A Descriptive Grammar of Shilluk 9780997329529

Citation preview

A Descriptive Grammar of Licensed under Creative Commons Attribution Non-Commercial No Derivatives License

Shilluk ISBN 978-0-9973295-2-9

Bert Remijsen & Otto Gwado Ayoker

A Descriptive Grammar of

Shilluk Bert Remijsen & Otto Gwado Ayoker

Language Documentation & Conservation Department of Linguistics, UHM Moore Hall 569 1890 East-West Road Honolulu, Hawai’i 96822 USA http://nflrc.hawaii.edu/ldc University of Hawai’i Press 2840 Kolowalu Street Honolulu, Hawai’i 96822-1888 © All texts and images, Bert Remijsen and Otto Gwado Ayoker, 2018. © All chapters are licensed under Creative Commons Licenses. Attribution: No Derivatives License. Cover photo: Red sorghum at Tonga, South Sudan, Bert Remijsen, 2013. Library of Congress Cataloging in Publication data ISBN 978-0-9973295-2-9 http://hdl.handle.net/10125/24647

A Descriptive Grammar of Shilluk

1

Introduction This special publication of Language Documentation and Conservation presents descriptive analyses on topics in the grammar of Shilluk, a NiloSaharan language spoken primarily in South Sudan. A salient characteristic of Shilluk is that it is rich in fusional morphology. That is, stem-internal changes, particularly in terms of tone and vowel length, have a high functional load in the paradigms of verbs and nouns. From 2008 onwards, we have built up a detailed understanding of these contrasts and their role in the grammar. Accountability is a central concern in documentary and descriptive linguistics (Himmelmann 1998, Woodbury 2003), and it is one that has determined the design of this publication in various ways. One way we are ensuring that our description is accountable is by including sound examples. We do this because the phenomena themselves are sounds; transcriptions based on the sounds are hypotheses (cf. Rice 2014). Aside from making the work more accountable, sound examples embedded in publications make the phenomena more accessible, reducing the threshold between the reader and an unfamiliar language. This is particularly important when the relation between a transcription and the sound it represents is hard to interpret, as is the case with tone. For example, on the basis of the transcription in (1), the great majority of readers will not have a clear sense of how the tone pattern of this monosyllabic word sounds. In fact, it carries a High Fall to Mid, which is realised as a shallowly falling contour in the upper half of a speaker’s tonal range. This melodic form is clear and concrete from the sound example, to the extent that a reader could mimic the melody they hear to develop a grip on the melodic shape. ́ ʌt ̪̄ jʌ̂ʌ boat:p ‘Boats.’

We believe it is important to make our descriptive analysis of Shilluk phonology accountable and accessible in this way, because research that is not accountable is of limited value to the study of language, and accessibility is crucial to get the message across. That is why we embed sound examples in relation to all numbered illustrations (see Chapter 1). We will add chapters as our understanding warrants it. This incremental model of publication is also motivated by accountability. As anyone who has engaged in it knows, descriptive research is by its nature a long-term endeavour. In our case, even though we have been investigating Shilluk A DESCRIPTIVE GRAMMAR OF SHILLUK

A Descriptive Grammar of Shilluk

2

grammar for ten years now, there are still phenomena that belong in a comprehensive description but of which we do not yet have sufficient understanding. In this regard, the work on the grammar of Dinka by Torben Andersen has been an inspiring model to us. Over a period of several decades, from Andersen (1987) onwards, he has published research articles describing many parts of Dinka grammar, to the effect that, at this point, the amalgam of these publications approaches the scope of coverage of a traditional descriptive grammar. However, journal articles are restricted in their length, and a report that takes 80 pages, as our descriptive analysis of the base paradigm of transitive verbs does in the first chapter, is problematic in that respect. By publishing in an electronic-only manner, we are not limited in this way. A final accountability measure is peer review. Each chapter is reviewed independently, that is, with the review process managed by the journal editor. We take this opportunity to thank Nick Thieberger and the anonymous reviewers involved for giving us the opportunity to contribute to the documentation and description of the Shilluk language in the best way we can imagine.

References Andersen, Torben (1987). The Phonemic System of Agar Dinka. Journal of African Languages and Linguistics 9, 1–27. Himmelmann, Nikolaus (1998). Documentary and descriptive linguistics. Linguistics 36, 161–195. Rice, Keren (2014). Sounds in grammar writing. In Toshihide Nakayama and Keren Rice (eds.) The art and practice of grammar writing. Language Documentation & Conservation, Special Publication 8, 69–89. Woodbury, Tony (2003). Defining documentary linguistics. Language Documentation and Description 1, 35–51.

A DESCRIPTIVE GRAMMAR OF SHILLUK

Language Documentation & Conservation Special Publication No. 14 – A Grammar of Shilluk Chapter 1: Forms and functions of the base paradigm of Shilluk transitive verbs by Bert Remijsen & Otto Gwado Ayoker http://nflrc.hawaii.edu/ldc/sp14 http://hdl.handle.net/10125/24777

Forms and functions of the base paradigm of Shilluk transitive verbs Bert Remijsen & Otto Gwado Ayoker, University of Edinburgh

Abstract • This chapter offers a descriptive analysis of the morphological

forms that make up the base paradigm of Shilluk transitive verbs, and also of the functions that are expressed through them. With respect to morphological exponence, tone and vowel length play a central role, both in marking the functions and in distinguishing a total of seven different verb classes. As for the functions, they are syntactic voice, subject marking, and tense-aspect-modality (TAM). These functions interact with one another and with other aspects of the syntax of the clause. For example, Imperfective aspect is only available in Object voice, and certain TAM forms interact with focus marking. We pay special attention to syntactic alignment, a topic on with earlier analyses diverge. Older studies distinguish between active and passive voices (Westermann 1912, Tucker 1955). More recently, the passive has been reinterpreted as an ergative construction (Miller & Gilley 2001). We find that the construction at the center of the controversy has all the morphosyntactic properties of a passive, but not the information-structural characteristics. The scope of this chapter is restricted to the base inflectional paradigm. This means that it does not cover the many derivations which present inflectional paradigms that are largely parallel to the base paradigm. For the sake of clarity and accountability, sound examples are embedded in relation to each of the numbered illustrations.

Licensed under Creative Commons Attribution Non-Commercial No Derivatives License

ISBN 978-0-9973295-2-9

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

2

1 Introduction This chapter describes the morphosyntactic structure of Shilluk clauses that are headed by transitive verbs. This topic is pivotal to the analysis of Shilluk grammar as a whole, both in relation to the syntax and in relation to the morphology. With regards to the syntax, the verb, as the head of the predicate, determines the structure of the clause as a whole. This is especially true in Shilluk, because it is a head-marking language. This means, among others, that the roles of the main arguments to the predicate are signposted morphologically on the verb, i.e., the head, rather than through case marking on the arguments, i.e., the dependents.1 In relation to the morphology as well, the transitive verbs are central to an accurate understanding of Shilluk grammar, because this set of lexical roots presents the richest morphological paradigms. The important role of the transitive verb system to the study of Shilluk is evident from the scientific record: the transitive verb system has been the main object of investigation in several studies (Tucker 1955; Miller & Gilley 2001, 2007; Remijsen, MillerNaudé & Gilley 2016), more so than any other aspect of Shilluk grammar. The verb forms involve intersecting dimensions of morphological and lexical information: • Seven verb classes, which differ in terms of vowel length and tone; • Morphological marking for Voice (three levels), Tense-Aspect-Modality [TAM] (seven levels), and Subject marking (seven levels); marked through vowel length, tone and affixation. To gain insight into the morphology of Shilluk transitive verbs, it is important to understand that the same morphophonological markers – vowel length and tone – express both lexical and morphological information. And this morphophonology of tone and length is not only crucial, but also highly complex: Shilluk presents three levels of vowel length and nine syllable-level tone categories, all of which appear in the paradigm of transitive verbs. To convey these phenomena effectively and accountably, sound examples are embedded in relation to the illustrations. The chapter is structured as follows. We start out describing the system of transitive verb classes (Section 2). Then we lay out the functions that are expressed through the base paradigm. These functions are three in number, and each of these constitutes the topic of a section: Voice (Section 3), Subject 1 Case-marking on pronouns in post-verbal position is a rare instance of dependent-marking (cf. Section 4). SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

3

marking (Section 4), and Tense-Aspect-Modality [TAM] (Section 5). Certain combinations of Voice and Tense-Aspect-Modality require the presence of one of a set of constituents elsewhere in the clause. We refer to this as ‘syntactic licensing’; this phenomenon is described in Section 6. Section 7 describes in detail the inflectional marking of the above-mentioned functions in relation to the seven verb classes. The base paradigm is summarized in full using exemplar verbs in Appendix A, and schematically in Appendix B. The scope of this chapter is limited in a number of ways. First, the description is limited to the base paradigm. That is, transitive verbs additionally present several derived paradigms, such as benefactive and antipassive. These are parallel to the base paradigm, in that they present inflections marking the same three functions of Voice, Subject marking and TAM. These derived paradigms of transitive verbs are beyond the scope of this chapter. With respect to the functions, we limit ourselves to a characterisation that motivates the functional interpretation of the morphological pattern. What we will not do is describe these functions in their own right. For example, in Section 3 we describe how several levels of tense, aspect and modality are inflectionally marked on the verb. But we do not describe tense, aspect and modality in their own right, an endeavour that encompasses lexical and syntactic perspectives in addition to the morphological angle. Also beyond the scope are transitive clauses involving complex predicates, non-declarative clauses, and defective transitive verbs.

1.1 Relation to earlier work

Our understanding of the morphophonological forms that make up the transitive verb paradigm largely follows the analysis presented in Remijsen et al. (2016); in the course of this chapter we will point out discrepancies. The most important point of divergence is that, in the current study, we distinguish between the base paradigm and derived paradigms. Specifically, we argue that Applicative voice and Subject voice are part of base inflectional paradigm, on a par with Object Voice, whereas operations such as Benefactive and Antipassive are best interpreted as derivations. In contrast, Remijsen et al. (2016) treated all these morphological operations on a par. This innovation in the analysis, i.e., of distinguishing between inflection and derivation, is motivated by the recognition that morphological marking for Voice, Subject marking and TAM recurs across derivations. For example, there are subjectmarked forms in the base paradigm, and also in the benefactive and the antipassive. The interpretation that the Applicative voice is inflectional is SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

4

supported by the same line of argument: it recurs across derivations. This insight is inspired by investigations in related languages, especially Andersen (1992–1994) on Dinka, and Reid et al. (2016) on Nuer. A central topic on which earlier studies disagree is the nature of syntactic alignment in Shilluk. While Westermann (1912) and Tucker (1955) distinguish between active and passive voices, Miller & Gilley (2001) invoke an ergative analysis for the latter. In our investigations into the voice system, we find support for the former analysis on the basis of the morphosyntactic characteristics. At the same time, we agree with Miller & Gilley (2001) that the construction hypothesized to be a passive in older work is unmarked in an information-structural sense. Our analysis of TAM largely follows Miller & Gilley (2007), and in particular we adopt the hypothesis of a contrast between evidential and non-evidential past tense forms. One expansion is the No Tense form. This level of TAM has not been investigated in detail before, even though it recurs with great frequency in Shilluk speech.

1.2 Methodology

The second author is a native speaker of the Shilluk language, more specifically from Tonga, a town at the southwestern edge of Shilluk-speaking region. Decisions on which structures are grammatical and which are not are based on his native-speaker intuitions. During the first six years of our investigations, in the course of intermittent research between 2008 and 2013, we studied the transitive verb system primarily through controlled elicitation, and only to a lesser extent on the basis of spontaneously uttered speech. In this way, we developed a detailed understanding of the morphophonological forms, while our knowledge of the functions remained rudimentary. During this period, the analysis of the morphophonology of verbs was supported through phonological and phonetic investigations with at least fifteen speakers. The results of these studies on the sound system are reported in Remijsen, Ayoker & Mills (2011), Remijsen & Ayoker (2014), and Remijsen, Ayoker & Jørgensen (to appear). By 2012, our analysis of the morphophonological forms was largely settled, and our methodology shifted to the study of spontaneous speech – especially narratives, but to a lesser extent also songs (Remijsen & Ayoker 2015). Since then, the grammatical analysis of narratives has played a central role in our investigations, bringing up phenomena we would not have thought of exploring through controlled elicitation. This process whereby the analysis develops through the study of narratives is open-ended. At the time of writing SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

5

this chapter, we have carried out detailed grammatical analysis of twelve narratives, totalling over 60 minutes, collected from eight speakers. Examples drawn from this documentary play an important role in making the descriptive analysis accountable (cf. Woodbury 2003). These and other forms of spontaneously uttered speech are signposted as such by means of a ^ adjacent to the example number. So far, fieldwork by the first author amounts to thirty weeks in Khartoum, Juba, and Tonga. In addition, the second author has spent three months in Edinburgh in 2009. Since 2015, however, the most important context of our investigations is not working in person, but rather through Skype, which has enabled us to intensify our collaboration considerably. From September 2015 onwards over a three-year period, we have conducted narrative analysis and controlled elicitation through a schedule of three sessions per week, on average.

1.3 The Shilluk sound system

In this section, we briefly summarize the main features of Shilluk phonology, along the lines of the analyses in Remijsen et al. (2011) and Remijsen & Ayoker (2014). This summary is relevant to the study of the transitive verbs, because stem-internal alternations, especially in terms of vowel length and tone, are central to the morphological marking. The majority of monomorphemic native Shilluk words consist of a single closed syllable, e.g. càm ‘eat’, kwʌ̄n ‘porridge’, lʊ̀ʊt ̪ ‘stick’ and ŋǒoom ‘awl’. Derivational and inflectional affixes result in more complex phonological word shapes. Consider, for example, the instrument nominalisation góooc-ɪ ̄ɪ ‘machete [hit-instr]’ or the cardinal á-dʌ̀k ‘three [card-third]’. The consonant inventory includes ten plosives and five nasals, structured orthogonally at five places of articulation: /p,b,m; t ̪,d̪,n̪; t,d,n; c,ɟ,ɲ, k,g,ŋ/. Voicing in plosives is distinctive in the syllable onset only; the realization of plosive codas varies allophonically in voicing and in manner (Remijsen et al. 2011); we represent them with the voiceless plosive character. The remaining consonants are the liquids /l,r/ and the semivowels /w,j/. The latter are the only consonants that can combine into complex onsets – e.g. tjɛ́l ‘foot:p’. The vowel inventory includes two sets, which differ in specification for Advanced Tongue Root (ATR). The –ATR vowels are /ɪ,ɛ,a,ɔ,ʊ/; the +ATR vowels are /i,e,ʌ,o,u/. Stem vowels display a ternary contrast between short, long, and overlong (Remijsen et al., to appear). We represent them using one, two, and three vowel characters, respectively. Affix vowels are short or long. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

6

The stem syllables of content words display nine distinctive tone patterns. There are three level tonemes: Low (L) /cv̀c/, Mid (M) /cv̄c/, and High (H) / cv́c/; four falling contour tones: Low Fall (LF) /cv̂c/, High Fall (HF) /cv̂ć /, Late Fall (LHF) /cv́c/̀ , and High Fall to Mid (HFM) /cv̂ć /̄ ; and two rising contour tones: Low Rise (LR) /cv̌c/ and High Rise (HR) /cv̄č /. Because the inventory is so big, especially in terms of contour tones, several tone categories are represented through a combination of diacritics, and in the case of two contours – the Late Fall and the High Fall to Mid – these diacritics are distributed over two segmental characters: the first vocalic character and the coda. These are merely conventions in answer to the challenge of representing nine distinct syllable-level tone categories. That is, all transcriptions for tone make reference to the surface-phonological specification for tone of the syllable as a whole. Affixes only carry the level tones, i.e., Low, Mid, and High. To the best of our knowledge, there is only one morpheme that represents an exception to the generalization that every syllable is specified for tone.2

2 Transitive verb classes Shilluk transitive verbs present a remarkable uniformity in their phonological form: the root consists of a single closed syllable. The composition of this syllable is summarized by the template /C(j/w)V(V)C/. That is, the root vowel is either short or long, and clustering of consonants is restricted to the onset, where either of the semivowels /w,j/ may follow another consonant. Among over 700 transitive verbs in our lexicographic data, there is only one that does not conform to this template: {òr} ‘send’, which lacks an onset. A total of seven classes can be distinguished, based on alternations in terms of vowel length and tone (Remijsen, Miller-Naudé & Gilley 2015, 2016). This section is dedicated to the description of these classes. As we lay out this system, we make reference to inflections for Voice (e.g. Subject voice), TAM (e.g. Past tense) and subject marking (e.g. 2nd singular). These functions determine the morphophonological specification in terms of which the verb classes diverge. Here we refer to them without motivating them; they will be discussed in depth in the following sections. A first dimension on which the verb classes diverge is vowel length in the stem syllable. This is illustrated in Table 1. In this table, the lexical 2 The exception at issue is an allomorph of the focus marker à. While à is Low-toned in most environments, it is toneless when it follows immediately after the verb stem. There it copies the tonal target of the syllable to its left. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

7

length of the root vowel is illustrated by the stem form in the Subject voice Past tense form, whereas morphological lengthening of the stem vowel is illustrated by the Past tense subject-marked for 2nd singular. Note that there are three patterns of alternation. First, there are verbs that have a short vowel lexically, and that do not display morphological lengthening of the stem vowel anywhere in their paradigm. The verb {ŋɔ̀l} ‘cut’, shown in Table 1, has a short vowel throughout its morphological paradigm. We refer to these as Fixed Short verbs. As seen from Table 1, {ŋɔ̀l} the Subject voice Past tense form and the Past tense 2nd singular form are indistinguishable from one another. Second, there are verbs that have a short root vowel, but which appear with increased vowel length in many inflections, including the past tense 2nd singular. This pattern, which we refer to as Short with Grade, is illustrated in Table 1 by {càm} ‘eat’. Third, there are verbs that have a long root vowel, and which also undergo morphological lengthening. These are the ‘Long (with Grade)’3 verbs; {lɛ̀ɛŋ} ‘throw’ is one of them. Table 1. The three patterns of vowel length alternation in transitive verbs. The lexical length of the stem vowel is illustrated by the Subject voice (sv) Past tense form, and morphological lengthening by the Past tense 2nd singular. Verb classes Example

sv past past 2sg

Fixed Short {ŋɔ̀l} ‘cut’

á-ŋɔ̀l á-ŋɔ̀l

Short with Grade {càm} ‘eat’

á-càm á-càaam

Long {lɛ̀ɛŋ} ‘throw’

á-lɛ̀ɛŋ á-lɛ̀ɛɛŋ

Note that morphological lengthening in Shilluk is ‘overlengthening’: if a verb displays morphological lengthening, then it lengthens to the third level of vowel length, i.e., overlong, irrespective of whether the root vowel is short or long, i.e., CVC, CVVC→ CVVVC. As a result, overlengthening is a neutralizing process: on the basis of a verb form with an overlong stem vowel, e.g. in the Past tense form 2nd singular, one cannot predict whether the Subject voice Past tense form has a short vowel or a long one. In this respect, Shilluk is different from Dinka, where short and long stem vowels both lengthen morphologically by one level of vowel length, i.e., CVC → CVVC and CVVC → CVVVC (Andersen 1990). Summarizing the vowel length alternations in the base paradigm of 3 The qualification ‘with Grade’ is superfluous in relation to Long verbs: they all display vowel length alternation. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

8

transitive verbs, we find a) Fixed Short verbs; b) Short with Grade verbs, that alternate between a short stem vowel and an overlong one; and c) Long verbs, alternating between long and overlong stem vowels. Vocalic overlength invariably expresses inflection in transitive verbs.4 It is worthwhile to note that the alternation between short and long stem vowels does not occur in the base paradigm.5 The patterns of vowel length alternation in the stem syllable are in part predictable on the basis of vowel quality. Long verbs come with any vowel quality. But verbs with a short root vowel, i.e., the Fixed Short and Short with Grade classes, display an interaction. Fixed Short verbs have closed or halfopen root vowels, but not the open vowels /a,ʌ/. In addition, the vowel is not preceded by a complex onset. Short with Grade verbs, in contrast, either have an open root vowel, i.e., /a,ʌ/, or a vowel preceded by a semivowel in the onset, e.g. {gwɔ̂ɲ} ‘scratch’.6 The second dimension on which the transitive verbs separate into lexical classes is tone. The verbs in Table 1 all belong to Low classes, which is marked by the low-tone diacritic in the morpheme representation of these verbs: {ŋɔ̀l}, {càm}, {lɛ̀ɛŋ}. But there are also Low Fall verbs, and the difference between Low and Low Fall verb classes is crossed orthogonally with the three patterns of length alternation, yielding six classes. This is shown in Table 2, which displays the same inflections as Table 1, but now includes the Low Fall verbs. Note that the latter have a Low Fall on the stem syllable in Subject voice Past and Past 2nd singular, whereas the Low verbs have the Low tone instead.

4 This is to be expected, because overlength in Shilluk, as in other West Nilotic languages, is the diachronic outcome of suffix vowels lost through compensatory lengthening (Andersen 1990). 5 However, this alternation between short vs. long vowels is attested elsewhere in the morphology of transitive verbs. Some verbs with a long stem vowel mark the antipassive derivation through a combination of stem-internal changes which includes a shortening of the stem vowel. For example, {bûuk} ‘cover with powder’ has the Antipassive Past á-bûk (cmp. transitive Subject voice Past á-bûuk), and {cʊ̂ʊl} ‘pay for’ has the Antipassive Past á-cût (cmp. transitive Subject voice Past á-cʊ̂ʊl). The derivational morphology of transitive verbs is beyond the scope of this chapter. 6 The relevance of vocalic complexity to the morphological quantity alternation in Shilluk is reminiscent of a similar interaction in Anywa, a closely-related language (Reh 1996). SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

9

Table 2. Evidence for the orthogonal crossing of tone (Low vs. Low Fall) with vowel length (Fixed Short vs. Short with Grade vs. Long) in the verb class system. Fixed Short

Short with Grade

Long

Verb classes

Low

Low Fall

Low

Example

{ŋɔ̀l} ‘cut’

{lɛ̂ŋ} ‘drum’

{càm} ‘eat’ {mʌ̂l} ‘roast’

{lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’

sv past past 2sg

á-ŋɔ̀l á-ŋɔ̀l

á-lɛ̂ŋ á-lɛ̂ŋ

á-càm á-càaam

á-lɛ̀ɛŋ á-lɛ̀ɛɛŋ

Low Fall á-mʌ̂l á-mʌ̂ʌʌl

Low

Low Fall á-mâat ̪ á-mâaat ̪

There is one additional verb class, which we label the High Fall class. It is included in Table 3, where it is illustrated by {mấal} ‘praise’. This pattern of tonal alternation is only found on Long verbs, i.e., it does not appear on Fixed Short verbs or on Short with Grade verbs. Note that, in the Subject voice Past tense form, the tonal specification on the stem syllable for this class is the same as that of the Low Fall verbs, whereas it patterns with the Low verbs in Past 2nd singular. In certain other inflections, its specification deviates from both the Low and Low Fall verbs. One such inflection is the Object voice Imperfective, where the stem syllable carries a High Fall, after which the class is named. Table 3. The difference in tone and vowel length between the seven classes of transitive verbs, illustrated by Subject voice Past, Past 2nd singular, and Object voice Imperfective. Fixed Short

Short with Grade

Long

Verb classes

Low

Example

{ŋɔ̀l} ‘cut’ {lɛ̂ŋ} ‘drum’ {càm} ‘eat’ {mʌ̂l} ‘roast’ {lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’ {mấal} ‘praise’

sv past

á-ŋɔ̀l

past 2sg á-ŋɔ̀l

ov impf ʊ̀-ŋɔ̀l-ɔ̀

Low Fall á-lɛ̂ŋ

á-lɛ̂ŋ

ʊ̀-lɛ̂ŋ-ɔ̀

Low á-càm

á-càaam

Low Fall á-mʌ̂l

á-mʌ̂ʌʌl

ʊ̀-càaam-ɔ̀ ʊ̀-mʌ̂ʌʌl-ɔ̀

Low á-lɛ̀ɛŋ

á-lɛ̀ɛɛŋ

ʊ̀-lɛ̀ɛɛŋ-ɔ̀

Low Fall

High Fall

á-mâat ̪

á-mâal

ʊ̀-mâaat ̪-ɔ̀

ʊ̀-mấaal-ɔ́

á-mâaat ̪

á-màaal

Across their paradigm, Low, Low Fall and High Fall verbs each appear with a range of different specifications for tone. Before considering the phenomena, we set out an opposition between two scenarios. One possible situation is that a transitive verb has a lexical specification for tone, and that the tonal specifications it appears with in its various inflections can be interpreted as the compositional outcome of the integration of this lexical or underlying SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

10

specification with inflectional tones. We refer to this as the Compositional Tone scenario. The alternative scenario is that a verb appears with a given set of tonal specifications across its paradigm, but that these specifications cannot plausibly be related through morphophonological processes. In this second scenario, the paradigm does not offer a basis to postulate an underlying or lexical specification. We refer to this as the Set Tone scenario. Both analyses are adequate in a descriptive sense, but the Compositional Tone scenario offers an explanation of how the tonal specifications of different inflections in the paradigm are related to one another, and would be more attractive for that reason (explanatory adequacy), if the data offer support for it. As it turns out, the situation is mixed. We will show that some inflections offer compelling evidence for Compositional Tone, whereas others do not. Starting with the former, the Low and Low Fall classes present inflections in which lexical and inflectional specifications are both expressed. Consider the data in Table 4, which presents Subject voice and Applicative voice (xv) forms, both in Past tense. The Applicative inflection involves overlengthening for verbs that are not Fixed Short. As for tone, the Low classes have a Mid tone on the stem in this inflection, and the Low Fall a High Fall to Mid. (The High Fall verbs pattern along with the Low class here.) Table 4. Evidence for Compositional Tone, from Subject voice Past tense and Applicative voice Past tense. Verb classes

Fixed Short Low

Low Fall

Short with Grade Low

Low Fall

Long Low

Low Fall

High Fall

Example {ŋɔ̀l} ‘cut’ {lɛ̂ŋ} ‘drum’ {càm} ‘eat’ {mʌ̂l} ‘roast’ {lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’ {mấal} ‘praise’ sv past á-ŋɔ̀l

xv past á-ŋɔ̄l

á-lɛ̂ŋ

́ ̄ á-lɛ̂ŋ

á-càm

á-cāaam

á-mʌ̂l

́ ʌl ̄ á-mʌ̂ʌ

á-lɛ̀ɛŋ

á-lɛ̄ɛɛŋ

á-mâat ̪

á-mấaat ̪̄

á-mâal

á-māaal

Note that the Low verbs have a lower specification for tone than the Low Fall verbs in the Subject voice Past tense: Low vs. Low Fall, respectively. This difference is retained, at a higher tonal register, in the Applicative voice, where they have Mid and High Fall to Mid, respectively. If we conceive of the Shilluk tone realization as involving three heights, numbered from 1 (low) to 3 (high), then a change from Low to Mid is a change from 1 to 2, and a change from Low Fall to High Fall to Mid can be conceived of as a change from 21 to 32. In both cases, the register is increased by 1 level. In this way, the tonal specification in the Applicative expresses both inflection and verb SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

11

class. The Compositional Hypothesis receives additional supported when the Applicative Past is inflected further for a 2nd singular subject, which is added in Table 5. This inflection adds a High target to the right of the specification of the Applicative voice Past tense. In the case of the Low verbs, which have a Mid tone in the Applicative Past, the addition of a High target yields a High Rise, i.e., ̄ + ́ → .̄̌ Or, in the numeric representation introduced above, 2+3 → 23. Table 5. Further evidence for Compositional Tone, from Subject voice Past tense, Applicative voice Past tense, and Applicative voice Past tense 2nd singular. Fixed Short

Short with Grade

Long

Verb classes

Low

Example

{ŋɔ̀l} ‘cut’ {lɛ̂ŋ} ‘drum’ {càm} ‘eat’ {mʌ̂l} ‘roast’ {lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’ {mấal} ‘praise’

sv past á-ŋɔ̀l xv past á-ŋɔ̄l xv past 2s á-ŋɔ̄ľ

Low Fall á-lɛ̂ŋ ́ ̄ á-lɛ̂ŋ á-lɛ́ŋ

Low á-càm á-cāaam ̌ am á-cāa

Low Fall á-mʌ̂l ́ ʌl ̄ á-mʌ̂ʌ á-mʌ́ʌʌl

Low á-lɛ̀ɛŋ á-lɛ̄ɛɛŋ á-lɛ̄ɛ̌ ɛŋ

Low Fall á-mâat ̪ á-mấaat ̪̄ á-máaat ̪

High Fall á-mâal á-māaal ̌ al á-māa

In the case of the Low Fall verbs, which have a High Fall to Mid in the Applicative Past, the addition of a High target yields a High level tone, i.e., ̂́ ̄ + ́ → ́, or numerically, 32+3→3. It is not surprising that a three-target configuration would be simplified in this way, i.e., for the sequence of High Fall to Mid followed by High to simplify to level High, because time pressure is very high in a three-target configuration (cf. Zhang 2001, Xu & Sun 2002). ́ ́) can be observed And in fact, the same simplification process ( ̂́ ̄ + → elsewhere in the grammar. This is shown in (1). This illustration shows possessive noun phrases. The possessed term, i.e., the head, is inflected for taking a possessor, an inflection we label pertensive following Dixon (2010). In the case of a suffixed noun, such as dɔ́ɔɔr-ɔ́ ‘axe-s’, pertensive is marked by a suffix -ɪ. In (1a), where the possessor is grammatically singular, the pertensive suffix has a Mid tone. When the possessor is grammatically plural, as in (1b), this suffix is invariably High-toned. Note how, in (1b), the addition of the High-toned suffix results in a High tone on the stem. Suffixless nouns, such as dúup ‘mouse’, display the same tonal alternation. The High Fall to Mid of the pertensive singular turns into a level High tone in the pertensive plural.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs (1)

a. c.

dâa dɔ̂ɔ́ r̄-ɪ ̄ twɔ́ɔŋ exsp.foc axe-prt Twong ‘There is Twong’s axe.’

dâa dû́uup̄ twɔ́ɔŋ exsp.foc mouse-prt Twong ‘There is Twong’s mouse.’

12

b.

dâa dɔ́ɔr-ɪ ́ mʌ́ʌn exsp.foc axe-prt:p woman.p ‘There is the women’s axe.’

d.

dâa dúuup mʌ́ʌn exsp.foc mouse:prt:p woman.p ‘There is the women’s mouse.’

The alternation in (1c,d) is identical to the one in Table 5, between High Fall to Mid tone and High tone in Applicative vs. Applicative 2nd singular. This suggests that the High tone on the stem syllable of Low Fall verbs inflected for Applicative 2nd singular can be derived in a compositional manner from the Applicative voice form without subject marking, through a word-internal simplification process, whereby ́̂ ̄ + ́→ .́ However, there are other inflections where the specification cannot be derived in a compositional manner. A first illustration of this appears in Table 6, which shows the Object voice Past tense form. In this inflection, all transitive verbs have the High Fall (/ ̂́/). Table 6. Neutralisation of tone contrast in the Object voice Past tense, as evidence for Set Tone. Fixed Short Verb classes Low Low Fall

Short with Grade Low

Low Fall

Long Low

Low Fall

High Fall

Example {ŋɔ̀l} ‘cut’ {lɛ̂ŋ} ‘drum’ {càm} ‘eat’ {mʌ̂l} ‘roast’ {lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’ {mấal} ‘praise’ sv past á-ŋɔ̀l ov past á-ŋɔ̂ĺ

á-lɛ̂ŋ ́ á-lɛ̂ŋ

á-càm á-cấm

á-mʌ̂l á-mʌ̂ĺ

á-lɛ̀ɛŋ á-lɛ̂ɛ́ ŋ

á-mâat ̪ á-mấat ̪

á-mâal á-mấal

A second illustration of non-compositional specification can be observed in the relation between Past and No Tense forms of the Subject Voice. These are shown in Table 7. Across classes, the Past tense is marked by the prefix á-, whereas the No Tense form is not affixed. Note that the Low Fall verbs present the same tone on the stem on both of these TAM forms – and so do the High Fall verbs, which pattern along with the Low Fall verbs in these inflections. In contrast, the Low verbs do differ in the tonal specification of the stem between Past and No Tense: the Past tense has a Low on the stem; the No Tense has a High Fall to Mid. Clearly, the tonal specifications in Past and No Tense forms cannot be related to one another in a compositional manner across verb classes. These data also show that verb classes do not have a monopoly on tones: whereas the High Fall to Mid identifies the Low classes in Subject voice SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

13

No Tense (Table 7), the same specification for tone is found on the Low Fall classes in the Applicative (Table 4). Table 7. Further evidence for Set Tone, from Subject voice Past vs. Subject voice No Tense. Fixed Short Verb classes Low Low Fall

Low

Short with Grade

sv past á-ŋɔ̀l sv nt ŋɔ̂ĺ ̄

cấm̄

Low Fall

Long

Low

Low Fall

High Fall

á-lɛ̀ɛŋ

á-mâat ̪

á-mâal

Example {ŋɔ̀l} ‘cut’ {lɛ̂ŋ} ‘drum’ {càm} ‘eat’ {mʌ̂l} ‘roast’ {lɛ̀ɛŋ} ‘throw’ {mâat ̪} ‘drink’ {mấal} ‘praise’ á-lɛ̂ŋ

lɛ̂ŋ

á-càm

á-mʌ̂l

mʌ̂l

lɛ̂ɛ́ ŋ̄

mâat ̪

mâal

Finally, as a third illustration of non-compositional specification, we already noted, in relation to Table 3, that the High Fall class pattern along with the Low class in some inflections, with the Low Fall class in others, and go their own way elsewhere. We conclude that the phenomena do not align fully with either the Compositional Tone scenario nor with the Set Tone scenario. Instead, we find that some inflections are related in a compositional manner, whereas others cannot. The evidence of compositionality in the Low and Low Fall classes suggests that verbs belonging to these classes actually do have an underlying specification from which several inflections can be derived in a compositional manner. In relation to the High Fall class, there is no strong evidence for compositionality. There is no phonological basis for the tone-based division into classes. But there is a semantic tendency: verbs that typically take a human semantic ́ t ̪} ‘greet’, {mấan} ‘hate’, etc., tend to belong to the High object, such as {mʌ̂ʌ Fall class. We underline that this is merely a tendency. In the remainder of this section, we illustrate the difference between verb classes using full-sentence examples involving minimal-pair verbs. First we contrast Low and Low Fall classes; then we contrast Low Fall and High Fall classes. The Low verb {kɔ̀l} ‘disturb’ and the Low Fall verb {kɔ̂l} ‘take out’ are both Fixed Short. As seen from (2a,b), they are indistinguishable from one another in the Object voice Past tense. In contrast, the Past tense 2nd singular reveals the difference between the tone classes: here {kɔ̀l} ‘disturb’ has the Low toneme (2c), whereas {kɔ̂l} ‘take out’ has the Low Fall (2d). In the Applicative voice (glossed xv), {kɔ̀l} ‘disturb’ has the Mid toneme (2e), whereas {kɔ̂l} ‘take out’ has the High Fall to Mid (2f). Finally, (2g) and (2h) show how the SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

14

difference between verb class is maintained when the stem is inflected further for 2nd singular. (2)

a. gwôk á-kɔ̂ĺ dog pst-disturb:ov ‘Somebody disturbed the dog.’

b. kʊ̂́ʊʊt-̪̄ ɔ̄ á-kɔ̂ĺ thorn-s pst-take.out:ov ‘Somebody took out the thorn.’

c. gwôk á-kɔ̀l dog pst-disturb:2s ‘You disturbed the dog.’

d. kʊ̂́ʊʊt-̪̄ ɔ̄ á-kɔ̂l thorn-s pst-take.out:2s ‘You took out the thorn.’

e. ʊ̀tjáaɲ á-kɔ̄l gwôk bell pst-disturb:xv dog ‘Smb. disturbed the dog with a bell.’

f.

g. ʊ̀tjáaɲ á-kɔ̄̌l gwôk bell pst-disturb:xv:2s dog ‘You disturbed the dog with a bell.’

ŋǒoom á-kɔ̂ĺ ̄ kʊ̂́ʊʊt-̪̄ ɔ̄ awl pst-take.out:xv thorn-s ‘Smb. took out the thorn with an awl.’

h. ŋǒoom á-kɔ́l kʊ̂́ʊʊt-̪̄ ɔ̄ awl pst-take.out:xv:2s thorn-s ‘You took out the thorn with an awl.’

Illustration (3) evidences the difference between the High Fall and Low Fall tone classes, using the minimal set of the High Fall verb {lʊ̂́ʊɲ} ‘take turns’ (3a,c) and the Long Low Fall verb {lʊ̂ʊɲ} ‘pluck’ (3b,d). The verb {lʊ̂́ʊɲ} ‘take turns’ could be mistaken for a Low verb based on the Past 2nd singular form (3a). However, the Object voice Imperfective (3c) reveals that it is a member of the Long / High Fall class: only members of this class have a High Fall on the verb stem in this inflection. (3)

a.

c.

t ̪āal á-lʊ̀ʊʊɲ cook:infa pst-take.turns:2s ‘You took turns cooking.’

t ̪āal ʊ̀-lʊ̂́ʊʊɲ-ɔ́ cook:infa impf-take.turns ‘Somebody is taking turns cooking.’

b. gjɛ̀ɛɛn-ɔ̀ á-lʊ̂ʊʊɲ chicken-s pst-pluck:2s ‘You plucked the chicken.’ d. gjɛ̀ɛɛn-ɔ̀ ʊ̀-lʊ̂ʊʊɲ-ɔ̀ chicken-s impf-pluck ‘Somebody is plucking the chicken.’

3 Voice In the study of morphosyntax, the concept of voice refers to ways a verb form may mark the relation between the event that the verb expresses and the semantic roles that are expressed by its arguments (Payne 1996:169). In English, for example, if the verb form of a declarative clause is in active voice (e.g. ate), the language user infers that the preverbal argument represents the semantic subject; and if the verb form is in passive voice (e.g. was eaten), they infer that SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

15

the preverbal argument expresses the semantic object.7 Shilluk transitive verbs present three voices that are marked through inflection in the base paradigm: Object voice (OV), Subject voice (SV), and Applicative voice (XV). In Section 3.1, we lay out the main structural characteristics of the system. In Section 3.2, we provide a more detailed description and present evidence from narratives. Section 3.3, finally, covers the wider phenomenon of syntactic alignment, in the context of earlier work in which the Object voice has been interpreted as a passive (Westermann 1912, Tucker 1955) and as an ergative construction (Miller & Gilley 2001).

