed by me as a native speaker have been cross-examined with other native speakers for grammaticality checks.

In addition to the data generated based on my native speaker’s introspection, I also carried out fieldwork within the language community aimed at ascertaining possible dialectal differences in the syntactic properties of the topic under investigation in the major dialectal areas of Dagbani: Nanunli, Nayahali and Tomosili (all dialects of Dagbani spoken in different geographical locations). The focus was to gather data on the topic and also to confirm some structures that had been generated. The essence of the field work was to get first hand natural data and also to get grammaticality judgements and contextual appropriateness of some sentences intended to be used as part of the analysis. Considering the fact that I am not conversant particularly with the Nanunli dialect, the fieldwork was also intended to offer me the needed opportunity to test for any possible syntactic differences that may be in any of the dialects and how to account for such disparities in my dissertation. The mode of data collection was mainly recordings of tales and administering of questionnaires. Given the fact that there were no noticeable syntactic differences detected among the three dialects, I assume that whatever generalizations are made here on the syntactic properties of wh-questions/fragment answers and related phenomena investigated in this dissertation, represent all three dialects of the language.

The writing system used in this dissertation largely draws inspirations from the writing rules (Dagbani Orthography) spelled out in an approved document that was produced in 1997. As Olawsky (1999) notes, Dagbani like most of the Ghanaian languages uses the Latin alphabet in its orthography together with some other phonetic symbols which include: ɛ, ɔ, ŋ, ɣ, ʒ. In the current orthography, Dagbani uses seven vowel phonemic symbols and twenty-seven consonantal phonemic symbols, consisting of six diagraphs and twenty-one monographs. Notwithstanding the fact that the writing system used in this dissertation is based on the existing approved orthography, there are two conditions under which I defy some of the spelling rules: (i) when they are not technically motivated and pose threat to my syntactic analysis. This includes the for instance the orthographic convention that the post-verbal particles and should be ←24 | 25→written together with the verb. This is syntactically not viable given my analysis that these particles are spell-outs of functional heads. The second condition under which I do not consider these writing rules is where it is apparent that (ii) the recommendation to use certain vowels is not phonetically motivated but supposingly recommended to create distinction among certain words which are segmentally the same.

1.4 Theoretical insights of the analysis

Though the dissertation is empirically motivated, I have also adopted various syntactic theoretical frameworks to account for certain phenomena analyzed in the work. This was found to be necessary because using theoretical concepts to account for linguistic phenomena in languages remains one of the key objectives in the study of language as a science. I have, therefore, employed various theoretical tenets to account for the empirical facts on the marking of number in wh-phrases, the distribution of wh-phrases in the derivation of wh-questions and the derivation of fragment answers. The formal accounts offered are couched within the theoretical notions of Chomsky’s (1995) idea of a feature-based account in Minimalist syntax (on accounting for the distribution of wh-phrases in the formation of wh-questions), and in the study of fragment answers, the theoretical framework adopted is Merchant’s (2001, 2004, 2008) PF-theory of ellipsis.

A fundamental claim of Chomsky’s (1995) Minimalist Programme (MP) is that there is a Language Faculty, (FL) which comprises two systems labeled the Cognitive and Performance systems. Whereas the former is responsible for the storage of information, the latter ensures that the stored information is accessed and used in the cognitive system. The theory further has it that the interactions of these two salient systems of language are reflected in the articulatory-perceptual system (A-P hereafter) and the conceptual-intentional system (C-I). This then yields the most crucial interface levels of grammar which include: the Logical Form (LF) and the Phonetic Form (PF). While the LF is connected to the C-I, PF is to AP. He further explains that whereas the PF form of a linguistic expression deals with how the structure is pronounced, the Logical Form (LF) deals with the linguistic aspect of meaning of a linguistic expression. This architecture of the FL makes feasible the correlation between articulation and semantics. It is further asserted that movement/extraction of every linguistic element is motivated by the need to check an abstract linguistic feature. A distinction is made between strong and weak features, as well as between interpretable and uninterpretable. The strong/weak distinction is based on the question as to whether features can be gotten rid off through covert/overt syntactic operations. The ←25 | 26→strong uninterpretable features have to disappear to avoid crashes in the derivation. Whereas strong features are checked off in the overt syntax, their weak counterparts are checked via covert (LF) syntactic operations. This idea shall be crucial in accounting for the ex-situ and in-situ wh-questions variation in Dagbani. Also relevant in this model of the MP is the idea that features need to enter into a checking relation in order to be licensed yielding the notion of Specifier-Head configuration in the model of Minimalism adopted for this work.

The theoretical framework adopted in the study of fragment answers in Dagbani is Merchant’s (2001, 2004, 2008) PF-theory of ellipsis. Merchant shows that the derivation of fragment answers involves two processes: A-bar movement together with PF-deletion. He demonstrates that the PF-deletion is triggered by a formal feature in the numeration called [E];-feature (Merchant 2004: 670) which licenses the deletion mechanism in fragment answers. According to Merchant, the [E]-feature has phonological, semantic and syntactic characteristics. Though I show that the PF-theory of ellipsis does not entirely account for the empirical facts of the derivation of Dagbani fragments without some challenges, in a whole it is an appealing theoretical tool in handling the empirical material presented.

In the discussion that follows, I present a brief discussion on some syntactic properties of Dagbani which are relevant to later discussion on the syntax of wh-questions. I discuss the basic clause structure, aspect, tense and focus marking (discussing both term and predicate focus).

1.5 Word order and clause structure of Dagbani

This section briefly outlines some syntactic properties of Dagbani. This is found necessary because it is assumed that such a grammatical sketch of relevant syntactic issues will be key in the interpretation and understanding of the data used1. In section 1.5.1, I present an overview of the tense, aspect, negation, and modality systems of Dagbani, whereas section 1.5.2 establishes the syntactic status of kà and ń in Dagbani. In section 1.5.3, I present a theory of focus that underpin the current study as well as a discussion on the pragmatic uses of Dagbani focus constructions and finally, section 1.5.4 is devoted to an overview of predicate focus in Dagbani.

←26 | 27→

1.5.1 The aspect and tense system of Dagbani

In this subsection, I give a brief overview of the aspect and tense systems of the Dagbani. Dagbani has an underlying SVO word order as its basic clause structure. In the canonical clause structure, the verb precedes the indirect and direct object as well as the adverbials as illustrated in (2a) and (2b).

Aspect according to Comrie (1976: 3) is defined as the “different ways of viewing the internal temporal constituency of a situation.” In the view of Bhat (1999: 43) “aspect indicates the temporal structure of an event, i.e. the way in which the event occurs in time (on-going or completed, beginning, continuing, or ending, iterative or semelfactive, etc).” Aspect as a category of verbs has been argued to be a binary linguistic feature composed of the perfective and imperfective aspect. Imperfective aspect typically denotes states or activities that cannot, or have not yet been completed, whereas perfective aspect implies that the event or action has been completed. Aspect is a crucial grammatical feature of the verb in Dagbani, since almost every verb is marked for this property. The language makes a distinction between two aspectual forms: the perfective, and the imperfective. As shown in Issah (2015a) and Hartmann & Issah (2018) whereas the imperfective aspect can be marked by [t];, [d], [n] and [ɾ], the perfective aspect is not expressed morphologically. Whereas the latter refers to a completed action/event (3a), the former refers to an ongoing action (3b).

In addition to this basic distinction in the perfective and imperfective aspectual system, the Dagbani verb further exhibits a distinction between conjoint and ←27 | 28→disjoint marking, Hartmann & Issah (2018) and Issah (2015a).2 Whereas in the conjoint form, the imperfective form is extended by the suffix -i, in the disjoint form, it is expressed by -a, see (4)3.

Following the proposal in Hartmann & Issah (2018), I contend that a glide [j]; is phonologically inserted due to the lack of an overt aspectual perfective marker in order to create a suffixed syllable of the form CV. The distinction between the perfective conjoint and disjoint is illustrated in (5a) and (5b) respectively.

In the analysis of the aspectual system of Dagbani, there are two basic assumptions: (i) suffixation is created via head movement (ii) these suffixes project the functional projection head labeled the Aspectual Phrase (AsP hereafter) that is beneath tense and above VP in the hierarchy of the Dagbani clause structure. On the syntactic status of these conjoint/disjoint markers, I adopt the proposal by Hartmann & Issah (2018: 6) who analyze the CJ/DJ markers as functional projections that are above AspP and which ‘demarcates a phrase edge in Dagbani.’ Adopting the terminology of Halpert (2012), Hartmann & Issah (2018) refer to this proposed functional projection as a Licensing Phrase (LP).

←28 | 29→

Whereas aspect is marked by suffixation, tense is marked syntactically by use of independent syntactic elements. These have generally been called preverbal particles in the Mabia literature because they generally precede the verb within the clause structure. The existence of preverbal articles has been argued to be a key property of the clause structure of the Mabia (Gur) languages of which Dagbani is one (cf. Bodomo 1993, 1994, 1997a, 1997b, 2001, 2018, Olawsky 1999), Gurenɛ (Atintono 2004, 2013), Kusaal (Abubakari 2011), Safaliba (Schaefer & Schaefer 2004) inter alia. These particles are used to express several grammatical functions.

The preverbal particles specify time reference points of verbs in the past: ‘yesterday’, dàà (remote past) ‘two or more days ago’, ‘earlier today’, in the present: yòlí ‘just’, ‘still, yet’, in the habitual: ‘normally’, or in the future: and its negative form . Though superficially, they seem to be adverbials, in Hartmann & Issah (2018), counter arguments against their analysis as adverbs are outlined including: distributional difference between adverbials and time depth markers, and the syntactic fact that the two can co-occur within the sentence structure. For details on this, readers may contact Hartmann & Issah (2018) and references cited therein. The use of these time depth (tense markers) is illustrated in the contrast in interpretation of the (a) and (b) versions of (6) and (7).

Some syntactic characteristics of the T element which also serve as counterevidence for its analysis as adverbial include the fact that (i) it cannot be questioned, (ii) it cannot be tropicalized/focused, and finally (iii) it is not possible to coordinate it.

In accounting for the derivation of tense, I take inspiration from languages like English where the sentence is argued to be headed by T as outlined in ←29 | 30→Adger (2003). In addition, I assume that the Dagbani time depth markers (tense particles) are also in T head. Thus, although the subject is base-generated at vP, it is A-moved to Spec, TP because of an EPP feature on T that needs to be fulfilled. Adopting a proposal made in Hartmann & Issah (2018), I contend that AspP and TP are functional projections that are higher up in the structure, above VP. As they assume, the V moves out to Asp whereas T hosts the temporal markers which are analyzed as tense markers. In the same vein, Hartmann & Issah (2018) propose the realization of the functional heads that code the Licensing Phrase, adopting the proposal of Halpert (2012). The proposed structure in (8b) accounts for the fact that CJ/DJ projects a functional phrase just as the aspectual markers based on the account of Hartmann & Issah (2018).

Negation is marked using the particles and for the non-future and future tense, respectively. Though there are other verbs which are used to code negation (negative verbs), my focus here is on the preverbal elements. The negation particle always follows the subject and precedes the verb. However, when there is a tense head within the clause structure, the negation marker can only follow the tense marker and precedes the verb, in that strict order (Issah in prep.)

←30 | 31→

Following postulations in Issah (in prep.), I assume that the negation particles are heads of a functional negation phrase (NegP). The NegP c-commands the VP and is c-commanded by any other functional projection in the inflectional domain.

Mood, just as tense and negation is not expressed morphologically (dedicated verb form), but rather syntactically via the use of special syntactic elements. For instance, is used to express conditions and occurs before the VP within the clause. In Dagbani, mood can be expressed in the domains of necessity, possibility/conditionals, and desirability. Below, I illustrate this using the conditional mood as shown in the contrast in (10a) and (10b).

The particles that code mood are analyzed heads of a functional category, modality phrase (ModP). Given the empirical fact presented in (10c) where the modal particle immediately precedes the subject, I propose that modality ←31 | 32→phrase is projected higher up in the clause structure than the tense, aspect and negation and the event phrasal categories.

Given the syntactic assumptions that tense, aspect, negation and mood are phrasal categories and project functional categories, and the proposal that the CJ/DJ alternation also projects the Licensing Phrase, adopting the proposal of Halpert (2012), the proposed clause structure of Dagbani is given in (11) considering all these functional heads within the hierarchy.

With the linear order of the clause structure proposed in (11), the assumption is that the highest projected functional layer is the TP which c-commands all other functional projections.

To summarize, the section has provided the reader with the clause structure of Dagbani which is shown to be made up of several functional elements proposed to be phrasal categories including tense, aspect, negation, and mood. In the section that follows, I take a brief overview of the syntactic status of and ń, which are salient syntactic elements in the derivation of ex-situ wh-questions as I show later in this dissertation.