3.1 The structure of the voice system

Consider the question-answer sequences in (4). In each of these, the answer clause is headed by a form of the verb {càm} ‘eat’. The preverbal argument is the same in each case: the 3rd plural pronoun gɛ́. However, the semantic role of this argument is different in each case. In (4a), the verb is in Object Voice, and here gɛ́ refers to the semantic object, i.e., the vegetables. In (4b), the verb is in Subject voice, and now gɛ́ stands for to the semantic subject, i.e., people. In fact, the use of Subject voice is determined further by information structure, in that the event referred to by the verb is part of the shared framework of reference. This is marked by the underlining; we will come back to this in Section 3.2.2. In (4c), the verb is in Applicative Voice, and here gɛ́ refers to the semantic role of instrument (spoons). We label the voices after the constituent expressed before the verb, be it the semantic subject (Subject voice), semantic object (Object voice), or a different semantic role (Applicative voice). Subject voice is not marked in the morpheme glosses, as it displays the lexical root to the greatest extent (cf. Section 2). (4)

́ a. lùm á-gwɔ̂k kɪ ̀ dɪ ̄ gɛ́ á-cấm grass:p pst-make:ov prp how pr3p pst-eat:ov ‘What was done with the vegetables?’ ‘They were eaten.’ b. ɟɪ ̀ɪ á-càm ŋɔ̄ people pst-eat what ‘What did the people eat?’

gɛ́ á-càm lùm pr3p pst-eat grass:p ‘They ate vegetables.’

7 Throughout this chapter, we use the term semantic subject for the more agentive of the two arguments of a transitive verb, and the term semantic object for the less agentive one. The former term leaves vague whether the semantic subject is an Agent or an Experiencer. And likewise, the latter leaves vagye whether the semantic object is a Patient, a Goal or a Destination. Also, these terms are not specific as to whether these entities are expressed as core or peripheral arguments, which is often useful in the descriptive analysis. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs c. pʌ̂t á-gwɔ̂ɔ́ ɔk̄ ŋɔ̄ spoon:p pst-make:xv what ‘What were the spoons used for?’



16

gɛ́ á-cāaam lùm pr3p pst-eat:xv grass:p ‘With them the vegetables were eaten.’

As seen from these examples, syntactic voice is marked on the stem syllable of the verb, rather than through affixation. The forms of {càm} ‘eat’ in (4) are summarized in Table 8, alongside the corresponding forms of two other verbs. Irrespective of the specification for voice, the verbs carry a prefix á-, which marks Past tense. Voice is marked on the stem syllable through vowel length and tone. These patterns of exponence will be described in detail in Section 7, in conjunction with the other functions that are marked on the verb. At this point, it suffices to say that, in the Past tense and without subject marking, the Object voice stem form invariably has a High Fall / ́̂ /. In the Subject voice form, the tonal specification is either Low / ̀ / or Low Fall / ̂ /, depending on the lexical class the verb belongs to; and the Applicative voice form has either ̄ again as a function of verb class. Unless the Mid / ̄ / or High Fall to Mid / ́̂ /, verb belongs to a Fixed Short class, there is morphological lengthening of the stem vowel in the Applicative voice, whereas the other two voices display the lexical vowel length. Table 8. The three voices of transitive verbs in Past tense, illustrated by three verbs: Fixed Short Low {ŋɔ̀l}, Short with Grade Low {càm} ‘eat’, and Low Fall {mâat ̪} ‘drink’. {ŋɔ̀l} ‘cut’ Object voice

Subject voice

Applicative voice

á-ŋɔ̂ĺ á-ŋɔ̀l á-ŋɔ̄l

{càm} ‘eat’

á-cấm

á-càm á-cāaam

{mâat ̪} ‘drink’

á-mấat ̪

á-mâat ̪ á-mấaat ̪̄

The examples in (4) reveal several important characteristics of the Shilluk voice system. To begin with, note that the voice inflection of the verb reveals the semantic role of the preverbal argument: from the voice inflection on the verb, the language user can tell whether this argument expresses the semantic subject (be it Agent or Experiencer), the semantic object (be it Patient or Goal), or another semantic role, in this case Instrument. In this sense of its semantic role being signposted on the verb, the preverbal argument has a special status, i.e., it is privileged. Which semantic role is expressed in the preverbal slot is dependent on the discourse context: the preverbal argument tends to express an argument that is already part of the framework of SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

17

reference shared by speaker and hearer(s). This can be seen from the examples in (4), where each of the precursor questions sets up a different semantic role of the event as known information, which can then be referenced pronominally in the reply. Because the preverbal argument expresses shared information, it is appropriate to refer to this constituent as the topic. That is, the topic can be defined as the preverbal constituent, whose semantic role that is signposted inflectionally on the verb, and which typically expresses shared information. Also, note that while the semantic role of the topic is evident from the verb forms in (4), it is not evident from the argument itself: the 3rd plural personal pronoun gɛ́ displays no case marking. And while there is one morphosyntactic context in which personal pronouns are case marked (see Section 4), it is a general characteristic of Shilluk that there is no case marking at all on nouns or on noun phrases. That is, whether a noun or noun phrase represents the semantic subject, semantic object, or any other semantic role, these functions are not morphologically marked on the noun or noun phrase, neither in the topic position, nor elsewhere in the clause. This is illustrated in (5). In (5a,b), the noun djɛ̀l ‘goat’ is the semantic object – as a topical core argument preceding the verb in (5a), and as a core argument following the verb in (5b). In (5c,d), the same noun is the semantic subject – as a topic in (5c), and in a prepositional phrase in (5d). Importantly, there is no case marking on the noun djɛ̀l in any these positions. 8 (5)

a. djɛ̀l á-cấm ɪ ɪ̄ mʌ̂́ʌʌt ̪ goat pst-eat:ov prp:p friend:p ‘The friends ate the goat.’8

b. mʌ̂́ʌʌt ̪ á-càm djɛ̀l friend:p pst-eat goat ‘The friends ate the goat.’

c. djɛ̀l á-càm lùm goat pst-eat grass:p ‘The goat chose to eat grasses.’

d. lùm á-cấm ɪ ̀ɪ djɛ̀l grass:p pst-eat:ov prp goat ‘The goat ate grasses.’

Finally, illustration (4) shows that the voice system affects valence, the number of core arguments expressed in the clause. When the verb is in Object voice, as in (4a), the semantic object in the topic slot represents the sole core argument. In the other two voices, there are two core arguments. When the verb is in Subject voice, as in (4b), the semantic subject appears in the topic slot, and the semantic object follows the verb. In Applicative voice (4c), the core arguments that are expressed are as follows: a semantic role other 8 The choice for an active translation of this sentence into English will be explained in Section 3.2.1. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

18

than either semantic subject or semantic object is expressed in the topic slot, and the semantic object follows immediately after the verb. Any additional argument(s) to the clause are expressed as peripheral constituents. This means a) that they can be freely omitted; and b) if they are expressed, they are expressed through a prepositional phrase. The constituent expressing the semantic subject in clauses headed by a verb in Object voice or Applicative voice is a case point. Both in (5a) and in (5d), the prepositional phrase expressing the semantic subject can be omitted.9 The preverbal topic can also be omitted in main clauses, i.e., there is topic drop, but this phenomenon is restricted to 3rd singular topics. This is illustrated in (6). The noun càak ‘milk’ is grammatically plural.10 Hence it cannot be omitted in the answer, even if it is an established topic, as in (6a). Instead, it needs to be represented by a pronoun (gɛ́). In contrast, when the grammatically singular noun mɔ̀k-ɔ̀ ‘alcohol’ is used in the same context, as in (6b), topic drop is grammatical, and in fact its expression through a pronoun is ungrammatical.11 (6)

́ a. câak á-gwɔ̂k kɪ ̀ dɪ ̄ milk.p pst-make:ov prp how ‘What was done with the milk?’

gɛ́/*Ø á-mấat ̪ pr3p pst-drink:ov ‘It was drunk.’

́ á-gwɔ̂k kɪ ̀ dɪ ̄ b. mɔ̀k-ɔ̀ alcohol-s pst-make:ov prp how ‘What was done with the alcohol?’

*ɛ́/Ø á-mấat ̪ pr3s pst-drink:ov ‘It was drunk.’

3.2 Evidence for the voice system

In Section 3.1, the voice system is laid out using elicited examples. For the sake of accountability, we now provide evidence for the functions of the three voices using examples drawn from narratives, i.e., spontaneous speech (marked by ^ adjacent to the example number throughout this chapter). We also develop the descriptive analysis of the functions of these three voices. Object voice is discussed in Section 3.2.1, Subject voice in Section 3.2.2, and Applicative voice in Section 3.2.3. 9 The preposition ɪ ɪ̀  / ɪ ̄ɪ has two forms, reflecting number. If its argument is grammatically singular, it is Low-toned (ɪ̀ɪ); and if its argument is grammatically plural, it takes a Mid tone (ɪ ̄ɪ). 10 Many liquids are grammatically plural in Shilluk. This is a characteristic property of NiloSaharan languages (Zwarts 2007). 11 In serializations involving a 3rd person subject, the subject is referenced pronominally to the left of the second predicate head, and this pronoun cannot be dropped. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

19

3.2.1 Object Voice

In Section 3.1, we explained that Object voice is used when the semantic object is the discourse topic. This is correct, but it is not the whole story: the Object voice form and its associated constituent order, i.e., Object-Verb, are also used when no constituent stands out in terms of information structure (cf. Miller & Gilley 2001). These two functions are described and evidenced in turn in this section. The use of Object voice when the semantic object is topical is illustrated in (7) using an example drawn from a narrative. The referent of the noun phrase jấāmɛ̂j́ ̄-āa ‘my mother and her friends’ is introduced in the first clause, where it is dislocated and then resumed pronominally. The same referent is the topic in the second clause, where it is represented by the 3rd plural pronoun ́ t ̪} ‘greet’, which gɛ́. Here its referent represents the semantic object of {mʌ̂ʌ appears in Object voice. (7)^

jấā mɛ̂j́ ̄-āa gɛ́ bếeen̄ -ɔ̄ kɪ ̀ pôot ̪-ɪ ́ wɔ́n ass mother:prt-1s pr3p come-nevp prp territory-prt.p pr1pex gɛ́ á-mʌ̂́ʌt ̪ ɪ ̀ɪ ján pr3p pst-greet:ov prp pr1s ‘My mother and her friends, they had come from our place. I greeted them.’

As noted above, Object voice is also the unmarked morphosyntactic structure. That is, when the discourse context does not privilege any semantic role, then the verb is in Object Voice and the semantic object appears in the topic slot. This is evidenced in (8). The initial sentence is included to clarify the discourse context: here two friends become established as a topic. The second clause is reciprocal. Here the same referent, the two friends, represents both the semantic subject and the semantic object. As they are identical in reference, semantic subject and semantic object are equally topical, and therefore discourse structure does not dictate which should appear as the topic. Here, the unmarked constituent order can emerge: note that the verb is in Object voice, and the preverbal argument expresses the semantic object. The fact that Object voice is unmarked in terms of information structure explains why it is often more appropriate to translate Shilluk clauses with Object voice into English using active voice rather than passive voice. (8)^

cʊ̄ʊl gɛ́-kɪ ́ bʊ̌ʊl á-bèeet gǎa mʌ̂́ʌʌt ̪ kéeer à á-piiit ̂́ ̄ ̪ gɛ̂n Chol pr3p-prp Bol pst-stay 3p:nomp friend:p since rel pst-grow.up:xv pr3p.n ́ ̄ dwɔ̂ɔ́ ŋ rɪ ́ɪ-gɛ́n mấar ɪ ɪ̄ gɛ́n kɪ ́ mɛ̂n refl-pr3p love:ov prp:p pr3p prp idp.s big ‘Chol and Bol were friends from childhood. They love each other greatly.’ SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

20

Negative evidence on this unmarked word order is presented in (9). The question ‘What happened?’ sets up the answer as a whole as new information. In the answer to this question, Object Voice is felicitous, but Subject Voice is not. (9)

áa ŋɔ̄ à á-cwôop lùm á-cấm ɪ ̄ɪ ɟɪ ̀ɪ whq what foc pst-happen grass:p pst-eat:ov prp:p people ‘What happened?’ *ɟɪ ̀ɪ á-càm lùm people pst-eat grass:p ‘People ate vegetables.’

3.2.2 Subject Voice

A first condition on the felicitous use of Subject voice in Shilluk is that it is only used if the semantic subject of the clause is topical, as established through the preceding discourse. Negative evidence on this was presented in (9). In that example, Subject voice cannot be used felicitously because the semantic subject is not topical. Positive evidence from spontaneous speech is presented in (10). In this narrative, Nyikango, the founder of the Shilluk nation, has just fallen out with his elder brother. At this point in the story, he is the central character and therefore topical, which is critical to the felicitous use of Subject voice in the second clause. (10)^ kɛ̂ɲ à á-kɛ̂t ̪ líɲ-gɛ́ ʊ́-tèeŋ-ɔ̀ time:cs rel pst-go conflict-3p impf-become.hard

ɲɪ ́kāaaŋɔ̄ á-kwàɲ kwóp-ɪ ́ d̪wốoot ̪-ɔ̀ Nyikango pst-choose discussion-prt.p depart-inf ‘When their feud worsened, Nyikango opted for discussing departure [with other people in the community, to see who would be willing to move away with him].’

If the topical status of the semantic subject were the only factor determining the use of Subject Voice, we would come across it frequently in discourse. Crucially, this is not the case. Tucker (1955:432), writes that “[t]he Shilluk Transitive Verb has two Voices Active [= our Subject voice] and Passive [our Object voice], the latter being on the whole more in use than the former in every day conversation.” And Westermann (1912:78) writes: “[Shilluk people] generally prefer to speak in the passive [= Object] voice”. We share Westermann’s assessment that the use of Object voice predominates over the use of Subject voice. This suggests that the use of Object voice is constrained beyond the information-structural status of the semantic subject. On this issue, Miller SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

21

& Gilley (2001) write that Subject voice is pragmatically marked, and they hypothesize the following two functions. The first function they hypothesize is choice: “AVO order [our Subject voice] is used to indicate that the agent chose to accomplish an action with respect to a particular goal” (Miller & Gilley 2001:36). The second function is that Subject voice conveys “contrastive focus on the immediately following post-verbal constituent” (Miller & Gilley 2001:36). They illustrate these functions of choice and contrastive focus using the examples in (11a) and (11b), respectively. We display them using our own transcriptions and glosses, but with their translations. (11)

a.

ɲâan-dấaɟ ̄-ɔ̄ á-ràk bjɛ́l young-woman-s pst-grind.coarsely grain:p ‘The woman chose to grind the durra.’

b. ɲâandấaɟ ̄-ɔ̄ á-ràk a bjɛ́l young.woman-s pst-grind.coarsely foc grain:p ‘The woman ground the durra (not another grain).’

Miller & Gilley (2001) do not distinguish Subject Voice (their AVO) from Object Voice (their OVA) in terms of morphophonological form. So while they transcribe the Subject voice verb form in (11a,b) as á-ˈrākk ,̀ they use the same transcription for the corresponding Object voice form. In contrast, we hypothesize that the Subject Voice and Object Voice forms are different. For example, corresponding to the Subject voice in (11a,b), which we transcribe as á-ràk, the Object voice form is á-rấk. We include an example with the latter form in (12), and recommend that the reader plays and compares the sound files associated with (11a) vs. (12), paying particular attention to the verb forms. (12)

bjɛ́l á-rấk ɪ ̀ɪ ɲâan-dấaɟ ̄-ɔ̄ grain:p pst-grind.coarsely:ov prp young-woman-s ‘The woman ground the durra.’

In the remainder of this section, we will first describe how the use of Subject voice is marked, and then point out other morphosyntactic structures that are used when the semantic subject is topical. In relation to the first issue, we can start out from the functions of ‘choice’ and ‘contrastive focus’, which were hypothesized in relation to Subject voice in Miller & Gilley (2001). To begin with, we argue that contrastive focus is central to the function of the Subject Voice. This is evident from the fact that Subject Voice can be used when the topical subject is inanimate. This is illustrated in (13). This sentence conveys contrastive focus: the pot is singled out relative to other objects that the wind SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

22

might have broken, but didn’t. Note that, because the wind is inanimate, the hypothesized function of a choice is not available: lacking volition, the wind is not an Agent. Instead, the subject in the topic slot can be characterized as a Cause or Force. This suggests that, of the two functions postulated by Miller & Gilley, contrastive focus is the more widely applicable one. In fact, contrastive focus often implies choice when the semantic subject is an agent, to the effect that there is no need to hypothesize choice as a function separate from contrastive focus. (13)

jɔ̀ɔɔm-ɔ̀ á-nʌ̀k pǔk wind-s pst-kill clay.pot ‘The wind broke the pot.’ [rather than something else]

Focus and information structure in general represents a comprehensive topic in its own right. At this point, we limit ourselves to defining Subject Voice as a marker of focus in the sense of Krifka (2008:248): the property of Subject voice is a Focus property, because Subject voice signals that alternatives to the referent of one of the internal arguments, especially the one expressing the semantic object, are relevant to the interpretation of the clause. The way these alternatives are relevant can be characterized as exclusion. For example, the use of Subject voice in (11) entails not only that the pot was broken, but also that the alternatives were not broken. That is, the use of Subject voice leads to truth-conditional differences in the interpretation of the clause, in the same way the use of only in conjunction with a sentence accent in English (Krifka 2008:244). Similary, the clause in (10) entails that Nyikango did not go for other courses of action. This interpretation is evident from the fact that the clause in (13) is felicitous in response to a question that presents an alternative semantic object, as in (14). It is equally felicitous in response to jɔ̀ɔɔm-ɔ̀ á-nʌ̀k ŋɔ̄ ‘What did the wind destroy?’ Crucial here is the notion of singling out relative to alternatives. (14)

jɔ̀ɔɔm-ɔ̀ á-nʌ̀k lwɔ̂ɔl? wind-s pst-kill gourd ‘Did the wind break the gourd?

pât ̪, á-nʌ̀k pǔk neg pst-kill clay.pot No, it broke the pot.’ [rather than anything else]

The information-structural interpretation of Subject voice interacts with the expression of pragmatic focus through a/à. Using the framework of Krifka (2008), we hypothesize that this marker is about the management of the common ground: it is concerned with the way the common ground should develop. It can be associated with a variety of arguments to the clause. In SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

23

(15), the verb is in Subject voice, and there is pragmatic focus on the semantic subject. Crucially, there is no expression of semantic focus in relation to the object here, that is, this interaction does not entail that the wind broke only the pot, rather than anything else. (15)

́ ̄ à á-nʌ̀k pǔk áa mɛ̂n whq who foc pst-kill pot ‘Who broke the pot?

jɔ̀ɔɔm-ɔ̀ à á-nʌ̀k pǔk wind-s foc pst-kill pot The wind broke the clay pot.’

Importantly, Subject voice is not used when the semantic subject is topical, while the remainder of the clause represents new information. In that information-structural scenario, transitive verbs can appear in one of several other constructions. These are illustrated in (16) and (17). In each of these interactions, the question sets up the wind as the topic. The answers illustrate three different constructions that can be used in this information-structural setting, none of which involve Subject voice. First, a semantic subject that has been introduced previously in discourse can be marked pronominally on the verb through subject marking. This is shown in (16). Note that, when the verb is marked for subject, then the preverbal argument does not need to be topical. We describe subject marking in Section 4. The fact that it is the form of the verb which signposts whether the preverbal argument is topical is in line with the overall head-marking nature of Shilluk morphosyntax. (16)

́ áa ŋɔ̄ à á-gwɔ̂k ɪ ̀ɪ jɔ̀ɔɔm-ɔ̀ whq what foc pst-make:ov prp wind-s ‘What did the wind do?

pǔk á-nʌ̀ʌʌk-ɛ́ clay.pot pst-kill-3s It broke the pot.’

Second, there are two valency-decreasing derivations, whereby the semantic subject is retained as the sole core argument: the ambitransitive and the antipassive. They are illustrated in (17a) and (17b), respectively. The answers in both of these examples display topic drop – as noted above, singular topics may be omitted in main clauses. When these derivations are used, the semantic object appears as a peripheral argument, i.e., in a prepositional phrase. These derivations are beyond the base paradigm; they will be described in a later chapter, dedicated to the derivations of transitive verbs. At this point it suffices to say that the use of these constructions affects TAM and the definiteness of the internal argument that follows the verb. With respect to TAM, the ambitransitive conveys that the event took place once, and the antipassive often conveys a durative aspect. In (17b), the use of the antipassive SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

24

is incongruous with the breaking of a single pot.12 As for definiteness, both with the antipassive and with the ambitransitive, the expression of the semantic object as a peripheral argument conveys that it is indefinite, or only affected in part, e.g. when the referent is a mass noun (cf. Fillmore 1977, Hopper & Thompson 1980) (17)

a.

jɔ̀ɔɔm-ɔ̀ á-gwɔ̂k à ŋɔ̄? wind-s pst-make foc what ‘What did the wind do?

b. jɔ̀ɔɔm-ɔ̀ á-gòook kɪ ́ ŋɔ̄ wind-s pst-make:atp prp what ‘What was the wind doing?

á-nʌ̀k-ɪ ̀ kɪ ́ pǔk pst-kill-amb prp clay.pot It broke a pot.’ (among other pots) á-nʌ̀ʌʌk kɪ ́ pû́k̄ -ɪ ̄ pst-kill:atp prp clay.pot:p It was breaking pots.’

In conclusion, Subject Voice is used when the semantic subject is topical. But its use is further constrained, in that it conveys semantic focus (in the sense of Krifka 2008) on the object. If the semantic subject is topical but there is no semantic focus on the object, then other constructions are used: a subjectmarked form or a derived intransitive.

3.2.3 Applicative Voice

The Applicative Voice form of the verb is used in conjunction with the topic slot being filled by an argument expressing a semantic role other than the semantic subject or the semantic object. These roles include Instrument, Reason, and Deictic setting (Location or Time). However, they do not include the Beneficiary or the Destination, as these are expressed using productive derivations for Benefactive and Spatial deixis, which are beyond the scope of this chapter. When the Applicative voice is used, the semantic object immediately follows the verb, as a core argument that is obligatorily expressed. The semantic subject can be expressed optionally, by means of an optional peripheral argument, marked by the preposition ɪ ɪ̀  / ɪ ̄ɪ. In terms of valence, then, the verb marked for Applicative voice is accompanied by two internal arguments. Of these two, only the topic can be omitted, specifically if it is singular (topic drop). A first example from a narrative is presented in (18). The verb jwɔ̄ɔɔk is inflected for Applicative voice. The preverbal argument expresses the semantic role of Instrument (the bull to be used in a commemoration ceremony). The semantic object follows the verb.

12 This particular example is also compatible with the interpretation that the Antipassive derivation has an iterative function. To the best of our knowledge, this is not the case. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

25

(18)^ wɪ ̀ɪj ɲɪ -̄ jɔ̄ɔm jwɔ̄ɔɔk mʌ̂́ʌʌt-̪̄ ɪ ̄ɪ allow bull-cattle.with.white.forehead cry:xv friend:prt-2s ‘Let the bull with the white forehead be used to mourn your friend.’

A second example involves a serial verb construction. In (19), the topic of this clause is a woman called Apwodho. She is referenced before the first constituent verb in the serialisation (á-wéekɪ ̀), and again, resumptively, by the pronoun ɛ́, immediately before the second constituent verb, cʊ̂́ʊʊl,̄ which is in Applicative voice. Apwodho constitutes the semantic role of Instrument, used to compensate the hippo, the semantic object that follows the verb. (19)^ ápwǒoot ̪ɔ́ ɲɪ ̄ ɟɪ ̀ɪw ɲɪ ̄ mīi ʊ́kwɛ́r Apwodho daughter:prt Jiw daughter:prt father:prt Okwer á-wéekɪ ̀ ɲɪ ́kāaaŋɔ̄ ɛ́ cʊ̂́ʊʊl ̄ a pʌ̀ʌʌr pst-give:bnf:ov Nyikango pr3s pay.for:xv foc hippo ‘Apwodho, the daughter of Jiw, the sister of Okwer, was given to Nyikango in compensation for for the hippo.’

Illustration (20) shows the use of Applicative voice with the topic expressing the semantic role of a Reason. This sentence is a particular kind of questionword question, in which the Reason, marked by dɛ̄, appears before the verb, along with an independent pronoun referring to the queried entity, and the focus marker à. So the topic in this sentence is jí dɛ̄ à ‘why, in relation to you’. The referent of the pronoun in the topical Reason is invariably resumed in the clause, in this case by the 2nd singular pronoun jín. (20)^ jɪ ́ dɛ̄ à gɔ̄ɔɔc jɪ ́n kɪ ̀ nɔ̀w pr2s whq foc hit:xv pr2s prp like.that ‘Why did they beat you like that?’

Illustration (21) shows the use of Applicative Voice with the topic slot expressing a Location: kɛ̂ɲ à ɪ ̄ɪc-ɛ̄ ‘a place inside which’. As in the earlier examples illustrating Applicative voice (18–20), the verb is followed by the semantic object. (21)^ kɛ̂ɲ à ɪ ̄ɪc-ɛ̄ cāaam d̪òk ɛ́ dâa pwốoot-ɔ́ pāa dwāat-á place:cs rel inside-3s eat:xv cattle sub exsp.foc beat-inf neg want-1s ‘A place where cows are eaten, while there is beating, I don’t want it.’

Our corpus includes many instances of Applicative Voice in relative clauses. To the best of our knowledge, the constituent in the topic slot is the only argument that can function as a common argument with a superordinate clause. Illustration (22) shows the use of the Applicative Voice in a relative SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

26

clause, which is headed by cāaam ‘eat:xv’. The common argument, àkɛ̄ɛlɔ́ ‘sorghum noodles’, expresses the semantic role of Instrument in the subordinate clause, in addition to representing the semantic object of the main clause. (22)^ kấā ʊ̀-tʌ̀ʌʌt-ɔ̀ kɪ ́ àkɛ̄ɛlɔ́ mɛ́ cāaam mɔ̀ɔk conj impf-cook:atp prp sorghum.noodles rel.s eat:xv kind.of.fish ‘Then she cooks sorghum noodles to eat the fish with.’

In summary, the Applicative voice is a morphosyntactic operation whereby a semantic role other than subject or object is expressed as a core argument in the topic slot. As we will see the Applicative Voice form is also obligatorily used in conjunction with two particular levels of TAM: Sequential Past and Conditional. These will be discussed in Section 5.6 and 5.7, respectively.

3.3 Syntactic alignment in Shilluk 3.3.1 Evaluation of earlier analyses

Subject voice and Object voice have long been interpreted as active voice and passive voice, respectively (Westermann 1912:78, Tucker 1955:432). This interpretation is compelling from the angle of the morphosyntax. First, the verb stem is morphologically unmarked in Subject voice, and marked in Object voice. This is shown in (23), which contrasts Subject voice (23a) and Object voice (23b) constructions involving the same verb. In (23a), á-lɛ̀ɛŋ displays the underlying specification for tone of the Long Low verb {lɛ̀ɛŋ} ‘throw’. In the Object voice form (23b), in contrast, the verb stem is morphologically marked for voice: the High Fall on á-lɛ̂ɛ́ ŋ overwrites the lexical specification, to the effect that the difference in lexical specification between verb classes is neutralised in this inflection (cf. Table 6). (23)

a.

twɔ́ɔŋ á-lɛ̀ɛŋ tɔ́ŋ Twong pst-throw spear ‘Twong threw the spear.’

b. tɔ́ŋ á-lɛ̂ɛ́ ŋ ɪ ̀ɪ twɔ́ɔŋ spear pst-throw:ov prp Twong ‘Twong threw the spear.’

Second, there is the syntactic evidence. In the Subject voice, both the semantic subject and the semantic object are expressed as core arguments. This can be seen from (23a). In the corresponding Object voice construction, (23b), the semantic subject is expressed as a peripheral argument, which can be freely omitted. In sum, the Object voice presents all morphosyntactic characteristics of a passive construction (cf. Dixon 2012:206). SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

27

In contrast, Miller & Gilley (2001) have argued that the Object voice is an ergative construction. However, their analysis that syntactic alignment in Shilluk is ergative is based on the assumption that “there is no formal marker of passive voice indicated on verbal morphology” (Miller & Gilley 2001:52). Indeed, they transcribe and gloss the Subject voice and Object voice forms identically (e.g. Miller & Gilley 2001:36). However, as seen from a comparison of the forms in (23a) vs. (23b), (11a) vs. (12), and Table 6, these two patterns of syntactic alignment are morphologically distinct from one another. Additional evidence against the ergative analysis comes from the prepositional phrase which Miller & Gilley (2001) interpret as an ergative constituent. While the analysis of this constituent is outside the study of transitive verbs, this phenomenon is important to the interpretation of syntactic alignment, and for this reason we will present our findings here, in Section 3.3.2. Following that, in Section 3.3.3, we consider another topic that has a bearing on syntactic alignment, namely the marking of verb arguments in transitives vs. intransitives.

3.3.2 The status of the constituent headed by ɪ̀ɪ / ɪ̄ɪ

Miller & Gilley (2001) interpret the ɪ ̀ɪ/ɪ ̄ɪ constituent as ergative. If this analysis is correct, it should display characteristics of a syntactic subject. In the following subsections, we discuss four phenomena that have a bearing on this question.

3.3.2.1 Omissibility

In Subject voice and Object voice alike, the topic can be dropped in a main clause, on the condition that it is grammatically singular. This is illustrated in (24) for Subject voice. Note that the singular subject can be dropped in (24a), but not the plural one in (24b): if it is not expressed by a full noun, there needs to be a pronoun. (24)

a.

twɔ́ɔŋ/*ɛ́/Ø á-lɛ̀ɛŋ tɔ́ŋ Twong/pr3s pst-throw spear ‘Twong / He threw the spear.’

b. mʌ́ʌn/gɛ́/*Ø á-lɛ̀ɛŋ tɔ́ŋ women.p/pr3p pst-throw spear ‘The women / They threw the spear.’

The same applies to the topic expressing the semantic object in Object voice: it can be omitted if it is singular, but not if it is plural. This is shown in (25). SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs (25)

a.

tɔ́ŋ/*ɛ́/Ø á-lɛ̂ɛ́ ŋ spear/pr3s pst-throw:ov ‘Somebody threw the spear / it.’

28

b. tɔ̂ŋ/gɛ́/*Ø á-lɛ̂ɛ́ ŋ spear:p/pr3p pst-throw:ov ‘Somebody threw the spears /them.’

In contrast, the prepositional phrase that expresses the semantic subject in Object voice can be freely omitted. This is illustrated in (26). (26)

a.

tɔ́ŋ á-lɛ̂ɛ́ ŋ ɪ ̀ɪ twɔ́ɔŋ / Ø spear pst-throw:ov prp Twong ‘Twong / Somebody threw the spear.’

b. tɔ́ŋ á-lɛ̂ɛ́ ŋ ɪ ̄ɪ mʌ́ʌn /Ø spear pst-throw:ov prp:p woman.p ‘Women / They threw the spear.’

In summary, the omission of the ɪ ɪ̀  / ɪ ̄ɪ constituent differs from topic drop. The former is completely unrestricted; the latter is limited to 3rd singular topics. This difference is in line with the interpretation that the ɪ ɪ̀  / ɪ ̄ɪ constituent is a peripheral argument, and not a core argument whose omission is restricted.

3.3.2.2 Co-occurrence with subject marking

The hypothesis that the ɪ ɪ̀  / ɪ ̄ɪ constituent is ergative implies that it represents the syntactic subject. However, this is not necessarily the case, as seen from the examples in (27). They display subject-marked verb forms, in addition to the ɪ ɪ̀  / ɪ ̄ɪ constituent. In (27a), which is a spontaneously uttered utterance, the verb is intransitive; in the elicited example in (27b), the verb is transitive. (27) a.^ à-rjɛ́ɛɛr-ɛ́ ɪ ̀ɪ lɛ̂t ̪-ɪ ̀ tɔ́ŋ seqp-shout-3s prp pain-prt spear ‘And he shouted because of the pain from the spear.’

b. lùm á-càm gɛ̂n ɪ ̀ɪ kʌ̄c grass:p pst-eat pr3p.n prp hunger ‘They ate plants because of hunger.’

From this evidence it is clear that we need to postulate a peripheral ɪ ɪ̀  / ɪ ̄ɪ constituent which expresses the semantic role of a Cause or Force. Given the similarity between a Cause and Force on the one hand and Agent, it is parsimonious to postulate that there are not two homophonous markers with very similar meanings but instead just a single one.

3.3.2.3 The binding of anaphora

If the ɪ ɪ̀  / ɪ ̄ɪ constituent is ergative, its binding properties should be akin to those of core arguments. If, on the contrary, the binding properties of the ɪ ɪ̀  / ɪ ̄ɪ constituent are akin to those of prepositional phrases, then this supports the interpretation that it is a peripheral argument. In most contexts, the binding of anaphora in Shilluk is governed by SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

29

linearity: the interpretation of an anaphoric element is bound by a constituent that has been uttered before. This is illustrated in (28) by an example drawn from a narrative. The pronominal suffix in the noun phrase mɛ̂j́ ̄-ɛ̄ obligatorily refers to the mother of the aforementioned daughter. It cannot refer to somebody else’s mother. (28)^ kấā ɲāaan-ánɪ ́ pwɔ̂ɔ́ c ɪ ̀ɪ mɛ̂j́ ̄-ɛ̄ conj daughter-cs-def thank:ov prp mother:prt-3s ‘Then this daughteri was thanked by heri/*k mother.’

When the same constituent mɛ̂j́ ̄-ɛ̄ appears in the topic slot, as in the elicited example in (29), then the interpretation of the pronoun cannot be bound by the ɪ ɪ̀  / ɪ ̄ɪ constituent. That is, Abac cannot be the referent of the pronoun in mɛ̂j́ ̄-ɛ̄ ‘his/her mother’, it has to be somebody else’s mother.13 (29)

mɛ̂j́ ̄-ɛ̄ á-pwɔ̂ɔ́ c ɪ ̀ɪ ábác mother pst-thank:ov prp Abac ‘Abaci thanked his/herk/*i mother.’

The same state of affairs holds with reflexives: the interpretation of an anaphoric element in the topic cannot be bound by the ɪ ɪ̀  / ɪ ̄ɪ constituent. This is shown in (30). (30)

a.

(ábác) rɛ̄ɛ á-lɪ ̂ɪɪt-ɛ̀ ̪ Abac refl-3s pst-see-3s ‘Abaci/Shei looked at herselfi.’

b. *rɛ̄ɛ á-lɪ ɪ̂́ t ̪ ɪ ̀ɪ ábác refl-3s pst-see:ov prp Abac ‘Abaci looked at herselfi.’

In itself, this does not challenge the analysis that the ɪ ɪ̀  / ɪ ̄ɪ constituent is a core argument, as the same linearity effect can be observed with Subject voice. This is shown in (31). Here again, a pronominal referent later in the clause is bound by an earlier noun (31a), but the reverse relation does not hold (31b). (31)

a.

ábác á-pwɔ̂ɔc mɛ̂j́ ̄-ɛ̄ Abac pst-thank mother:prt-3s ‘Abaci thanked heri mother.’

b. mɛ̂j́ ̄-ɛ̄ á-pwɔ̂ɔc ábác mother:prt-3s pst-thank Abac ‘His/herk/*i mother thanked Abaci.’

All of the evidence considered so far suggests that linearity is the key factor determining the binding of anaphora. There is one construction, however, where a pronoun is licensed by a following argument within the same clause, 13 Abac is a name given to women. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

30

and where it matters whether this argument is core or peripheral (cf. Legate 2012). We already mentioned in Section 3.2.3 that question-word questions formed using dɛ̄ have in the topic slot the semantic role of Reason. The referent of this Reason is indexed pronominally in the preverbal topic, and it needs to be licensed by a core argument to the right of the verb. (As the topic is neither the semantic subject nor the semantic object, the verb is in Applicative voice, with or without subject marking.) In (32a); the pronoun in the Reason is licensed by the semantic object of the verb; in (32b), it is licensed by the semantic subject, subject-marked syntactically. (32)

a. gɛ́ dɛ̄ à mʌ̄ʌʌt ̪ gɛ́n ɪ ̀ɪ ɟấaak̄-ɔ̄ pr3p why foc greet:xv pr3p prp chief-s ‘Why is the chief greeting them?’ b. gɛ́ dɛ̄ à mʌ̄ʌt ̪ gɛ̂n ɟấaak̄-ɔ̄ pr3p why foc greet:xv pr3p.n chief-s ‘Why are they greeting the chief?’

However, the pronoun in the topic cannot be licensed by the semantic subject, if the latter is expressed in the ɪ ɪ̀  / ɪ ̄ɪ constituent. This is shown in (33). Note that the only difference between (32b) and (33) is the manner in which the semantic subject is expressed. (33)

*gɛ́ dɛ̄ à mʌ̄ʌʌt ̪ ɟấaak̄-ɔ̄ ɪ ̄ɪ gɛ́n pr3p why foc greet:xv chief-s prp:p pr3p ‘Why are they greeting the chief?’

Additional cases of this construction provide further support that the difference between core and peripheral arguments is critical here. Beneficiaries are expressed solely as core arguments, and these can be queried in the dɛ̄ constituent (34a), just as semantic objects can (34b).14 Illustration (34b) additionally shows that the queried referent can be expressed as a full noun in the preverbal constituent (kwʌ̄n). (34)

a. jí dɛ̄ à t ̪ʌ́lɪ ̀ jín kwʌ̄n pr2s why foc cook:bnf:ov pr2s porridge ‘Why is the porridge cooked for you?’

b. ɛ́ dɛ̄ kwʌ̄n à t ̪ʌ́lɪ ̀-ɪ ̄ɪ jín pr2s why porridge foc cook:bnf:ov-3s.obl pr2s ‘Why is the porridge cooked for you?’ 14 The examples in (34) display the benefactive derivation, which is beyond the scope of this chapter. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

31

Instruments, now, can be expressed either as a core argument or through a prepositional phrase. In the former case, that is, if the Instrument is expressed as a core argument, this argument can license the pronominal referent in the topic slot. This is shown in (35a), where the Instrument is expressed through a 3rd singular suffix on the verb.15 But when the Instrument is expressed as a prepositional phrase, headed by kɪ ́, it cannot license the pronoun in the topic: (35b) is ungrammatical. (35)

a. ɛ́ dɛ̄ pâal à cāaam-ɪ ̄ɪ kwʌ̄n pr3s why spoon foc eat:xv-3s.obl porridge ‘Why is the porridge eaten with a spoon?’

b. *ɛ́ dɛ̄ pâal à cāaam kwʌ̄n kɪ ́ ɛ́n / gɔ̀n pr3s why spoon foc eat:xv porridge prp pr3s / pr3s.obl ‘Why is the porridge eaten with a spoon?’

In summary, the question construction with dɛ̄ presents evidence that the ɪ ɪ̀  / ɪ ̄ɪ constituent is not a core argument. Like the constituent headed by kɪ ́, and unlike core arguments, the constituent headed by ɪ ̀ɪ/ɪ ̄ɪ cannot license the topic.

3.3.2.4 Evidence from -ɔ

A fourth argument regarding the prepositional status of the ɪ ɪ̀  / ɪ ̄ɪ constituent can be drawn from the distribution of the suffix -ɔ, which is part of the inflectional marking of Imperfective and Non-Evidential Past. The presence of this suffix is sensitive to the nature of the following argument, crucially, whether it is a core argument or a peripheral one. We will illustrate this using a verb form in the Spatial / Centrifugal derivation.16 A crucial characteristic of this derivation is that the Destination is represented morphosyntactically as an optional core argument. Note how the final -ɔ̀ is present in (36a), where the verb is in sentence-final position, but absent in (36b) when it is followed by the destination argument. (36)

a. kwʌ̄n ʊ́-câaam-ɔ̀ porridge nevp-eat:fug:ov ‘Smb. apparently went to eat porridge.’

b. kwʌ̄n ʊ́-câaam kàl porridge nevp-eat:fug:ov compound ‘Smb. apparently went to the compound to eat porridge.’

The verb-final -ɔ is not dropped before any following argument. Notice how it is present when the verb is followed by a prepositional phrase headed 15 This suffix -ɪ ̄ɪ is not to be confused with the preposition ɪ ɪ̀  / ɪ ̄ɪ. 16 The Spatial / Centrifugal derivation is not described in its own right in this chapter. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

32

by kɪ ̀ that expresses a deictic constituent, as in (37a), and likewise when it is followed by a prepositional phrase headed by kɪ ́, which expresses an instrument, as in (37b). (37)

a.

kwʌ̄n ʊ́-câaam-ɔ̀ kɪ ̀ lɛ́w porridge nevp-eat:fug:ov prp dry.season ‘Smb. apparently went to eat porridge in the dry season.’

b. kwʌ̄n ʊ́-câaam-ɔ̀ kɪ ́ pâal porridge nevp-eat:fug:ov prp spoon ‘Smb. apparently went to eat porridge using a spoon.’

Interestingly, if a grammatical marker is associated with a core argument, the -ɔ is dropped. As noted above, inanimate destinations, such as kàl in (36b), are not morphosyntactically marked. But animate destinations are marked by jìi/jǐii ‘a(nimate) d(estination)’. Its use is illustrated in (38), which can be compared with (36b). As the destination is a core argument, the verb does not have the suffix -ɔ; that is, the form with -ɔ is ungrammatical here. (38)

kwʌ̄n ʊ́-câaam / *ʊ́-câaam-ɔ̀ jìi bɔ̀ɔɔt ̪-ɔ̀ porridge nevp-eat:fug:ov ad craftsman-s ‘Smb. apparently went to the craftsman to eat porridge.’

From the above it is clear that the presence vs. absence of the suffix -ɔ depends on the status of the following argument. It is dropped if the verb is followed by a core argument, equally so if this core argument is preceded by a grammatical marker. Having established the heuristic, we can examine the state of affairs with the ɪ ɪ̀  / ɪ ̄ɪ constituent As seen from (39), if the ɪ ɪ̀  / ɪ ̄ɪ constituent follows immediately after the verb, the suffix is present, just as when the verb is followed by kɪ ̀ (37a) or kɪ ́ (37b), whose status as preposition is not contentious. This state of affairs is in line with the interpretation that ɪ ɪ̀  / ɪ ̄ɪ is a preposition marking a peripheral argument. (39)

kwʌ̄n *ʊ́-câaam / ʊ́-câaam-ɔ̀ ɪ ̀ɪ bɔ̀ɔɔt ̪ɔ̀ porridge nevp-eat:fug:ov prp compound ‘The craftsman apparently went to eat porridge.’

Evidence re. this phenomenon based on a narrative is presented in (40a). This sentence displays ôr-ɔ̀, the patient-oriented infinitive nominalization of the Spatial / Centrifugal derivation of the verb {òr} ‘send’. Like the NonEvidential Past and the Imperfective, this infinitive form carries the -ɔ suffix. In (40a), which is drawn from a narrative, the verb is immediately followed by the destination, a core argument. This explains the lack of the suffix -ɔ. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

33

In contrast, if the destination is left out, so that the infinitive is immediately followed by ɪ ̀ɪ, then this suffix is not dropped. This is shown in the elicited example in (40b). The realization of this sentence without -ɔ̀ on the verb is not grammatical. (40)

́ wɔ́ á-cɛ̂k kɪ ́=à ôr jìi-ɪ ̄ɪ ɪ ̀ɪ mɛ́j-wɔ́n a.^ ʊ̀ conj pr1pex pst-aux:ov prp=foc send:spt ad-2s prp mother-1pex ‘And our mother has sent us (to come) to you.’

́ b. ʊ̀ wɔ́ á-cɛ̂k kɪ ́=à ôr-ɔ̀ / *ôr ɪ ̀ɪ mɛ́j-wɔ́n conj pr1pex pst-aux:ov prp=foc send:spt prp mother-1pex ‘And our mother has sent us (to come).’

3.3.3 Case marking in transitives vs. intransitives

There is no case marking on the noun-phrase arguments of transitive verbs. But there is on pronominal arguments. This phenomenon is restricted a) to the expression of the semantic subject, i.e., nominative case; and b) to the position following the verb. We will describe this phenomenon in detail in Section 4. At this point, we make the comparison with intransitives. Case marking on pronominal arguments of transitive verbs is illustrated in (41), using the 3rd plural pronoun. This pronoun is unmarked when it expresses the semantic subject in the topic slot, as in (41a): this form, gɛ́, with High tone, is also used when the same pronoun expresses any other semantic role in the preverbal position. But when the same pronoun expresses the semantic subject following the verb, it is marked for case through tone. This is shown in (41b). (41)

a. gɛ́ á-càm djɛ̀l pr3p pst-eat goat ‘They ate the goat.’

b.

djɛ̀l á-càm gɛ̂n goat pst-eat pr3p:n ‘They ate the goat.’

Importantly, the same tonal case marking is found in relation to intransitives. Illustration (42a) displays the unmarked word order of intransitives, which is Subject-Verb. The pronoun is unmarked, just as it is in (41a), where the verb is transitive. There are some constructions that displace the subject of an intransitive from the preverbal position position; time adverbials are one of them. This operation is marked on the verb. As seen from (42b), when this happens the subject of an intransitive verb is case-marked in the same way as when the semantic subject of a transitive verb is positioned after the verb (41b).

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs (42)

a.

gɛ́ á-lʌ̂ʌʌɲ pr3p pst-disappear ‘They disappeared.’

34

b. kàa á-lʌ̂́ʌʌɲ̄ gɛ̂n sub pst-disappear:vs pr3p:n ‘When they disappeared, […].’

In conclusion, nominative case marking on pronouns does not single out the semantic subject of transitives; it equally applies to the semantic subject of intransitives.

3.3.4 Conclusion

Morphosyntactically, the Object voice has all the characteristics of a passive construction. It is morphologically marked on the verb, and the semantic subject is demoted. And yet, in terms of information structure, the Object voice is very much unlike a passive. Consider Dixon’s assessment (2012:222) that “[a]n active construction is always functionally unmarked, and is used in neutral circumstances. Passive or antipassive will only be employed to meet some specific syntactic, semantic or pragmatic purpose.” In Shilluk, it is the way around: the Object voice is used when the clause as a whole represents new information, and it is the Subject voice that is more constrained in terms of information-structural conditioning. For this reason, we agree with Miller & Gilley (2001) that the passive analysis does not offer an optimal match with the Shilluk phenomena, and this is why we have used Subject voice and Object voice instead. One part of the puzzle that has not been taken into consideration in this discussion in earlier work is the status of the Applicative voice. Like the Object voice, Applicative voice is morphologically marked on the verb, and it is used when a semantic role other than the subject is topical. In both voices, the semantic subject is expressed as a peripheral argument, using ɪ ɪ̀  / ɪ ̄ɪ. In this sense, the Shilluk voice system is akin to that of Austronesian languages, a parallel recently observed in relation to Dinka (Erlewine, Levin & van Urk 2017). In this context, it is worthwhile to note that, just as in those voice systems, the topic slot is privileged, in that extraction under relatvization is limited to this position. This helps to explain why the Applicative voice is primarily found in relative constructions: it is the only way to make a relative clause if the shared argument expresses neither the semantic subject nor the semantic object of the relative clause. An example is presented in (43). kɛ̂ɲ à pwōoot ɟɪ ̀ɪ kɪ ́ lʊ̀ʊt ̪ (43)^ t ̪ʌ̂ʌʌw lʌ̂w kɪ ̀ bɔ̂ɔl-ɪ ̀ date:p better prp front-prt.s place:cs rel hit:xv people prp stick ‘Desert dates are better than a place where people get beaten with a stick.’ SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

35

This characterization of syntactic alignment in Shilluk offers greater descriptive adequacy than both the active/passive analysis and the ergative analysis. However, it is certainly not a perfect correspondence, because Subject voice in Shilluk is heavily conditioned in information-structural terms.

4 Subject marking A transitive verb can be marked for its semantic subject. This marking is illustrated in Table 9, which shows the full range of subject-marked forms for Short with Grade {càm} ‘eat’, in Past tense. Note that there is a) a general pattern of subject marking, which is based in form on the Subject voice form, and b) subject marking based on the Applicative voice form. With general subject marking, the preverbal constituent expresses the semantic object; with Applicative subject marking, it expresses a semantic role other than subject or object. The preverbal argument never expresses the semantic subject when there is subject marking. That is, subject marking is not an instance of agreement marking: pronominal marking of the subject is in complementary distribution with the expression of the subject as a full noun phrase (cf. Creissels 2006). This means that Subject voice and subject marking are very different from one another in a morphosyntactic sense. The paradigms for the two types of subject marking – general and Applicative – are described in Sections 4.1 and 4.2, respectively. Table 9. Subject marking in the past tense of Short with Grade {càm} ‘eat’.

1st sg.

2nd sg. 3rd sg.

1st pl. inc.

1st pl. exc. 2nd pl. 3rd pl.

Subject marking Subject marking (general pattern), Subject marked Applic. voice (general pattern) with focus

Subject marked Applic. voice, with focus

á-càaam-á á-càaam á-càaam-ɛ́ á-càm wāa á-càm wɔ̂n á-càm wûn á-càm gɛ̂n

á-cām a jân á-cām a jîn á-cām a ɛ̂n á-cām a wāa á-cām a wɔ̂n á-cām a wûn á-cām a gɛ̂n

á-càm a jân á-càm a jîn á-càm a ɛ̂n á-càm a wāa á-càm a wɔ̂n á-càm a wûn á-càm a gɛ̂n

á-cāaam-á ̌ am á-cāa á-cāaam-ɛ́ á-cām wāa á-cām wɔ̂n á-cām wûn á-cām gɛ̂n

In relation to the formal realization, the key thing to note about subject marking is that it is of an inflectional nature in part of the paradigm, but syntactic elsewhere. That is, in some forms, the subject marker is realized SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

36

as a bound morpheme on the verb, through affixation and/or stem-internal changes. In other forms, the marker is best conceived as an independent pronoun, case-marked through tone for its role as a semantic subject, i.e., nominative. This difference is represented in Table 9 through hyphenation. As seen from this table, we postulate that plural subject marking is invariably syntactic. Note that the plural subject markers can invariably be separated form the verb by the focus marker. In the singular, we find syntactic subject marking when there is a focus marker associated with the pronominal subject, and inflectional subject marking when there is not. In the following subsections, we will present a descriptive analysis of this system of pronominal subject marking. There are two important questions to be addressed here. First, there is the formal status of the subject markers: is the marking inflectional or syntactic? We will motivate the distinction between inflectional vs. syntactic subject marking using the following criteria, which relate to the degree of cohesion and mutual dependence between the verb stem and the pronominal marker: • Is the pronominal subject marker seperable from the verb stem, i.e., can another element intervene? If so, this supports the interpretation that it is an independent word. If not, this support the interpretation that it is an affix. • Does the verb stem present a particular form when it is followed by the pronominal marker? If so, this supports the interpretation that we are dealing with inflection. • Does the pronominal marker present a particular form when it appears in a juncture with the verb? And is this form dependent of the class of the verb?17 Affirmative answers to these criteria support the interpretation that we are dealing with inflection. Second, there is the question of the function, which is especially relevant to the general subject-marked forms.

4.1 Subject marking (general pattern)

Subject marking in Shilluk mixes characteristics associated with Subject voice with those associated with Object voice. The morphological form of a subject-marked verb is predictable on the basis of the corresponding Subject voice form, but the associated syntactic structure of the clause is the same 17 We thank an anonymous reviewer for pointing this criterion out to us. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

37

as that of Object Voice. This is illustrated in (44). Illustration (44a) presents an example of a clause headed by a subject-marked verb form. Note that the stem of the verb, i.e., càm, is identical to the Subject voice stem, which is also càm (44b), and different from the corresponding Object Voice stem, which is cấm (44c). In spite of this formal similarity between the subject marked form (44a) and the Subject voice form (44b), the semantic object preceeds the verb in the subject-marked construction, just as it does in Object Voice (44c), and markedly different from the corresponding Subject voice construction (44b). These are the essential characteristics of the general pattern of subject marking: morphologically, it is based on the Subject Voice form, but like Object voice it displays object-verb constituent order. (44)

a. djɛ̀l á-càm gɛ̂n goat pst-eat pr3p:n ‘They ate the goat.’

b. djɛ̀l á-càm gɛ́n goat pst-eat pr3p ‘The goat ate them.’

c. djɛ̀l á-cấm ɪ ɪ̄ gɛ́n goat pst-eat:ov prp:p pr3p ‘They ate the goat.’

The only difference distinguishing the subject-marked construction in (44a) from the corresponding Subject voice construction in (44b) is the case marking on the pronominal subject in the former: the Low Fall on gɛ̂n signposts unambiguously that this pronoun expresses the semantic subject. This case marking, i.e., nominative, is only found on pronominal subjects: there is no nominative case marking on nouns. In any other context, the personal pronouns are High-toned – see e.g. (44b,c). It is not the case that the pronoun gɛ́n is case-marked for nominative whenever it expresses the semantic subject. This can be inferred from (44c), where it expressed the semantic subject in a prepositional phrase. The same point is illustrated further in (45a,b), where the same pronoun expresses the semantic subject as a core argument preceding the verb. (45)

a. gɛ́ á-càm djɛ̀l pr3p pst-eat goat ‘They ate the goat.’

b. gɛ́ á-cʌ̀m-ɪ ̀ kɪ ́ djɛ̀l pr3p pst-eat-amb prp goat ‘They ate some of the goat.’

While clauses with subject marking and those with Subject voice do not match up in terms of constituent order, there is a similarity in terms of informationstructure. Note that, for a semantic subject to be expressed pronominally, it needs to be part of the framework of reference shared by speaker and hearer. In other words, the semantic subject is inherently topical in subject marked clauses, even though it does not appear in the preverbal slot. As for the information-structural status of the preverbal constituent in clauses with subject marking, we find that it can represent new information. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

38

This is illustrated by the narrative example in (46). The semantic subject of the second clause, a man called Chol, is well-established as the central character of the narrative at this point. In the first clause, he is referenced pronominally ́ ʌt-ɛ̄ ̪̄ ‘friend:prt-3s’). In the second clause, he is as a noun possessor (mʌ̂ʌ still referenced pronominally, through subject marking on the verb (cwɔ̀ɔɔl̄ ‘Abac, his wife’) preceeds the ɛ̀ ‘call:3s’). The semantic object (ábác cjếek-ɛ̄ verb. Crucially, however, its referent is not topical in the information-structural sense: this is the first time in the narrative that Abac has been referred to. (46)^ rùu-ɪ ̀ wɔ̄ɔw kɛ̂ɲ à á-bếeen̄ jwɔ̂ɔ́ k̄-ɪ ̄ mʌ̂́ʌʌt-̪̄ ɛ̄, arrive:dvn-prt situation time:cs rel pst-come:xv mourn:inf-prt friend:prt-3s kấā ábác cjếek̄-ɛ̄ cwɔ̀ɔɔl-ɛ̀ conj Abac wife:prt-3s call:3s ‘As the time arrived to commemorate his friend, he called Abac, his wife.’

This example shows that, in subject marked clauses, the preverbal argument does not need to represent shared information. In this respect subject-marked clauses are fundamentally different from those without subject marking: in the latter the preverbal argument is an established discourse topic (cf. Section 3). As seen from (46), this is not necessarily the case when subject marking is involved. The remainder of this section presents a descriptive analysis of the morphosyntactic nature of subject marked constructions. The key question is: how should the sequence of a verb followed by the subject marker be interpreted? One possibility is to interpret it as a suffix. In this analysis, the verb in (44a) is not á-càm but rather á-càm-gɛ̂n, i.e., inflected for 3rd plural. Such inflectional interpretations have been advanced in earlier work (Miller & Gilley 2001; Remijsen, Miller-Naudé & Gilley 2016). If we were to start out from the axiom that subject marking applies uniformely across levels of person, number and clusivity, then this inflectional analysis is inevitable, because, as will be pointed out below, subject marking is clearly inflectional in forms marked for a singular subject. Another argument in support for the inflectional analysis is the restricted distribution of the casemarked form of the personal pronouns. The case-marked forms are not used whenever the pronouns express the semantic subject, but only when they do so as an internal argument following the verb. However, we argue against this analysis for a juncture like á-càm + gɛ̂n, because the verb can be separated from the subject marker, specifically by the focus marker. This is illustrated in (47a). This function morpheme can appear in a variety of syntactic positions in the clause, suggesting that it is not a SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

39

bound morpheme. For example, the focus marker can also be inserted between a OV verb and its semantic subject, expressed through a prepositional phrase (47b). (47)

a.

djɛ̀l á-càm a gɛ̂n goat pst-eat foc pr3p:n ‘They ate the goat.’

b. djɛ̀l á-cấm a ɪ ̄ɪ gɛ́n goat pst-eat:ov foc prp:p pr3p ‘They ate the goat.’

In fact, all seven of the levels of number, person and clusivity can be used in this way, i.e., with the focus marker positioned between the the verb stem and the subject marker: á-càm a jân / jîn / ɛ̂n / wāa / wɔ̂n / wûn / gɛ̂n (cf. Table 9). The verb is in the Subject voice form, and does not interact with the subject marking in any way. Based on these considerations, we interpret subject marking in these forms as being syntactic in nature, involving a form the personal pronoun that is case-marked for nominative. As seen from Table 9, singular subjects are marked on the verb in a different way when there is no intervening focus marker. In the case of the verb {càm} ‘eat’, the forms are á-càaam-á, á-càaam, and á-càaam-ɛ́, respectively. The alternative construction, i.e., syntactic subject marking, is only found when the focus marker intervenes between the verb and the subject marker. This is illustrated in (48).18 (48)

a.

djɛ̀l á-càaam-á goat pst-eat-1s ‘I ate the goat.’

b. djɛ̀l á-càm a jân goat pst-eat foc pr1s:n ‘I ate the goat.’18

Above we noted that the morphological form of the verb that is used with syntactic subject marking is identical to the Subject voice form. Inflectional marking of singular subjects is derived from the same stem form, modified as follows: first, the stem vowel is in the long vocalic grade; second, the specification for tone on the suffix in the 1st and 3rd singular forms depends on the class the verb belongs to. These suffixes (-a,-ɛ) are High-toned when the verb belongs to one of the Low classes, and Low otherwise. For example, the Long Low verb {lɛ̀ɛŋ} ‘throw’ has the Past tense 1st singular form á-lɛ̀ɛɛŋá, whereas the Long High Fall verb {mʌ̀ʌt ̪} ‘greet’ has á-mʌ̀ʌʌt̪-à in the same inflection. Finally, in the case of the 2nd singular, the pattern of marking is purely stem-internal, i.e., there is no suffix. 18 Here again, case marking through tone on the pronoun is crucial for to the interpretation of semantic argument structure. If the pronoun is not case-marked for nominative, the sentence is interpreted as Subject Voice, with a topical semantic subject: djɛ̀l á-càm a ján ‘The goat ate me.’ SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

40

It makes sense to treat syntactic and inflectional subject marking together, because these constructions are treated in the same way in the grammar. We will come back to this in Section 6 and in Section 6.2. On the basis of the criteria set out in the introduction to Section 2, we interpret as inflectional the markers of singular subjects without focus marking (cf. Table 9). First, these subject markers are inseparable from the verb stem. Second, the form of the verb stem interacts with subject marking: unless the stem is Fixed Short class, the stem vowel displays morphological lengthening. Third, the 1st and 3rd subject markers are heavily reduced forms of the personal-pronoun forms, and their specification for tone interacts with the verb class system. Finally, in the case of the 2nd singular, the marking is purely stem-internal, which precludes a syntactic interpretation altogether. Fourth, the interaction between verb class and the specification for tone on the pronominal subject marker lends support to the interpretation that the latter is a suffix.

4.2 Subject marking based on Applicative Voice

Like the Applicative voice forms, the subject-marked Applicative voice forms are used when the topic expresses a semantic role other than subject or object, and in other constructions that require the use of Applicative voice (see Sections 5.6, 5.7). And also just as the Applicative voice form, a subjectmarked Applicative voice form is invariably followed by a core argument expressing the semantic object. This is evidenced by the narrative example in (49). At issue here is the second clause: note how the topic slot accommodates an instrument (ɲɪ ́ŋ-ɪ ̀ ‘name:p-cs’, the common argument of the main clause and the subordinate), and the semantic object is expressed following the verb (pɪ ́ɲ ‘ground’). (49)^ gɛ́ cwɔ̂l-ɪ ̀ gáa àkǒool běeen, à jɪ ́ŋ ɲɪ ́ŋ-ɪ ́ d̪wʌ̀ʌŋ mɔ́ɔɔ-gɛ́n pr3p call-iter pr3p:nomp leader:p all rel nomp name.p-prt status idp.p:prt-3p ɲɪŋ̄́̂ -ɪ ̄ à ɲấk̄ gɛ̂ pɪ ́ɲ name-cs rel fight:xv pr3p:n ground ‘They were all called leaders, which were their titles, the titles they used fighting for the land.’

In relation to the general pattern of subject marking, we have argued for an inflectional interpretation of singular subject marked forms when there is no focus marking, and for a syntactic interpretation otherwise. We postulate the same analysis in relation to subject marking in the Applicative voice. The SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

41

relevant forms are illustrated in Table 10. Critical to our interpretation is the interaction with focus marking. The singular subject marked applicative voice forms without focus marking involve a pronominal form that is either heavily reduced or stem-internal. These two units cannot be separated by the focus marker. That is, if the focus marker intervenes, a different stem form is used. These arguments (inseparability; the reduced nature of the pronominal element) support an inflectional interpretation of subject marking in relation to the singular forms without focus marking. Table 10. The paradigm of subject-marked Applicative voice forms, illustrated by {càm} ‘eat’. Subject marked Applic. coice á-cāaam-á ̌ am á-cāa á-cāaam-ɛ́ á-cām wāa á-cām wɔ̂n á-cām wûn á-cām gɛ̂n

Subject marked Applic. voice, with focus á-cām a jân á-cām a jîn á-cām a ɛ̂n á-cām a wāa á-cām a wɔ̂n á-cām a wûn á-cām a gɛ̂n

Elsewhere, i.e., when the subject is plural and/or there is a focus marker, we find a different stem form. Illustration (50a) presents elicited data showing that the focus marker can intervene between the stem and the pronominal element; (50b) shows the parallelly with general subject marking, which equally allows for the focus marker to intervene. (50)

a. pâal á-cām (a) gɛ̂n kwʌ̄n spoon pst-eat:xv foc pr3p:n porridge ‘They / They used the spoon to eat porridge.’

b. kwʌ̄n á-càm (a) gɛ̂n porridge pst-eat:ov foc pr3p:n ‘They / They ate the porridge.’

Just as in relation to general subject marking, we use seperability as the critical argument to determine which instances of Applicative subject marking are inflectional and which are syntactic. The evidence is more finely balanced, however, as the subject-marked stem form displays the short vocalic grade (e.g. á-cām gɛ̂n), whereas the Applicative voice on which these subjectmarked forms are based is in the long vocalic grade (e.g. á-cāaam). We consider the evidence from separability to be decisive. It may well be that diachronically, the forms are becoming inflectional. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

42

5 Tense-Aspect-Modality Seven different levels of Tense-Aspect-Modality (TAM) are expressed inflectionally in the verb paradigm. Many other dimensions of TAM are expressed in other ways, e.g. lexically, through clause-level markers, auxiliaries, and serialisation. The scope of this section is restricted to the former, i.e, to the morphological expression of TAM. These seven levels are illustrated in Table 11 for the Fixed Short Low verb {càm} ‘eat’. Note that the inflections for TAM involve affixation, vowel length, and tone. Past, No Tense and Future tend to have the same stem form, and are distinguished by prefixes: á- for Past, ʊ́- for Future, and none for the No Tense form. Note how these TAM prefixes cross orthogonally with the voice marking on the stem. Table 11: The levels of TAM marking, illustrated by Short with Grade {càm} ‘eat’. Subject Voice Past (OV) No-tense (OV) Future (OV) Non-evidential past (OV) Imperfective (OV) Sequential past Conditional

á-càm cấ̀m̄ ʊ́-cấm̄ ʊ́-càaam

Object Voice á-cấm ́ câm ́ ʊ́-câm

ʊ́-càaam-ɔ̀ ʊ̀-càaam-ɔ̀

Applicative Voice á-cāaam cāaam ʊ́-cāaam ʊ́-càaam à-cāaam ʊ̀-cāaam

Imperfective and Non-Evidential Past share the same stem form, which involves the long vocalic grade, and also the same prefix, but they differ in the tonal specification of this prefix. It is ʊ̀- for Imperfective, and ʊ́- for Non-evidential past. The Imperfective is only available for the Object voice. Sequential past and Conditional, finally, are only available with Applicative voice form. Both have a Low-toned prefix: à- for Sequential past, ʊ̀- for Conditional. In the following subsections, we briefly describe and illustrate each of the seven inflectionally marked dimensions of TAM, focusing on their function. A detailed description of how these levels are marked in the different combinations of voice, subject marking and verb class will follow in Section 7.

5.1 Past Tense Illustration (51) evidences the use of the past-tense form of the verb.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

43

́ (51)^ jɛ́n̪ mɔ́k-ánɪ ́ á-mấat ̪ ɪ ̀ɪ ján gɛ́ bǎa á-bîikjɛ̀ l tree:p idp.p-def pst-drink:ov prp pr1s pr3p nomp card-six ‘Those medicines I took six of them.’ (lit. Those medicines I drank, they were six.)

5.2 Future Tense

Had the sentence in (51) been uttered with future tense reference, the sentence would have been as in (52) below. The only difference is the TAM prefix on the verb. (52)

́ jɛ́n̪ mɔ́k-ánɪ ́ ʊ́-mấat ̪ ɪ ̀ɪ ján gɛ́ bǎa á-bîikjɛ̀ l tree:p idp.p-def fut-drink:ov prp pr1s pr3p nomp card-six ‘Those medicines, I will take six of them.’

Illustration (53) shows the use of Future Tense in a narrative. (53)^ kấā kɛ̀-lɔ́k já ʊ́-jwɔ̂ɔ́ k, mǔuuc kɪ ́ wâat ̪ conj sub-turn pr1s fut-cry:ov give:2s prp bull ‘And when my final funeral rite will be held, offer a bull.’

5.3 No Tense

In Object Voice, the No Tense form has the same stem shape as the corresponding Past and Future tense forms, but then without affixation. Consider the illustration in (54), which is drawn from the same narrative as as (51). Here we find the same stem form as in (51), i.e., mấat ̪, now without a TAM prefix. Whereas (51) has specific time reference for past, there is no specific time reference in (54). (54)^ ʊ̀ gɛ́ mấat ̪ gɛ́ bǎa á-rjɛ̄ɛw á-rjɛ̄ɛw á-rjɛ̄ɛw conj pr3p drink:ov pr3p ncop card-second card-second card-second ‘And they are taken (lit. drunk) two of them three times per day (lit.: two two two.)’

Illustration (55) presents a second example of the No Tense form, now with Subject Voice. ́ ɲʊ̂ʊt ̪ à (55)^ mɛ̂n kóookɪ ̄ɪ mɪ ɪ́̂ ̄ jàp-ɪ ̀ cám idp.s show foc reward idp.s:prt:s search:inf-prt eat:infa

kɪ ̀ bấaaŋ̄ ŋàan mɛ̌ɛɛkɔ́ prp behind person.cs other ‘This one (story) shows the outcome of looking for food from another person.’

The use of the No Tense is felicitous only if the clause contains a licensing SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

44

constituent. In (54) this is a serialisation; in (55) it is the focus marker. This interaction between the use of the No Tense inflection of the verb and other constituents of the clause will be described in Section 6.

5.4 Non-evidential Past

The past tense is characterised more accurately as evidential past, because, alongside it, the inflectional paradigm includes another TAM form that has several functions relating to non-evidentiality, inference, and contestation. We gloss this morphological form as Non-Evidential Past; the functions are summarised briefly below. This phenomenon is described in detail in Miller & Gilley (2007). Our investigations corroborate their analysis. A first function of the Non-Evidential Past is illustrated in (56). The verb á-rùm ‘pst-think’ conveys that the assertion in the subordinate clause is based on conjecture, rather than on eyewitness observation. Hence the use of the past tense form á-cấm is ungrammatical in the subordinate clause. Instead, the verb is inflected for Non-Evidential Past, which signposts that the assertion is non-evidential. Accordingly, the use of the non-evidential past is obligatory with clauses that are inherently non-evidential, such as {kwìc} ‘not know’. (56)

gɛ́ á-rùm kɪ ̀nɪ ̀ kwʌ̄n ʊ́-càaam-ɔ̀ / *á-cấm pr3p pst-think quot porridge nevp.ov-eat / pst-eat:ov ‘They thought that the porridge was eaten.’

Similarly, Non-Evidential Past is used when describing a hypothetical situation in the past, i.e., a counterfactual or subjunctive conditional in the sense of Kaufmann (2006). This is illustrated in (57). The context here is that a man killed a hippo, the hunting of which is restricted in Shilluk culture. Against this background, the sentence in (57) expresses the hypothetical situation whereby the man had killed a kind of antelope instead, which would not have been problematic. Note that the verb in the subordinate clause is Non Evidential Past, in particular the Subject voice form, which does not carry affixes. (57)

kɛ̀-lɔ́k ɛ́ nʌ̀ʌʌk a ŋɛ́r, pāa dɪ ́ cʊ̂́ʊl ɪ ̀ɪ ɛ́n sub-turn pr3s kill:nevp foc white.eared.kob neg irr pay.for:ov prp p3rs ‘If he had killed a white-eared kob, he would not have to pay for it.’

The Non-Evidential Past has other functions that are related to nonevidentiality. One of these is that it conveys perfective and/or inference. Consider the example in (58), which is drawn from a narrative, in which a SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

45

woman describes a scary encounter she had when she was a child. She came across a strange creature in a field. In the cited utterance, she describes how this creature was covered in a black cloth, and she uses the Non-Evidential Past, here with subject marking (kùm-ɛ̀ ‘cover:nevp-3s’). The verb is used here with perfective meaning, i.e., the covering event is presented as completed: she reports the outcome of it (the creature being covered), rather than the event in process (the covering in process), which is inferred. (58)^ gìn-ánɪ ́ rɪ ́ɪ-gɔ̀ kùm-ɛ̀ kɪ ́ ʊ́gɔ̂t́ ̪ mɛ́-lʊ́ʊʊc̀ thing.cs-def refl-pr3s.obl cover:nevp-3s prp cloth mdf-black:ctg ‘That thing, it had covered itself with a dark cloth.’

This perfective-aspect function of the Non-Evidential Past is also in evidence in (59). Here a speaker described a glass bottle standing on a table, without its cap. (59)^ àd̪ʌʌ ́ t ̪ ʊ́-jɛ̀ɛɛp-ɔ̀ bottle nevp.ov-open ‘Somebody opened the bottle (and left it open).’

The perfective meaning of the Non-Evidential Past is clear in comparison with the use of the Past Tense in the same environment. This can be seen from (60); these sentences differ only in the TAM marking on the verb in the subordinate clause, which is Non Evidential Past in (60a), and Past in (60b). The main clause implies eyewitness observation. In (60a), the use of the Non Evidential Past in the subordinate clause conveys that the agent witnesses e.g. emptied bowls, and infers that the porridge was eaten up, i.e., an earlier non-witnessed event. In contrast, the use of the Past Tense in the subordinate clause in (36b) is felicitous in a situation whereby the event represented by the subordinate clause (the eating) is on-going at the time when the event of the main clause takes place. (60)

a.

́̂ ̪ á-lɪɪt ɪ ̄ɪ gɛ́n kɪ ̀nɪ ̀ kwʌ̄n ʊ́-càaam-ɔ̀ pst-see:ov prp.p pr3p comp porridge nevp.ov-eat ‘They saw that the porridge had been eaten up.’

́̂ ̪ ɪ ̄ɪ gɛ́n kɪ ̀nɪ ̀ kwʌ̄n á-cấm b. á-lɪɪt pst-see:ov prp:p pr3p comp porridge pst-eat:ov ‘They saw that the porridge was being eaten.’

In other words, the use of the Non Evidential Past in (60a) conveys – through its perfective function – that the time setting of the inferred event in the subordinate clause precedes the past tense time setting of the main clause SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

46

(pluperfect). This interpretation equally applies in (58): the visit to the field is set in the past, and the inferred covering of the creature is inferred to have taken place before that. A third function of the Non-Evidential Past is contestation. This interpretation is available in relation to events the speaker is highly likely to have an eyewitness account on, e.g. an event they participated in themselves. Illustration (61) shows corresponding sentences; in (61a) the verb in the subordinate clause is in the Non-Evidential Past; in (61b) it is in the Past tense. The use of Non-Evidential Past in (61a) conveys that the speaker disagrees with what was asserted about them. (61)

a.

gɛ́ á-kôoop kɪ ̀nɪ ̀ kwʌ̄n càaam jân pr3p pst-say:atp quot porridge eat:nevp pr1s:n ‘They said that I ate the porridge.’ (and I disagree)

b. gɛ́ á-kôoop kɪ ̀nɪ ̀ kwʌ̄n á-cấm ɪ ̀ɪ ján pr3p pst-say:atp quot porridge pst-eat:ov prp pr1s ‘They said that I ate the porridge.’

5.5 Imperfective

As the terms suggests, the imperfective inflection represents the event as an on-going process. Illustration (62) presents an example from a song. The first line presents a generic statement, and here the verb is marked by the habitual marker ɲɪ ́. In the second line, the verb ʊ̀-cɔ̂ɔɔŋ-ɔ̀, from Low Fall {cɔ̂ɔŋ} ‘dance’ is in the Imperfective. This clause refers to the practice whereby an age set of young men dances with an age set of young women from a different village.19 Note that the referent event has no specific end point.20 ́ ̪̄ (62)^ ɲɪ ̄ɪɟáak mấaat-āa, mʌ́ʌn ɲɪ ́ tʌ̂ŋ kɪ ́=à d̪wʌ̀ʌʌŋ bùul, pấat ̪ kɪ ́ jāat ̪, Nyijak friend:prt-1s women hab win.over:ov prp=foc dress drum neg prp tree20 ‘Nyijak my friend women are won over by dance attire, not by magic,’ kwâa kɪ ɪ̄ l ʊ́-bɔ̂w ʊ̀-cɔ̀ɔɔŋ-ɔ̀ ɪ ɪ̄ wɔ̀ɔɔp descendant:prt Kil Obow impf-dance prp:p young.man:p ‘(We) young men dance with the descendant of Kil Obow.’

A clause whose verb is inflected for imperfective does not necessarily convey present tense setting. This is illustrated in (63), which is drawn 19 In this case, the age set of the women is referred to by their leader, i.e., the descendant of Kil Obow. 20 The meaning of jāat ̪ ‘tree, plant’ extends to ‘medicine, cure, magic’ (e.g. ɔ̄ɔt-jāat ̪ ‘housetree’ means hospital). SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

47

from the same song. Here we find, in the first clause, the Imperfective used in a context that has past-tense time reference. This past tense reference is revealed in the subsequent clause, where the same event is referenced through the same verb, this time inflected for past tense. Evidence liks this supports the interpretation of the function of this inflection as Imperfective, i.e., as aspect rather than tense. This interpretation goes back to Miller & Gilley (2001). (63)^ gâaar-ɔ̀ ʊ̀-bàaaɲ-ɔ̀ ɪ ̄ɪ mʌ́ʌn, ʊ́-múr ɲɪ ́ŋ-ɪ ́ gɛ́ lʊ́ʊʊc̀ dead.wood-s impf-reject prp:p women masc-vagina face-prt:p pr3p black:ctg ‘The women were rejecting the dead-wood men, bastards whose faces are dirty.’

wʊ̄ʊ-mʌ́ʌn á-bʌ̀ɲ-ɪ ̀, pàláan̪-ɪ ̀ à nɪ ̀ ɲâaŋ, gɪ ̀n à bûut kɪ ̀ nâam women-women pst-reject-amb Fulani-cs rel like crocodile thing:cs rel lie prp river ‘The women rejected: “The Fulani are like a crocodile, something that lies in the river.’

The imperfective is rare in our corpus, and we do not yet fully understand why. One relevant observation is that the addition of the focus marker to a clause whose verb is in the Imperfective is ungrammatical. This restricts the functional range of the Imperfective. Another is that there are other constructions that leave tense unspecified. One such construction involves the use of the habitual marker ɲɪ ́, as in (62). Another is the No Tense form of the verb (cf. Section 5.3).

5.6 Sequential Past

The use of the Sequential Past is felicitous when the event expressed by the clause follows in close succession after another event. This is illustrated in (64). At this point in the narrative, the storyteller has just laid out how the protagonist has reported the death of his friend to the head of the village during the night. The use of the sequential past conveys that the event of the clause took place in sequence after this. It can often be translated well into English using ‘and then’. (64)^ kɪ ̀ mwɔ̄ɔ̌ ɔl à-cwɔ̄ɔɔl ɟɪ ̀ɪ gɛ́-kɪ ́ ɲɪ ́ɪɪ-mīii bʊ̌ʊl ɪ ̀ɪ ɟấaak̄-ɔ̄ prp morning seqp-call:xv people pr3p-prp offspring-mother:prt Bol prp chief-s ‘[And he went straight to the chief of their village. He woke him up, and he told him the story of how an animal had killed his friend Bol in the forest.] After that, in the morning, the chief called all the people, and the brothers of Bol.’