1.5.2 On the syntactic status of and ń in Dagbani

Considering the fact that the particles and ń or its phonological variants /ḿ) are very salient in the syntactic characterization of wh-questions (particularly in the ex-situ strategy), it is crucial that we establish their exact syntactic status since that would be decisive in determining the clause structure to be assigned the ex-situ wh-questions. Following the proposals of earlier research findings on Dagbani focus constructions (cf. Fiedler 2007, Fiedler & Schwarz 2005, Fiedler et al. 2010, Hudu 2006, 2012, Issah 2008, 2012, 2013b, Schwarz & Fiedler 2007), I refer to these particles as focus markers. This claim is supported by question-answer pairs as a test for focus on syntactic elements. The question-answer pairs as a standard diagnostic test for determining focus elements in natural languages is well known in the literature on information structure (cf. Aboh 2007, Ameka 1992, 2010, Dik 1978, Krifka 2004, 2007 inter alia). For instance, Ameka (1992: 5) claims that “a felicitous answer to a content question would be a focused constituent since it would provide information that would be a substitute for the interrogative word.” Boadi (1974) corroborates this assumption of Ameka when he also asserts that focus constructions are answers to an interrogative word fronting construction in a question-answer pair. In the data that follow, I use the question-answer pair, questioning both arguments and non-arguments in the ←32 | 33→clause structure. I consider subject arguments (12), object arguments (13) and adjuncts (14).

Three descriptive generalizations that can be made based on the question-answer pair data presented in (12) through (14) are: (i) the particles and ń are overtly expressed both in the ex-situ wh-questions that trigger the focus context, as well as in the corresponding ex-situ answer provided. Whereas in the questions the particles follow the wh-phrases, in the answers, they follow the syntactic elements that substitute for the wh-phrases, and (ii) with the exception of the subject question in (12), object arguments and adjuncts have two forms of answers ←33 | 34→(which I shall call in-situ and ex-situ forms throughout the rest of the dissertation) and (iii) the distribution of the particles and ń seems to be regulated by some considerations since they are in complementary distribution. Each of these aforementioned issues shall be attended to in the rest of the dissertation.

My syntactic account of and ń is that they are focus markers that head a projected functional focus phrase (FocP) within the clausal left periphery. Thus, constituents that have these particles immediately following them are A-bar moved from the canonical positions to the Spec, FocP for pragmatic/semantic reasons. This movement is proposed to be triggered by the need to check an uninterpretable focus feature (Chomsky 1995, Kiss 1998) that the extracted syntactic elements carry and the syntactic requirement for and ń to be expressed overtly so as to create the Spec-Head structural configuration required for feature checking (see Chomsky 1995). A proposed structural representation of the focus construction is shown in (15) based on Issah (2012) and Issah & Hartmann (2018).

As assumed for focus markers in other African languages (see Hartmann & Zimmermann 2012) for the Bura copula focus marker án, conclusions that can be drawn about these two syntactic elements: and ń are as follow: (i) their overt presence results in a structural bi-partition of a clause into a focus part (invariably to the left) and the backgrounded/presupposed component (which is to the right), (ii) the particles induce the (contrastive) focus interpretation into clauses within which they are found. In conclusion, the first function is syntactic whereas the second is pragmatic/semantic. These two notions associated with these focus markers shall later be relevant to our discussion on the derivation of ex-situ wh-questions. For now, I do not comment on the inventory of the focus heads and their distribution. This is given a systematic discussion in chapter three, section 3.3.3 where I offer a novel account of what determines the spell-out of these functional heads in the derivation of ex-situ wh-questions departing ←34 | 35→from the so-called subject/non-subject asymmetry. Having proposed that the particles and ń are focus heads, in the next section, I present a theory of focus that underpins the current research and the pragmatic uses of focus.

1.5.3 A theory of focus and pragmatic uses of focus constructions

Given the fact that the notion of focus is crucial to later discussions in this dissertation, this subsection outlines a theory of focus that underpins this current research work. In line with proposals in the study of focus constructions (Aboh 2004, 2007, Ameka 1992, 2010) among several others, I take the notion of focus to be a universal category of information structure (IS). Although the information structural notion of focus has been defined differently by varied linguists, in this work, I adopt the definition of Rooth (1985) who asserts that focus on a constituent α ([α]F) invokes a set A of alternatives to α, indicating that members of A are under consideration. Rooth (1985) also asserts that focus may have varied pragmatic functions depending on the interaction of α with other alternatives in a given discourse. For instance, it is asserted in the literature e.g. Ameka (1992), Hartmann & Zimmermann (2012), Rooth (1985) and Zimmermann & Onea (2011) inter alia that some of the pragmatic uses of focus include: new information, corrections, confirmations, selective and contrastive foci usages. According to Zimmermann & Onea (2011: 1663), a focus constituent X expresses new-information if α introduces an element of A into the common ground, and if the alternatives to α have not been explicitly introduced in the preceding discourse, as in (16a). They further opine that a focus constituent X is used correctively if α competes with one or more elements of A for introduction in the Common Ground, where α’s competitors have been explicitly mentioned in the preceding discourse (16b). They are of the view that the idea of Common Ground (CG) is understood as the mutually shared knowledge of the interlocutors in a discourse in line with the proposal of Stalnaker (1973), among others. In addition, Zimmermann & Onea (2011: 1663) are of the view that a focus constituent X is said to have been used selectively if α introduces an element of A into the common ground, and α is chosen from a restricted subset of A, the members of which have been explicitly mentioned in the preceding context (16c). Finally, Zimmermann & Onea (2011: 1663) assert that a focus constituent X is used contrastively if α is juxtaposed to one or more elements of A that are denoted by constituents Y, Z, …in the preceding discourse, where Y, Z,…are of the same syntactic category and denote into the same semantic word field as X (16d).

←35 | 36→

It is worthy of note that focus constructions in Dagbani also have varied pragmatic uses including new information focus (17) selective focus (18), corrective focus (19) and contrastive focus (20).

The fact that both (17A1) and (17A2) are felicitous as answers to the question in (17Q), I will assume that the particle is compatible with in-situ focus in Dagbani. The in-situ focus stratety, I argue in this dissertation induces new information focus. The function and distribution of is explored further in section 1.5.4 on my discussion of predicate focus.