The verb stem is that of the Applicative voice, and the prefix is Low-toned à-, whereas the past tense marker á- carries a High tone. Just as in other constructions involving the Applicative voice form, the semantic object follows SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

48

after the verb. In (64), this is ɟɪ ̀ɪ gɛ́-kɪ ́ ɲɪ ́ɪɪ-mīii bʊ̌ʊl ‘the people and Bol’s brothers’. Illustration (65) presents a second example of the Sequential Past, now with the verb additionally inflected for subject marking. (65)^ kấā ʊ́gɪ ̀ik rɛ́ŋ(-ɪ ́) a jǐii gɛ́n, à-rwɔ̄ɔm gɛ̂ gɔ̀n kɪ ́ tɔ̂ɔɔŋ conj buffalo run-fug:nt foc ad:p pr3p seqp-hit.simult:xv pr3p:n pr3s.obl prp spear:p ‘The buffalo ran towards them, and then they hit it simultaneously with their spears.’

5.7 Conditional

A clause headed by a verb inflected for the conditional TAM form expresses a hypothetical state in the present or in the future. The verb form has the stem form of the Applicative Voice, and a Low-toned prefix ʊ̀-. Syntactically, the verb appears in clause-initial position, and the semantic object follows the verb. The syntactic positioning of the object following the verb is a characteristic that the conditional inflection shares with the Sequential past and with other clauses that involve the Applicative voice form. Two examples are presented in illustration (66). (66)

a.

ʊ̀-kēě el djêk, gɛ́ ʊ́-cʊ̂́ʊl ɪ ̀ɪ jín cond-spear:xv:2s goat:p pr3p fut-pay.for:ov prp pr2s ‘If you spear goats, you will have to pay for them.’

́ b. ʊ̀-kēě el djêk, gɛ́ cɛ̂k kɪ ́=a cʊ̂ʊʊl-ɔ̀ cond-spear:xv goat:p pr3p aux:ov:nt prp=foc pay.for-inf ‘If somebody spears goats, they have to be paid for.’

The conditional, as a morphological form of the verb, is not used with past-tense reference. This is illustrated in (67). The sentence in (67a) is ungrammatical: it includes a condition that is set in the past and marked morphologically through the conditional inflection. Instead, this meaning is expressed using the subordinate clause structure of a time adverbial, as in (67b). (67)

a.

*ʊ̀-kēeel lʌ̂ʌj, rîŋ-ɔ́ á-ɲwấak cond-spear:xv animal meat-s pst-share:ov ‘If somebody speared an animal, the meat was shared.’

b. kɛ̂ɲ à á-kēeel lʌ̂ʌj, rîŋ-ɔ́ á-ɲwấak time:cs rel pst-spear:xv animal meat-s pst-share:ov ‘When / If somebody speared an animal, the meat was shared.’

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

49

6 Syntactic licensing Most verb forms that do not carry prefixation for TAM require the presence of another element within the clause that supplies the aspectual setting. We refer to this as ‘syntactic licensing’ of the verb form. Among the levels of TAM outlined in Section 5, there are two that do not (always) involve a TAMmarking prefix. One is the No Tense form, which is never marked for TAM; the other is the Non-Evidential Past, which involves a TAM prefix in some combinations of Voice and TAM, but not in others. The phenomenon is illustrated in (68), which shows a) a Subject voice No Tense form; b) an Object voice No Tense form; c) a subject marked No Tense form; and d) a Non-Evidential Past form. If the focus marker a/à is not there in each of these clauses, they are all ungrammatical, in the sense that these sentences are judged to be incomplete or unfinished by native speakers. In this way, the focus marker licenses the use of these verb form that do not carry a TAM prefix. And there are several other licensers; we will go into them further on in this section. (68)

a. dɛ̄ɛŋ cấm̄ *(a) lùm Deng eat:nt foc grass:p ‘Deng eats the vegetables.’

b. *lùm *(à) cấm grass:p foc eat:ov:nt ‘The vegetables are being eaten.’

c. lùm *(à) cáaam̀ -ɛ̀ grass:p foc eat:nt-3s ‘He eats the vegetables.’

d. dɛ̄ɛŋ càaam *(a) lùm Deng eat:nevp foc grass:p ‘Deng apparently eats the vegetables.’

The base-paradigm forms that require syntactic licensing are listed in Table 12. As noted above, it is the No Tense form and the Non-Evidential Past where the verb may appear without affixation. Table 12 illustrates, for each combination of Voice and subject marking, whether it is grammatical without syntactic licensing. As seen from this table, the No Tense forms require syntactic licensing in all combinations of Voice and Subject marking. In relation to the subject-marked forms, it does not matter whether the marking is inflectional or syntactic. For example, there is *lùm cáaam̀ -ɛ̀, with the subject marked inflectionally, and the corresponding plural *lùm cấm̄ gɛ̂n, where the subject is marked syntactically. Both are ungrammatical, indicating that, for the grammar, it does not make a difference whether Subject marking is inflectional or syntactic.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

50

Table 12: Grammaticality/ungrammaticality of different combinations of voice, subject marking and TAM without syntactic licensing in declarative clauses. No Tense

Non Evidential Past

Subject Voice

*cấm̄ lùm eat:nt grass:p

*càaam lùm eat:nevp grass:p

Subj.-marked Subject Voice

*lùm cáaam̀ -ɛ̀ grass:p eat:nt-3s

✓ lùm càaam ɛ̂n grass:p eat:nevp pr3s:n

Object Voice

*lùm cấm grass:p eat:ov

Irrelevant, carries TAM prefix

Applicative Voice

*pâal cāaam lùm spoon eat:xv grass:p

Irrelevant, carries TAM prefix

Subj.-marked Applic. Voice

*pâal cāaam-ɛ́ lùm spoon eat:xv:nt-3s grass:p

✓ lùm càaam ɛ̂n grass:p eat:nevp pr3s:n

As for the Non-Evidential Past forms, these carry a TAM prefix in the Object Voice (e.g. ʊ́-càaam-ɔ̀) and in the Applicative Voice form (e.g. ʊ́-càaam). As a result, these forms do not require syntactic licensing. In contrast, syntactic licensing is required for the Subject voice form. Surprisingly, the NonEvidential Past subject-marked forms are grammatical in the absence of TAM inflection, i.e., they do not require syntactic licensing (cf. Table 12). This is illustrated by the narrative example in (69).21 (69)^ gì twôooc ɛ̂n ɪ ɪ̄ c ʊ́gɔ̂t́ ̪ pr3p.obl tie:spat:nevp pr3s:n inside cloth ‘He had tied them inside the cloth.’

How can the need for syntactic licensing be explained? The data suggest that declarative clauses – i.e., statements rather than commands or questions – need a specification for TAM, and for aspect in particular. Most often, this requirement is fulfilled through inflection. However, it can equally be fulfilled through the syntax or through the lexicon. As for the syntactic licensers, the one that occurs the most frequently is the focus marker à/a. Another is the infinitival adverb. These will be discussed in Sections 5.1 and 5.2, respectively. But there are others as well, and the following examples reveal that the common characteristic of syntactic licensers is that they specify aspect. Consider the sentence in (70a). The verb is in the No Tense form, and it is ungrammatical, in the sense that it is judged ‘incomplete’. 21 In this example, the verb appears in a derivation for spatial deixis. This is orthogonal to the phenomenon at issue. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

51

However, if the internal argument is reciprocal, as in (70b), it is grammatical. Reciprocity implies that the event extends over time, thereby specifying an aspectual scope. Concretely, in this example, there are at least two instances of annoying. (70)

a.

*kùl kɔ̂ĺ ɪ ̄ɪ gɛ́n Kul annoy:ov prp:p pr3p ‘They disturb Kul.’

b. rɪ ́ɪ-gɛ́n kɔ̂ĺ ɪ ̄ɪ gɛ́n refl-pr3p annoy:ov prp:p pr3p ‘They are disturbing one another.’

Another syntactic licenser is serialisation.22 This is illustrated in (71a,b). In (71a), the No Tense Object voice form is licensed by an adjectival predicate component; in (71b), the No Tense Subject voice form is licensed by an intransitive predicate. Note that, in both cases, serialisation offers an aspectual setting for the referent of No Tense verb form, by relating it to the referent of the second constituent of the serialisation. (71)

a. kwʌ̄n cấm ɛ́ léeet ̪̀ porridge eat:ov pr3s hot:ctg ‘The porridge is eaten while hot.’

ʊ́-cùŋɔ̀ b. ábác cấm̄ kwʌ̄n ɛ́ Abac eat:nt porridge pr3s impf-stand ‘Abac eats porridge while standing.’

The above examples illustrate syntactic licensing. Alternatively, the specification for TAM may be satisfied lexically. Some transitive verbs regularly appear without either TAM marking or a licenser in the syntactic environment, even in declaratives. Consider the examples in (72). The verbs {mấar} ‘love’ in (72a) and {cɛ̀t ̪} ‘detest’ in (72b) can be used in the No Tense form, without syntactic licensing. What distinguishes them from verbs like {kɔ̀l} ‘annoy’ in (70a) and {twấaɲ} ‘betray’ (72c) is that ‘love’ and ‘detest’ are dispositions that have temporal extent, inherent to the lexical meaning, which is not the case for ‘annoy’ and ‘betray’. (72)

a.

twɔ́ɔŋ mấar Twong love:ov ‘Twong is loved.’

b. twɔ́ɔŋ cɛ̂t́ ̪ Twong detest:ov ‘Twong is detested.’

c.

*twɔ́ɔŋ twấaɲ Twong betray:ov Twong is being betrayed.’

Other verbs that pattern along with {mấar} ‘love’ and {cɛ̀t ̪} ‘detest’ ́̂ ́ t ̪} include {mấan} ‘hate’, {wɔ̂ɔ́ r} ‘respect’, {bɪɪw} ‘despise, disrespect’, {ŋʌ̂ʌ ́ ́ ́ ‘trust’, {wɔ̂ɔj} ‘neglect’, {bʊ̂ʊr} ‘hold in higher esteem’, {lʊ̂ʊt} ‘surpass’, {bwɔ̂ɔ́ n} ‘consider unworthy’, {tjấam} ‘defeat’, {mʊ̂ʊt} ‘distrust, envy’, {kʊ̂ʊr} ‘protect’. Aside from the last two and {cɛ̀t ̪}, all of these belong to 22 Serialisation can be defined by the presence of two lexical roots serving as predicates within the same clause. In such constructions, any tense marking is expressed on the first verb, and the topic is resumed by a pronoun before the second constituent head. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

52

the Long High Fall class. As noted in Section 2, verbs that typically take a human semantic object tend to be part of this class. At the same time, this phenomenon is not determined by verb class. First, the set includes {mʊ̂ʊt} ‘distrust, envy’, {kʊ̂ʊr} ‘protect’, and {cɛ̀t ̪} ‘detest’, all three of which belong to classes other than Long High Fall. Second, there are also Long High Fall verbs that typically take a human semantic object, and which nonetheless require syntactic licensing, such as {twấaɲ} ‘betray’. Evidently, it is not about verb class. Instead, the shared characteristic of verbs that can appear without inflectional TAM and without syntactic licensing is that their lexical semantics express an emotion or attitude that is not punctual, but instead extends over time. This fits with the hypothesis that the crucial requirement is aspectual setting, which can be satisfied lexically, through inflection, or by syntax. The requirement of syntactic licensing does not apply to imperatives and yes/no-questions. These non-declarative speech acts can be used systematically without tense marking.

6.1 Focus marking

The role of focus marking in syntactic licensing is evidenced by the narrative example in (72). The verb is in the Subject voice No Tense form, and the focus marker is associated with the argument that follows the verb (kóookɪ ̄ɪ ‘reward’). ́ ɲʊ̂ʊt ̪ a kóookɪ ̄ɪ mɪɪ̂́ ̄ (73)^ mɛ̂n jàp-ɪ ̀ cám kɪ ̀ bấaaŋ̄ ŋàan mɛ̌ɛɛkɔ́ idp.s show foc reward idp.s:prt search:inf-prt eat:infa prp behind person other ‘This one (story) shows the outcome of looking for food from another person.’

A clause can include no more than a single instance of the focus marker. When a constituent following the verb is focused, as in (48), the focus marker is positioned before it; and when the topic is focused23, the focus marker appears between the topic and the verb For the No Tense form to be syntactically licensed in this way, it does not matter where in the clause the focus marker appears. This is illustrated in (74). All four of the sentences are ungrammatical if the focus marker is left out. The 23 The focus marker can be associated with the preverbal constituent, which we have analysed as the topic. This is problematic in terms of the axiom that topic and focus are mutually exclusive, i.e., that a given constituent cannot represent both (Kroeger 2004:151). It may be that we are dealing with a contrastive topic in such situations. This question calls for a detailed investigation of Shilluk information structure in its own right. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

53

structure becomes correct if the focus marker is associated with any argument, be it core, as in (74a,b), or peripheral, as in (74c,d). (74)

a.

dɛ̄ɛŋ cấm̄ a kwʌ̄n Deng eat:nt foc porridge ‘Deng eats porridge.’

b. kwʌ̄n à cấm ɪ ̀ɪ dɛ̄ɛŋ porridge foc eat:ov:nt prp Deng ‘Deng eats the porridge.’

c.

kwʌ̄n cấm a ɪ ̀ɪ dɛ̄ɛŋ porridge eat:ov:nt foc prp Deng ‘Deng eats the porridge.’

d. kwʌ̄n cấm kɪ ́ à pâal porridge eat:ov:nt prp foc spoon ‘The porridge is eaten with a spoon.’

An aside on the form of the focus marker. It has two allomorphs. It is Lowtoned, i.e., à, in most environments, including those in (74,b,d), but toneless a when it follows immediately after the verb. In the latter environment, it gets its specification for tone by copying it from the end target of the preceding syllable. This means that it is realised with Mid tone in in the juncture cấm̄ a (74a), but with Low tone in cấm a (74c). The fact that it is associated with the constituent to its right in terms of its function in the clause but to the constituent to its left in a phonological sense fits a widely observed crosslinguistic pattern (Himmelmann 2014). In conclusion, focus marking interacts with the TAM system in Shilluk. Similar phenomena have been reported in Hyman & Magaji (1970), Hyman & Watters (1984) and Andersen (1988). Hyman & Magaji describe how in Gwari, tenses other than perfective are found with a focus marker or without it. Perfective tense in contrast requires the presence of the focus marker. Hyman & Watters develop the analysis of these and similar phenomena, by distinguishing between focus in its widely-used information-structural sense, as opposed to focus as a syntactic phenomenon. Andersen (1988) reports that in Päri, a language closely related to Shilluk, clauses in which the verb is marked for past tense cannot have the focus marker. Andersen concludes that the past tense marker is itself a focus marker. Here the notion of a grammatical feature of focus, as postulated in Hyman & Watters (1984), can be useful: if focus is conceived of as a grammatical feature, and a sentence can include only one instance of this feature, the ungrammaticality of the the focus marker in clauses that carry the past tense marker follows. The above description suggests that in Shilluk, it is TAM marking that is crucial, and that the presence of the focus marker satisfies this requirement, on a par with with morphological TAM marking, lexically inherent aspectual characteristics, and other syntactic licensers. It remains to be investigated what the interaction between focus marking and the No Tense form means for information structure. In a clause with a SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

54

verb in Past tense, the focus marker is solely an information-structural device, highlighting information. In a clause with a verb in No Tense it licenses the verb, so that it can conceivably be interpreted as a TAM marker. While the No Tense and in some cases the Non Evidential Past can be licensed by the focus marker, the Imperfective also interacts with focus marking, but in the opposite way: the cooccurrence of Imperfective inflection on the verb with the focus marker is not grammatical. This is illustrated in (75). This sentence was uttered without focus marking in a narrative. It is not grammatical for the focus marker to be included here, be it on the semantic object in the topic slot or on the semantic subject in the prepositional phrase. (75)^ kwâa kɪ ̄ɪl ʊ́bɔ̂w (*à) ʊ̀-cɔ̀ɔɔŋ-ɔ̀ (*a) ɪ ̄ɪ wɔ̀ɔɔp descendant:prt Kil Obow foc impf-dance foc prp:p young.man:p ‘(We) young men dance with the descendant of Kil Obow.’

6.2 Infinitival adverb

A verb form without TAM can be licensed by an infinitival adverb. The phenomenon is illustrated in (76a,b), in each case in the second clause, which is the main clause. In these examples, the main clause is headed by a subject̀ ). This marked verb form, which is in the No Tense form (cáaam̀ -ɛ̀, náaak-ɛ̀ subject-marked verb form is followed by an infinitive of the same verb root. The morphological characteristics of the infinitive will be described in Section 7.6. This infinitive, which repeats the subject-marked verb, is the constituent that provides syntacticlicensing for the No Tense form. It is used as an adverb, expressing emphasis on the event expressed by the verb, as suggested by the fact that this constituent is repeated. That is, just as in the case of the focus marker à/a, we find again here a connection between focus marking and TAM (cf. Andersen 1988, Hyman & Watters 1984). (76) a.^ kɛ̀ ɲwáaal-ɪ ́ɪ, jí cáaam̀ -ɛ̀ à càaam-ɔ̀ sub touch:xv:nt:2s-3s.obl pr2s eat:nt-3s foc eat-inf ‘And if you touch him, he will eat you up.’

b.

jɛ̀l-ù gɛ́n, gɔ̀ nʌ́ʌʌk̀-ɛ̀ à nʌ̀ʌʌk-ɔ̀ separate-2p.imp pr3p 3s.obl kill:nt-3s foc kill-inf ‘Separate them! He is surely about to kill him.’

The use of the infinitival adverb is only grammatical if the verb form carries subject-marking. In this context, it is worthwhile to note that it does not matter whether the subject is marked through inflection, as in (76) or SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

55

syntactically, as in (77). This equivalence shows that syntactic subject marking is equivalent to inflectional subject marking. (77)

kwʌ̄n cấm̄ gɛ̂ à càaam-ɔ̀ porridge eat pr3p.n foc eat-inf ‘They are definitely going to eat porridge (right now).’

As for the expression of TAM, the use of the infinitival adverb with the No Tense form conveys that the event is about to happen. Aside from the No Tense, this constructuion can also be used in Non-Evidential Past (78a), in Future tense (78b), and in Past tense (78c). (78)

a. já á-lìŋ ɪɪ́̂ kwʌ̄n càaam gɛ̂ à càaam-ɔ̀ pr1s pst-hear:atp indir porridge eat:nevp pr3p.n foc eat-inf ‘I heard that they definitely ate the porridge.’

à càaam-ɔ̀ b. kwʌ̄n ʊ́-cấm̄ gɛ̂ porridge fut-eat pr3p:n foc eat-inf ‘They will definitely eat porridge.’ c. á-kwìc-á kwìc-ɔ̀ PST-not.know-1S not.know-INF ‘I totally did not know.’

The interpretation that this construction conveys focus on the verb is supported by the fact that, while the focus marker can be included in the clause, it can only mark the infinitival adverb, which is co-referent with the verb. This is illustrated in (79). Note that the sentence is ungrammatical if the focus marker appears before the verb, where it would mark the topic, and following the verb, where it would mark the pronominally expressed semantic subject. In contrast, the focus can be added before the infinitival adverb. (79)

kwʌ̄n (*à) cấm̄ (*a) gɛ̂ (à) càaam-ɔ̀ porridge foc eat:nt foc pr3p.n foc eat-inf ‘They are definitely going to eat porridge (right now).’

Andersen (1988) describes a similar construction in Päri. Illustration (80) is cited from Andersen (1988:293). Note that, like the Shilluk constructions in (76–79), the Päri example has the verb repeated at the end of the clause, and Andersen labels this constituent as a verbal adverb. The main difference between the Shilluk construction and its Päri counterpart is that in Shilluk, the infinitival adverb construction is available only if the verb carries subject marking.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs (80)

56

jòobì kêel ùbúrr-ì kèel-ò buffalo shoot Ubur-erg shoot:suf ‘Ubur will shoot the buffalo.’

The two constructions also present a similarity similar in terms of focus marking. Andersen notes that Päri clauses in which the infinitival adverb appears cannot take focus marking, and he infers that the infinitival adverb is a focus marker itself. As noted above, the corresponding Shilluk construction does allow for the presence of a focus marker, but it can only be associated with the infinitival adverb. This indicates that in Shilluk as well, focus on the verb is implicated when the infinitival adverb is used, a hypothesis which is also supported by native-speaker interpretations of the meaning sentences that present this construction.

7 The base inflectional paradigm

In the preceding sections, four factors have been introduced that together determine the structure of the base inflectional paradigm. They are Verb class (Section 2), Voice (Section 3), Subject marking (Section 4), and Tense-AspectModality (Section 5). The last of these factors interacts with syntactic licensing (Section 6), which is not itself a factor in the inflectional paradigm. Now we will build on this groundwork to describe the inflectional paradigm.

7.1 The Subject Voice forms

Table 13 illustrates the Subject voice forms of the inflectional paradigm, by verb class and by Tense-Aspect-Modality (TAM). The Subject voice forms are available in four levels of TAM: Past, Non-Evidential Past, No Tense, and Future. Across verb classes, Past tense is marked by the prefix á-, and Future tense by the prefix ʊ́-. The Non-Evidential Past displays the long grade wof the stem vowel (unless the verb is Fixed Short). Table 13: The Subject voice forms by TAM and verb class. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’.

NEvP NoTns Future

Low

ŋɔ̀l ŋɔ̂ĺ ̄ ʊ́-ŋɔ̂ĺ ̄

Fixed Short Fall

lɛ̂ŋ lɛ̂ŋ ʊ́-lɛ̂ŋ

Short with Grade

Low

càaam cấm̄ ʊ́-cấm̄

Fall

mʌ̂ʌʌl mʌ̂l ʊ́-mʌ̂l

Low

lɛ̀ɛɛŋ lɛ̂ɛ́ ŋ̄ ʊ́-lɛ̂ɛ́ ŋ̄

Long

Low Fall

mâaat ̪ mâat ̪ ʊ́-mâat ̪

High Fall

mấaal mâal ʊ́-mâal

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

57

The tonal specification on the stem syllable depends on the verb class. The three Low Fall verb classes display their lexical specification (Low Fall) on the stem syllable in all four TAM forms. It follows that in the case of Low Fall verbs that are Fixed Short, there is syncretism between Non-Evidential Past and No Tense forms, because these verbs do not display a morphological vowel length alternation. For the three Low verb classes, we find their lexical specification (Low) on the stem syllable in Past and in Non-Evidential Past, but not in the No Tense and Future forms, which have a High Fall to Mid. The High Fall verbs, finally, pattern along with the Low Fall verbs in all inflections other than the Non-Evidential Past. It is only the latter inflection that reveals the specification that is specific to this verb class, i.e., High Fall. Finally, the No Tense and Non-Evidential Past forms require syntactic licensing (cf. Section 6). Note that there is no single form in the Subject voice paradigm that displays the lexical specifications for both tone and vowel length. Instead, we find the lexical specification for tone of the verb root – Low, Low Fall or High Fall – reflected in the Non-Evidential Past, and the lexical specification for vowel length – Short or Long – in the other three levels of TAM (Past, No Tense and Future).

7.2 The Object Voice forms

The Object voice part of the base paradigm, summarized in Table 14, presents an additional level of TAM, as compared to the Subject voice and to the subject-marked forms: the Imperfective (cf. Section 5.5). The Object voice forms have one stem form in Past, No Tense, and Future, and another in NonEvidential Past and Imperfective. In Past, No Tense, and Future, the stem vowel is in the short vocalic grade; and the tone on the stem syllable is the High Fall for all verb classes, i.e., replacing the lexical specification. In the Non-Evidential Past and Imperfective, the stem vowel is in the long vocalic grade, and here the tone reflects the lexical specification: Low for Low verbs, Low Fall for Low Fall verbs, and High Fall for High Fall verbs. The affixes marking Past and Future are á- and ʊ́-, respectively, just as in Subject Voice. Non-Evidential Past and Imperfective have ʊ́- and ʊ̀-, respectively, and the difference in tone is the only difference between these verb forms. The latter two TAM levels also have a weakly-realised suffix -ɔ, Low-toned for all classes other than Long High Fall, where it is High-toned instead. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

58

Table 14: The Object voice forms by TAM and verb class. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Fixed Short

No Tense Past Future NEvP Impf

Low

ŋɔ̂ĺ á-ŋɔ̂ĺ ʊ́-ŋɔ̂ĺ

ʊ́-ŋɔ̀l-ɔ̀ ʊ̀-ŋɔ̀l-ɔ̀

Fall

́ lɛ̂ŋ ́ á-lɛ̂ŋ ́ ʊ́-lɛ̂ŋ

ʊ́-lɛ̂ŋ-ɔ̀ ʊ̀-lɛ̂ŋ-ɔ̀

Short with Grade

Low

cấm á-cấm ʊ́-cấm ʊ́-càaam-ɔ̀ ʊ̀-càaam-ɔ̀

Fall

Low

ʊ́-mʌ̂ʌʌl-ɔ̀ ʊ̀-mʌ̂ʌʌl-ɔ̀

ʊ́-lɛ̀ɛɛŋ-ɔ̀ ʊ̀-lɛ̀ɛɛŋ-ɔ̀

mʌ̂ĺ á-mʌ̂ĺ ʊ́-mʌ̂ĺ

lɛ̂ɛ́ ŋ á-lɛ̂ɛ́ ŋ ʊ́-lɛ̂ɛ́ ŋ

Long

Low Fall

mấat ̪ á-mấat ̪ ʊ́-mấat ̪ ʊ́-mâaat ̪-ɔ̀ ʊ̀-mâaat ̪-ɔ̀

High Fall

mấal á-mấal ʊ́-mấal ʊ́-mấaal-ɔ́ ʊ̀-mấaal-ɔ́

For the sake of backward compatibility, it is worthwhile to note that our descriptive analysis of the Object Voice Imperfective and Non-Evidential Past forms deviates from earlier work. We postulate that the High Fall class has a High Fall on the stem followed by a High tone on the suffix in the Object ́ ʌt̪-ɔ́ ‘impf-greet’ Voice Imperfective and Non-Evidential Past forms, e.g., ʊ̀-mʌ̂ʌ ́ ʌt ̪-ɔ́ ‘nevp:ov-greet’, respectively. In contrast, Remijsen, Millerand ʊ́-mʌ̂ʌ Naudé & Gilley (2015, 2016) postulated a High Fall to Mid followed by Mid in ́ ʌt-ɔ̄ ́ ʌt-ɔ̄ ̪̄ and ʊ́-mʌ̂ʌ ̪̄ . these forms, i.e., ʊ̀-mʌ̂ʌ Of all the levels of TAM that are available for Object voice, the No Tense forms are the only ones without a prefix, and they require syntactic licensing. This is illustrated in (81a): the addition of a focus marker renders the sentence grammatical. Without it, the sentence is ungrammatical (incomplete). In relation to the Imperfective, we find the opposite interaction: this level of TAM cannot combine with the focus marker. This is shown in (81b), which is ungrammatical with a focus marker, and fine without it. (81)

a.

kwʌ̄n cấm kɪ ́ *(à) pâal porridge eat:ov prp foc spoon ‘The porridge is eaten with a spoon.’

b. kwʌ̄n ʊ̀-càaam-ɔ̀ kɪ ́ (*a) pâal porridge impf-eat prp foc spoon ‘The porridge is being eaten with a spoon.’

7.3 The general subject-marked forms

In Section 4 we described how there are two sets of subject-marked forms: one set that is used with the semantic object preceeding the verb (Section 4.1), and another set that is used with a semantic role other than subject and object in that position (Section 4.2). We refer to the former as the general subjectmarked forms, and the latter as Applicative voice subject-marked forms. In this SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

59

subsection, we describe the former, i.e., the general subject-marked forms, and the description is limited to inflectional subject marking. The general subject-marked forms are formally derived from the Subject voice forms, rather than from the Object voice forms. This analysis is supported by several pieces of evidence. First, there is the interaction with TAM marking: the general subject-marked forms come in Past, Non-Evidential Past, No Tense and Future, precisely those TAM levels available for Subject voice. In contrast, Object voice additionally combines with an additional level of TAM: the Imperfective. Second, the morphological forms of the general subject-marked inflections are predictable on the basis of the corresponding Subject voice forms, but not on the basis of the Object Voice forms. With respect to syntactic subject marking, the same form is used as in Subject voice, followed by the casemarked pronominal form: jân, jîn, ɛ̂n, wāa, wɔ̂n, wûn, gɛ̂n. With respect to inflectional subject marking – i.e., in the 1st, 2nd and 3rd singular, and without focus marking (cf. Table 9 and Section 4.1) – these forms as well are derived from the Subject voice forms. This can be seen from Table 15. It shows the Subject voice form, the Object voice form, and the 1st singular subjectmarked form, in each case in Past and in No Tense. Note that, in relation to the Low Fall classes, the stem syllable of the Subject voice form has the Low Fall both in Past tense and in No Tense form, whereas in Object voice, the stem syllable has the High Fall in the same TAM levels. Crucially, the 1st singular has the same specification as the Subject voice (Low Fall) in both TAM levels. Table 15: Subject voice, Object voice and 1st singular subject-marked forms, in Past and No Tense, by verb class. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Fixed Short

SV, Past SV, NT

OV, Past OV, NT

Low

á-ŋɔ̀l ŋɔ̂ĺ ̄

á-ŋɔ̂ĺ ŋɔ̂ĺ

1st sg., Past á-ŋɔ̀l-á ̀ 1st sg., NT ŋɔ́l-à

Short with Grade

Fall

Low

Fall

lɛ̂ŋ

cấm̄

mʌ̂l

á-lɛ̂ŋ ́ á-lɛ̂ŋ ́ lɛ̂ŋ

á-lɛ̂ŋ-à lɛ̂ŋ-à

á-càm

á-mʌ̂l

cấm

á-mʌ̂ĺ mʌ̂ĺ

cáaam̀ -à

mʌ̂ʌʌl-à

á-cấm

Low

á-lɛ̀ɛŋ lɛ̂ɛ́ ŋ̄

á-lɛ̂ɛ́ ŋ lɛ̂ɛ́ ŋ

á-càaam-á á-mʌ̂ʌʌl-à á-lɛ̀ɛɛŋ-á lɛ́ɛɛŋ̀-à

Long

Low Fall

High Fall

mâat ̪

mâal

á-mâat ̪

á-mấat ̪ mấat ̪

á-mâaat ̪-à mâaat ̪-à

á-mâal á-mấal mấal

á-màaal-à màaal-à

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

60

Then consider the Low verbs. The Subject voice forms have a different specification for tone in Past vs. No Tense, but the specification for tone in the Object voice forms does not differ between these inflections. Again, the 1st singular forms pattern along with the Subject voice, presenting different inflection in Past vs. No Tense. In relation to High Fall verbs, finally, the tonal specification of the subject-marked forms is not predictable on either the Subject voice forms or the Object voice forms. In summary, the presence vs. absence of an alternation in the specification for tone on the stem syllable in the subject-marked forms parallels the presence vs. absence of the same alternation in the Subject voice forms. In contrast, the specification for tone in Object voice has no bearing on the specification for tone in subject-marked forms. In the remainder of this subsection, we will describe inflectional subject marking systematically, by level of TAM. Table 16 displays the inflected forms marked for subject in the Past tense. If the verb class displays morphological lengthening, the stem vowel is in the long grade. The 1st and 3rd person singular are marked by a suffix -a and -ɛ, respectively. The specification for tone on this suffix is High in the case of the Low classes, and Low otherwise. Table 16. Inflectional SV subject marking in the Past tense, by verb class and person. The Subject voice is included for the sake of comparison. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Fixed Short

2nd sg. 3rd sg. Subj. Voice

Low

á-ŋɔ̀l á-ŋɔ̀l-ɛ́ á-ŋɔ̀l

Fall

Short with Grade

Low

á-lɛ̂ŋ á-càaam á-lɛ̂ŋ-ɛ̀ á-càaam-ɛ́ á-lɛ̂ŋ á-càm

Fall

á-mʌ̂ʌʌl á-mʌ̂ʌʌl-ɛ̀ á-mʌ̂l

Low

á-lɛ̀ɛɛŋ á-lɛ̀ɛɛŋ-ɛ́ á-lɛ̀ɛŋ

Long

Low Fall

á-mâaat ̪ á-mâaat ̪-ɛ̀ á-mâat ̪

High Fall

á-màaal á-màaal-ɛ̀ á-mâal

Subject-marking in the Non-Evidential Past forms is purely syntactic. That is, we find the Subject Voice Non-Evidential Past form, followed by the case-marked form of the personal pronoun. However, there is an interesting difference between the Non-Evidential Past Subject Voice forms and the NonEvidential Past forms with syntactic case marking. The former need to be syntactically licensed, but the latter does not. For example, (82a), with the verb in Subject voice, is ungrammatical in the absence of a syntactic licenser. Adding the focus marker renders the structure grammatical. In contrast, the SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

61

subject marked Non Evidential Past form in (82b) is grammatical, with or without focus marker.24 (82)

a. twɔ́ɔŋ lɛ̀ɛɛŋ *(a) lʊ̀ʊt ̪ Twong throw:nevp foc stick ‘Twong apparently threw the stick.’

b. lʊ̀ʊt ̪ lɛ̀ɛɛŋ (a) ɛ̂n stick throw:nevp foc pr3s.n ‘S/He / S/He apparently threw the stick.’

Table 17 shows the forms for inflectional subject marking in the No Tense form, for each verb class. As is the case for inflectional subject-marking in the Past tense, the stem vowel is in the long grade (unless the verb is Fixed Short). Table 17. Inflectional subject marking in the No Tense, by verb class and person. The Subject voice is included for the sake of comparison. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Fixed Short

2nd sg. 3rd sg. Subject Voice

Low ŋɔ́l ̀ ̀ ŋɔ́l-ɛ̀ ŋɔ̂ĺ ̄

Fall

lɛ̂ŋ lɛ̂ŋ-ɛ̀ lɛ̂ŋ

Short with Grade

Low

cáaam̀ cáaam̀ -ɛ̀ cấm̄

Fall

mʌ̂ʌʌl mʌ̂ʌʌl-ɛ̀ mʌ̂l

Low

lɛ́ɛɛŋ̀ lɛ́ɛɛŋ̀-ɛ̀ lɛ̂ɛ́ ŋ̄

Long

Low Fall

mâaat ̪ mâaat ̪-ɛ̀ mâat ̪

High Fall

màaal màaal-ɛ̀ mâal

The Low Fall classes have the same specification for tone on the stem syllable as the corresponding Subject voice forms. In the case of the Low classes, in contrast, the corresponding Subject voice forms have a High Fall to Mid, and the subject-marked forms have a Late Fall. We can explain this alternation in terms of a regular morphophonological process, which we will first illustrate in the morphology of nouns. When a Low tone is added to the right of a High Fall to Mid, the result is a Late Fall. This is shown in (83). The constructstate form of a noun (83a,c) is used when the noun is followed by modifier that is not a possessor. In the case of a suffixless noun like gwôk ‘dog’, this inflection is stem-internal (83a); in the case of a suffixed noun like kʊ̂́ʊʊt-̪̄ ɔ̄ ‘thorn-s’, it involves the suffix -ɪ (83c). This inflection forms the basis for inflection for proximal demonstrative. Proximal demonstrative is marked by adding a Low target to the construct-state form. This can be seen most clearly in a suffixed noun, as in the comparison between (83c) and (83d). 24 On this basis, it could be argued that nominative case marking is a syntactic licenser. However, case marking does not have this characteristic in relation to the No Tense forms. A form like *kwʌ̄n cấm̄ wɔ̂n ‘We are eating porridge.’ is incomplete and in that sense ungrammatical, in spite of the case-marked personal pronoun. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

62

In a suffixless paradigm, the same sequence, i.e., ̂́ ̄ + ̀, yields ́ ̀. This is illustrated in (83a,b) (83)

a.

gwốooŋ̄ tɛ̂ɛk dog:cs strong ‘A strong dog’

b. gwóooŋ̀ dog:cs:dem ‘This dog.’

c.

kʊ̂́ʊn̪̄ -ɪ ̄ bɛ̂t́ ̪ thorn:cs sharp ‘A sharp thorn.’

d. kʊ́ʊn̪ -ɪ ̀ thorn:cs:dem ‘This thorn.’

The alternation at issue in the Low classes in Table 17, between the Subject voice form of a Low-toned verb such as cấm̄ and its 2nd singular subject marked forms, can be explained in terms of the same process, whereby the addition of a Low tone to the right of a High Fall to Mid yields a Late Fall. In the case of the High Fall class, the tonal specification of the stem syllable in the Subject voice No Tense form, which is Low-toned, differs in a non-compositional way from the Subject voice No Tense form, which carries a Low Fall. Instead it displays the same tone as the subject marked Past tense form (cf. Table 16). The No Tense forms in Table 17 additionally form the basis corresponding inflectionally subject marked Future tense forms, which have the prefix ʊ́- and are otherwise identical. The latter do not need to be licensed syntactically. In contrast, the subject-marked No Tense forms need to be licensed syntactically, irrespective of whether the subject is marked inflectionally or syntactically. In the remainder of this subsection, we describe a stem alternation that affects the inflectionally subject marked forms of verbs belonging to Low classes. As seen from Tables 15-17, verbs belonging to the Low Fall and High Fall classes display the same stem form and suffixation in Past and No Tense, i.e., the only difference is the presence vs. absence of the Past tense marker á-. In the case of the Low verb classes, inflectional subject marking involves a different specification in Past on the one hand and No Tense on the other (cf. Tables 15-17). For example, for the 1st singular of {càm} we find á-càaam-á (past) but cáaam̀ -à (No Tense). Of these two, the stem form used in the Past tense, displays the lexical specification for tone of the verb, i.e., Low. In fact, we also find the stem form and suffix specified for tone as in the Past tense, but without this prefix, e.g. càaam-á. One morphosyntactic environment in which this form is found is the yes-no question, as in (84a,b). Another is an imperative form, as in (84c). The use of kwáaaɲ̀ in (84a,c) and kwáaaɲ̀ -à in (84b) is ungrammatical. (84)

a.

kwàaaɲ-ɛ́ t ̪ɔ́ɔl ̀ take-3s rope ynq ‘Is he taking the rope?’

b. kwàaaɲ-á t ̪ɔ́ɔl ̀ take-1s rope ynq ‘Shall I take the rope?’

c.

kwàaaɲ take:2s ‘Take (it)!’

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

63

This phenomenon is specific to the Low classes: in the Low Fall and High Fall classes, there is no difference in terms of the tonal specification of the stem syllable between subject-marked Past and No Tense. Beyond these non-declarative speech acts, the stem form of the Past tense form but without the Past Tense prefix is also used in a number of other morphosyntactic constructions. One such construction involves the conjunction kấā, which marks sequential past, i.e., its function is akin to that of the Sequential Past inflection (cf. Section 5.6). If there is no subject marking, a transitive clause marked by this conjunction has the Object Voice No Tense form. This is illustrated by the narrative example in (85). Note that there is no TAM marking on the verb, i.e., kấāis a syntactic licenser, which does not come as a surprise, because tense (past) and aspect (consecutive) are expressed by the conjunction. (85)^ kấā làaaw-ɔ́ kwấɲ ɪ ̀ɪ dɪ ̀wʌ̀ʌʌt conj cloth-s take:ov prp Diwaat ‘And then Diwaat took power (lit.: the cloth).’

With subject marking, one would expect to find the No Tense subject marked form, in this case kwáaaɲ̀ -ɛ̀. Instead, we find kwàaaɲ-ɛ́. This is illustrated in (86), which resembles the spontaneous example in (85), but is marked for a 3rd singular subject. (86)

kấā làaaw-ɔ́ kwàaaɲ-ɛ́ / *kwáaaɲ-ɛ̀ conj cloth-s take-3s take:nt-3s ‘And then s/he took the cloth.’

The same happens when various syntactic TAM markers are involved. One of these is the irrealis marker dɪ ́. As seen from (87), with subject marking we find again the stem and suffix specified for tone as in the Past tense, but without the prefix. (87)

làaaw-ɔ́ dɪ ́ kwàaaɲ-ɛ́ / *kwáaaɲ-ɛ̀ cloth-s irr take-3s / take:nt-3s ‘S/He would have taken the cloth.’

How can this extended use of the Past tense subject-marked stem form be explained? It is worthwhile to note here that, for six of the seven verb classes, the lexical specification for tone of a transitive verb is revealed in the Subject voice Past tense (e.g. á-càm). This suggests that, in the case of the Low verbs, the Subject voice No Tense form (e.g. cấm̄ ), and the inflectional subject marking that is derived from it (e.g. cáaam̀ -à) result from a subsequent SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

64

diachronic development, which has not replaced the older subject-marked form (e.g. càaam-á) in a variety of syntactic contexts.

7.4 The Applicative Voice forms

The key syntactic characteristics of the Applicative voice are that the preverbal argument, the topic, is neither the semantic subject nor the semantic object, and that semantic object follows immediately after the verb (cf. Section 3.2.3). The forms are presented in Table 18. Applicative voice combines with the following levels of TAM: No Tense, Past, Future, Non-Evidential Past, Sequential Past and Conditional. All of these inflections involve the long vocalic grade of stem vowel, unless the paradigm is Fixed Short. With respect to tone, there are two stem forms: one in the Non-Evidential Past, and the other in all of the other TAM forms. The Non-Evidential Past displays the lexical specification of the verb root, i.e., Low for the Low verbs, Low Fall for the Low Fall verbs, and High Fall for the High Fall verbs. The tonal specification in the other TAM levels is predictable on the basis of this lexical specification for all classes other than High Fall. That is, if the lexical specification is Low, then the levels of TAM other than Non-Evidential Past have a Mid tone on the stem syllable in Applicative voice; and if the lexical specification is Low Fall, then the TAM levels other than Non Evidential Past have a High Fall to Mid. The High Fall verbs, finally, pattern along with the Low classes, i.e., they also have a Mid tone. So the Applicative Voice stem form is the same for No Tense, Past, Future, Sequential Past and Conditional. What distinguishes them is the TAM prefix: á- for Past, à- for Sequential Past, ʊ́- for Future, and ʊ̀- for conditional. Finally, the Non-Evidential Past is marked by the prefix ʊ́- just as in the Object Voice. Table 18. The Applicative voice forms by TAM and verb class. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’

Past Seq. Past Future Conditional Non-Ev. Past

Low

Fixed Short

á-ŋɔ̄l à-ŋɔ̄l ʊ́-ŋɔ̄l ʊ̀-ŋɔ̄l ʊ́-ŋɔ̀l

Fall

́ ̄ á-lɛ̂ŋ ́ à-lɛ̂ŋ̄ ́ ̄ ʊ́-lɛ̂ŋ ́ ʊ̀-lɛ̂ŋ̄ ʊ́-lɛ̂ŋ

Short with Grade

Low

á-cāaam à-cāaam ʊ́-cāaam ʊ̀-cāaam ʊ́-càaam

Fall

́ ʌl ̄ á-mʌ̂ʌ ́ à-mʌ̂ʌʌl ̄ ́ ʌl ̄ ʊ́-mʌ̂ʌ ́ ʊ̀-mʌ̂ʌʌl ̄ ʊ́-mʌ̂ʌʌl

Low

á-lɛ̄ɛɛŋ à-lɛ̄ɛɛŋ ʊ́-lɛ̄ɛɛŋ ʊ̀-lɛ̄ɛɛŋ ʊ́-lɛ̀ɛɛŋ

Long

Low Fall

á-mấaat ̪̄ à-mấaat ̪̄ ʊ́-mấaat ̪̄ ʊ̀-mấaat ̪̄ ʊ́-mâaat ̪

High Fall

á-māaal à-māaal ʊ́-māaal ʊ̀-māaal ʊ́-mấaal

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

65

In the Non-Evidential Past, the only difference between the Applicative voice form and Object voice form is that the latter has the suffix -ɔ. For example, in the case of {càm} ‘eat’, there is ʊ́-càaam in Applicative voice and ʊ́-càaam-ɔ̀ in Object voice. However, evidence from elsewhere in the grammar shows that the presence vs. absence of this suffix depends on whether the verb is followed by a core argument (see Section 3.3.2.4). This is invariably the case when the verb is inflected for Applicative voice, and never so when it is inflected for Object voice. Hence the two forms could be considered syncretic, the difference falling out from the interaction with the syntactic context.

7.5 The subject-marked Applicative Voice forms

Applicative voice forms can be marked for the semantic subject. In Section 4.2 we argued that subject marking Applicative voice form is inflectional in nature in the singular forms when there is no focus marker involved. Table 19 lays out the subject-marked Applicative voice forms in these singular forms, in the No Tense level of TAM. Just as in the Applicative voice without subject marking, the Past, Future, Sequential Past, and Conditional levels of TAM differ only in the addition of a prefix: á- in the case of Past, ʊ́- in the case of Future, à- in the case of Sequential Past, and ʊ̀- in the case of Conditional. Table 19. The subject-marked Applicative voice forms, in No Tense. Each class is represented by one verb. The Applicative voice form is included for the sake of comparison. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Fixed Short

2nd sg. 3rd sg. Applic Voice w. synt. Subj marking Applic Voice

Low

Fall

Short with Grade

Low

Fall

Low

Long

Low Fall High Fall

ŋɔ̄ľ ŋɔ̄l-ɛ́

lɛ́ŋ lɛ́ŋ-ɛ́

̌ am cāa cāaam-ɛ́

mʌ́ʌʌl mʌ́ʌʌl-ɛ́

lɛ̄ɛ̌ ɛŋ lɛ̄ɛɛŋ-ɛ́

máaat ̪ máaat ̪-ɛ́

̌ al māa māaal-ɛ́

ŋɔ̄l ŋɔ̄l

́ ̄ lɛ̂ŋ ́ ̄ lɛ̂ŋ

cām cāaam

mʌ̂ĺ ̄ ́ ʌl ̄ mʌ̂ʌ

lɛ̄ɛŋ lɛ̄ɛɛŋ

mấat ̪̄ mấaat ̪̄

māal māaal

The morphological shape of the subject-marked Applicative inflections is predictable on the basis of the Applicative voice form, which is also displayed in Table 20. In terms of tone, the Low and the High Fall classes have a Mid tone in all forms other than 2nd singular. This is the same specification as in the Applicative voice. In the 2nd singular, a High target is added to the SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

66

stem syllable to the right of the lexical specification, yielding a High Rise. The Low Fall classes, which have a High Fall to Mid in the Applicative voice, display a High tone in the form inflected for a singular subject. As explained in Section 2, the addition of a High toned suffix to a stem that has a High Fall to Mid regularly changes the High Fall to Mid tone on the stem to High. In this way, the tonal specification in these inflected form can be derived in a compositional manner. As for vowel length, the forms with inflectional subject marking display the long vocalic grade. In contrast, the forms with syntactic subject marking have the short vocalic grade (cf. Table 9). The Non-Evidential Past subject-marked Applicative forms are identical to the general subject-marked forms, described in Section 7.3. This is illustrated in (88). Note that verb is the same in (88a) vs. (88b), even though (88a) has the verb with general subject marking – the preverbal argument represents the semantic object – while (88b) displays applicative syntactic alignment, i.e., the preverbal argument represents a semantic role other than subject or object.25 (88)

a.

kwʌ̄n càaam jân porridge eat:nevp pr1s:n ‘I would have eaten porridge.’ (Others assert this, I disagree.)

b. pâal càaam jân kwʌ̄n spoon eat:nevp pr1s:n porridge ‘I would have used a spoon to eat porridge.’ (Others assert this, I disagree.)

While the forms are the same, there is an interaction with focus here. When the preverbal argument expresses the semantic object, then either topic or the semantic subject can be marked for focus. This is shown in (89a), which shows variants on (88a). But if the topic expresses a semantic role other than subject or object, than only this argument can be marked for focus, as in (89b), which shows the variants on (89b). (89)

a.

kwʌ̄n (à) càaam (a) jân porridge foc eat:nevp foc pr1s:n

b. pâal (à) càaam (*a) jân (*à) kwʌ̄n spoon foc eat:nevp foc pr1s:n foc porridge

7.6 The patient-oriented infinitive

The base paradigm includes an infinitive, which refers to the event denoted by the source verb. As will be demonstrated below, this form is a noun in a morphosyntactic sense. In spite of the fact that this derivation changes the word class, we treat it as part of the base inflectional paradigm, for the 25 The function of the Non-Evidential Past is explained in Section 5.4. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

67

following reasons. First, derived verb paradigms (e.g. spatial, benefactive, etc.) also present infinitives, so that the infinitive of the base paradigm can be seen on a par with the infinitives of the derived paradigms. Second, the stem form of this infinitive is identical to a form of the base inflectional paradigm. In other words, this form relates in a predictable and productive way to the base paradigm. Patient-oriented infinitives occur frequently: it is used in small-clause constructions and with auxiliary verbs. We first describe its use, and then its morphological form. In the narrative example in (90), the patient-oriented infinitive is gûuur-ɔ̀, from {gûur} ‘grind’. It is used here in a small-clause construction marked by bɛ̄ɛ, which expresses a goal. This example reveals that this infinitive is a noun: it is inflected for being the possessed term in a possessive noun phrase (pertensive). The possessor, àwấac ‘sour dough’, expresses the semantic object of the verb. (90)^ cʊ̄ʊt ̪-ɛ̄ kấā kɛ́t ̪ à bɛ̄ɛ gûur-ɪ ̀ àwấac end:prt-3s conj go.away foc prp grind:inf-prt sour.dough ‘So, after that she went away to grind (sorghum grain to make) sour dough.’

Illustration (91) shows how a patient-oriented infinitive form is used in a clause headed by the auxiliary verb {cɛ̀k}. This auxiliary is devoid of semantic content, but it does inflect for Voice and TAM. In (91), it is in Object voice and in No Tense. The infinitive appears in a prepositional phrase, marked by kɪ ́.26 ́ (91)^ jấā àkwàrɪ ̀còoot-ɪ ̄ɪ cɛ̂k kɪ ́=à gûuur-ɔ̀ those.of k.o.herb-apl aux:ov:nt prp=foc grind-inf ‘Herbs like akwaricoto [a bitter kind of herb] were ground.’

In form, the patient-oriented infinitive is identical to the Imperfective form of the transitive verb, but without the prefix. This can be seen from Table 20. Note how, for each of the seven classes of transitive verbs, the stem syllable of the base form of the patient-oriented infinitive is the same as that of the imperfective verb form, in terms of both tone and vowel length.

26 When the preposition kɪ́ is followed by the focus marker à, they coalesce into a single syllable [kấā]. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

68

Table 20. Forms illustrating the patient-oriented infinitive, by verb class, and its relation to the rest of the paradigm, in terms of stem length alternation. Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, {mấal} ‘praise’. Fixed Short

Verb, Impf. Inf., base Inf., Pert. sg.

Low

ʊ̀-ŋɔ̀l-ɔ̀ ŋɔ̀l-ɔ̀ ŋɔ̀l-ɪ ̀

Fall

Short with Grade

ʊ̀-lɛ̂ŋ-ɔ̀ lɛ̂ŋ-ɔ̀ lɛ̂ŋ-ɪ ̀

Low

Fall

Low

Long

Low Fall

High Fall

ʊ̀-càaam-ɔ̀ ʊ̀-mʌ̂ʌʌl-ɔ̀ ʊ̀-lɛ̀ɛɛŋ-ɔ̀ ʊ̀-mâaat ̪-ɔ̀ ʊ̀-mấaal-ɔ́ càaam-ɔ̀ mʌ̂ʌʌl-ɔ̀ lɛ̀ɛɛŋ-ɔ̀ mâaat ̪-ɔ̀ mấaal-ɔ́ càm-ɪ ̀ mʌ̂l-ɪ ̀ lɛ̀ɛŋ-ɪ ̀ mâat ̪-ɪ ̀ mấal-ɪ ̀

The patient-oriented infinitive displays the inflectional paradigm of a noun. That is, as any other noun, the patient-oriented infinitive has pertensive27, and construct state inflections, and the formation of these inflections is regular and predictable. This is shown in Table 21, which shows the inflectional paradigm of the bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’, which is a suffixed noun to begin with, and that of two patient-oriented infinitives. Note that the quantity alternation in the stem vowel between the noun base and the inflected forms is the same in the infinitives and the underived noun. Table 21. Noun paradigms for bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’, and two patient-oriented infinitives. bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’ mâaat ̪-ɔ̀ ‘drink:inf’ mấaal-ɔ́ ‘praise:inf’

Pertensive, singular possessor Pertensive, plural possessor Construct state

bɔ̀ɔt ̪-ɪ ̀ bɔ̀ɔt ̪-ɪ ́ bɔ̀ɔn̪-ɪ ̀

mâat ̪-ɪ ̀ mâat ̪-ɪ ́ mâan̪-ɪ ̀

mấal-ɪ ̀ mấal-ɪ ́ mấal-ɪ ̀

The quantity alternation in the stem vowel of patient-oriented infintives is predictable from the quantity alternation of the the source verb. For example, {mâat ̪} ‘drink’ is a member of the Long Fall class, i.e., the stem vowel alternates between long and overlong in the base paradigm of the verb. We find the same quantity alternation in the paradigm of the infinitive (cf. Table 21). Similarly, the Fixed Short root {ŋɔ̀l} ‘cut’, has the patient-oriented infinitive ŋɔ̀l-ɔ̀ ‘cut:inf’, and retains this short vowel throughout its nominal paradigm: as seen from Table 20, the corresponding pertensive with singular possessor is ŋɔ̀l-ɪ ̀ ‘cut:inf-prt’. Finally, a Short with Grade verb like {càm} 27 The term pertensive (Dixon 2010) refers to an inflection marking the possessed (head) term of a possessive noun phrase. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

69

‘eat’ has an overlong vowel in the base form of the patient-oriented infinitive: càaam-ɔ̀ ‘eat:inf’. The pertensive form of this infinitive is càm-ɪ ̀ ‘eat:inf-prt’ (cf. Table 21). In each case, the stem vowel in the paradigm of the patientoriented infinitive noun alternates between the same levels of vowel length as in the base paradigm of the source verb. We label this infinitive ‘patient-oriented’, because there is another infinitive form, which is agent-oriented. The difference between the two is clear when they head a possessive noun phrase. Consider the case of {càm} ‘eat’. The patient-oriented infinitive is càaam-ɔ̀, and the agent-oriented one is cám. Possessive constructions with both of these are presented in (92). In (92a), the possessive construction is headed by a patient-oriented infinitive, càm-ɪ ̀, which is the pertensive inflection of càaam-ɔ̀, and the possessor term expresses the semantic object of the verb. In contrast, in a possessive construction headed by the agent-oriented infinitive, as in (92b), the possessor term expresses the semantic subject of the verb. (92)

a.

càm-ɪ ̀ gjɛ̀ɛɛn-ɔ̀ eat:inf-prt chicken-s ‘The eating of a chicken’

b. cấaam̄ gjɛ̀ɛɛn-ɔ̀ eat:infa:prt chicken-s ‘The eating by a chicken’

8 Conclusion This chapter has laid out the main structural properties of Shilluk main clauses headed by a transitive verb. The base paradigm of transitive verbs is characterised by a great degree of interaction between factors. In Section 8.1, we summarize interactions at the level of the forms, and in Section 8.2 on interactions at the level of the functions.

8.1 Fusional forms

The base paradigm of Shilluk verbs presents morphological marking for Voice, TAM, and subject. It is a central characteristic of the system that these functions are marked in a fusional manner. On the one hand, various functions may be expressed on a single syllable. For example, máaat ̪ ‘drink:xv:2s’ is the 2nd singular No Tense Applicative voice form of {mâat ̪} ‘drink’ – both Applicative voice and 2nd singular are marked on the stem syllable. Second, even when a function is marked by an affix, there often is a change in the stem syllable as well. For example, á-càm ‘pst-eat’ and ʊ́-cấm̄ ‘fut-eat’ are the SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

70

Past Tense and Future tense forms of {càm} ‘eat’, respectively, both in Subject voice. Aside from the prefix, the difference in tense is also marked through tone on the stem. Of the three above mentioned factors, Voice is marked purely steminternally. TAM is expressed primarily through prefixes. Subject marking varies between syntactic marking, suffixal marking, and purely stem-internal marking. The most versatile morphophonological parameter is tone. Shilluk has an inventory totaling nine different contrasting specifications in stem syllables, and this system of contrasts is heavily exploited in the morphology. In fact, the rich inventory is undoubtedly the diachronic outcome of diachronic changes in the morphophonology (cf. Andersen 1990). It is insightful to compare the role of tone in the morphology of Shilluk verbs with the situation in Dinka, a closely-related language which also presents a rich system of morphological marking. Andersen (1992–1994) shows how in Dinka tone, vowel length, voice quality are used to mark derivation and inflection. For example, in relation to the word we̤é c ‘kick:pet:2s’ “Kick it hither!”, Andersen (1992–1994:61) postulates the layers (strata) of derivation that are shown in Table 22. Table 22. A schematic representation of the layers of derivation of we̤é c ‘kick:pet:2s’, according to Andersen (1992–1994:61). Root Derivation (centripetal) Inflection (2nd singular)

Voice quality

Length

Tone

Modal Breathy Breathy (unchanged)

Short Long Long (unchanged)

Fall Low High

Shilluk’s richer tonal inventory allows for deeper strata to be transparent in the surface form. To illustrate this, consider again the word máaat ̪ ‘drink:xv:2s’, the 2nd singular No Tense Applicative voice form of the Low Fall verb {mâat ̪} ‘drink’, and alongside it lɛ̄ɛ̌ ɛŋ ‘throw:xv:2s’, the corresponding form of {lɛ̀ɛŋ} ‘throw’, shown in Table 23. Both forms are part of the base inflectional paradigm, i.e., there is no derivation involved here. Note that both the Applicative voice forms and the 2nd singular Applicative voice forms reveal the underlying specification for tone of the verb. As explained in Section 2, Applicative voice raises the tonal register from Low to Mid in the case of {lɛ̀ɛŋ} ‘throw’, and from Low Fall to High Fall to Mid in the case of {mâat ̪} ‘drink’. The 2nd singular subsequently adds a High end SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

71

target, yielding a High Rise or a High tone, depending on what this High tone attaches to. This means that tone can be ‘spent’ more than once in the morphology, without the top layer overwriting the layer or layers below it completely. Table 23: A schematic representation of the layers of derivation of two Shilluk verb forms. {lɛ̀ɛŋ} ‘throw’ Root Applicative voice 2nd singular

lɛ̀ɛŋ lɛ̄ɛɛŋ lɛ̄ɛ̌ ɛŋ

{mâat ̪} ‘drink’ mâat ̪ mấaat ̪̄ máaat ̪

8.2 Interactions between functions Just as formal exponence is non-concatenative in Shilluk, the functions that are expressed equally interact in several ways. We will summarize these interactions out in relation to each of the three functions (Voice, TAM, Subject marking). If the morphological operation of Voice were the sole expression of topicality, we would find Object voice used if the semantic object is topical, Subject voice used if the semantic subject is topical, and Applicative voice if a different semantic role is topical. While the first and the last of these generalisations do hold, the relation between Subject voice and information structure is more complex. As seen fromTable 24, we hypothesize that it is only when the semantic subject and the referent event of the verb are both part of the framework of reference shared by speaker and hearer that Subject voice is used. If only the Subject is topical, then a variety of other structures are used, including two valency-decreasing operations, in which the semantic object can be expressed as a peripheral argument. This shows that the expression of Voice interacts with subject marking and with the valency changing operations of antipassive and ambitransitive. We commend this topic of the relation between construction type / verb form on the one hand and information structure on the other hand for further research.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

72

Table 24. The relation between topic and the choice of verb form, in relation to transitive verbs. Which constituent is topical?

Choice of verb form

None (all new information) Semantic object Semantic role other than subject or object Semantic subject Semantic subject and Semantic object Semantic subject and verb

Object voice Object voice Applicative Subject marking, Antipassive, Ambitransitive Subject marking Subject voice

In relation to the controversy on the Shilluk voice system, we conclude that Object voice displays the morphosyntactic characteristics of passive voice (cf. Westermann 1912, Tucker 1955). However, unlike a passive, it is unmarked in an information-structural sense (cf. Miller & Gilley 2001). In turn, Subject voice displays the morphosyntactic characteristics of active voice, but it is marked in an informational structural sense. TAM interacts with focus marking and with Voice. The No Tense form and most Non Evidential Past forms require a syntactic licenser, of which focus is the one that appears with the greatest frequency. In addition, some levels TAM only appear in one particular voice. Specifically, the Imperfective is only available in Object voice, but not in Subject voice and Applicative voice, and the Sequential Past and Conditional only appear in Applicative voice. Subject marking also interacts with focus marking. Inflectional subject marking is restricted to singular subjects, and to contexts where there is no focus marker. In contrast, the focus marker appears syntactically between the verb and the case-marked subject marker, suggesting that the latter is not phonologically integrated with the verb.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

73

Acknowledgements

We gratefully acknowledge the Leverhulme Trust, which funded this project through the research grant “A descriptive analysis of the Shilluk language” (RPG-2015-055). We also thank the speakers who produced the narratives from which the spontaneously uttered sound examples are drawn. Aside from the second author, they are Maria Bocay Onak, Teresa Akic Awanh, Peter Mojwok Yor, and Andrew Wanh Mayik. We are grateful to Laura Arnold, who provided detailed feedback ahead of the review process, and also to the two anonymous reviewers for Language Documentation & Conservation. Their feedback has been very useful to improve the chapter. Finally we thank the editor, Nick Thieberger, for his support for this project, and Christine Bruderlin, for her work on the layout.

Abbreviations

The following abbreviations are used in the glosses. AMB Ambitransitive APL Associative plural ATP Antipassive AUX Auxiliary verb BNF Benefactive BULL Marker of male cattle terms CARD Cardinal CONJ Conjunction CS Construct state CTG Contingent form of adjective DEF Definite DEM Demonstrative EXSP Existential predicate marker FOC Focus FUG Centrifugal deixis FUT Future IDP Independent pronoun IMP Imperative IMPF Imperfective INDIR Indirect speech marker INF Infinitive INFA Agent-oriented infinitive IRR Irrealis ITER Iterative MASC Masculine nominalizer MDF Modification marker N Nominative NEG Negation marker NEVP Non-evidential past SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

NOMP NT OBL OV P PST PET PRP PRT QUOT REFL REL S SEQP SUB SV WHQ XV YNQ

74

Copula for nominal predicates No tense Oblique pronoun Object voice Plural Past Centripetal deixis Preposition Pertensive Quotative Reflexive Relativizer Singular Sequential past Subordination marker Subject voice wh-question marker Applicative voice Yes/no question marker

Bibliography

Andersen, Torben (1988). Ergativity in Päri, a Nilotic OVS language. Lingua 75, 289–324. Andersen, Torben (1990). Vowel length in Western Nilotic languages. Acta Linguistica Hafniensia 22, 5–26. Andersen, Torben (1992-1994). Morphological stratification in Dinka: On the alternations of voice quality, vowel length and tone in the morphology of transitive verbal roots in a monosyllabic language. Studies in African Linguistics 23(1), 1–63. Creissels, Denis (2006). A typology of subject and object markers in African languages. In F.K. Erhard Voeltz (ed.) Studies in African linguistic typology. Amsterdam: John Benjamins, 43–70. Dixon, R.M.W. (2010). Basic Linguistic Theory – volume 2. Oxford University Press. Dixon, R.M.W. (2012). Basic Linguistic Theory – volume 3. Oxford University Press. Erlewine, Michael Y., Theodore Levin & Coppe van Urk (2017). Ergativity and Austronesian-type voice systems. In Jessica Coon, Diane Massam, & Lisa Demena Travis (eds.). The Oxford handbook of ergativity. OUP. Fillmore, Charles J. (1977). The case for case reopened. In Peter Cole & Jerold M. Sadock (eds.) Syntax and semantics – volume 8: Grammatical relations. SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

75

Academic Press, 59–81. Himmelmann, Nikolaus P. (2014). Asymmetries in the prosodic phrasing of function words: another look at the suffixing preference. Language 90, 927–960. Hopper, Paul J., & Sandra A. Thompson (1980). Transitivity in grammar and discourse. Language 56, 251–299. Hyman, Larry M. and Magaji, Daniel J. (1970). Essentials of Gwari grammar. Ibadan: University of Ibadan-Institute of African Studies. Hyman, Larrry M. & John R. Watters (1984). Auxiliary focus. Studies in African linguistics 15(3), 234–273. Kaufmann, Stefan (2006). Conditionals. In Keith Brown (ed.) Encyclopedia of Language and Linguistics, volume 3, 3–6. Elsevier. Kroeger, Paul R. (2005). Analyzing syntax: a lexical-functional approach. Cambridge University Press. Krifka, Manfred. (2008). Basic notions of information structure. Acta Linguistica Hungarica 55, 243–276. Legate, Julie A. (2012). Subjects in Acehnese and the nature of the passive. Language 88, 495–525. Miller, Cynthia L. and Leoma G. Gilley (2001). Evidence for ergativity in Shilluk. Journal of African Languages and Linguistics 22: 33–68. Miller, Cynthia L. and Leoma G. Gilley (2007). Evidentiality and mirativity in Shilluk. In Mechthild Reh & Doris L. Payne (eds.). Advances in NiloSaharan linguistics: Proceedings of the 8th International Nilo-Saharan Linguistics Colloquium,University of Hamburg, August 22–25, 2001. 191– 206. Cologne: Rüdiger Köppe Verlag. Payne, Thomas E. (1997). Describing morphosyntax: a guide for field linguists. Cambridge University Press. Reh, Mechthild (1996). Anywa Language – Description and Internal Reconstructions. Nilo-Saharan 11. Cologne: Rüdiger Köppe Verlag. Reid, Tatiana, Irina Monich, Matthew Baerman, Oliver Bond & Bert Remijsen (2016). Nuer morphophonology: The verbal paradigm. P-workshop presentation, University of Edinburgh. Remijsen, Bert, Otto G. Ayoker, and Timothy Mills (2011). Shilluk. Journal of the International Phonetic Association 41(1), 111–125. Remijsen, Bert & Otto G. Ayoker (2014). Contrastive tonal alignment in falling contours in Shilluk. Phonology 31(3), 435–462. Remijsen, Bert, Cynthia L. Miller-Naudé & Leoma G. Gilley (2015). Steminternal and affixal morphology in Shilluk. To appear in Matthew SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

76

Baerman (ed.) The Oxford Handbook of Inflection. Oxford University Press, 577–596. Remijsen, Bert & Otto G. Ayoker (2015). Collection of Shilluk narratives and songs. Edinburgh Datashare [multimedia data collection, online at https://datashare.is.ed.ac.uk/handle/10283/425]. Remijsen, Bert, Cynthia L. Miller-Naudé & Leoma G. Gilley (2016). The morphology of Shilluk transitive verbs. Journal of African Languages and Lingusitics 37, 201–245. Remijsen, Bert, Otto G. Ayoker, & Signe Jørgensen (to appear). Three-level vowel length in Shilluk. Submitted. Xu, Yi & Sun, Xuejing (2002). Maximum speed of pitch change and how it may relate to speech. Journal of the Acoustical Society of America 111, 1399– 1413. Tucker, Archibald N. (1955). The verb in Shilluk. Mitteilungen des Instituts für Orientforschung 3: 421–462. Westermann (1912). The Shilluk people – their language and folklore. Negro Universities Press [reprinted 1970, United Presbyterian Church of North America]. Woodbury, Tony (2003). Defining documentary linguistics. Language Documentation and Description 1, 35–51. Zhang, Jie (2001). The effects of duration and sonority on contour tone distribution – Typological survey and formal analysis. UCLA PhD dissertation. Zwarts, Joost (2007). Number in Endo-Marakwet. In Mechthild Reh and Doris L. Payne (eds.). Advances in Nilo-Saharan linguistics: Proceedings of the 8th International Nilo-Saharan Linguistics Colloquium, University of Hamburg, August 22–25, 2001. Cologne: Rüdiger Köppe Verlag, 191–206.

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

77

Appendix A. Paradigm tables Table A.1. The three voices (separate panels), by TAM (rows), and verb class (columns). Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. Subject marking is not included. SUBJECT VOICE

Low

ŋɔ̂ĺ ̄ á-ŋɔ̀l ʊ́-ŋɔ̂ĺ ̄

No Tense Past Future Non-Ev. Past ŋɔ̀l OBJECT VOICE

Low

No Tense Past Future Non-Ev. Past Imperfective

ŋɔ̂ĺ á-ŋɔ̂ĺ ʊ́-ŋɔ̂ĺ ʊ́-ŋɔ̀l-ɔ̀ ʊ̀-ŋɔ̀l-ɔ̀

APPLIC. VOICE

Low

No Tense Past Future Seq. Past Conditional Non-Ev.Past

ŋɔ̄l á-ŋɔ̄l ʊ́-ŋɔ̄l à-ŋɔ̄l ʊ̀-ŋɔ̄l ʊ́-ŋɔ̀l

Fixed Short

Short with Grade

Fall

Low

Fall

lɛ̂ŋ á-lɛ̂ŋ ʊ́-lɛ̂ŋ lɛ̂ŋ

cấm̄ á-càm ʊ́-cấm̄ càaam

mʌ̂l á-mʌ̂l ʊ́-mʌ̂l mʌ̂ʌʌl

Fixed Short Fall

lɛ̂ŋ́ á-lɛ̂ŋ́ ʊ́-lɛ̂ŋ́ ʊ́-lɛ̂ŋ-ɔ̀ ʊ̀-lɛ̂ŋ-ɔ̀ Fixed Short Fall

lɛ̂ŋ́ ̄ á-lɛ̂ŋ́ ̄ ʊ́-lɛ̂ŋ́ ̄ à-lɛ̂ŋ́ ̄ ʊ̀-lɛ̂ŋ́ ̄ ʊ́-lɛ̂ŋ

lɛ̂ɛ́ ŋ̄ á-lɛ̀ɛŋ ʊ́-lɛ̂ɛ́ ŋ̄

Low Fall

High Fall

lɛ̀ɛɛŋ

mâat ̪ á-mâat ̪ ʊ́-mâat ̪ mâaat ̪

mâal á-mâal ʊ́-mâal mấaal

Low

lɛ̂ɛ́ ŋ á-lɛ̂ɛ́ ŋ ʊ́-lɛ̂ɛ́ ŋ ʊ́-lɛ̀ɛɛŋ-ɔ̀ ʊ̀-lɛ̀ɛɛŋ-ɔ̀

Low Fall

High Fall

mấat ̪ á-mấat ̪ ʊ́-mấat ̪ ʊ́-mâaat ̪-ɔ̀ ʊ̀-mâaat ̪-ɔ̀

mấal á-mấal ʊ́-mấal ʊ́-mấaal-ɔ́ ʊ̀-mấaal-ɔ́

Fall

Low

Low Fall

High Fall

ʊ́-mʌ̂ʌʌl

lɛ̄ɛɛŋ á-lɛ̄ɛɛŋ ʊ́-lɛ̄ɛɛŋ à-lɛ̄ɛɛŋ ʊ̀-lɛ̄ɛɛŋ ʊ́-lɛ̀ɛɛŋ

á-mấaat ̪̄ ʊ́-mấaat ̪̄ à-mấaat ̪̄ ʊ̀-mấaat ̪̄

māaal á-māaal ʊ́-māaal à-māaal ʊ̀-māaal ʊ́-mấaal

Short with Grade

Low

cấm á-cấm ʊ́-cấm ʊ́-càaam-ɔ̀ ʊ̀-càaam-ɔ̀

Fall

mʌ̂ĺ á-mʌ̂ĺ ʊ́-mʌ̂ĺ ʊ́-mʌ̂ʌʌl-ɔ̀ ʊ̀-mʌ̂ʌʌl-ɔ̀

Short with Grade

Low

cāaam á-cāaam ʊ́-cāaam à-cāaam ʊ̀-cāaam ʊ́-càaam

Long

Low

mʌ̂ʌ́ ʌl ̄ á-mʌ̂ʌ́ ʌl ̄ ʊ́-mʌ̂ʌ́ ʌl ̄ à-mʌ̂ʌ́ ʌl ̄ ʊ̀-mʌ̂ʌ́ ʌl ̄

Long

Long

mấaat ̪̄

ʊ́-mâaat ̪

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

78

Table A.2. Inflections for general subject marking by TAM (separate panels), person, and verb class (columns). Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. PAST

Low

1st sg. 2nd sg. 3rd sg.

á-ŋɔ̀l-á á-ŋɔ̀l á-ŋɔ̀l-ɛ́

NO TENSE 1st sg. 2nd sg. 3rd sg. FUTURE 1st sg. 2nd sg. 3rd sg.

Fixed Short

̀ ŋɔ́l-à ŋɔ́l ̀ ̀ ŋɔ́l-ɛ̀

Low

Low

Fall

Low

Low Fall

High Fall

á-lɛ̂ŋ-à á-lɛ̂ŋ á-lɛ̂ŋ-ɛ̀

á-càaam-á á-càaam á-càaam-ɛ́

á-mʌ̂ʌʌl-à á-mʌ̂ʌʌl á-mʌ̂ʌʌl-ɛ̀

á-lɛ̀ɛɛŋ-á á-lɛ̀ɛɛŋ á-lɛ̀ɛɛŋ-ɛ́

á-mâaat ̪-à á-mâaat ̪ á-mâaat ̪-ɛ̀

á-màaal-à á-màaal á-màaal-ɛ̀

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

lɛ̂ŋ-à lɛ̂ŋ lɛ̂ŋ-ɛ̀

cáaam̀ -à cáaam̀ cáaam̀ -ɛ̀

mʌ̂ʌʌl-à mʌ̂ʌʌl mʌ̂ʌʌl-ɛ̀

lɛ́ɛɛŋ̀-à lɛ́ɛɛŋ̀ lɛ́ɛɛŋ̀-ɛ̀

mâaat ̪-à mâaat ̪ mâaat ̪-ɛ̀

màaal-à màaal màaal-ɛ̀

Fixed Short

̀ ʊ́-ŋɔ́l-à ʊ́-ŋɔ́l ̀ ̀ ʊ́-ŋɔ́l-ɛ̀

Long

Fall

Fixed Short

Low

Short with Grade

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

ʊ́-lɛ̂ŋ-à ʊ́-lɛ̂ŋ ʊ́-lɛ̂ŋ-ɛ̀

ʊ́-cáaam̀ -à ʊ́-cáaam̀ ʊ́-cáaam̀ -ɛ̀

ʊ́-mʌ̂ʌʌl-à ʊ́-mʌ̂ʌʌl ʊ́-mʌ̂ʌʌl-ɛ̀

ʊ́-lɛ́ɛɛŋ̀-à ʊ́-lɛ́ɛɛŋ̀ ʊ́-lɛ́ɛɛŋ̀-ɛ̀

ʊ́-mâaat ̪-à ʊ́-mâaat ̪ ʊ́-mâaat ̪-ɛ̀

ʊ́-màaal-à ʊ́-màaal ʊ́-màaal-ɛ̀

Table A.3. Inflections for applicative subject marking by person, and verb class (columns). Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. The forms given are in the No Tense form. Past, Future, Sequential Past, and Conditional levels of TAM differ only in the addition of a prefix: á- in the case of Past, ʊ́- in the case of Future, à- in the case of Sequential Past, and ʊ̀- in the case of Conditional. SUBJ, APPLIC

Low

1st sg. 2nd sg. 3rd sg.