←36 | 37→

The data in (17) illustrate new information focus. It is worthy to point out that this use of focus can be coded in both the ex-situ strategy (17A1) and the in-situ strategy as well (17A2). The availability of two focus positions in Dagbani is accounted for by assuming that whereas in-situ presents just new information, the ex-situ strategy is dedicated for contrastive information. This will be given a more systematic discussion in chapter six. In (18) we have an illustration of selective focus. Whereas (18a) introduces the two elements nìmdí ‘meat’ and shínkááfà ‘rice’ into the CG, in the following sentence in (18b), one of the introduced elements nímdì ‘meat’ is chosen/selected. In this context too, we observe that the only strategy that is available is the ex-situ strategy, confirming my analysis for Dagbani, that the in-situ position is dedicated for new information focus. In (19) and (20), I exemplify corrective and contrastive foci respectively. One key characterization is that in all the varied pragmatic usage of focus, only the new information focus (17) seems to be compatible with the in-situ strategy.

In Dagbani, I demonstrate that the relevant properties of focus are primarily morphosyntactic, at least with available empirical evidence on term and predicate foci. Adopting the idea of a theory of focus by Jackendoff (1972), I contend that the syntactic correlate of focus is F-marking and that invariably, syntactic nodes can be freely annotated with a privative syntactic feature F(ocus). Focus in Dagbani will be demonstrated to be reflected in the morphosyntactic characterization of the language, a property that has been established in the literature to be phenomenal in most (West) African languages.

To sum up, this subsection presented a theory of focus that will be used for the analysis in this dissertation. Also, the particles kà and ń are analyzed as focus markers. A theory of focus was also presented as well as the pragmatic uses of focus in Dagbani. I showed that focus could be used to code varied uses including the realization of new information focus, selective focus corrective focus and contrastive focus. Knowledge of the syntactic status of these two ←37 | 38→particles is key to our understanding of the structure of wh-questions in subsequent chapters. In the next subsection, I present an overview of predicate focus marking in Dagbani.

1.5.4 Predicate focus in Dagbani

In the previous section, I outlined the marking of term focus on adjuncts, and subject and object arguments. This section gives an overview of the predicate focus system of Dagbani. I demonstrate here that focusing of the Verb or the Verb Phrase, (V and VP hereafter) cannot be marked using the ex-situ focus strategies as discussed for subject and non-subject argument constituents in Section 1.5.3. Rather narrow focus on V and focus on VP is always realized in-situ (the canonical positions) in Dagbani. The claim is made that narrow focus on V and focus on VP is marked using the particles and for Vfoc and VPfoc respectively. This is illustrated in (21) using the question-answer pair strategy where (A1) is for V focus whereas (A2) illustrates VP focus.

The illicitness of (21A3) and (21A4) is interpreted to mean that the fronting of predicates for focus marking is disallowed in Dagbani. The ungrammaticality of (21A4) suggests that the focusing of fronted nominalized verb is not allowed ←38 | 39→once a copy of the nominalized verb is left within the TP internal base position of the verb. This is in contrast to the grammaticality of (21A5) where fronting of a focused nominalized verb with a modal verb in the TP internal structure is possible because a copy of the nominalized verb is not spelt-out within the base position of the verb. Note that although nominalized verbs seem to be allowed for fronting in the coding of focus, it is only in the context of modal verbs suggesting that further investigation is required on this in Dagbani. The ban on the focusing of predicates within the left periphery of the clause is not discussed further here since it is given a more systematic attention later in section 6.3. Thus the main empirical observation is that narrow VP and V focus are licensed in just one environment, the in-situ position, a position that is associated with new information focus in this dissertation.

It is worthy of note that the particle can also occur in a context in which only the DP complement of the verb is marked for new information focus (22A1). This is taken to mean that the particle also correlates with in-situ object and adjunct focus. This is, however, not valid for the particle which does not mark focus on full DP focus (A3), as exemplified in the question-answer pairs in (22).

The account of the function of as in the context of (22) is that it marks in-situ (new information) on the NP, kàwáná ‘maize.’ In non-subject ex-situ wh-questions, it is always obligatory when an in-situ answer is given. This is discussed in chapter six on the structure of answers to wh-questions.

←39 | 40→

Having shown that there is empirical motivation to argue that the use of these particles is pragmatically triggered, I take a closer look at their distributional properties. The post-verbal particle requires an obligatory overt DP complement or adjunct (23a, 23b) and so its occurrence clause finally yields ill-formed sentences.

In Gurenɛ, another Mabia (Gur) language spoken in the Upper East Region of Ghana, Dakubu (2000) identifies a phonologically and syntactically similar particle and proposes that:

when la follows the verb, it marks Focus on the entire predicate, that is, it asserts the concreteness or factivity of the VP-the verb together with its Complement. --- it never occurs with an intransitive verb or a verb whose Complement (which may be an NP, a pronoun, a locative NP or an entire clause is not expressed. (Dakubu 2000: 61)

Also, Bodomo (1997a) and Hiraiwa & Bodomo (2008) make a similar observation for Dagaare, a genetically related Mabia language, and conclude that almost every main (declarative) clause in Dagaare has the focus particle la, and that the particle requires an overt DP/NP complement or adjunct.

Given the assumption that marks VP-focus, or DP/adjunct focus in the in-situ position as so far shown, the expectation should be that it should with incompatible with subject focus. This fact is borne out considering the contextual inappropriateness of the sentence (24A2) as an answer to (24Q) which requires a subject focus.

←40 | 41→

Pursuing an analysis according to which particle only marks focus within the VP, complement of the verb or adjuncts, I assume that the subject is located within the Spec, TP, for which reason cannot mark focus on the subject. This explains why the sentence in (24A2) is contextually inappropriate. The sentence that is felicitous in this context where the focused constituent is the subject Napari, is (24A1).6

Following from the discussion so far on the contexts in which the occurrence of and is felicitous, and the proposal that they code new information focus, the schema in (25) is proposed to represent their distribution.

Given the current analysis which seeks to suggest that and are predicate focus markers and require either the verb or the VP to be focal, it means that there is evidence to argue that Dagbani has a lower focus projection within the clause structure.

Notice that although full DPs are incompatible with in the clause structure of Dagbani as illustrated in (21A1), weak object pronouns do occur with it, even though they can only precede the particle. They occur between the verb and the particle (26A1). When the weak pronominal element follows , rather than being located in between the and the verb, the resulting structure is illicit as illustrated by the question answer pair in (26A2). Also, when emphatic pronouns occur to the left of the particle, the resulting structure is also illicit (26A3).

←41 | 42→

Just as discussed on the V-focus marker in Dagbani and its interaction with weak object pronouns, similar structural fact holds for the particle in terms of its occurrence with weak pronouns. It was previously demonstrated that the post-verbal requires an obligatory DP complement or an adjunct to follow it. Notwithstanding this fact, in cases that the DP complement is a weak pronoun, it must precede . This is shown in the question-answer pairs in (27) and (28). The illicitness of (27A2) versus grammaticality of (27A3) further shows the difference in the syntax of when it occurs with full DP objects (27A3) and weak pronoun objects (27A2).