ŋɔ̄l-á ŋɔ̄ľ ŋɔ̄l-ɛ́

Fixed Short

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

lɛ́ŋ-á lɛ́ŋ lɛ́ŋ-ɛ́

cāaam-á cāǎ am cāaam-ɛ́

mʌ́ʌʌl-á mʌ́ʌʌl mʌ́ʌʌl-ɛ́

lɛ̄ɛɛŋ-á lɛ̄ɛ̌ ɛŋ lɛ̄ɛɛŋ-ɛ́

máaat ̪-á máaat ̪ máaat ̪-ɛ́

māaal-á māǎ al māaal-ɛ́

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

79

Appendix B. Paradigm tables, schematic Table B.1. Schematic representation of the morphological marking of the three voices (separate panels), by TAM (rows), and verb class (columns). For each class in each inflection, the table shows affixes, stem tone and vowel lengthening (::). Subject marking is not included. Fixed Short

SUBJECT VOICE

Low

No Tense Past Future Non-Ev. Past

HFM á- L ʊ́- HFM L

OBJECT VOICE

Low

No Tense Past Future Non-Ev. Past Imperfective

HF á- HF ʊ́- HF ʊ́- L -ɔ̀ ʊ̀-L -ɔ̀

APPLIC. VOICE

Low

No Tense Past Future Seq. Past Conditional Non-Ev.Past

M á- M ʊ́- M à- M ʊ̀- M ʊ́- L

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

LF á- LF ʊ́- LF LF

HFM á- L ʊ́- HFM L ::

LF á- LF ʊ́- LF LF ::

HFM á- L ʊ́- HFM L ::

LF á- LF ʊ́- LF LF ::

LF á- LF ʊ́- LF LF ::

Fixed Short

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

HF á- HF ʊ́- HF ʊ́- LF-ɔ̀ ʊ̀- LF -ɔ̀

HF á- HF ʊ́- HF ʊ́- L :: -ɔ̀ ʊ̀-L :: -ɔ̀

HF á- HF ʊ́- HF ʊ́- LF :: -ɔ̀ ʊ̀- LF :: -ɔ̀

HF á- HF ʊ́- HF ʊ́- L :: -ɔ̀ ʊ̀-L :: -ɔ̀

HF á- HF ʊ́- HF ʊ́- LF :: -ɔ̀ ʊ̀- LF :: -ɔ̀

HF á- HF ʊ́- HF ʊ́- HF :: -ɔ́ ʊ̀- HF :: -ɔ́

Fixed Short

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

HFM á- HFM ʊ́- HFM à- HFM ʊ̀- HFM ʊ́- LF

M :: á- M :: ʊ́- M :: à- M :: ʊ̀- M :: ʊ́- L ::

HFM :: á- HFM :: ʊ́- HFM :: à- HFM :: ʊ̀- HFM :: ʊ́- LF ::

M :: á- M :: ʊ́- M :: à- M :: ʊ̀- M :: ʊ́- L ::

HFM :: á- HFM :: ʊ́- HFM :: à- HFM :: ʊ̀- HFM :: ʊ́- LF ::

M :: á- M :: ʊ́- M :: à- M :: ʊ̀- M :: ʊ́- L ::

SPECIAL PUBLICATION NO. 14 , 2018

Forms and Functions of the Base Paradigm of Shilluk Transitive Verbs

80

Table B.2. Schematic representation of general subject marking by TAM (separate panels), person, and verb class (columns). Each class is represented by one verb: {ŋɔ̀l} ‘cut’, {lɛ̂ŋ} ‘drum’, {càm} ‘eat’, {mʌ̂l} ‘roast’, {lɛ̀ɛŋ} ‘throw’, {mâat ̪} ‘drink’, and {mấal} ‘praise’. PAST

Fixed Short

Low

1st sg. 2nd sg. 3rd sg.

á- L -á á- L á- L -ɛ́

NO TENSE

Long

Fall

Low

Fall

Low

Low Fall

High Fall

á- LF -à á- LF á- LF -ɛ̀

á- L :: -á á- L :: á- L :: -ɛ́

á- LF :: -à á- LF :: á- LF :: -ɛ̀

á- L :: -á á- L :: á- L :: -ɛ́

á- LF :: -à á- LF :: á- LF :: -ɛ̀

á- L :: -à á- L :: á- L :: -ɛ̀

Fixed Short

Short with Grade

Long

Fall

Low

Fall

Low

Low Fall

High Fall

1st sg.

LHF -à

LF -à

LHF :: -à

LF :: -à

LHF :: -à

LF :: -à

L :: -à

3rd sg.

LHF -ɛ̀

LF -ɛ̀

LHF :: -ɛ̀

LF :: -ɛ̀

LHF :: -ɛ̀

LF :: -ɛ̀

L :: -ɛ̀

2nd sg.

FUTURE 1st sg. 2nd sg. 3rd sg.

Low

Short with Grade

LHF

Low

LF

Fixed Short

ʊ́- LHF -à ʊ́- LHF ʊ́- LHF -ɛ̀

LHF ::

LF ::

Short with Grade

LHF ::

LF ::

Long

L ::

Fall

Low

Fall

Low

Low Fall

High Fall

ʊ́- LF -à ʊ́- LF ʊ́- LF -ɛ̀

ʊ́- LHF :: -à ʊ́- LHF :: ʊ́- LHF :: -ɛ̀

ʊ́- LF :: -à ʊ́- LF :: ʊ́- LF :: -ɛ̀

ʊ́- LHF :: -à ʊ́- LHF :: ʊ́- LHF :: -ɛ̀

ʊ́- LF :: -à ʊ́- LF :: ʊ́- LF :: -ɛ̀

ʊ́- L :: -à ʊ́- L :: ʊ́- L :: -ɛ̀

Table B.3. Schematic representation of applicative subject marking by person (rows), and verb class (columns). The forms given are in the No Tense form; Past, Future, Sequential Past, and Conditional levels of TAM differ only in the addition of a prefix: á- in the case of Past, ʊ́- in the case of Future, à- in the case of Sequential Past, and ʊ̀- in the case of Conditional. SUBJ, APPLIC Low 1st sg. 2nd sg. 3rd sg.

M -á HR M -ɛ́

Fixed Short

Short with Grade

Fall

Low

H -á H H -ɛ́

M :: -á HR :: M :: -ɛ́

Long

Fall

Low

Low Fall

High Fall

H :: -á H :: H :: -ɛ́

M :: -á HR :: M :: -ɛ́

H :: -á H :: H :: -ɛ́

M :: -á HR :: M :: -ɛ́

SPECIAL PUBLICATION NO. 14 , 2018

Language Documentation & Conservation Special Publication No. 14 – A Grammar of Shilluk Chapter 2: Inflectional morphology and number marking in Shilluk nouns by Bert Remijsen & Otto Gwado Ayoker http://hdl.handle.net/10125/24778

CHAPTER 2

Inflectional morphology and number marking in Shilluk nouns Bert Remijsen & Otto Gwado Ayoker, University of Edinburgh

Abstract • This chapter offers a descriptive analysis of two topics in the morphology of Shilluk nouns: the inflectional paradigm, and number marking. Aside from the base form, the inflectional paradigm includes the following four forms: a) pertensive with singular possessor; b) pertensive with plural possessor; c) construct state; and d) proximal demonstrative. All of these can be interpreted as instances of head marking, which is characteristic of Shilluk morphosyntax in general (cf. Chapter 1). Following a description of the morphosyntactic functions of the base form and the four inflections, we describe in detail the patterns of morphophonological exponence through which the inflections are expressed. This pattern of exponence includes vowel length, tone, nasalisation, floating quantity, and suffixation. Floating quantity is of particular note: this marker has not been postulated in earlier work. Overall, we find that the inflectional paradigm is largely productive and regular. In contrast, the morphological marking for number is neither regular nor productive, and this is why we do not consider it to be part of the inflectional paradigm. The newly discovered marker of floating quantity supports Gilley’s (1992) tripartite analysis of number marking for Shilluk. For the sake of clarity and accountability, sound examples are embedded in relation to each of the numbered illustrations.

Licensed under Creative Commons Attribution Non-Commercial No Derivatives License

ISBN 978-0-9973295-2-9

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

2

1 Introduction 1.1 Scope and structure

This chapter presents descriptive analyses of two dimensions that are central to the morphology of nouns in Shilluk. One is the inflectional paradigm; the other is the marking of number, which we will argue to be a separate matter. It is the latter topic, the marking of number, which has been the focus of much of the earlier work on the morphology of Shilluk nouns (Gilley 1992, 2000; Storch 2005; Reid 2009; Remijsen, Miller-Naudé & Gilley 2015; Xu 2017). All of these studies note the complexity of number marking in Shilluk. The same characteristic has been observed in closely-related languages such as Dinka (Ladd, Remijsen & Manyang 2009) and Nuer (Frank 1999, Baerman 2012), and also in relation to the Nilo-Saharan languages as a whole (Dimmendaal 2000). We will show that number marking in Shilluk is neither regular nor productive, and for these reasons we interpret it as derivational rather than inflectional (cf. Haspelmath 1996:47). As for the inflectional paradigm of nouns, the following four functions are expressed: a) pertensive with singular possessor; b) pertensive with plural possessor; c) construct state; and d) proximal demonstrative. This inflectional paradigm has not yet been described in full; it is center stage here. The common characteristic of these inflections is that all four represent instances of head marking. That is, they are inflections on the head of the noun phrase, signposting a functional meaning in relation to a constituent within its own syntactic domain (cf. Creissels 2009). For example, pertensive signposts that the head noun is accompanied by a possessor. It is insightful to contrast head marking with dependent marking, familiar from languages like Latin, where the nominal inflections known as cases express a particular dependency relation of the marked noun to a head, be it the head of the clause (e.g. accusative), or to the head of the noun phrase (e.g. genitive). In Shilluk, in contrast, it is the head of the syntactic domain that is morphologically marked, rather than the dependent. This head-marking nature is not specific to nouns; it is equally characteristic of Shilluk verbs, where three voices signpost the nature of the dependents. This voice system is described in Chapter 1. Various instances of morphological marking that are found on nouns are beyond the scope of this chapter, as it covers a) inflections that are specific to nouns; and b) number marking. This excludes morphological marking for vocative and for associative plural: both of these functions are marked at SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

3

the right edge of the noun phrase, irrespective of whether the domain-final constituent is a noun, a verb or an adjective. However, because they are most often marked on nouns, we will briefly illustrate the functions of these morphological processes in this chapter, without describing their formal realization in detail. Also beyond the scope of this chapter are compounding, borrowing, and various derivations through which nouns are formed on the basis of roots that belong to other lexical categories. Finally, we exclude the morphological marking of sex and maturity on cattle terms. This part of the nominal morphology has been described in detail in Martin (2018). All of these morphological processes are orthogonal to the inflectional system described in this chapter. That is, a compound noun like ɔ̄ɔt-jāat ̪ ‘hospital’, a borrowing like bɛ̂ɛ́ k ‘bag’, a deverbal nominalization like à-mǎk ‘captive’, and a cattle term marked for sex such as á-dík-ɔ̀ ‘cow (female, mature) with deep red hide color’ – all of these display the inflectional morphology described in this chapter. The chapter is structured as follows. In Section 2, we present an overview of the functions that are marked through inflection. As noted above, they are pertensive with singular possessor, pertensive with plural possessor, construct state, and demonstrative. At the end of that section, we will also illustrate the morphological marking for associative plural and for vocative, even though these two phenomena are not specific to nouns. The following two sections, Section 3 and Section 4, are dedicated to the description of the morphophonological forms that express the four inflections. In line with the above interpretation of the nature of number marking, the paradigms of singular and plural nouns will be treated separately: Section 3 describes the inflectional paradigm marking the four functions outlined above in relation to singular nouns, and Section 4 describes the same paradigm in relation to plurals. The morphological marking of number on nouns is described in Section 5. A brief summary of Shilluk phonology and the conventions used in transcriptions can be found in the introduction to Chapter 1.

1.2 Methods

Our descriptive analysis is based primarily on a dataset of around 1600 Shilluk nouns. This is counting singulars and plurals separately; there are over 900 singular nouns, the remainder being plurals. The dataset consists of two types of data: a) transcriptions of nouns in various inflections, and b) over 5000 sound clips of noun targets elicited in sentence frames. Both are publicly available as part of archived lexicographic materials (Remijsen, Ayoker & SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

4

Martin 2018). The dataset and the resulting analysis of Shilluk nouns were built up gradually, over a ten year period, as we studied Shilluk through controlled elicitation and through the analysis of narrative text. The narratives were useful to establish the functions of the various inflections. In this chapter, illustrations drawn from this corpus of narratives are marked using a ^ following the illustration number; in each case, the name of the narrative and the time interval of the relevant extract in that recording are specified between square brackets. The audio recordings and annotations of the narratives in question are publicly available online, through Edinbugh DataShare (Remijsen & Ayoker, no date). We used controlled elicitation to determine the formal realization. Here we describe the frames used most in controlled elicitation. Whenever we encountered a new noun, we collected three forms: the base form, the 1st singular possessed form, which is based on the pertensive, and the demonstrative. We collected these three forms for those levels of number that exist for that lexeme, be it singular, plural, or both. This is illustrated in (1) for the nouns àd̪ʌʌ ́ t ̪ ‘bottle’, which has both singular and plural forms. As seen from (1), the frame sentence we used is headed by the irrealis existential predicate marker dɪ ́ɪ, which takes a noun phrase argument.1 Because of it general meaning, i.e., existential, this predicate can accommodate any noun as its argument (cf. Pike 1948). In addition, the same frame sentence can be used with singular and plural nouns: the predicate marker is not marked for number. (1)

a.

kɛ̀ dɪ ́ɪ àd̪ʌʌ ́ t̪ cnd exsp:irr bottle ‘If there is a bottle.’

d.

kɛ̀ dɪ ́ɪ àd̪ʌt́̂ -̪̄ ɪ ̄ cnd exsp:irr bottle-p ‘If there are bottles.’

b.

́̂ ʌt-̪̄ āa kɛ̀ dɪ ́ɪ àd̪ʌʌ cnd exsp:irr bottle-1s ‘If there is my bottle.’

e.

kɛ̀ dɪ ́ɪ àd̪ʌt́ ̪-áa cnd exsp:irr bottle:p-1s:p ‘If there are my bottles.’

c.

kɛ̀ dɪ ́ɪ àd̪ʌʌ ́ ʌǹ̪ μ cnd exsp:irr bottle:cs:dem ‘If there is this bottle.’

f.

kɛ̀ dɪ ́ɪ àd̪ʌt́ ̪-ɪ ̀ cnd exsp:irr bottle:p-dem ‘If there are these bottles.’

We used the sound recordings to ensure that our transcriptions are accurate. This was important in relation to tone and vowel length, because Shilluk phonology is particularly rich in terms of both of these dimensions of 1 The symbol μ stands for mora or weight unit; we use it as a representation of floating quantity. Its role in Shilluk morphophonology is described in Sections 3.3 and 4.3. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

5

phonological contrast. Aside from its ‘broad spectrum’ applicability in semantic terms, the frame sentence (kɛ̀ dɪ ́ɪ [target]) presents characteristics that are conducive to establishing the phonological characteristics of the target word with accuracy. First, the target noun appears in utterancefinal position, where its realization is maximally salient thanks to final lengthening. Second, because the utterance is so short, the fundamental frequency (f0) range is wide, which helps in the identification of tone patterns. Third, the specification for High tone on dɪ ́ɪ is like a tuning fork: it offers a reference of the f0 level at the high end of the speaker’s range, and this reference is helpful in determining the specification for tone of the noun target, be it in impressionistic auditory analysis or in the inspection of the f0 pattern. There is one phonological contrast in the realm of tone that cannot be discriminated easily using the kɛ̀ dɪ ́ɪ frame: Low Fall vs. High Fall. In the context of an immediately preceding High tone target, the Low Fall is very similar in its tonal realization to the High Fall. Preceded by a Low tone target, in contrast, the difference between Low Fall and High Fall is salient, but here the realization of the Low Fall is similar to that of the Low. Hence, the tonal specification of nouns that have an early-aligned falling melody on the stem syllable cannot be determined accurately in the kɛ̀ dɪ ́ɪ context, and for these nouns we have used a second preceding context, namely dɪ ̂-a, the realis existential predicate marker, which is phonetically realized as [dâa]. In this frame sentence, the noun is immediately preceded by a Low tone target. The allophonic range serves as the heuristic to determine unambiguously whether a noun stem carries a Low Fall as opposed to a High Fall. This is illustrated in (2), which shows how the Low, the Low Fall and the High Fall are realized in the two preceding contexts. The embedded sound illustrations reveal that, following after High-toned dɪ ́ɪ, the melody of mʌ̂ʌt ̪, which carries a Low Fall (2c), is akin to that of mʊ̂́ʊt ̪, which carries the High Fall (2e). In contrast, following after dɪ ̂-a, which ends in a Low target, the melody on mʌ̂ʌt ̪ in (2d) is similar to that of lʊ̀ʊt ̪ in (2b). By collecting nouns with an earlyaligned falling melody in both of these frames, Low Fall and High Fall can be distinguished with accuracy.

SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns (2)

6

a.

kɛ̀ dɪ ́ɪ lʊ̀ʊt ̪ cnd exsp:irr stick ‘If there is a stick.’

b.

dɪ ̂=a lʊ̀ʊt ̪ exsp=f stick ‘There is a stick.’

c.

kɛ̀ dɪ ́ɪ mʌ̂ʌt ̪ cnd exsp:irr friend ‘If there is a friend.’

d.

dɪ ̂=a mʌ̂ʌt ̪ exsp=f friend ‘There is a friend.’

e.

kɛ̀ dɪ ́ɪ mʊ̂́ʊt ̪ cnd exsp:irr banana ‘If there is a banana.’

f.

dɪ ̂=a mʊ̂́ʊt ̪ exsp=f banana ‘There is a banana.’

1.3 The shape of Shilluk nouns

In this subsection, we describe the word structure of the base forms of Shilluk nouns. The base form is the form within the inflectional paradigm which tends to reflect the root most closely, albeit not completely.2 We do not make a distinction here between native monomorphemic nouns and other types of nouns (i.e., derived nouns, compounds, loan words), because the inflectional morphology applies in the same way to all nouns. The shape of the base form is particularly important because so much of Shilluk morphology is stem-internal. The base form, along with the other forms of the inflectional paradigm, has at its center a stem that consists of a single syllable.3 The phonotactic structure of this stem is C(w/j)V(V)(V)C. That is, the template of the stem includes an onset, which may be complex, in which case the second element is a semivowel (either /j/ or /w/); a vocalic nucleus, which is short (V), long (VV), or overlong (VVV) (Remijsen, Ayoker & Jørgensen 2019); and a coda consonant. A sizeable proportion of Shilluk nouns, both singular and plural ones, display this monosyllabic template in the base form. Illustration (3) presents several examples of this template. (3)

d̪ɔk ́ ‘mouth’

lɛ̄k ‘teeth’

bjɛ́l ‘grain’

mɛ̂ɛl ‘drought’

lwáak ‘byre’

ràaw ‘millet’

jʌ̂́ʌʌtμ̪̄ ‘boats’

mwɔ̄ɔ̌ ɔlμ ‘morning’

ŋ̀uuur ‘lion’

Onsetless nouns are very rare in the native vocabulary – the only cases we ̄ ̄ ‘deep hole in ground-s/-p’ and know of are: ɔ̄ɔt/ɔ̂ɔt ‘house:s/:p’ and ốl-ɔ̀/ốl-ɪ òook ‘wickerwork fish traps’. Open-syllable nouns, e.g. t ̪ʌ̌ʌʌ ‘desert date’, are also uncommon. 2 In suffixed nouns, the base form may display morphological lengthening (see Section 3.2). 3 That is, unless it is a compound, in which case there is more than one stem syllable. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

7

The monosyllabic stem that is at the center of the base form can be flanked by a suffix and a prefix. We will describe each of these in turn, starting with suffixes. In singular base forms, the suffix -ɔ is common, as in twốooŋ̄-ɔ̄ ‘insect (generic)-s’ and bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman-s’, and many plural base forms present the suffix -ɪ, as in pû́k̄-ɪ ̄ ‘big clay pot-p’ and pùk-ɪ ̀ ‘turtle-p’. The only other suffix that is found in base forms is the derivational suffix -ɪ ̄ɪ, which derives instrument nouns from transitive verbs, as in góooc-ɪ ̄ɪ ‘machete’ (from {gɔ̀ɔc} ‘hit’) and ŋíc-ɪ ̄ɪ ‘symbol’ (from {ŋɪ ̀c} ‘recognize’). The monosyllabic stem may be preceded by another syllable.4 The inventory of these prefixes is highly restricted: only two of them are found with frequency in the lexicon, both of them vocalic: a- and ʊ-. For each of these, there are hundreds of nouns that are formed with them.5 We will discuss these two prefixes in turn. It will become clear that the phenomena run parallel. Some nouns formed with prefix a- are transparently derived. For example, Low-toned à- serves to derive nouns referring to the undergoer of the event referenced by a transitive verb, as in (4a). And High-toned á- derives cardinal numbers from ordinal numbers, as in (4b). High-toned á- and Low-toned àalso derive cattle nouns referring to cows, in the sense of mature female cattle, as in (4c). (4)

a.

à-mǎk pat-catch ‘captive’ à-ɲwɔ̌m pat-marry ‘married woman’

b.

á-kjɛ̀l crd-first ‘one’ á-dʌ̀k crd-third ‘three’

c.

á-dùk fem-grey ‘grey cow’ à-cwíil fem-many.colours ‘cow with many colours’

In other cases, there is a clear semantic relation between a word with a prefix and a word without one, but the semantic relation cannot be defined as easily. For example, álwɛ̂ɛ́ ɛt-ɔ̄ ‘crab’ is related to lwɛ̂ɛ́ ɛt-ɔ̄ ‘finger’, and áwāaac-ɔ̄ ‘kind of plant with bitter fruit’ to wâc ‘bitter’. Finally, there are words with initial a- that are not obviously related to a word without prefix. Examples 4 There are words that present two syllables before the stem, in a particular morphological derivation: cattle terms in which both gender and maturity are marked (Martin 2018). Martin (2018:44,45) lists examples such as ɲāa-dɪ ̄-lwāl ‘light red bull calf (dim-bull-light.red)’ and ɲāa-ʊ́-gwɛ̂l ‘bull calf with many colours (dim-bull-many.colours)’. However, this could also be interpreted as a compound noun. 5 Segmentally identical prefixes, i.e., a- and ʊ-, with a various specifications for tone, have a high functional load in Shilluk verb morphology, marking levels of tense-aspect-modality (cf. Chapter 1). SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

8

include áŋwʌ̄ʌk ‘oryx’ and ápɪ ̂ɪj ‘ice’. Either there was a derivational base and it has been lost subsequently, or there never was a derivational base, for example in the scenario of borrowing, and the Shilluk word was formed with initial a- to begin with. It is worthwhile to note that restrictive relative clauses are formed using à. Hence, it is plausible for nouns beginning with this segmental sequence to represent reinterpretations of what was once a relative clause. This diachronic path, i.e., for nouns to develop diachronically out of headless relative clauses, has been noted in various other Nilotic languages, including Maa (Vossen 2015) and Turkana (Dimmendaal to appear). The situation is similar in relation to ʊ-. Some derivations with this prefix have a transparent semantic relation with the source. For example, nouns marked with ʊ̀- can refer to the result of the action referenced by the source verb, as in (5a). Also, again drawing on Martin (2018), Low-toned ʊ̀- is a morphological marker of mature female cattle, as in (5b), and High-toned ʊ́marks a male member of a group, or a bull (5c). (5)

a.

ʊ̀-pɛ̀ɛt nom-cut.to.dry ‘dried fish’ ʊ̀tút nom-tie.together ‘knot’

b.

ʊ̀-bwôoor-ɔ̀ fem-brown ‘brown cow’ ʊ̀-bɔ̂ɔ́ w fem-white ‘white cow’

c.

ʊ́-cʊ̂́ʊʊlɔ̀ mal-Shilluk ‘Shilluk man’ ʊ́-wéer mal-horns.upwards ‘bull with horns pointing upwards’

There are also cases where there is a semantic relation with a root, although it is difficult to pinpoint, as in the case of ʊ́-cǔɲ ‘liver’, which is related to cúɲ ‘emotion’.6 Finally, there are words with the prefix ʊ- for which no derivational base is known, such as ʊ́gǐik ‘buffalo’ and ʊ́pǔun ‘cake’. Just as is the case with comparable phenomena involving a-, either there was a derivational base and it has been lost subsequently, or there never was a derivational base to begin with, and the word was formed in analogy with the derivational use of this prefix. Other prefixes have a much lower functional load. We note dɪ ̄-, observed in seven nouns, including dɪ ̄-kɔ́k ‘hoe’ and dɪ ̄-lāal ‘bronze’. The derivation is transparent in most cases.7 Most of these refer to instruments. In addition, 6 Many languages express emotional states with reference to body parts (see e.g. Klamer 2001). Thus, the similarity between the Shilluk words for ‘emotion’ and ‘liver’ is not surprising. 7 For example, dɪ ̄-kɔ́k ‘hoe’ is transparently derived from the transitive verb {kɔ̂k} ‘hoe’. In the case of dɪ ̄-lāal ‘bronze’, the sequence */lāal/ is not found as a monosyllabic noun, but it is found in á-lāal ‘flat-topped acacia’, presumably with reference to the reddish colour of its bark. This is likely, because another variety of acacia, namely ácàaar-ɔ̀ ‘yellow-bark acacia’, is SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

9

Martin (2018), a study on Shilluk cattle nouns, reports many terms referring to bulls, such as dɪ ̄-dùk ‘grey bull’ and dɪ ̄-bwōoor ‘brown bull’. The final two prefixes are ɲɪ- and ɲaa-. The form ɲɪ ̄ is used almost exclusively in relation to cattle terms, where it is found both in bull terms like ɲɪ ̄-bɔ̂ɔ́ ŋ ‘white bull’ and heifer terms ɲɪ ̄-bɔ̂ɔ́ w ‘white heifer’ (Martin 2018:43). Outside this domain, it is used to refer to offspring, as in ɲɪ ̄-rʌ̄t ̪ ‘prince’ and ɲɪ ̄-wâaac-ɔ̀ ‘cousin’. The situation for ɲaa- is similar. It is found in terms referring to immature cattle and in kinship relations, e.g. ɲāa-dɪ ̄-dùk ‘grey bull calf’ and ɲáa-dùk ‘grey heifer’ (Martin 2018:43). In summary, the set of nominal prefixes is restricted to a-, ʊ-, dɪ-, ɲɪ-, ɲāa. Of this set, only a- and ʊ- occur with high frequency in the lexicon. Interestingly, all five of these prefixes figure prominently in the marking of sex and maturity on cattle terms. For the analysis of inflectional morphology, the prefixes to the monosyllabic stem are irrelevant, in that the morphophonological processes apply identically to nouns that have prefixes as to nouns that do not. In other words, nouns with prefixes do not present separate inflectional paradigms, irrespective of whether the derivational function of a prefix is a) transparent, productive and regular, as in à-mǎk ‘captive’, b) transparent but not productive and regular, as in ʊ́-cǔɲ ‘liver’, and c) even if there is no synchronically transparent derivational relation at all, as in the case of of ʊ́gǐik ‘buffalo’. These affixes in Shilluk mirror comparable phenomena in other Nilo-Saharan languages, many of which present prefixes whose synchronic function is not always clear (Dimmendaal to appear).

2 Shilluk noun inflections and their functions Apart from the base form, nouns appear in four inflections: • • • •

pertensive with singular possessor pertensive with plural possessor construct state demonstrative

This set of inflections defines nouns as a lexical category in Shilluk. That is, the paradigm listed above distinguishes nouns from members of the two other major lexical categories, i.e., verbs and adjectives, which present different inflectional paradigms. Specifically, verbs are regularly inflected for tensealso named with reference to the color of its bark: câar means ‘light’. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

10

aspect-modality, voice and subject (Chapter 1), and adjectives have a stagelevel form. In this section we describe the functional range of the base form and each of these four inflections.

2.1 The base form

If a noun is not modified, it appears in the base form. This is illustrated by the narrative example in (6). There are two nouns here that do not have any modifiers. One is lʌ̂́ʌʌj ‘animals’ in the first clause; the other is ʊ́gǐik ‘buffalo’ in the third. (6)^

cjʌ̂ŋ á-kjɛ̀l gɛ́ á-mɔ̂ɔɔl bɛ̄ɛ dwāar kɪ ́ lʌ̂́ʌʌj day crd-first pr3p pst-go.early sub hunt:inf.a prp animal:p ‘One day, they went early to hunting animals, kấā gɛ́ ʊ́-rɔ̂ɔɔm-ɔ̀ kɪ ́ ʊ́gǐik mɛ́ dwɔ̂ɔ́ ŋ conj pr3p impf-meet prp buffalo rel.s big and they encounter a big buffalo,

kấā ʊ́gǐik rɛ́ŋɪ ́=a jǐii gɛ́n conj buffalo run:dest:nt=f ad:p pr3p and the buffalo runs towards them.’ [KeepTheSecret 26.9-33.3]

In addition, the base form is used with two particular modifiers. One of these is a constituent marked by mɛ́/mɔ́, which marks modifiers and relative clauses of indefinite referents.8 Note that, in the second clause in (6), the buffalo is introduced as ʊ́gǐik mɛ́ dwɔ̂ɔ́ ŋ ‘a big buffalo’. This is the first mention of this entity in the narrative, and hence it is indefinite. The allomorph mɛ́ is used if the head is grammatically singular, as in (6) and (7a), and the allomorph mɔ́ is used if the head is grammatically plural, as in (7b). (7)

a.

dɪ ̂=a ʊ́gǐik mɛ́ dwɔ̂ɔ́ ŋ exsp=f buffalo rel.s big ‘There is a big buffalo.’

b.

dɪ ̂=a ʊ́gìiik mɔ́ dɔ̀ɔŋ-ɔ̀ exsp=f buffalo:p rel.p big-p ‘There are big buffalos.’

Nouns that do not have a modifier are not marked for indefiniteness by mɛ́/ mɔ́, nor by any other function morpheme. However, the inflection for voice on the verb heading a clause is determined by information structure, and in particular by what the topic is. In this way, the voice of the verb conveys whether referents are to be interpreted as definite within the discourse (see Section 3.1 of Chapter 1). 8 We know of two nouns that have forms that are specific to junctures with mɛ́ /mɔ́. One is ɟâal-ɔ̀ ‘man’, which appears as ɟâal when followed by mɛ́. The other is ɟɔ̂k ‘men’, which appears as jâak ‘men’ when followed by mɔ́. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

11

Illustration (8) presents two additional narrative examples illustrating the same facts. In (8a), ʊ́gɔ̂t ̪ mɛ́ lʊ́ʊʊc̀ ‘a black cloth’ is the first reference to this entity. In (8b), gìc-ɔ̀ ‘something’ is inherently (i.e., lexically) indefinite. In both cases, the modifier marked by mɛ́/mɔ́ is an adjective.9 It may equally be a verb or a relative clause. (8)

a.^ gìn-ánɪ ́ rɪ ́ gɔ̀ kùm-ɛ̀ kɪ ́ ʊ́gɔ̂t ̪ mɛ́ lʊ́ʊʊc̀ something:cs-def refl pr3s.obl cover:nevp-3s prp cloth rel.s black:ctg ‘That thing, it had covered itself with a black cloth.’ [DownWithIllness 42.6-46] ́ ̄ dwɔ̂ɔ́ ŋ t ̪wɔ̂ɔw mɛ́ lɛ̂t ̪ càaarɔ̀ kɪ ́ mɛ̂n b.^ bǎa gìc-ɔ̀ nomp something-s rel.s painful very prp idp.s big anyway ‘It is something very painful, massive really.’ [SweetnessOfTheDrum 36.7-38.3]

The other modifier with which the base form is used is a cardinal number. This is illustrated in (6) by cjʌ̂ŋ á-kjɛ̀l ‘one day’. Cardinal numbers belong to the lexical category of nouns, on the basis of their inflectional paradigm. This means that junctures such as cjʌ̂ŋ á-kjɛ̀l ‘one day’ and ʊ́gìiik á-dʌ̀k ‘three buffaloes’ are noun-noun junctures.

2.2 The pertensive forms

Possessive noun phrases express a range of semantic relations between a possessed term, which is the head, and a possessor term, which is a modifier. Aside from the most obvious one, i.e., ownership (e.g., Anna’s book), in many languages possessive noun phrases may express part-whole relationship (e.g., the cover of the book), kinship (e.g., Anna’s mom), the attribute of an entity (e.g., the duration of the event), orientation / location (e.g., the side of the mountain), or association (e.g., Anna’s manager) (cf. Dixon 2010:262-263). In Shilluk, all of these semantic relations are most often expressed using a possessive construction in which the possessed term is inflected. Following Dixon (2010:268), we refer to this morphological marking as ‘pertensive’. Importantly, pertensive refers to morphological marking of the possessive relationship on the possessed term, which is the head of the possessive noun phrase.10 The Shilluk pertensive is illustrated in the narrative examples in (9). In (9a), pjēeen ʊ́gǐik ‘hide of the buffalo’, has pjēeen as its head, and ʊ́gǐik as 9 This is ascertained on the basis of the morphological paradigm, which is different for verbs vs. adjectives: verbs inflect for TAM, voice and subject; adjectives inflect for stage-level. 10 In contrast, ‘genitive’ refers to marking on the possessor, the modifier in the possessive noun phrase. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

12

the possessor-noun modifier. The possessive relationship is marked on the head: pjēeen̄ μ is the pertensive form of pjēn ‘hide’. The possessor term is not morphologically marked. There are three other pertensive forms in the same sentence. In the case of jếeep̄-ɛ̄, the base form of the noun is jép ‘tail’, which has the pertensive form jếeep̄μ. Here the possessor is expressed pronominally, through the 3rd singular suffix -ɛ̄, which attaches to the pertensive inflection of the noun. (9)

a.^ kấā pjēeenμ̄ ̌̄ ʊ́gǐik gɛ́-kɪ ́ jếeep̄-ɛ̄ kwấɲ ɪ ̄ɪ ɲɪ ́ɪɪmiii bʊ̌ʊl conj hide:pert buffalo pr3p-prp tail:prt-3s take:ov:nt prp.p siblings:prt Bol ‘And then Bol’s siblings took the hide of the buffalo and its tail.’ [KeepTheSecret 142.3-145.4] b.^ já ɲɪ ́ kit̂́ ̪ɪ ̀=à ɔ̄ɔt-jāaatμ̪ kấaalμ̄ àbʊ̂́ʊn jǐii jấā sìstèrâat pr1s hab carry:pet=f hospital:prt camp:prt priest ad:p apl order.of.nuns ‘They used to take me to the hospital of the Catholic mission (lit.: to the hospital of the camp of the priest, to those of the order of the nuns.’ [DownWithIllness 120.2-123.6]

The clause in (9b) presents an embedding of one possessive noun phrase inside another. The more deeply embedded possessive noun phrase is kấaalμ̄ àbʊ̂́ʊn which refers to a Catholic mission post, literally, ‘the camp of the priest’. The base form of the word meaning ‘camp’ is kâal. In (9b), however, this noun appears as kấaalμ̄ , with an overlong vowel and a High Fall to Mid specification for tone. These suprasegmental features are the exponents of the pertensive inflection. The noun kâal is in the pertensive because it has the possessor modifier àbʊ̂́ʊn ‘priest’. The last noun, the possessor, is not morphologically marked; it appears in the base form – rather than in its pertensive form, which would be àbʊ̂́ʊʊn̄ μ. The noun ɔ̄ɔt-jāaat ̪ ‘hospital:prt’ is the pertensive form of ɔ̄ɔt-jāat̪ ‘hospital’.11 It is modified by the possessor modifier kấaalμ̄ àbʊ̂́ʊn; this is why ɔ̄ɔt-jāat ̪ appears in the pertensive. For singular nouns, there are in fact two pertensive forms: one used with singular possessors, and the other used with plural possessors. They are illustrated in (10) using the nouns ʊ́gǐik ‘buffalo’ in (10a,b,c), and gwôk ‘dog’ in (10d,e,f). For these two nouns, the pertensive with singular possessor and the one with plural possessor both have an overlong stem vowel, and they additionally involve tonal exponence. The grammatical number of the possessor is expressed through tone. If the possessor is plural, this is marked on the possessed term through the addition of a High tone target, which is added to the lexical specification for tone. Hence, ʊ́gìiikμ, the pertensive form 11 This is a compound of the words ɔ̄ɔt ‘house’ and jāat ̪ ‘plant, tree, medicine’. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

13

used with a singular possessor, has a Low tone on its stem syllable (10b), but ʊ́gǐiik, which is followed by a plural possessor, has a Low Rise on its stem syllable (10c). Similarly, gwôk ‘dog’ has the High Fall to Mid when the possessor is singular (10e), whereas it has a High tone when the possessor is plural (10f). In this way, High Fall to Mid plus High yields High, and Low plus High yields Rise.12 (10)

a.

dɪ ̂=a ʊ́gǐik exsp:f buffalo ‘There is a buffalo.’

d.

dɪ ̂=a gwôk exsp:f dog ‘There is a dog.’

b.

dɪ ̂=a ʊ́gìiikμ twɔ́ɔŋ exsp:f buffalo:prt Twong ‘There is Twong’s buffalo.’

e.

dɪ ̂=a gwốook̄μ twɔ́ɔŋ exsp:f dog.s:prt Twong ‘There is Twong’s dog.’

c.

dɪ ̂=a ʊ́gǐiik mʌ́ʌn exsp:f buffalo:prt:p women ‘There is the women’s buffalo.’

f.

dɪ ̂=a gwóook mʌ́ʌn exsp:f dog.s:prt:p women ‘There is the women’s dog.’

The examples in (10) show that the tonal specification of the pertensive forms is determined by the number of the possessor, rather than by a contextual tone process: the singular possessor twɔ́ɔŋ (a person’s name) and the plural noun mʌ́ʌn ‘women’ are identical in terms of tonal specification: both possessors are High-toned. The morphological rather than phonological nature of the tonal contrast is equally in evidence when the following possessors are Low-toned, as in (11). (11)

a.

dɪ ̂=a ʊ́gìiikμ bɔ̀ɔɔt ̪-ɔ̀ exsp:f buffalo:prt craftsman ‘There is the craftsman’s buffalo.’

c.

dɪ ̂=a gwốook̄μ bɔ̀ɔɔt ̪-ɔ̀ exsp:f dog.s:prt.s craftsman-s ‘There is the craftsman’s dog.’

b.

dɪ ̂=a ʊ́gǐiik ɟɪ ̀ɪ exsp:f buffalo:prt:p people ‘There is the people’s buffalo.’

d.

dɪ ̂=a gwóook ɟɪ ̀ɪ exsp:f dog.s:prt.p people ‘There is the people’s dog.’

In the case of plural nouns, there is only one pertensive form. For example, the plural noun gwốook ‘dogs’ has the pertensive gwók-ɪ́, which is used both with a singular possessor, e.g. gwók-ɪ ́ bɔ̀ɔɔɔt ̪-ɔ̀ ‘the craftsman’s dogs’ and with a plural possessor gwók-ɪ ́ ɟɪ ̀ɪ ‘the people’s dogs’ The pertensive stem forms are also used when the possessor is expressed pronominally. Here again, the specification for tone on the possessed term marks the number of the possessor, e.g. gwốook̄-ɛ̄ ‘her/his dog’ vs. gwóook 12 The same morphophonological processes can be found in the verb system. See Chapter 1, illustration (1), and associated discussion. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

14

gɛ́n ‘their dog’. Note that the specification for tone when the noun is marked by the singular pronominal suffix is the High Fall to Mid (gwốook̄-ɛ̄), just as in (11c), where the possessor noun in singular. Similarly, the stem carries a High tone when the noun is followed by a plural pronominal possessor (gwóook gɛ́n), just as in (11d), where the possessor noun is plural.