←42 |

From the data in (27), it is evident that the weak pronouns and full DPs have different syntactic relations with the post-verbal lá. It is worthy of mention that emphatic pronouns do not precede the particle as exemplified by the illicitness of the sentences in (28).

From the aforementioned facts on the distribution of weak pronouns on one hand, and emphatic and full DPs on the other and their interaction with the focus heads and , the following descriptive generalizations/summaries are made about the syntax of and in (29).

Based on the empirical facts discussed above on the distribution of these particles, I propose the structural representation in (30) in an attempt to offer a unified account of the distribution of these particles.

←43 |

To account for the derivation of the syntax of weak pronouns and the particles and , I assume that the weak object pronouns are clitics to the verb.7 With this assumption, I contend that weak pronouns invariably incorporate into the verb and move higher up to the AspP position. Thus, verb together with the weak pronoun raises to a higher functional projection. The cliticized pronoun is, therefore, assumed to undergo successive cyclic movement to Foc and then finally lands at Asp. The assumption that the verb movement proceeds through the focus head raises a formal problem since the focus head position is filled by the low focus marker. In addressing this formal challenge, it is assumed that the verb has to excorporate when moving on to Asp.

To conclude this section, it has been argued that unlike term focus which can be coded using the ex-situ strategy, predicate focus only employs the in-situ strategy. This explains why narrow focus on V and focus on VP is always realized in in-situ. I also showed that aside the marking predicate focus, it also correlates with in-situ object and adjunct focus marking. It was further demonstrated that when weak pronouns occur with the and particles, they are clicticized to the verb and move cyclically to AspP via Foc.

1.6 The structure of the dissertation

This dissertation is structured into seven chapters. In chapter two, I discuss the inventory and internal structure of wh-phrases which are analyzed as question operators in Dagbani. This chapter also provides empirical evidence to show that the inventory of wh-phrases proposed in earlier works is not exhaustive and therefore, motivates a novel inventory of the Dagbani wh-phrases. One of the central aspects of this chapter is that, aside the typological discussion on wh-phrases, it also offers a theoretical account of number as a functional projection within the DP with special focus on suppletion. Finally, I further provide structural arguments to show that there is a distinction between wh-phrases that serve as pronouns and those that are determiners.

Chapter three addresses in detail the formation of wh-questions with focus on the ex-situ strategy. I demonstrate that all categories of wh-operators (both subjects and non-subject) are compatible with this strategy of forming ←44 | 45→wh-questions. I further show that the ex-situ strategy of forming wh-questions is characterized by two main processes: (i) A-bar moving the wh-phrase from the canonical position to the clausal left periphery and (ii) mandatorily introducing the focus heads and ń, which must immediately follow the extracted wh-phrase. I argue that in overt wh-movement, the extracted wh-phrase mandatorily c-commands an overtly focus head in the formation of ex-situ wh-questions otherwise the derivation will crash. The parallel distribution between of ex-situ focused constituents and extracted wh-phrases is used as an argument to support an analysis according to which wh-movement is an instance of focus movement. In addition, I show that Dagbani exhibits an asymmetric pattern in the extraction of wh-phrases regarding the presence of traces/resumptive pronouns as reflexes of movement. For instance whereas movement of a wh-phrase from an embedded subject position mandatorily requires a resumptive pronoun, that of matrix subject gives rise to a trace. I account for this asymmetry in the extraction out of the local and embedded subject positions. The distribution of the two focus heads and ń is also investigated. In contrast to previous studies on Dagbani focus constructions that suggest that the distribution of the two focus heads and ń is regulated by the syntactic category of the extracted wh-phrase, I show that the earlier proposal is not viable and offer a novel account which is supported empirically.

In chapter four, I provide a syntactic analysis of wh-in-situ questions in Dagbani. I show that wh-in-situ questions are interpretable as wh-questions and not as echo-questions. I demonstrate that unlike the ex-situ strategy, the in-situ strategy is not compatible with all categories of wh-operators. For instance, I show that whereas non-subject wh-phrases can occur in the in-situ positions, their subject counterparts are disallowed from occurring in the in-situ position. Beside these descriptive accounts, I propose a minimalist analysis following Chomsky (1995) and Huang (1982), that although the in-situ wh-questions appear superficially to be in the canonical base position, they undergo covert (LF) movement at the post syntactic level for their semantic interpretation. The movement analysis is supported using the fact that wh-extraction is constrained by island effects (Ross 1967). I also offer an analysis of the ban on in-situ subject wh-questions. Finally the chapter also provides counterarguments for the proposal in previous studies that in-situ wh-questions mark contrastive focus using the particle lá.

In chapter five, empirical arguments in favor of the movement analysis proposed for ex-situ and in-situ wh-questions in chapters three and four are presented. Given the fact that wh-questions are argued to involve movement, one diagnostic tool that is used to motivate this kind of syntactic account is locality ←45 | 46→constraints. I, therefore, argue that if we assume a movement analysis in the derivation of wh-questions, then it should be backed by empirical evidence. I present the ban on movement out of relative clauses, coordinate structures and complex NP structures (syntactic domains which are typically assumed to be islands for movement), as language internal arguments for the movement account. I further show that employing the resumption strategy does not repair island violations in any of these syntactic configurations. However, for the coordinate structure, there is an exception in which movement is allowed, (that is Across-the-Board phenomena (ATB)) where movement of a constituent out of all the conjuncts of a coordinate structure is assumed to apply at once. I thus, conclude that since both the ex-situ wh-in-situ questions cannot evade island violations, their derivation can be accounted for via a movement analysis.

Chapter six provides an account of the structure of answers to wh-questions and syntactic derivation of fragments. The chapter shows that although it is possible for wh-questions and their congruent answers to share structural parallelisms, such a correlation is not a strict condition for licensing congruency. On the syntactic derivation of fragments, I propose that their derivation involves two main processes: A-bar movement together with an elliptical phenomenon following the PF theory of ellipsis by Merchant (2001, 2004) among others. This movement plus ellipsis is motivated by some empirical evidence drawn from the works of Merchant (2001, 2004), among others. Topics also discussed in this chapter include: asymmetries in fragments and their semantic recoverability, the motivation for the fact that fragment answers are sensitive to locality constraints. The idea of PF-deletion and the ellipsis feature is also discussed in this chapter.

Finally, in chapter seven, I conclude the dissertation by way of summarizing all the preceding chapters. It also outlines the contributions/significance of the dissertation and highlight potential areas for future research.

1.7 Interim summary

In conclusion, the goal of this chapter was to provide the reader with some background information on the genetic, geographical and basic grammatical properties of Dagbani. It was demonstrated that Dagbani is tonal language with a two phonemic tone system and has an SVO basic word order. The clause structure of the language has varied functional elements projecting various functional phrasal categories including tense, aspect, negation, the conjoint/disjoint paradigm and mood. I also offered a discussion on a theory of focus within which the analysis on focus is cast, the focus system of the language and the strategies employed to mark term and predicate focus, showing that whereas the former ←46 | 47→involves movement, the latter is marked in the in-situ positions. The organization of the dissertation, the source of the data and orthography used in the Dagbani data presentation and the theoretical insights of the analysis were also presented. The syntactic status of and ń which are salient for later discussion in the dissertation was also discussed. I established based on their contexts of use that these particles are analyzable as focus markers which are analyzed as heads of a functional focus phrase projected in the clausal left periphery. Focus fronting in Dagbani however, is not associated mainly with exhaustive or contrastive interpretation since new information focus can also be coded via left peripheral movement. This idea of the morphosyntax, semantics and pragmatics of focus will be pursued furthers in chapters three and four. The next chapter lays the foundation of our investigation of the syntax of wh-questions by taking a close look at the inventory, internal structure and other grammatical properties of Dagbani wh-questions, which are argued to constitute an idiosyncratic syntactic property of wh-questions.

←47 | 48→←48 | 49→

1 Given the fact that most of the syntactic issues discussed are based on previous research works, readers are referred to consult those works and references cited therein for in-depth discussions.

2 Saanchi (2003) makes a similar observation for Dagaare, where the aspectual marker determines the possibility or otherwise of some overt material following the verb.

3 For a formal account of the conjoint and disjoint verb form in Dagbani, I refer readers to Hartmann & Issah (2018), whereas Issah (2015a) offers a descriptive overview of the phenomenon.

4 The difference in verb morphology of bó-r-í and bó-r-á ‘want’ here is attributable to the proposed disjoint/conjoint alternation in the Dagbani verb form which was introduced briefly under the discussion on aspect marking. For details readers should see Issah (2015b) for a descriptive overview of the phenomenon and Issah & Hartmann (2018) for a detailed formal analysis.

5 The particle is discussed more systematically in Hudu (2012) and Issah (2008, 2013).

6 Hudu (2012) seeks to analyze the particle as a contrastive focus marker. On my discussion of in-situ wh-questions in chapter four, I present counter evidence to this analysis of as a contrastive focus marker. For detailed discussion of counterarguments against la as an aspectual marker in Dagbani, readers are referred to Hudu (2012) and Issah (2008, 2013).

7 I take notice of the fact that in a genetically related language such as Dagaare, a similar observation is accounted for using the notion of object shift by Hiraiwa and Bodomo (2008:249–250). My current account of the syntax of weak pronouns and post-verbal particles and using the clicticization approach is just one of choice and not one approach being better than the other.

Chapter two The grammar of the Dagbani interrogative DP

2.0 Introduction

The goal of this chapter is to present a systematic account of the interrogative Determiner Phrase (DP) of Dagbani. Given the fact that wh-phrases are core elements of Dagbani wh-questions, a thorough investigation of their inventory, internal structure, morpho-syntactic characterizations and other salient grammatical properties is essential. Additionally, knowledge of the grammar of wh-phrases will be crucial to understanding the discussion of wh-questions in the following chapters. The grammatical properties the wh-phrases discussed include number marking, the distinction between human/nonhuman and lexical ambiguities. I further show that [±] animate is not relevant in pronominal wh-phrases. The chapter also reviews earlier works on the topic including Issah (2015a) and Olawsky (1999). Olawsky’s inventory has gaps (because he misses some wh-phrases) and misinterprets one form (ŋùn). Issah (2015a) also fails to provide an exhaustive list. I further provide an alternative analysis where I offer an exhaustive list of the wh-phrases. Additionally, I clarify the use/meaning of some wh-phrases that were misinterpreted in the previous studies.

My contribution in this chapter is both empirical and theoretical. From the empirical perspective, I offer a full list of wh-phrases in Dagbani, indicating gaps in the previous literature on this topic and offering alternative analyses which are supported empirically. From a theoretical perspective, I offer a structural analysis of the interrogative DP, proposing the projection of a functional number phrase (NumP) based on the characterization of the wh-phrases. I also offer a theoretical account of suppletion which is demonstrated to be present in number marking of at least, one of the wh-phrases.

The chapter is structured as follows: Section 2.1 introduces the background information on wh-questions in languages and their role in the formation of wh-questions. Crucial in the discussion in this section is the claim that the wh-phrases which are indeed always obligatory in the formation of wh-questions are the syntactic elements that contribute an interrogative interpretation of the clauses in which they appear as wh-questions. In section 2.2, I review some previous studies on Dagbani wh-phrases unravelling some shortcomings and offering alternative proposals thereof. Section 2.3 discusses some grammatical properties of Dagbani wh-phrases including: the semantic distinction between ←49 | 50→[±human], number marking and ambiguity. Conclusions and a summary of the chapter are given in section 2.4.

2.1 Wh-phrases as question operators in wh-questions

Wh-phrases have been established to be a salient syntactic element that induce interrogative force of wh-questions in natural languages (Cheng 1997, Chomsky 1995, Huang 1982 inter alia). According to Siemund (2001), Cysouw (2004), among others, the wh- in the name of these questions is motivated by the fact that most of the words that introduce this category of questions in the English language are signaled by wh as in what, which, who/whom, when, where, and why. Other writers such as Ameka (1992, 2010) and Torrence & Kandybowicz (2015) also use expressions like question words and content questions to refer to wh-words and wh-questions respectively. Boadi (1990) on the other hand, uses the expression question word/phrase to refer to this group of words in his discussion of wh-questions in Akan, while Siemund (2001) refers to this type of questions as constituent interrogatives and words that signal them as interrogative phrases. Cysouw (2004: 3) argues that, although there are quite a number of languages which have predictable morphological forms for their wh-words, as with the English wh, this could not be assumed to be universal since a good number of languages equally deviate from such a morphological regularity.