2.3 The construct state form and the proximal demonstrative

A fourth form in the inflectional paradigm of nouns is the construct state form. It is used in a variety of constructions in which the head noun is modified by a constituent that is neither a) a possessor (in which case the pertensive inflections are used), b) nor marked by mɛ́ / mɔ́ (in which case the base form is used), and c) nor is it modified by a cardinal numeral (which also goes with the base form). One of these is illustrated in (12a), where the head noun is directly followed by an adjective. The same inflection is used when the noun is modified by a demonstrative (e.g. gwốooŋ̄=àcà ‘that dog’), a verb modifier (gwốooŋ̄ mʌ̀ʌʌt ̪-ɔ̀ ‘the drinking dog’), or a restrictive relative clause.13 The key characteristic distinguishing the construct state form from the pertensive is nasalisation of any root-final plosive. This can be seen from the comparison between the construct state form in (12a) and the pertensive with singular possessor form in (12b). Note that the forms of gwôk in (12a) vs. (12b) are identical but for the nasalization in (12a). (12)

a.

dɪ ̂=a gwốooŋ̄ dwɔ̂ɔ́ ŋ exsp:f dog:cs big ‘There is a big dog.’

b.

dɪ ̂=a gwốook̄μ twɔ́ɔŋ exsp:f dog:prt Twong ‘There is Twong’s dog.’

The proximal demonstrative inflection is found on the head noun of a noun phrase when proximal demonstrative represent the only modification, as in (13a). If there is any other modifier within the the noun phrase following the noun, the proximal demonstrative is expressed as a morpheme with segmental content, which is ɛ̀n if the head is grammatically singular, and ɛ̀k if the head is grammatically plural. These morphemes appear at the end of the end of the noun phrase, as in (13b). A noun inflected for proximal demonstrative differs from a noun inflected for construct state primarily in terms of tone, through the addition of a Low tone target. Added to the construct state form gwốooŋ̄, for instance, this results in a Late Fall specification for tone: gwóooŋ̀μ. A second marker of the proximal demonstrative is floating quantity. This feature is phonetically realized only in a particular following context, which we 13 Non-restrictive relative clauses are marked by mɛ́ / mɔ́, which conveys indefiniteness. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

15

describe in Section 3.3. If the word is in utterance-final position, as in (13), then floating quantity is not realized. (13)

a.

dɪ ̂=a gwóooŋ̀μ exsp=f dog:dem ‘There is this dog.’

b.

dɪ ̂=a gwốooŋ̄ à tɛ̂ɛk ɛ̀n exsp=f dog:cs rel strong dem.s ‘There is this strong dog.’

Illustration (14) presents examples from a narrative illustrating the use of the construct state form. Here the noun pâac ‘village’ appears three times in this inflection, which is pâaan.14 In (14a), this noun is modified by the definiteness-marking suffix -ánɪ́. Then, further on in the same narrative (14b), the same noun is modified by a restrictive relative clause and by an adjective. (14)

a.^ pâaan-ánɪ ́ à á-péekɪ ̀ ɪ ̀ɪ ɲɪ ́kāaaŋɔ̄ kɪ ̀ tjɛ̀ɛl-ɪ ̀ mâal village:cs-def f pst-settle prp Nyikango prp leg-cs first ‘That (aforementioned) village is where Nyikango settled down first.’ [TheDescendantsOfWaang 173.6-175.4] b.^ á-pêekɪ ̀=a àjɪ ̄ɪcd̪ấaɟ ̄-ɔ̄ past-settle=f Ayijdhajo

pâaan à cwɔ̂lɪ ̀-áa pâaan dwɔ̂ɔ́ ŋ àjɪ ̄ɪcd̪ấaɟ ̄-ɔ̄ ànàn village:cs rel call:iter-arg village:cs big Ayijdhajo now ‘He settled in Ayijdhajo, the village that is now called the big village Ayijdhajo.’ [TheDescendantsOfWaang 179.9-183.1]

2.4 Inflection at the level of the noun phrase

The inflections listed in the previous sections are specific to nouns: they cannot be marked on members of any other word classes. In addition, there are two instances of morphological marking that occur on noun phrases, specifically on the rightmost constituent within the NP. As a consequence, they are most often found on nouns but they can also be marked on members of other lexical categories. One of these is the vocative; the other is the associative plural. In this chapter, we limit ourselves to describing their function. Section 2.4.1 deals with the vocative, and Section 2.4.2 with the associative plural. However, they are not included in the descriptive analysis of nominal morphological exponence in Sections 3 and 4, where the scope is restricted to inflectional marking specific to nouns. 14 This construct state form is irregular; the regular derivation would have been *pâaaɲ, i.e., with the same place of articulation as the base form pâac. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

16

2.4.1 Vocative

The vocative is used when a noun phrase is used as a term of address. Its use is illustrated by the narrative examples in (15). In (15a), the vocative noun phrase is jáaak gɔ́l ‘Men of this family!’. Without vocative marking, it would have been jáaak gɔ́l̀μ, with a Late Fall on the possessor (gɔ́l̀μ), which is inflected for proximal demonstrative. In (15b), mʌ́ʌt ̪ is a vocative form of mʌ̂ʌt ̪ ‘friend’. In both instances, the vocative contributes a High end target to the tonal specification. (15)

̪̄ wāa a.^ jáaak gɔ́l bá kɛ̂t́ = bɛ̄ɛ lɪ ̂ɪt ̪-ɪ ́ ɟấm-ɪ ̀ men:prt family:dem:voc hort go.away=pr.1p.inc:n sub look:inf-prt.p things-p ‘Men of this family, let’s go check out the things.’ [TheDescendantsOfWaang 428.6-430.3] b.^ cùŋ-ɪ ́ mʌ́ʌt ̪, tɔ́ŋ kʊ́ʊ pùt ̪ stand:2s friend:voc spear proh pull.out:2s ‘Wait my friend, don’t pull out the spear!’ [KeepTheSecret 56.1-58.2]

In the above examples, the vocative happens to be marked on nouns. However, this is not invariably the case: vocative marking is found on whichever constituent appears at the right edge of the noun phrase. This is illustrated in Table 1. Note that the head noun ɟùr ‘woman’ is marked for vocative only if it appears at the end of the noun phrase. Otherwise, the vocative is expressed on the modifier at the right edge of the noun phrase, be it a verb, an adjective, or a demonstrative. In each case, a High tone target is associated with the last word. The combination of this High tone with a Low tone target yields a Low Rise (e.g. ɟùr + ́ > ɟǔr); when combined with a Low Fall, the result is a High Rise (e.g. tɛ̂ɛk + ́ > tɛ̄ɛ̌ k). Table 1. Examples of the morphological marking of vocative in noun phrases in which the final constituent belongs to different lexical categories. NP final constituent

Noun phrase without vocative

Noun phrase with vocative

Translation of the vocative

Noun Verb

ɟùr ɟùr à nèeen-ɔ̀

‘Woman!’ ‘Woman who is watching!’

Demonstrative

ɟùr àcà

ɟǔr ɟùr à nèeen-ɔ̌ɔ ɟùr à tɛ̄ɛ̌ k

Adjective

ɟùr à tɛ̂ɛk

ɟùr àcǎaa

‘Strong woman!’

‘Woman over there!’

In this way, the marking of vocative is appended to the right of the lexicalmorphological specification of the target syllable. Even if an NP-final noun has a suffix already, the High target that marks vocative can be added on. In SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

17

̄ addition, final vowels are lengthened in the vocative. For example, wʌ̂́ʌʌt-āa ́ ̌ ̄ ‘son:prt-1s’ yields the vocative wʌ̂ʌʌt-āaa ‘son:prt-1s:voc’. This is further illustrated in Table 1 by nèeen-ɔ̌ɔ ‘watch-inf:voc’.

2.4.2 Associative plural

The associative plural serves the function of widening the scope of reference from a referent to entities that belong with it or are conceptually similar to it. A cross-linguistic analysis of this phenomenon can be found in Moravcsik (2003). In Shilluk, the associative plural construction involves the constituent jấā, which is followed by a noun phrase referring to the entity that is central to the set. The final word in this noun phrase is morphologically marked for associative plural. The phenomenon is illustrated in (16) by an example drawn from a narrative. The associative plural expression is jấā dɛ̄ɛ̌ ɛŋ. The base form of the name ‘Deng’ is dɛ̄ɛŋ, and jấā dɛ̄ɛ̌ ɛŋ refers to Deng and his associates, in this case his kinship group. (16)^ ácàaarɔ̀ pā̌aa jấā dɛ̄ɛ̌ ɛŋ Acaro village:prt:p group.of Deng:apl ‘Acaro, the village of the people of Deng, […]’ [DengsFish 4.5-8.1]

Cross-linguistically, associative plural constructions tend to be restricted to human referents (Moravcsik 2003:472). In Shilluk, however, the functional scope of the associative plural is not restricted in this manner. This is shown by the narrative example in (17). The noun àkwàrɪ ̀còoot-ɔ̀ refers to a particular bitter herb, and jấā àkwàrɪ ̀còoot-ɪ ̄ɪ extends the reference to the wider set of herbs that it belongs to, i.e., to other bitter herbs. ́ (17)^ jấā àkwàrɪ ̀còoot-ɪ ̄ɪ cɛ̂k kấa gûuur-ɔ̀ ʊ̀ ɲɪ ́ kît ̪-ɪ ̀ kɛ̂ɲ-ánɪ ́ group.of k.o.herb-apl aux:ov prp:f grind-inf conj hab carry-dest place:cs-def ‘Herbs like akwaricoto [a bitter kind of herb] were ground and then put on the place.’ [DownWithIllness 152.2-155.2]

So far, we have only considered the associative plural in noun phrases consisting solely of the head noun, i.e., without any modifiers. Now we will present evidence to support the interpretation that the associative plural is marked on the final constituent in the noun phrase. Consider the noun phrase bɔ̀ɔt ̪-ɪ ̀ pâac ‘craftsman of the village’ in (18a). When this noun phrase appears in an associative plural, the morphological marking appears on the possessor pâac ‘village’, which is realized pā̌aac, as in (18b).

SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns (18)

a.

dɪ ̂=a bɔ̀ɔt ̪-ɪ ̀ pâac exsp=f craftsman-prt.s village ‘There is a village craftsman.’

b.

18

dɪ ̂=a jấā bɔ̀ɔt ̪-ɪ ̀ pā̌aac exsp=f group.of craftsman-prt.s village:apl ‘There is a village craftsman and his associates.’

The same is true when the final constituent in a noun phrase is not a noun. This is illustrated in (19) for a noun phrase involving an adjective modifier. In (19b), the adjective tɛ̂ɛk displays the morphological marking of associative plural in the same way as the possessor pâac does in (18b). (19)

a.

dɪ ̂=a bɔ̀ɔn̪ -ɪ ̀ tɛ̂ɛk exsp=f craftsman-cs strong ‘There is a strong craftsman’

b.

dɪ ̂=a jấā bɔ̀ɔn̪ -ɪ ̀ tɛ̄ɛ̌ ɛk exsp=f group.of craftsman-cs strong:apl ‘There is a strong craftsman and his associates.’

2.5 Conclusion

The inflections that are specific to nouns are pertensive, construct state, and demonstrative. These are all instances of head-marking, i.e., they are inflections on the head of the noun phrase, signposting a morphological meaning in relation to a constituent within its own syntactic domain. This is characteristic of Shilluk grammar as a whole: within the verb phrase as well, the relations between the verb and its arguments are morphologically marked primarily on the head (voice), rather than on the verb arguments (case). Similar head-marking inflections on nouns have been reported for several other West Nilotic languages, including Dinka (Andersen 2002) and Anywa (Reh 1996). Given the head-marking character of the inflectional morphology of Shilluk, it is convenient that vocative and associative plural are not (necessarily) marked on the noun. If vocative and associative plural were to be marked on the head of the noun phrase, then they would be competing for expression with the construct state and pertensive. In (19b), for example, the noun meaning ‘craftsman’ would need to express both associative plural and construct state. This would be problematic, in particular considering the fact that Shilluk morphology is predominantly fusional rather than agglutinative.

3 Morphological exponence in the inflectional paradigm of singular nouns In Section 2, we described the functions of the different forms in the nominal paradigm. Now we will describe the patterns of morphological exponence through which these inflections are expressed – here in Section 3 in relation SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

19

to singular nouns, and in Section 4 in relation to plural nouns. In relation to singular nouns, five morphophonological parameters play a role in the expression of these inflections: suffixation, vowel length, tone, floating quantity, and nasalization of the stem coda. The role of each of these is described in a dedicated subsection, and the final subsection, Section 3.6, offers an overview of exceptions.

3.1 Suffixation

The great majority of singular nouns display one of two patterns: they may be suffixless throughout the nominal paradigm introduced in Sections 2.1-2.3, or they may be suffixed throughout this paradigm. In our dataset, which includes over 900 singular nouns, roughly 60 percent of singular nouns are suffixless, and most of the remainder are suffixed.15 These two patterns are illustrated in Table 2. The great majority of suffixed nouns carry the suffix -ɔ in the base form, and in the inflected forms they have the suffix -ɪ. Suffixal marking in the base form is described in Section 3.1.1; suffixal marking in the inflected forms is described in Section 3.1.2. Table 2. Illustration of the division between suffixless and suffixed singular nouns. Suffixless

Suffixed

Base

lwáak ‘byre’

bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’

Pertensive (sg)

lwấaakμ̄

bɔ̀ɔt ̪-ɪ ̀

Pertensive (sg)

lwáaak

bɔ̀ɔt ̪-ɪ ́

Construct state

lwấaaŋ̄

bɔ̀ɔn̪-ɪ ̀

Demonstrative

lwáaaŋ̀μ

bɔ̀ɔn̪-ɪ ̀

There is only one other suffix that is found in the base form of singular nouns: -ɪ ̄ɪ, which is derivational, deriving instrument nouns from transitive verbs. Just as nouns that have the suffix -ɔ in the base form, these derived nouns ending in -ɪ ̄ɪ also have the suffix -ɪ in the inflected forms. For example, ŋíc-ɪ ̄ɪ ‘symbolnom’ has the pertensive form ŋic̄̂́ -ɪ̄. In contrast, suffixless nouns do not take a suffix in the base form, nor in the inflections listed in Table 2. However, 15 Exceptions to these patterns will be described further along in this section. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

20

all nouns take suffixes marking pronominal possessors. We will describe this system in Section 3.1.3. If we consider suffixation in the base and in the inflected forms as independent factors, there are actually four different possible combinations: a) suffixless both in base and in inflections; b) suffixless in base but suffixed in inflections; c) suffixed in base but not in inflections; and d) suffixed both in base and inflections. In fact, all four of these combinations are attested. Each of them is illustrated by an example in Table 3. However, as noted at the beginning of this section, almost all nouns are either suffixless throughout or suffixed throughout. The other two patterns are very rare: we know of 23 nouns that are suffixless in the base form but suffixed in inflections, and seven that are suffixed in the base form but suffixless in inflections. We treat these patterns as exceptional, and discuss them in Sections 3.6.2 and 3.6.4, respectively. Table 3. Illustration of the orthogonal crossing between suffixation in base and suffixation in the inflected forms. Suffixless Suffixless Inflected forms Suffixed

Base

Suffixed

Base Pertensive (sg)

lʊ̀ʊt ̪ ‘stick’ lʊ̀ʊʊtμ̪

tôoor-ɔ̀ ‘socket’ tôoorμ

Base

kǐtμ ‘mountain’

lɛ̀ɛɛl-ɔ̀ ‘pebble’

Construct state

kìn-ɪ ̀

lɛ̀ɛl-ɪ ̀

Construct state Pertensive (sg)

lʊ̀ʊʊn̪ kìt-ɪ ̀

tôoor lɛ̀ɛl-ɪ ̀

3.1.1 Suffixal marking in the base form of singular nouns

In relation to suffixed singular nouns, the suffix -ɔ is found with either Low, Mid, or High specification for tone, i.e., -ɔ̀,-ɔ̄,-ɔ́. But the specifications for tone on this suffix are not orthogonally crossed with specifications for tone on the stem syllable. Table 4 lists, for each specification for tone on the suffix, which specifications are attested on the preceding stem syllable. As seen from this table, each specification for tone on -ɔ combines with between two and four specifications for tone on the stem syllable: -ɔ̀ occurs with Low, Low Fall, High Fall, and Late Fall on the stem; -ɔ̄ occurs with Mid and High Fall to Mid, Mid; and -ɔ́ combines with Low Fall, Low, Mid, and High. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

21

Table 4. Illustrations of the specifications for tone on the stem syllable that are attested with different specifications for tone on the the suffix -ɔ. -ɔ̀ (Low) Common Rare

-ɔ̄ (Mid)

bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’ mûn-ɔ̀ ‘neck’ tû́l-ɔ̀ ‘owl’ káaak-ɔ̀ ‘island’

ɟấaak̄-ɔ̄ ‘chief’

-ɔ́ (High)

bwōɲ-ɔ̄ ‘Arab(s)’

bjêeel-ɔ́ ‘sorghum plant’ pʌ̀ʌʌt-ɔ́ ‘bark’ àpāk-ɔ́ ‘wave’ dɔ́ɔɔr-ɔ́ ‘axe’

Approached the other way around, each specification for tone on the stem combines either with one or with two specifications for tone on -ɔ. This is shown in Table 5. If the stem syllable of a noun suffixed with -ɔ has the High Fall, then the following -ɔ suffix invariably has the Low tone. Similarly, High Fall to Mid on the stem invariably goes with Mid on the following -ɔ. If the stem syllable has either the Low, the Low Fall or the High, the suffix -ɔ can have carry either Low or High tone. Finally, Mid-toned stem syllables combine with Mid and High tone on the suffix -ɔ. Table 5. Illustrations of the specifications for tone on the suffix -ɔ that are available for different specifications on the stem syllable. Tone on stem

Attested specifications on suffix -ɔ

Low Low Fall High Fall High Fall to Mid Mid High

bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’, pʌ̀ʌʌt-ɔ́ ‘bark’ mûn-ɔ̀ ‘neck’, bjɛ̂ɛɛl-ɔ́ tû́l-ɔ̀ ‘owl’ ɟấaak̄-ɔ̄ ‘chief’ bwōɲ-ɔ̄ ‘Arab(s)’, àpāk-ɔ́ ‘wave’ káaak-ɔ̀ ‘island’, dɔ́ɔɔr-ɔ́ ‘axe’

It is worthwhile to note that there are no suffixed singular nouns with a Low Rise or a High Rise on the stem syllable, even though these specifications are found on suffixless nouns. That is, nouns pʌ̀ʌʌt-ɔ́ ‘bark’ and bjêeel-ɔ́ ‘stalk of grain’ may be phonetically realized with rising F0 on the stem syllable. Crucially, however, there is no evidence of contrast between CV̀ C-ɔ́ and CV̌ Cɔ́, and likewise between CV̂ C-ɔ́ and CV̌̄ C-ɔ́. Aside from -ɔ, the base form may alternatively end in the suffix -ɪ ̄ɪ. This is a derivational suffix, yielding instrument nouns derived from transitive verbs (e.g. ŋíc-ɪ ̄ɪ ‘marker, symbol’ < {ŋɪ ̀c} ‘recognize’, and kóook-ɪ ̄ɪ ‘reward, payment’ < {kɔ̀ɔk} ‘pay to’). The combination of tonal specifications of stem syllable and suffix – High on stem, Mid on suffix – is particular to this SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

22

derivation: as noted in relation to Table 4, if a noun ending in -ɔ has a Mid tone on the suffix, then the stem syllable carries either the High Fall to Mid tone or the Mid tone, but not a High tone. In the inflected forms, singular nouns whose base forms end in -ɪ ̄ɪ carry the suffix -ɪ, just as singular nouns that end in -ɔ do. For example, kóook-ɪ ̄ɪ ‘reward’ yields pertensive with singular ̄ ‘chief’ yields ɟấaŋ̄-ɪ ̄ in the same inflection. possessor kốoŋ̄-ɪ ̄, just as ɟấaak-ɔ̄

3.1.2 Suffixal marking in inflected forms (apart from pronominal possession)

Leaving aside the pronominal marking of possession – we will go into it in Section 3.1.3 – suffixed singular nouns have the suffix -ɪ in all of their inflected forms. The specification for tone on this suffix is determined completely by the morphology in some inflections, whereas in others there is an interaction between the morphological specification and the tonal specification of the stem syllable, in other words, the stem’s lexical specification. The suffix -ɪ is invariably Low-toned in the demonstrative, and High-toned in the pertensive with plural possessor, whatever the specification for tone on the stem syllable. This is illustrated in Table 6. In contrast, in the pertensive with singular possessor and in the construct state, the specification for tone of the suffix -ɪ is predictable from the tone of the stem: it is Low-toned following a stem that carries a Low or a Low Fall, and Mid-toned following a stem that carries a Mid or a High Fall to Mid – these four being the only specifications for tone the stem can carry in the pertensive with singular possessor and in the construct state. In other words, the tone on the suffix -ɪ in pertensive (sg) and construct state matches the end target of the tonal specification of the preceding stem syllable. This is illustrated in Table 6, which shows the construct state ́ ̄-ɪ̄, with a forms tîm-ɪ̀, with a Low-toned suffix following the Low Fall, and wûn Mid-toned suffix following the High Fall to Mid. The tonal alternations of the stem syllable in the inflected forms will be laid out in Section 3.4. Table 6. Illustrations of the relation between tone on the suffix -ɪ and tone on the stem syllable. Low Fall on stem Base Pertensive (sg) Pertensive (pl) Construct state Demonstrative

tîp-ɔ́ ‘shadow’ tîp-ɪ ̀ tîp-ɪ ́ tîm-ɪ ̀ tîm-ɪ ̀

High Fall to Mid on stem

wû́n̄-ɔ̄ ‘tethering rope’ wû́n̄-ɪ ̄ wún-ɪ ́ wû́n̄-ɪ ̄ wún-ɪ ̀

SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

23

3.1.3 Suffixal marking of pronominal possession

All singular nouns, i.e., whether they belong to a suffixless paradigm or a suffixed one, take suffixal markers to express pronominal possession for certain combinations of person, number and clusivity. These paradigms are illustrated in Table 7, in relation to the nouns in Table 2 above. The pronominally marked forms are based on the pertensive inflections. This can be seen from ̄ Table 7: note that the pronominally marked forms (e.g. lwấaak-āa, bɔ̀ɔt ̪āa) display the same level of vowel length as the pertensive inflections (e.g. lwấaak̄, bɔ̀ɔt ̪-ɪ̀), as opposed to the level of vowel length of the base form (lwáak, bɔ̀ɔɔt ̪-ɔ̀). Moreover, the forms marked for a singular pronominal ̄ possessor (e.g. lwấaak-āa, bɔ̀ɔt ̪-āa) display the same specification for tone on the noun stem as the pertensive with singular possessor (lwấaak̄, bɔ̀ɔt ̪-ɪ́). And likewise, the forms marked for plural pronominal possessors – e.g. lwáaakɛ́ɛ, bɔ̀ɔt ̪-ɛ́ɛ – display the same specification for tone on the noun stem as the pertensive with a plural nominal possessor: lwáaak, bɔ̀ɔt ̪-ɪ́.16 Table 7. Illustration of the expression of pronominal possession on suffixless and suffixed singular nouns. Suffixless

Suffixed

Base

lwáak ‘byre’

bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’

Pertensive (sg)

lwấaak̄μ

bɔ̀ɔt ̪-ɪ ̀

Pertensive (pl)

lwáaak

bɔ̀ɔt ̪-ɪ ́

1st sg

lwấaak̄-āa

bɔ̀ɔt ̪-āa

2nd sg

lwấaak̄-ɪ ̄ɪ

bɔ̀ɔt ̪-ɪ ̄ɪ

3rd sg

lwấaak̄-ɛ̄

bɔ̀ɔt ̪-ɛ̀

1st pl incl

lwáaak-ɛ́ɛ

bɔ̀ɔt ̪-ɛ́ɛ

1st pl excl

lwáaak wɔ́n

bɔ̀ɔt ̪-ɪ ́ wɔ́n

2nd pl

lwáaak wún

bɔ̀ɔt ̪-ɪ ́ wún

3rd pl

lwáaak gɛ́n

bɔ̀ɔt ̪-ɪ ́ gɛ́n

We interpret as inflectional the expression of pronominal possession when the number is singular, and also in the 1st plural inclusive. Note how, in suffixed paradigms, the suffixes marking 1st, 2nd, and 3rd sg., and also the 16 The tonal alternations of the stem syllable in the inflected forms will be laid out in Section 3.4. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

24

one marking 1st pl. incl., appear instead of the pertensive suffixes -ɪ ̀/-ɪ̄ (sg) and -ɪ́ (pl). We interpret this interaction with the stem as an indication of the inflectional nature of this juncture. We do not consider to be inflectional the pronominal expression of 1st plural exclusive, 2nd plural and 3rd plural, i.e., wɔ́n, wún, gɛ́n. These free pronominal forms combine with the full form of the pertensive with plural possessor. Moreover, the pronominal expression of 1st plural exclusive, 2nd plural and 3rd plural is identical to the independent pronouns used as clause arguments. In contrast, 1st singular -āa, 2nd singular -ɪ ̄ɪ, 3rd singular -ɛ̀/-ɛ̄, and 1st plural inclusive -ɛ́ɛ are different from the corresponding independent pronouns used as clause arguments, which are ján, jín, ɛ́n, and wāa, respectively. For the possessor-marking suffixes 1st singular -āa, 2nd singular -ɪ ̄ɪ, and 1st plural inclusive -ɛ́ɛ, the specification for tone is fixed. In the case of 3rd singular -ɛ̀/-ɛ̄, however, there is a tonal alternation. The allomorphy is determined in the same way as for -ɪ ̀/-ɪ ̄ in pertensive (sg) and construct state (cf. Section 3.1.2): the Low-toned allomorph follows stems that carry Low and Low Fall, and the Mid-toned allomorph follow stems that carry Mid or High Fall to Mid.17 Hence we find bɔ̀ɔt ̪-ɪ ̀ kùl ‘Kul’s craftsman’ and bɔ̀ɔt ̪-ɛ̀ ‘her/his craftsman’, both with Low-toned suffixes, but ɟấaŋ̄-ɪ ̄ ‘Kul’s chief’ and ɟấaŋ̄-ɛ̄ ‘her/his chief’, both with Mid-toned suffixes.

3.2 Vowel length

Singular nouns follow one of three patterns of length alternation in their inflectional paradigms (Remijsen, Miller-Naudé & Gilley 2015). We will lay out these patterns first in relation to suffixless nouns, and then in relation to suffixed nouns. The patterns of vowel length alternation in suffixless singular nouns are illustrated in Table 8. First, the stem vowel in the base form may be long (VV), as it is in the case of māac ‘fire’ in Table 8. All such nouns have an overlong stem vowel (VVV) in the inflected forms, which are illustrated here by the pertensive with singular possessor. Nouns that follow this pattern can be referred to as Long nouns. In contrast, if the stem vowel in the base form is short (V), there are two possibilities. For some nouns, the stem vowel remains short throughout its inflectional paradigm. The noun lɪ ́ɲ ‘war’ illustrates this pattern. These nouns can be referred to as Fixed Short. For other nouns with a short vowel in the base form, the stem vowel becomes overlong (VVV) in the inflected forms, as in the case of bâk ‘garden’. These are the Short with Grade nouns. 17 Low, Low Fall, Mid, and High Fall to Mid are the only specifications for tone that can be found on the stem syllable in the pertensive with singular possessor. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

25

Table 8. The three patterns of vowel length alternation in the stem syllable of suffixless singular nouns, each illustrated by one noun: lɪ ́ɲ ‘conflict’, bâk ‘garden’, and māac ‘fire’. Fixed short (V~V)

Short with grade (V~VVV)

Long (VV~VVV)

Base

lɪ ́ɲ

bâk

māac

Pertensive, sg. poss

lɪɲ̄́̂ μ

bâaakμ

māaacμ

Whether a suffixless singular noun with a short stem vowel in the base form is Fixed Short or Short with Grade is predictable on the basis of its vowel quality: if the vowel is /a/ or /ʌ/, i.e., if the vowel is low (open), then the paradigm is Short with Grade, and the same is true when the onset of the stem syllable is complex and includes a semivowel (/w/ or /j/). Otherwise, if the stem vowel is neither low nor preceded by a complex onset involving a semivowel, then the paradigm is Fixed Short.18 The other way around, i.e., approached from an inflected form such as the pertensive (sg), it is impossible to predict whether a form with an overlong vowel that either has the quality /a/ or /ʌ/ or is preceded by a complex onset with a semivowel is part of a Short with Grade paradigm – and therefore has a short stem vowel in the base form – or rather part of a Long paradigm – in which case the base form has a long stem vowel. For example, whereas the pertensive (sg) māaacμ goes with the base form māac, there is no way to predict that the base form is māac and not *māc. Suffixed singular nouns display the same three patterns of vowel length alternation in the stem syllable as suffixless singular nouns, i.e., Fixed Short, Short with Grade and Long. This is illustrated in Table 9. Note that pɪ ̂c-ɔ̀ ‘tethering pole’ has a short vowel throughout its paradigm. This is the Fixed Short pattern. In Short with Grade suffixed nouns, such as pâaal-ɔ̀ ‘knife’ in Table 9, the stem vowel alternates in length between short and overlong. The stem vowel is short in the inflected forms, and overlong in the base. Finally, if the stem vowel of a suffixed noun is long in the inflected forms, then it will be overlong in the base form. This is the Long pattern, illustrated in Table 9 by pʌ̀ʌʌt-ɔ́ ‘bark’. 18 We know of one exception to this generalization. The noun bác ‘amniotic sac’ has a short vowel in the stem, and a vowel /a/, and yet it is Fixed Short: the pertensive (sg) is bấc̄μ and the construct state báɲ̀. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

26

Table 9. The three patterns of vowel length alternation in the stem syllable of suffixed singular nouns, each illustrated by one noun. Fixed Short (V~V)

Short with Grade (V~VVV)

Long (VV~VVV)

Base

pɪ ̂c-ɔ̀ ‘tethering pole’ pɪ ̂c-ɪ ̀

pâaal-ɔ̀ ‘knife’ pâl-ɪ ̀

pʌ̀ʌʌt-ɔ́ ‘bark’ pʌ̀ʌt-ɪ ̀

Construct state

pɪ ̂ɲ-ɪ ̀

pâl-ɪ ̀

pʌ̀ʌn-ɪ ̀

Pertensive, sg. poss.

As in suffixless nouns, the Long pattern of alternation is found with any vowel quality, whereas whether a noun is Fixed Short or Short with Grade is predictable on the basis of vowel quality and the composition of the onset of the stem syllable. If, in the inflected forms, a short stem vowel is open, i.e., /a/ or /ʌ/, or if it is preceded by a complex onset, with semivowel /j/ or /w/, then the corresponding base form has an overlong vowel (Short with Grade). Otherwise, the stem vowel in the base form is short (Fixed Short). The vowel length alternation works the same way in nouns carrying the other suffix that is found in singular base forms, that is, in instrument nouns ending in -ɪ ̄ɪ (cf. Section 3.1.1). In the paradigms of those nouns as well, if the base form has an overlong stem vowel, then the inflected forms have either a long or a short stem vowel. For example, góooc-ɪ ̄ɪ ‘machete’ has the construct state form gốoɲ̄ -ɪ ̄, and lʌ́ʌʌt ̪-ɪ ̄ɪ ‘spindle’ has the construct state form lʌ̂́n̪̄-ɪ̄. The pattern of quantity alternation (Fixed Short, Short with Grade, Long with Grade) in the inflected forms of such instrument nouns is the same as that of the transitive verb from which the instrument noun is derived (cf. Remijsen, Miller-Naudé & Gilley 2016). Comparing the length of the stem vowel in suffixed and suffixless noun paradigms, the patterns of alternation between base form and inflected forms are the exact opposite. That is, the suffixless nouns are in the short grade in the base form, and lengthen in inflected forms (unless the paradigm is Fixed Short). This is illustrated in the top half if Table 10. The suffixed singular nouns, in contrast, are in the short grade in the inflected forms and lengthen in the base form (unless the paradigm is Fixed Short), as seen from bottom half of Table 10. So, suffixed singular nouns display the same three patterns of quantity alternation in the stem vowel as sufixless singular nouns (Fixed Short, Short with Grade, Long), but the direction of the alternation between the base and the inflected forms is the reverse. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

27

Table 10. The relation between stem vowel length alternation and suffixation in singular noun paradigms, illustrated by d̪ɔk ́ ‘mouth’, káŋ ‘trumpet’, gʌ̀ʌt ‘riverbank’, pɪ ̂c-ɔ̀ ‘tethering stick’, pâaal-ɔ̀ ‘knife’, bɔ̀ɔɔt ̪-ɔ̀ ‘craftsman’. Fixed Short (V~V) Suffixless base pert, sg poss Suffixed base pert, sg poss

d̪ɔk ́ ́̂ ̄ μ d̪ɔk pɪ ̂c-ɔ̀ pɪ ̂c-ɪ ̀

Short w. Grade (V~VVV) káŋ kấaaŋ̄μ pâaal-ɔ̀ pâl-ɪ ̀

Long (VV~VVV) gʌ̀ʌt gʌ̀ʌʌtμ bɔ̀ɔɔt̪-ɔ̀ bɔ̀ɔt̪-ɪ ̀

It is worthwhile to note that the same three patterns of quantity alternation or the lack of it, i.e., Fixed Short, Short with Grade, Long with Grade are also found in the inflectional paradigms of verbs and adjectives. This is illustrated in Table 11. Table 11. The three patterns of length alternation that are found for the stem vowel in transitive verbs and in adjectives, illustrated by the verbs {ŋɔ̀l} ‘cut’, {càm} ‘eat’, and {lɛ̀ɛŋ} ‘throw’, and by the adjectives mɛ̂t ‘tasty’, wâc ‘sour’, pɛ̂ɛt ‘smelling badly’.

Transitive verbs Adjectives

inf.a inf base ctg

Fixed Short (V~V)

Short w. Grade (V~VVV)

Long (VV~VVV)

ŋɔ́l ŋɔ̀l-ɔ̀ mɛ̂t ʊ̀-mɛ̀t-ɔ̀ɔ

cám càaam-ɔ̀ wâc ʊ̀-wàaac-ɔ̀ɔ

lɛ́ɛŋ lɛ́ɛɛŋ-ɔ̀ pɛ̂ɛt ʊ̀-pɛ̀ɛɛt-ɔ̀ɔ

Across the morphological paradigms of different word classes, stem forms followed by the suffixes -ɔ and -ɔɔ go with an overlong vowel in the stem, and stem forms followed by the suffix -ɪ never do (i.e., they have either a short vowel or a long one). Both generalisations are specific to content words that display a morphological alternation in vowel length, i.e., they do not apply if the paradigm is Fixed Short or Fixed Overlong.19

3.3 Floating quantity

The notion of floating quantity refers to a weight unit or mora (μ) that is not associated with any segmental content, and that associates across a word boundary, giving rise to increased duration of a vocalic prefix in the following 19 Fixed Overlong is an exceptional pattern of vowel length alternation in inflectional noun paradigms, found in a handful of nouns. It is described in Section 3.6.5. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

28

word (Remijsen 2018). This weight unit is an instance of morphological marking, that is, it is part of morphological exponence. In the inflectional paradigm of suffixless singular nouns, it is found in two inflections: the demonstrative (Section 3.3.1), and the pertensive with singular possessor (3.3.2). Finally, we show that floating quantity is unattested in the pertensive with plural possessor and in the construct state (Section 3.3.3). Floating quantity is not found in suffixed singulars.

3.3.1 Floating quantity in the demonstrative inflection

Illustration (20) displays the base form (20a,b) and the demonstrative form (20c,d) of the noun kùl ‘warthog’. In utterance-final context (20a,c), the two forms are phonetically indistinguishable in terms of segmental composition and quantity. The reader can assess this by playing the embedded sound examples. However, in the context of a following prefix á-, a past-tense marker, there is a difference in quantity: the prefix á- has a greater duration when it follows the demonstrative inflection (20d) than when it follows the base form (20b). This can be ascertained by listening to the embedded sound clips. In the embedded sound examples, the duration of the /á-/ prefix is 101 milliseconds (ms) in (20b), and 136 milliseconds in (20d). (20)

a.

[dàa kùl] dɪ ̂=a kùl exsp=f warthog ‘There is a warthog.’

b.

[kùl álɪ ̂ɪɪd̪à] kùl á-lɪ ̂ɪɪt ̪-à warthog pst-look.at-1s ‘I looked at the warthog.’

c.

[dàa kùl] dɪ ̂=a kùlμ exsp=f warthog:cs:dem ‘There is this warthog.’

d.

[kùl ǎalɪ ̂ɪɪd̪à] kùlμ á-lɪ ̂ɪɪt ̪-à warthog:cs:dem pst-look.at-1s ‘I looked at this warthog.’

We refer to the quantity that is realized on the following vocalic prefix as floating quantity, along the lines of the concept of floating tone (Welmers 1970, McKendry 2013). Just as a floating tone in a language like Mixtec, the realization of floating quantity in Shilluk is contingent on a docking site to the right, beyond the word domain, and it is not realized otherwise.20 We represent it using the superscript μ (mora), as in kùlμ ‘warthog:cs:dem’ in 20 Floating morae have been postulated before: see e.g. Hyman & Byarushengo (1984) on Haya, and van den Heuvel (2006) on Biak. A salient difference between the Shilluk phenomenon on the one hand and the phenomena reported in earlier studies is that only in Shilluk the floating weight unit is dependent for its realisation on a docking site across a word boundary. In contrast, in Haya and in Biak floating quantity associates within the same word. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

29

(20c,d). Measurements of the duration of the á- prefix following a noun that has floating quantity are around 40 percent longer, on average, than following a noun that does not have floating quantity. Aside from vowel duration, there is also a difference in melody between (20b) and (20d). The fundamental frequency (F0) starts rising from the end of the stem vowel in the base form kùl (20b), but only from the beginning of the prefix á- in the case of the demonstrative inflection kùlμ (20d). It appears that the floating quantity invariably carries through the specification for tone from the preceding stem syllable, for example, a Low tone after stem syllable that carries a Low or a Low Fall. For this reason, we do not postulate a tonal specification associated with the floating quantity.21 The demonstrative inflection is marked on nouns that are at the right edge of the noun phrase. In clause-initial position, the following constituent is therefore typically a verb. In addition to the Past tense marker á-, the other vocalic tense-aspect-modality prefixes, that is, ʊ̀- for Imperfective and ʊ́- for Future and Non-Evidential Past, also accommodate floating quantity. This is shown in (21), again contrasting base forms of nouns with corresponding demonstrative forms.22 The durations of the following prefix /ʊ-/ in (21a,c,e) are 93, 84 and 96 milliseconds, respectively; the corresponding durations of the same prefix in (21b,d,f), where floating quantity is involved, display durations of 147, 156, and 118 milliseconds. (21) a.