However, as I will show in the next section, there is no particular morphological form dedicated to this class of words/phrases in Dagbani. Nevertheless, I use the terms wh-phrases to refer to the question operators and wh-questions for the questions that are formed. The motivation for the use of these rather Anglo-centric terms is the fact that they have been used for languages other than Dagbani in which the various question operators do not display the kind of similarity in form present in English. Thus, whereas answers to wh-questions are usually analyzable as the set of all propositions, wh-phrases are interpreted as interrogative operators which have existential quantification (like indefinite DPs) in the linguistics literature (cf. Polleto 2005 inter alia). However, unlike other languages like English where these question operators can be used to express a whole range of other operator functions (relative operators) (Poletto 2005), Dagbani makes use of distinct class of operators as relativizers as evident in discussion in chapter 5.

Discussing the strategies that are employed for formation of wh-questions, Siemund (2001: 1018) claims that cross-linguistically, natural languages employ different devices for the marking of wh-questions. Some of the strategies include the use of intonation, question particles and inversion. Siemund further argues ←50 | 51→that out of the varied strategies identified cross-linguistically, the most prevalent device is the use of interrogative phrases, while the least common is the use of intonation. He argues that the latter is such insignificant a device in forming wh-questions that, even when it is present in a language, it is available only as an optional strategy. Siemund (2001: 1018) further asserts that although all natural languages can be argued to have interrogative phrases, their inventory as well as semantic distinction is a language-specific property. A similar view is expressed by Dryer (2005), who demonstrates that all natural languages, whether spoken or signed, have a set of wh-words which are employed to indicate constituent interrogatives. Like Siemund, Dryer also holds that the inventory of these wh-words is a language-internal property.

According to Siemund (2001: 1018), interrogative phrases are: “analyzed as placeholders or variables in a proposition to be filled or assigned a value by an answer.” Ultan (1978: 228–229) also holds the view that: “interrogative phrases are characteristic of all languages, that is, all languages have interrogative substitutes for nouns and a number of adverb-like words or phrases expressive of locative, temporal, enumerative, manner, purpose and other functions.” Siemund (2001: 1018) is also of the view that in constituent interrogatives, speakers generally “expect information that allows them to complete the interpretation of a proposition.” This probably explains why König & Siemund (2007: 291) argue that constituent interrogatives “receive answers that provide the kind of information specified by the interrogative phrase.”

Within the theoretical tradition of Minimalism (Chomsky 1995 and Hornstein et al. 2005: 258 inter alia), wh-phrases are interrogative operators. According to the authors, they are called operators because they give rise to A-bar movement dependencies and have a property of binding a variable. They further assert that any expression that contains these wh-phrases are referred to as ‘operator expression.’ According to Siemund (2001), wh-phrases are generally substitutes for various syntactic categories, including those with argument and non-argument functions. He argues that those wh-phrases that substitute for arguments are generally specified for subject and object functions whereas those substituting for non-argument functions are those that seek adverb-like information such as (i) location (ii) temporal setting/time (iii) manner and finally (iv) reason.

In the domain of Minimalist syntax, when I talk of a subject, I mean an argument that is generated in the specifier of vP (Spec, vP) and is A-moved to the Spec, TP to fulfil the grammatical feature of Extended Projection Principle (EPP) on T as proposed by Chomsky (1981). This grammatical property stipulates that every clause should have a subject. On the other hand, the object constituent is ←51 | 52→an internal argument, generally a DP/NP which is selected by the verb (a complement to the verb) whereas the adjunct is not an argument within the structure, given the fact that it is not selected by the verb. In the structure that follows, I give an internal structure for the proposal made on the distinction between arguments and adjuncts and their syntactic position within the basic clause structure.

In the data in (2) through (4), I illustrate how each of these syntactic elements are substituted for by wh-phrases using question/answer pairs. Accordingly, in (2) the wh-phrase ŋùní ‘who’ substitutes for the subject of the sentence Chentiwuni as evident in the corresponding felicitous answer in (2A1). In the same vein, in the question in (3), the wh-phrase ‘what’ is substituting for the object argument yílí ‘house’ as in the corresponding answers we have in (3A1) and (3A2). Finally in (4), bòndálí, ‘when’ is a substitute for an adjunct which is what is demonstrated in (4Q) and (4A1) where this wh-phrase is replaced with sóhàlá ‘yesterday.’

←52 |

It should be noticed that the particles and ń occur in both ex-situ wh-questions and their corresponding answers. This requirement for overtly expressed focus heads and ń is only in the ex-situ wh-questions and their corresponding answers but not in their in-situ counterparts as in (3A1) versus (3A2) and (4A1) vs (4A2).8 It should be noted that in the derivation of non-subject wh-questions, the wh-phrase can either be moved overtly/covertly to the clausal left periphery creating what has been termed as ex-situ and in-situ strategies. These two strategies are discussed in chapters three and four respectively. Readers should recall my proposal in the preceding chapter that whereas ká/ń encode focus within the functional layer in the left periphery, is a lower focus marker. These two markers are however, mutually exclusive in their distribution as evident in the illicit sentences in (3A3) and (4A3). The distributional asymmetry between subject and non-subject wh-phrases is discussed in detail in chapters three and four.

←53 | 54→

To sum up, it has so far been demonstrated that wh-phrases constitute an essential linguistic property of wh-questions in Dagbani. Their morphological presence is the primary device that signifies a particular structure as being interpreted as a wh-question since they primarily are question operators.9 In the section that follows, I give a systematic account of the inventory, internal structure and grammatical properties of Dagbani wh-phrases. I begin with a review of the previous studies of the semantic domains of the wh-phrases in section 2.2, followed by a novel account on their inventory.

2.2 Previous studies of Dagbani wh-phrases

To the best of my knowledge there is no known work that is entirely dedicated to a thorough investigation of the internal structure and grammatical characteristics of Dagbani wh- phrases. However, there are at least some researchers who have carried out some studies on aspects of the wh-phrases. Notable among some works that have discussed the inventory of wh- phrases include Issah (2015a) and Olawsky (1999). Both analyses, are however, not without some shortcomings. In this subsection, I investigate some of the shortcomings of the previous studies and offer alternative analysis.

Olawsky (1999) outlines seven distinct semantic categories of wh- phrases for Dagbani as shown in tab. 1. The data are adapted with some modification particularly the adding of the tonal marks since tone is not marked in Olawsky’s work.

Tab. 1: Dagbani wh-phrases according to Olawsky (1999: 71)

Interrogative phrases

Phonetic transcription











<wúlà >






< ŋùn>



<bɔ̀zùɣú >



a Olawsky’s representation of the nonhuman wh-phrases as bɔ̀ rather than is motivated by the prescription of the Dagbani Orthography Committee of 1977: 24 that ‘words that are not nouns or verbs have final/ɔ/’. In my representation therefore, I defy this orthographic convention considering that fact that it has no scientific motivation and has the potential of complicating linguistic analysis. Also phonetically, native speakers never realize the vowel quality as/ɔ/but rather/o/.

←54 | 55→

There is not only a failure to present an exhaustive list of the semantic domains of wh-phrases in Olawsky’s work, but also potentially interesting issues on wh-phrases that are not accounted for including: a lack of a systematic investigation into the morphological (internal) composition of the wh- phrases and a discussion on the grammatical properties of these wh-phrases. In the discussion that follows, I give systematic account of each of these shortcomings identified in Olawsky’s (1999) characterization of Dagbani wh-phrases.

As briefly pointed out above, one of the shortcomings of Olawsky’s study is that he does not provide an exhaustive list of the semantic distinctions/categories of the wh-phrases in Dagbani. He for instance, identifies only seven semantic distinctions of the wh-system as outlined in tab. 1, while ignoring the wh-phrases within the semantic domain of time, bòndálí ‘when’ quantity NP+álá ‘how many’ and cost/amount álá ‘how much’. However, these semantic distinctions, which are not accounted for in the inventory of Olawsky, do exist in the wh-phrases system of the language. This means that the inventory of wh-phrases presented by Olawsky (1999) is not exhaustive.

In the discussions of the inventory of the wh- phrases, I offer an alternative inventory of the wh-phrases according to which the language has ten distinct wh-phrases, eight of which are analyzed as wh-pronouns plus two wh-determiners. This, I contend, captures the full wh-phrase system of Dagbani. In addition, Olawsky misidentifies the wh- phrase that signifies possession by glossing ŋùn as ‘whose.’ However, the notion of ‘whose’ is not expressed using the monomorphemic wh-phrase ŋùn as Olawsky proposes. Contrary to his claim, the wh-phrase that expresses the notion of ‘whose’ uses an analyzable wh-phrase comprising the wh-phrase ŋùní+NP where the NP invariably refers to the possessed while the ŋùní ‘who’ substitutes for the possessor. This fact is crucial in my proposing a distinction between wh-pronouns and wh-modifiers/determiners.

In isolation therefore, it is factually not a promising analysis to assign ŋùn the semantic domain of possession ‘whose’ as done in Olawsky’s work. Not only is the semantic domain wrong, but also, ŋùn is just a phonological variant of ŋùní realized in fast speech where the front high vowel/í/is elided. The ungrammatical sentence in (5b) buttresses the argument that ŋùn does not denote ‘whose’ as the discussion of Olawsky (1999) seeks to portray.

←55 | 56→

Issah (2015a) also discusses aspects of wh-phrases in Dagbani, though not without some shortcomings. Issah’s classification of the wh-phrases, just like Olawky’s work, is also characterized by a failure to account for all the semantic distinctions in the language. For instance, his study excludes the wh-phrase for quantity and glosses álá ‘how much’ as being ambiguous denoting both cost and quantity. However, though álá admittedly denotes the concept of ‘how much’, it does not have multiple interpretation and so differs from the semantic category of quantity ‘how many’. I demonstrate that the concept of ‘how many’ unlike ‘how much’ is expressed with a complex wh-phrase which has the structure NP+ álá, where the wh-phrase álá substitutes for the numeral/quantification and the NP is the noun whose amount is interrogated. The distinction between two is illustrated by the data in (6).

The illicitness of sentence (6c) where álá is intended to read as ‘how many’ serves as an empirical evidence to my current proposal that álá cannot be interpreted as denoting quantity ‘how many’ alone as earlier proposed in the literature. I therefore conclude that the wh-phrase álá ‘how much’ which denotes cost/amount without a NP, is not ambiguous and therefore cannot be interpreted as ‘how many’ as proposed by the earlier works of Issah (2015a).

It is worthy of mention that this is somewhat different from what pertains in English where the distinction might come from the noun that is modified by álá ‘how much.’ For instance, it is not possible to ask ‘how many do you have’ in English without an elliptical interpretation. Let us consider the data in (7) where I show that in Dagbani, álá never occurs with mass nouns.

←56 | 57→

Based on data in (7), I contend that the use of álá has no elliptical interpretation in Dagbani.

Issah (2015a) just as observed of Olawsky (1999) is also silent about any further discussion on the formal properties of this class of the wh-phrases. I show later that there are some salient grammatical features of these wh-phrases, none of these features received any systematic discussion in Issah (2015a).

In summary, although it was established that Dagbani wh-phrases have received quite some appreciable level of linguistic investigation in previous studies, some shortcomings were nevertheless identified in these earlier works, including an inexhaustive list of the semantic distinctions of wh-phrases, misidentification of the semantic domains of wh-phrases and failure on the part of earlier scholars to discuss the grammatical properties of the wh-phrases. In the discussion that follows, I fill the analytical gaps in the incomplete list of the sematic domains of the wh-phrases and also discuss the salient grammatical properties such as number value specification, semantic distinction between human and nonhuman wh-phrase and lexical ambiguity in wh-phrases. I begin with a discussion of a novel account of the inventory of Dagbani wh-phrases in section 2.2.1 where I provide the first complete list of Dagbani wh-phrases, together with the semantic distinctions they make.

2.2.1 A novel account on the inventory of Dagbani wh-phrases

Contrary to the earlier assertions of Olawsky who identified only seven semantic distinctions for the wh-phrases as captured in tab. 1, I propose that there are ten distinct wh-phrases, which I argue captures the full system of the inventory of Dagbani wh-phrases as given in tab. 2. The semantic categorization of the wh- phrase as presented in tab. 2 is based on the semantic classification of Siemund (2001: 1023) with some modification.

Tab. 2: Novel proposal on inventory of Dagbani wh-phrases


semantic category


















ISBN (Book)
Publication date
2020 (May)
Berlin, Bern, Bruxelles, New York, Oxford, Warszawa, Wien, 2020. 238 pp., 6 tables.

Biographical notes

Samuel Alhassan Issah (Author)

Samuel Alhassan Issah is a senior lecturer at the College of Languages Education, University of Education, Winneba/Ghana. He holds a PhD in linguistics awarded by the Goethe University, Frankfurt am Main, and a Master of Philosophy in Theoretical Linguistics awarded by the University of Tromsoe, Norway. His research interests include information structure (focus realizations) of Dagbani and related languages, structure of Dagbani and related languages, the syntax of elliptical phenomena, and the syntax of anaphoric expressions.


Title: On the structure of A-bar constructions in Dagbani: Perspectives of «wh»-questions and fragment answers