[kùl ʊ́lɪ ̂ɪɪd̪à] kùl ʊ́-lɪ ̂ɪɪt ̪-à warthog fut-look.at-1s ‘I will look at the warthog.’

b.

[kùl ʊ̌ʊlɪ ̂ɪɪd̪à] kùlμ ʊ́-lɪ ̂ɪɪt ̪-à warthog:cs:dem fut-look.at-1s ‘I will look at this warthog.’

c.

[kùl ʊ́lɪ ̂ɪɪd̪ɔ] ̀ kùl ʊ́-lɪ ̂ɪɪt ̪-ɔ̀ warthog nevp:ov-look.at ‘Smb. apparently looked at the warthog.’

d.

[kùl ʊ̌ʊlɪ ̂ɪɪd̪ɔ] ̀ kùlμ ʊ́-lɪ ̂ɪɪt ̪-ɔ̀ warthog nevp:ov-look.at ‘Smb. apparently looked at this warthog.’

e.

[kùl ʊ̀lɪ ̂ɪɪd̪ɔ] ̀ kùl ʊ̀-lɪ ̂ɪɪt ̪-ɔ̀ warthog impf-look.at ‘Smb. is looking at the warthog.’

f.

[kùl ʊ̀ʊlɪ ̂ɪɪd̪ɔ] ̀ kùlμ ʊ̀-lɪ ̂ɪɪt ̪-ɔ̀ warthog impf-look.at ‘Smb. is looking at this warthog.’

21 In this context, it is worthwhile to note that the tone-bearing unit in Shilluk is the syllable rather than the mora (cf. Remijsen & Ayoker 2014:443). 22 For the Future tense forms (21a,b), we illustrate using forms that display with 1st singular person marking, just as we did in (20) in relation to the Past tense. The Non-Evidential Past and Imperfective forms cannot display subject marking. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

30

In contrast, if the verb is in the No Tense form, which does not involve a TAM prefix, then the floating quantity is not realised. This is illustrated by the sound examples in (22a,b) in relation to a consonant-initial stem. (22)

a.

́̂ kùl mâar a nɪɪɪn̄ warthog love:nt f sleep:inf ‘The warthog loves to sleep.’

b.

́̂ kùlμ mâar a nɪɪɪn̄ warthog:cs:dem love:nt f sleep:inf ‘This warthog loves to sleep.’

Floating quantity equally cannot dock across a word boundary onto a vowelinitial stem. This is illustrated in (23b), where the demonstrative inflection is followed by the No Tense form of an onsetless verb stem.23 Here as well, the floating quantity of the demonstrative inflection is not realized: there is no audible difference between (23a) and (23b) in the duration of the vowel of the verb stem as a function of floating quantity. As a result, the difference between the demonstrative inflection and the base form is not realized in any way in (23). It should be noted that verbs that lack an onset are extremely rare: {òr} ‘send’ is the only such verb we know of. (23)

a.

kùl ốr̄ a wʌ̂ɲ-ɔ̀ warthog send:nt f letter ‘The warthog sends a letter.’

b.

kùlμ ốr̄ a wʌ̂ɲ-ɔ̀ warthog:cs:dem send:nt f letter ‘This warthog sends a letter.’

Following the verb, a noun carrying a floating mora may be followed by an adverb that begins with a vocalic prefix (in the sense introduced in Section 1.3), such as áwàa ‘yesterday’, átīnɪ ́ ‘earlier on same day’, and ácɔ̄ɔɔn ‘this morning’. These equally display increased duration under the influence of the floating mora. In summary, the demonstratives of suffixless singulars are marked by a floating mora. This mora can dock only on the vocalic prefix of a following word, in which case this prefix displays increased duration.

3.3.2 Floating quantity in pertensive with singular possessor

Like the corresponding demonstrative inflection, the pertensive with singular possessor of suffixless singular nouns also displays floating quantity. When floating quantity marks pertensive with singular possessor, however, its ability to dock across a word boundary is more restricted. Nouns inflected for pertensive (sg) are invariably followed by a modifier noun phrase which expresses the possessor. Within this noun phrase, the head noun is in initial position. If this possessor noun has a vocalic prefix 23 Underlining marks argument focus. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

31

with vowel quality /a/ and specified for High tone, then the floating quantity of the pertensive (sg) can dock. This scenario is illustrated in (24a), where the name ábác follows the pertensive (sg) inflection of the suffixless noun gɔ̂ɔt ‘corner’. Illustration (24b) is included for the sake of comparison; here ábác is the internal argument of a verb in subject voice, and this verb form (á-lɪ ̂ɪɪt ̪) does not have floating quantity. Crucially, the initial vowel of ábác sounds saliently longer following the pertensive (sg) gɔ̂ɔɔtμ in (24a) than after the verb in (24b). This is evidenced by the embedded sound examples. The duration of the initial /á/ in ábác is 120 ms in (24a), where it is affected by floating quantity, and 87 ms in (24b), where it is not. (24)

a.

[dàa gɔ̀ɔɔt ǎabác] dɪ ̂=a gɔ̂ɔɔtμ á-bác exsp=f corner:prt fem-amniotic.sac ‘There is the corner of Abac.’

b.

[álɪ ̂ɪɪd̪ ábác] á-lɪ ̂ɪɪt ̪ á-bác pst-look.at:fug fem-amniotic.sac ‘S/he went to look at Abac.’

The intial vowel of ábác is a prefix, marking the derivation of a woman's name.24 However, the fact that the á- can indeed be analysed as a prefix does not matter for the docking of floating quantity. As explained in Section 1.3, we interpret all vowels preceding the closed monosyllabic stem as prefixes. In the case of the demonstrative form of singular suffixless nouns, any following vocalic prefix is a docking site for floating quantity. In relation to the pertensive (sg), floating quantity can only dock if a) the vowel phoneme of the vocalic prefix is /a/, and b) that prefix carries a High tone. The importance of these conditioning factors is illustrated in Table 12 and the associated sound examples. The first condition, i.e., that the vowel needs to be /a/, is illustrated in the top row by the pair á-tɔ̀r ‘Ator’ vs. ʊ́-tɔ̀r ‘Otor’. These are personal names, for a woman and a man respectively, both derived from tɔ̀r ‘wilderness’ through prefixation. Vowel quality is the only feature that distinguishes them. Playing the sound files, the reader can ascertain that this factor is critical. In the case of á-tɔ̀r, where the /á-/ constitutes a docking site, the duration of this prefix vowel is 120 ms following gɔ̂ĺ μ̄ ‘compound. prt’ and 108 ms following pấaalμ̄ . In the case of ʊ́-tɔ̀r, where the /ʊ́-/ does not represent a docking site, the duration of the /ʊ́-/ is 68 ms following gɔ̂ĺ μ̄ ‘compound.prt’ and 57 ms following pấaalμ̄ . 24 The name ábác ‘Abac’ is a woman’s name, referring to a person who is born with the amniotic sac (bác) around her. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

32

Table 12. Junctures between the pertensive (sg) forms gɔ̂́lμ̄ ‘compound:prt’ and pấaalμ̄ ‘spoon:prt’ and four possessor terms in which the floating quantity can vs. cannot dock onto a following vocalic prefix. Floating quantity can dock [gɔ̂ĺ ̄ gɔ̂ĺ μ̄

/ pấaal ̄ / pấaalμ̄

Floating quantity cannot dock ā̌atɔ̀r] á-tɔ̀r

house:prt / spoon:prt fem-wilderness ‘The house / spoon of Ator.’

[gɔ̂ĺ ̄ gɔ̂ĺ μ̄

/ pấaal ̄ / pấaalμ̄

ā̌abác] á-bác

house:prt / spoon:prt fem-amniotic.sac ‘The house / spoon of Abac.’

[gɔ́l gɔ̂ĺ μ̄

/ páaal / pấaalμ̄

ʊ́tɔ̀r] ʊ́-tɔ̀r

house:prt / spoon:prt mal-wilderness ‘The house / spoon of Otor.’

[gɔ̂ĺ ̄ gɔ̂ĺ μ̄

/ pấaal ̄ / pấaalμ̄

àcʌ̀k] à-cʌ̀k

house:prt / spoon:prt nom-compose ‘The house / spoon of the composer.’

The second condition is that the specification for tone of the following prefix has to be High. This is illustrated by à-cʌ̀k ‘composer (nom-compose)’. Even though the vowel is /a/, the prefix does not display a salient increase in duration: its duration in the embedded example is 96 ms following gɔ̂ĺ μ̄ and 86 following pấaalμ̄ . This indicates that the floating mora cannot dock.

3.3.3 No floating quantity in pertensive with plural possessor and construct state

As described in Section 2, the other two inflected forms that are specific to nouns are the pertensive with plural possessor and the construct state. These do not display floating quantity at all. The examples in Table 13 illustrate this in relation to the pertensive (pl). This table and the associated sound examples contrast the pertensive (sg) and pertensive (pl) forms of pâac ‘village’, which ̌ ar, respectively.25 Following the pertensive (sg) form pāaarμ, are pāaarμ and pāa which has floating quantity, the initial /a/ of álāal has a duration of 151 ms; in contrast, following the pertensive (pl), the initial /a/ of álʌ̂lɪ ̀ is only 108 ms long. This represents evidence that pertensive (sg) pāaarμ has floating quantity, ̌ ar does not. There is also a difference in tone: the whereas pertensive (pl) pāa floating quantity in the pertensive (sg) carries the specification for Mid tone from the preceding stem syllable, and this Mid tone replaces the specification for High tone on the initial syllable of álāal. That is, from the end of the vowel 25 The paradigm of the noun pâac ‘village’ is irregular. It is discussed in Section 3.6.6. SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

33

of pāaarμ to the middle of the vowel of the initial /a/ in álāal, the F0 remains virtually unchanged – the actual values are 150 Hz and 146Hz. In the case of ̌ ar álʌ̂lɪ̀, F0 increases by 20 Hz over the same domain (156 Hz to 176 Hz). pāa Table 13. The pertensive (sg) form pāaarμ ‘village:prt’ and the pertensive (pl) form pā̌aar ‘village:prt:p’, with different possessors. Pertensive (sg): floating quantity

Pertensive (pl): no floating quantity

pāaarμ

álāal ‘acacia’

pā̌aar

álʌ̂l-ɪ ̀ ‘acacia-p’

pāaarμ

ápwōooc-ɔ̄ ‘rabbit-s’

pā̌aar

ápwôoc-ɪ ̀ ‘rabbit-p’

Similarly, we have not found any evidence of floating quantity in the construct state inflecton. This is illustrated by the examples in (25), which contrast the use of the relativizer à, which requires the preceding head noun kùl ‘warthog’ to be in the construct state, with the use of focus marker à, which requires the preceding head noun to be in the base form. The paradigm of kùl ‘warthog’ is syncretic. Hence, evidence for the fact that these function morphemes indeed do condition the use of construct state and base form, respectively is presented in (25c,d): gwôk ‘dog’ does have different forms in these inflections. Crucially, the relativizer à in (25a,c) is not saliently longer than the phonologically identical focus marker in (25b,d). Specifically in relation to the sound examples embedded in (25), the durations of the relativizer in (25a,c) are 104 and 98 ms, respectively, and the durations of the segmentally identical focus marker in (25b,d) are 89 and 107 ms, respectively. (25) a.

c.

kùl à lʊ́ʊʊc̀ bǎa mjấā warthog:cs rel black:ctg nomp rel.s:prt:1s ‘The black warthog is mine.’

b.

́̂ -ɔ̄ kùl à mâar nɪɪɪn̄ warthog f love:nt sleep:inf ‘Specifically the warthog loves to sleep.’

gwốooŋ̄ à lʊ́ʊʊc̀ bǎa mjấā dog:cs rel black:ctg nomp rel.s:prt:1s ‘The black dog is mine.’

d.

́̂ -ɔ̄ gwôk à mâar nɪɪɪn̄ dog f love:nt sleep:inf ‘Specifically the dog loves to sleep.’

An alternative interpretation of the data in (25) is that the lack of evidence for floating quantity is due to the following environment, the Low toned relativizer à. Indeed, in relation to the pertensive (sg) inflection, the following vowel /a/ needs to be High-toned for floating quantity to dock. However, we know of no following context that presents this environment. This is because adjectival modifiers are invariably consonant-initial, and modifiers that go with the construct state tend to require the presence of the relativizer. For example, when a noun is modified by a verb through the use of the construct SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

34

state form, a modifier past-tense verb cannot follow immediately after the noun. This is shown in (26), where the relativizer cannot be omitted. (26)

dɪ ̂-a pùuuŋ *(à) á-mʌ̂ʌʌt ̪ [á:mʌ̂ʌʌt ̪] exsp=f turtle:cs rel pst-drink:atp:pst ‘There is the turtle that drank.’

In summary, we have no evidence for floating quantity in the construct state form, nor in the pertensive with plural possessor.

3.4 Tone

In this section we describe the specifications for tone of singular nouns, both in the base form (Section 3.4.1) and in the inflected forms (Section 3.4.23.4.5). In relation to the latter, each inflection comes with a specification for tone. They are listed in Table 14. In the case of the pertensive (pl), it is a High target, and in the case of the demonstrative, a Low target. In the case of the pertensive (sg) and the construct state, there are two possible specifications: Low and Mid; which of these a noun appears with depends on the specification for tone of the stem syllable. In Section 3.1.2, we described these specifications in relation to suffixed singulars, where they are associated with the suffix -ɪ. In the case of suffixless singulars, the specification for tone that is associated with an inflection appears on the stem syllable. However, it does not replace the lexical specification; rather, it is added to the right of the lexical specification of the stem, in a fusional manner (cf. Pike 1948). The interaction between the lexical and inflectional specifications is center stage in Sections 3.4.2-3.4.5. Table 14. Specifications for tone associated with different inflections of singular nouns. The difference between LF (L) and LF (H) is explained in Section 3.4.4. Inflection

Specification for tone

Pertensive (sg), and Construct state Pertensive (pl) Demonstrative

Low if stem syllable has Low, Low Fall (LF), or Low Rise Mid if stem has Mid, High, High Fall to Mid, High Rise, Low Fall (H) High Low

3.4.1 The tonal specification of singular base forms

Shilluk has a total of nine phonologically distinctive tone categories (Low, Mid, High, Low Fall, High Fall, Late Fall, High Fall to Mid, Low Rise, High Rise), but not all of these are found in singular base forms. Table 15 displays the patterns of vowel length and tone that are found on native non-derived SPECIAL PUBLICATION NO. 14 , 2019

Chapter 2: Inflectional morphology and number marking in Shilluk nouns

35

stems, crossing vowel length in the stem with tone on the stem syllable, for suffixless nouns with a stem vowel that is either short or long. Most base forms of suffixless singular nouns consist of a single closed syllable (Section 1.3), and the vowel is either short or long, but not overlong (Section 3.2).26 Aside from monosyllables, there are also many nouns that begin with an initial a- or ʊ-. As explained in Section 1.3, these initial syllables do not interact in any way with inflection, and for this reason they are not treated separately. Table 15. The lexical specifications for tone on the stem syllables of base forms of suffixless singular nouns. Tone on stem

V

VV

Low

ɟùr ‘young woman’

lʊ̀ʊt ̪ ‘stick’

Mid

kʌ̄c ‘hunger’

māac ‘fire’

High

àkúr ‘pigeon’

dúup ‘k.o. rodent’

Low Fall

dâk ‘cooking pot’

lʊ̂ʊl ‘steep river bank’

High Fall

àtwấk ‘peacock’

kɛ̂ɛ́ w ‘boundary’

Rise

kɔ̌cμ ‘hoe’

ʊ́gǐik ‘buffalo’

The High Fall is rare in native suffixless singular base forms: we know of only five native words that have this specification. It is found more often in loan words, such as mʊ̂́ʊt ̪ ‘banana’ ( á-lwɔ̄ɔk ‘pst-bathe:xvs’). And if an intransitive verb is underlyingly specified for Low Fall tone, then it gets a High Fall to Mid in the applicative with ́ r̄ ‘pst-fly:xvs’). The latter pathway is inversion (e.g. á-pʌ̂ʌr ‘pst-fly’ > á-pʌ̂ʌ also illustrated by the existential predicate marker dɪ ̂, which changes to dɪ ́̂ ̄ in the applicative with inversion. In contrast, the negative existential marker bʊ̄ŋ SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

32

carries a Mid tone to begin with, and therefore it cannot form the applicative

with inversion using either of these two regular pathways. That is, it has as its lexical specification a toneme that marks Applicative, and as a consequence

the tonal marking of Applicative applies vacuously, in the same way as English Jones may not get an additional sibilant to mark the genitive in Jones’ diary. When it comes to the interaction with licensing by the focus marker a,

however, bʊ̄ŋ does not mirror its affirmative antonym dɪ ̂, nor does it align

with verbs. Note that in (41-43), the No Tense form of bʊ̄ŋ is used without focus marking. This is further illustrated in (47). (47)^

́ ̄ ́ bʊ̄ŋ d̪ấaan̪ -ɔ̀ mɛ̂n à ʊ́-jʊ̂ʊt mɔ̂k nexp person-sg idp.sg mdf fut-find idp.pl:dxs ‘Nobody (there is not a person who) will find out about these things.’ [KeepTheSecret 237.5-239.3]

In fact, the addition of the focus marking in relation to the Theme argument of the No Tense form bʊ̄ŋ renders the clause ungrammatical. This is shown

in (48a), which is ungrammatical with a. This is the opposite of dɪ ̂, whose No Tense form requires syntactic licensing, and is typically found in a juncture

with the focus marker a, yielding dâa. This is shown in (48b); this sentence is ungrammatical without the focus marker a cliticised onto dɪ ̂. (48)

a. bʊ̄ŋ (*a) càak nexp foc milk.pl ‘There is no milk.’

b. dâa / *dɪ ̂ càak exsp=foc / exsp milk.pl ‘There is milk.’

As for other specifications of TAM, bʊ̄ŋ does not require syntactic licensing

here either, neither in Non-Evidential Past (49a), Past (49b), or Future (49c).

For the sake of comparison, dî requires syntactic licensing both in Past and in Non-Evidential Past, but not in the Future. (49) a.

bʊ̂́ŋ càak nexp:nevp milk ‘Apparently there was no milk.’

b. á-bʊ̄ŋ càak pst-nexp milk ‘There was no milk.’

c.

ʊ́-bʊ̄ŋ càak fut-nexp milk ‘There will be no milk.’

The fact that the negation marker bʊ̄ŋ appears to be incompatible with the

focus marker suggests that the meaning of the latter includes affirmation as part of its meaning.

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 33

4. Locative predications formed with nû́t

The morpheme nû́t is a locative predicate marker, whose meaning can be translated as ‘to be present’. In some contexts, it may appear to be synonymous to the existential predicate marker dɪ ̂. The example in (50) is a case in point; nû́t appears in the final clause, and we translate it as ‘Okwer’s sister is present’, rather than ‘Okwer has a sister’, even though the latter would be felicitous in this context. (50)^ ɟàal-ɔ̀ mɛ̀ɛɛk-ɔ́ á-ŋùnɪ ̀ pɪ ́ɲ kɪ ̂nɪ ̀ man-sg some-sg pst-murmur ground quot ŋàaan dɪ ̂ ɲɪ ̄-mɛ̄n ɲɪ ́ ɟàaat-ɔ̀  à dɪ ̄ person:cs exsp sister:3sg hab have.nothing-nevp foc how A man murmured: “How can someone who has a sister (be said to) have nothing?” ́ -ɪ ̄=ánɪ ́ oow, kwɔ̂p á-líŋ-ɪ ̀ ɪ ̄ɪ ɟɪ ̀ɪ kɪ ̂nɪ ̀ ɛ̀ɛɛ, áa ŋɔ̄ excl word.pl-cs=def pst-hear:ov prp.pl people quot yes whq what ʊ́kwɛ́r ɲɪ ̄-mɛ̄n nû́t Okwer sister:3sg locp Oow! People heard those words and said “Yes, is it not so? Okwer’s sister is present.” [TheDescendantsOfWaang 373.6-382.4]

The fine-grained difference in meaning between dɪ ̂ and nû́t is illustrated in (51). Note how dɪ ̂ expresses availability, whereas nû́t expresses presence. (51)

dɪɪ́̂ càak kàa  nùt d̪jàaŋ exsp:nevp milk when locp:xv:nevp cow ‘Apparently, there was milk when there was a cow.’

Syntactically, dɪ ̂ and nû́t differ clearly: dɪ ̂ is invariably followed by a core argument expressing the Theme; in contrast, nû́t is preceded by its argument, at least in the unmarked syntactic alignment. This is shown in (52a) vs. (52b), respectively. (52)

a. dâa ɲāaar-ɛ̄ exsp=foc daughter-3sg ‘S/He has a daughter.’

b.

ɲāaar-ɛ̄ nû́t daughter-3sg locp ‘His/her daughter is present.’

The syntactic order in a clause headed by nû́t, i.e., with its core argument preceding it in the unmarked syntactic alignment, is identical to that of a clause headed by an intransitive verb. This raises the question as to whether nû́t should not be analysed as a verb rather than a nonverbal predicate marker. One argument for the latter interpretation is tone: intransitive verbs SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 34

never have the High Fall. The inflectional paradigm lends further support to this interpretation – it is provided in Table 5. As seen from this Table, nû́t presents morphological marking for voice and TAM.16 Table 5. The paradigm of nû́t No Tense

Non-Evidential Past

Base

nû́t

nû́t-ɔ̀

Applicative

nūt

nùt

The paradigm of nû́t diverges from that of verbs in various ways. First, verbs additionally inflect for subject; second, the marking of TAM on nû́t is limited to No Tense and Non-Evidential Past. Verbs, in contrast, also present inflectional marking for Past and Future. In relation to nû́t, the latter meanings are expressed periphrastically, as illustrated for the Future in (53a). However there is a Non-Evidential Past, nû́t-ɔ̀, shown in (53b). As seen from all of the examples in this section, neither the No Tense form nor the NonEvidential Past require syntactic licensing. (53)

a. ɟấaak̄-ɔ̄ ʊ́-bèeet ɛ́ nû́t chief fut-stay pr.3sg locp ‘The chief will be present.’

b. ɟấaak̄-ɔ̄ nû́t-ɔ̀ chief-sg locp-nevp ‘Apparently the chief was present.’

The use of the inflections for Applicative voice is illustrated in (54), both in the No Tense form (54a) and in the Non-Evidential Past (54b). The latter is also illustrated in (51).17 (54)

a. ɛ́n à nūt ábác pr.3sg foc locp:xv Abac ‘This is why Abac is present.’

b. ɛ́n à nùt ábác pr.3sg foc locp:xv:nevp Abac ‘This is why Abac was apparently present.’

Often, nû́t is followed by a prepositional phrase that specifies the location. This is shown in (55); in each of these examples, the location is expressed in a prepositional phrase, and it is not possible to raise the location to a core 16 The Non-Evidential Past has two functions: one is evidentiality; the other is perfective (cf. Chapter 1, Section 5.4). 17 The presence of the focus marker in these examples is due to the expression of a Reason in the topic slot. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

35

argument slot. This is a notable difference with dɪ ̂, which allows for the expression of the location in the topic slot (cf. Section 2.1.2).

(55) a.^ ɛ́ɛ  d̪ấaan̪ -ɔ̀ nû́t kɪ ̀ ɪ ̄ɪc-ɪ ̄ɪ sub person-sg locp prp belly:prt-2sg ‘While you are pregnant (lit. while there is a person in your belly, […]’ [TraditionsOfWomen 36.2-37]

b.^ kwǎaa-wāaaŋ nû́t kɪ ̀ pā-d̪jàaŋ Kwawaang locp prp Padhyang ‘Kwawaang (lit. descendants of grandmother) are present in Padhyang.’ [TheDescendantsOfWaang 720.1-722.1] c.

càak nû́t kɪ ̀ rɪ ́ɪ àbūl-ɔ́ milk locp prp refl cake-sg ‘There is milk (present) in the cake.’

5. Locative predications formed with jɪ ̂ŋ

The morpheme jɪ ̂ŋ heads locative predications. Different from nû́t, it has a

valency of two, with the argument to its left expressing the subject and the

argument to its right its Location.18 Both are core arguments, i.e., there is no

preposition, and there are no other function morphemes, bound or otherwise, marking the fact that the constituent that follows jɪ ̂ŋ is a Location. This is illustrated in (56).

́ ̪̄ -ɪ ̄=ánɪ ́ (56) a.^ kàa á-kʌ́ʌʌl ̀ jɛ̂n já jɪ ̂ŋ pâac when pst-carry:pet:xv plant.pl-cs=def pr.1sg locp village ‘When the medicines were brought, I was in the village.’ [DownWithIllness 360.4-362.3] ́ ̄ b.^ ʊ́twɔ̀ɔɔcμ cɔ̂ĺ -ɔ̀ mɛ̂n à jɪ ̂ŋ nâam kʌ́ʌʌl ̀ k.o.plant:prt Shilluk-sg idp.sg mdf locp river bring:pet:ov kấā ɲɪ ́ rʌ̂́p conj hab burn:ov ‘Shilluk water hyacinth19 that is in the river was brought and it was burnt.’ [DownWithIllness 156.4-159.3] 18 The constituent order is different when jɪ ̂ŋ is morphologically marked for Applicative voice, an operation which involves a shift in the position of the subject. 19 The nominal possessor ‘Shilluk’ serves to refer specifically to an indigenous waterplant, which displays similarities with the invasive water hyacinth. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 36 c.^ d̪òk jâa   cấaɲɔ̀ cow.pl locp=foc Chanyo ‘The cows are in Chanyo.’ [TheDescendantsOfWaang 424.3-425.2] d.^ máa d̪òk-ɪ ́  gɛ́n jâa   jɛ̂t ̪-ɪ ́ ɟjɛ̂p-ɪ ́ gɛ́n sub cow:p-prt pr.3pl locp=foc inside.pl-prt pocket:pl-prt pr3.pl ‘[…], because their cows are in their pockets (i.e., they hold their assets in money, rather than acquiring cattle).’ [AchangVillagePart1 572.4-574.8]

When jɪ ̂ŋ is followed by the focus marker a/à, the two morphemes can coalesce into a single syllable, yielding jâa – in spite of the fact that jɪ ̂ŋ ends in a consonant. This is illustrated in (56c,d). This coalesced stem form appears in free variation with the juncture jɪ ̂ŋ a, which can be inserted in its place in (56c,d) without change in meaning. While the complement of jɪ ̂ŋ generally expresses a Location semantic role, in some cases this may extend semantically. This is shown in (57), where the complement of jɪ ̂ŋ is the noun phrase kǎaam gɛ́n ‘between them’, which in this case has a non-spatial meaning, through polysemic extension of kàam ‘distance’. (57)^ máa mếer̄-ɪ ̄ à á-jɪ ̂ŋ kǎaam gɛ́n ŋʌ́ʌɲɪ ̀ ɪ ̄ɪ  ɟɪ ̀ɪ běeen sub peace-cs mdf pst-locp distance:prt:pl pr.3pl know:ov prp.pl people  quant ‘Because everybody knew the harmony that was between them.’ [KeepTheSecret 138.6-141.5]

Table 6 displays the forms of jɪ ̂ŋ in different specifications for TAM and voice. Past and Future are marked through prefixes; both Non-Evidential Past and Applicative voice are marked through tone. The Non-Evidential Past form is syncretic, i.e., the same form is used across the two voices. Different from transitive verbs, there is no inflection for Object voice (compare with Chapter 1, Section 3.2.1). Table 6. The paradigm of jɪ ̂ŋ by TAM and voice. No Tense / Past / Future

Non-Ev. Past

Base

jɪ ̂ŋ / á-jɪ ̂ŋ / ʊ́-jɪ ̂ŋ

jɪŋ̂́

Applicative

jɪŋ̄̂́ / á-jɪŋ̄̂́ / ʊ́-jɪŋ̄̂́

jɪŋ̂́

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

37

The nonverbal predicate marker jɪ ̂ŋ requires syntactic licensing in No Tense, Past, and Non-Evidential Past. It is only in the Future tense that syntactic licensing is not required. This is shown in (58a,b,c,d) for the No Tense, Past, Future, and Non-Evidential Past, respectively. This pattern of syntactic licensing is identical to the one for dɪ ̂, which requires to be syntactically licensed in the same levels of TAM. (58)

a. d̪òk jɪ ̂ŋ *(a) cấaɲɔ̀ cow.pl locp foc Chanyo ‘The cows are in Chanyo.’

b. d̪òk á-jɪ ̂ŋ *(a) cấaɲɔ̀ cow.pl pst-locp foc Chanyo ‘The cows were in Chanyo.’

c.

d. d̪òk jɪŋ́̂ *(a) cấaɲɔ̀ cow.pl locp:nevp foc Chanyo ‘The cows apparently were in Chanyo.’

d̪òk  ʊ́-jɪ ̂ŋ  cấaɲɔ̀ cow.pl fut-locp Chanyo ‘The cows will be in Chanyo.’

The Applicative is used in a range of constructions in which an argument that would otherwise be expressed peripherally appears as a core argument in the preverbal topic slot. The subject is thereby displaced from the topic slot. Examples of the applicative are presented in (59), which is to be compared with the use of the base form in (56c) above. In (59a), the promoted topic argument expresses a Reason; in (59b), it is an expression of Time, with timeadverbial marker in the topic slot. Crucially, in both cases the presence of a promoted topic is signposted on jɪ ̂ŋ through morphological tone. (59)

a. dâa bôook-ɔ̀, ɛ́n à jɪŋ̄̂́ d̪òk cấaɲɔ̀ exsp=foc fear-sg pr.3sg foc locp:xv cow.pl Chanyo ‘There is fear, that is why the cows are in Chanyo.’ b. dâa   bôook-ɔ̀, kàa  jɪŋ̄̂́ d̪òk  cấaɲɔ̀ exsp=foc fear-sg when locp:xv cow.pl Chanyo ‘There is fear, when the cows are in Chanyo.’

As noted above, there is no separate morphological form for the Applicative voice Non-Evidential Past. Here the same form is found as in the NonEvidential Past form used with unmarked voice. This is shown in (60); the verb {kwìc} ‘not know’ in the main clause is lexically non-evidential, and hence requires the use of the Non-Evidential Past in the subordinate clause. The applicative nature of the construction is purely syntactic here, without morphological signposting on the head of the predicate. (60)

gìn  à pɛ̀r-ɛ̀ jɪŋ ̂́ d̪òk cấaɲɔ̀, kwíc-à thing:cs mdf reason-3sg locp:nevp cow.pl Chanyo not.know:nt-1sg ‘I don’t know the reason why the cows were in Chanyo.’ SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

38

The locative predicate marker jɪ ̂ŋ also inflects for subject. This inflectional marking is found when the topic slot holds a promoted argument, and the subject is displaced to the position following the verb. This means that the construction can be interpreted as a subject-marked Applicative. As in verbs, inflectional subject marking is limited to singular subjects, and it is not found in the Non-Evidential Past. The inflectional subject marking on jɪ ̂ŋ is illustrated for 1st, 2nd and 3rd person singular in (61a).20 If the subject is plural, we find instead the applicative form followed by the full form of the independent pronoun, case-marked for nominative.21 This is shown for 1st plural inclusive, 1st plural exclusive, 2nd plural and 3rd plural in (61b). These patterns mirror the situation in transitive verbs, where the subject-marked Applicative are formed in the same manner (Chapter 1, Section 4.2). (61)

a. ɛ́n à  jɪ ́ŋ-áa / jɪ ́ŋ / jɪ ́ŋ-ɛ́ pâac pr.3sg foc locp:xv-1sg / locp:xv:2sg / locp:xv-3sg village ‘Because of this I / you / they (sg.) are in the village.’ b. ɛ́n à jɪŋ̄́̂ wāa / wɔ̂n / pr.3sg foc locp:xv pr.1pl.inc:nom / pr.1pl.exc:nom / wûn /  gɛ̂n pâac pr.2pl:nom / pr.3pl:nom village ‘Because of this we (incl.) / we excl. / you (pl.) / they are in the village.’

In the Non-Evidential Past, there is no inflectional marking of subjects at all. This is in parallel with the situation in the verb system (cf. Chapter 1, page 66). This is illustrated in (62), which presents a 3rd singular subject associated with jɪ ̂ŋ in Past tense, where subject marking of singular subjects is inflectional (62a), and in Non-Evidential Past, where subject marking of singular subjects is syntactic (62b). Note that in the latter, the subject is expressed through the full form of the 3rd singular pronoun (ɛ́n), case-marked for nominative through a Low Fall. (62)

a. ɛ́n à á-jɪ ́ŋ-ɛ́ pâac pr.3sg foc pst-locp:xv-3sg village ‘Because of this he was in the village.’

20 The high tone on the stem follows compositionally from the combination of the applicative stem jɪŋ̄̂́ , which has a High Fall to Mid with a a High-toned suffix. That is, High Fall to Mid changes to High when followed by High within the same prosodic word in Shilluk in general. 21 The forms of the independent pronouns with nominative case marking are presented on page 35 of Chapter 1, in the second and fourth columns of Table 9. The forms without case marking are segmentally the same, but they have a High tone. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

39

b. ɛ́n à jɪŋ́̂  ɛ̂n pâac pr.3sg foc locp:nevp pr.3sg:nom village ‘Because of this he was apparently in the village.’

6. Headless locative predication

In addition to the locative predicate constructions with nû́t (Section 4) and jɪ ̂ŋ (Section 5), locative predications may be headless, i.e., without a dedicated function morpheme that heads the predicate. This structure involves the focus marker positioned before the noun phrase that expresses the Location.22 The phenomenon is illustrated in (63). (63)

gwôk à kâl dog foc compound ‘The dog is in the compound.’

Illustration (64) presents three examples of headless locative predications drawn from narratives. (64)

a.^ ʊ́gwɔ̂ɔ́ ɔk̄-ɔ̄ à wīc jāaan̪ àcɪ ́nɛ̂ɛ́ , àcàa jackal-sg foc top tree:cs dxs dxs ‘Jackal is up in the top of that tree, there.’ [Nyagul 71.7-73] b.^ áa ŋɔ̄  à kàaamμ wún kɪ ́  mʌ̂́ʌʌt-̪̄ ɪ ̄ɪ whq what foc distance:prt pr.2pl prp friend:prt-2sg ‘What is between you and your friend?’ [KeepTheSecret 209.2-211.3] c.^ ŋɛ̀ɛɲ-ɪ ́ d̪ấaan̪ -ɔ̀ à pôot ̪-ɪ ̀ bwōɲ-ɔ̄ jáaak gɔ́l multitude:prt person-sg foc land:prt Arabs-sg men:prt compound:prox:voc ‘A mass of people are in the land of the Arabs, men of this family.’ [AchangVillagePart3: 785.6-787.6]

When the focus marker is used in this manner, it signposts informationstructural prominence on the preceding topic. In addition, its use here can be understood in terms of the general requirement for clauses in Shilluk to be licensed for TAM, be it lexically, morphologically, or syntactically. The focus marker fulfills this function, and its use in headless locative predications can be attributed to this requirement. 22 Another language in which the focus marker may be used as a copula in non-verbal predicates in is Guinean Maninka (Vydrin 2020:88-90) SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

40

It is worthwhile to note that a headless locative predication are similar in form to a noun phrase with a nominal modifier that expresses a location, in the sense that the focus marker is homophonous with the modification marker. The two constructions are contrasted in the examples in (65). In (65a), repeated from (53), à is the focus marker, and gwôk à kâl is a well-formed clause. In contrast, the à in (65b) is the modification marker, and gwốooŋ̄ à kâl is a noun phrase with a locative modifier (cf. Chapter 5, Section 4.3). There is no ambiguity between these constructions, because the modification marker combines with the construct state form of the head noun. Hence, gwôk ‘dog’ appears in its base form when it is followed by the focus marker à (65a), and in its construct state form gwốooŋ̄ in (65b), where it is followed by the modification marker. (65)

a. gwôk à kâl dog foc compound ‘The dog is in the compound.’

b. gwốooŋ̄ à kâl bǎa mjấaā dog:cs mdf compound nomp mdf.sg:1sg ‘The dog that is in the compound is mine.’

Moreover, the fact that (65b) is grammatical fits with the interpretation that the focus marker in locative predications such as the ones in (63,64) serves to satisfy the requirement for TAM licensing. In a subordinate clause, this licensing is fulfilled by the matrix clause. This explains why, in (65b), kâl ‘compound’ may be a grammatical locative predicate by itself.

7. Nominal predications formed using bǎa

The morpheme bǎa is a copula that predicates a relation between the referents of two noun phrase constituents, positioned on either side of bǎa. The first argument is the topic. Pronominal 3rd singular topics are obligatorily dropped, unless they are marked for focus (cf. Chapter 1, page 18).23 Nominal predications marked using bǎa can have one of two functions. One is proper inclusion. In this use, the referent of the topic belongs to a set referenced by the noun phrase constituent that follows bǎa. Illustration (66) shows this function in a narrative; bǎa is used to describe phenomena: a man called 23 The limitation to topic drop, i.e., that a 3rd singular pronoun can appear in the topic slot if it is marked for focus, was not recognized in Chapter 1. This is illustrated in ɛ́n à á-mấat ̪ pr.3sg foc pst-drink:ov ‘That one (e.g. a particular bottle) was drunk’, where the 3rd singular pronoun is grammatical because it is followed by the focus marker. Without the focus marker, topic drop is obligatory: á-mấat ̪ past-drink:ov ‘It was drunk’. We are grateful to an anonymous reviewer of the current chapter for pointing this out to us. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 41

Deng, and the village Acaro. In both cases, the relation expressed is one of proper inclusion: Deng belongs to the set of hunters / fishermen, and Acaro belongs to the set of villages that are big and good. Characteristically for proper inclusion as a semantic relation, the noun phrase arguments that follow bǎa are not referential, i.e., it does not refer to a single real-world entity, but rather denotes a set. (66)^ ácàaarɔ̀ pāa jấā dɛ̄ɛ̌ ɛŋ Acaro  village:prt idp Deng:apl bǎa  pâac  mɛ́ dwɔ̂ɔ́ ŋ ʊ̀ bǎa pâac  mɛ́  dɔ́ɔɔc̀ nomp village mdf.sg big conj nomp village mdf.sg good:ctg ‘Acaro, the village of the people of Deng, is a big village and a good village.’ ́ ̄ =áa ʊ̀ mɛ̂n  dɛ̄ɛŋ kɪ ́ rɛ̄ɛ, conj idp.sg=nomp Deng prp rfl:3sg bǎa ɟàal jáp  kɪ ́  gìi kʌ̀ʌʌj-ɔ̀ nomp man search:infa prp something:cs protein.rich.food-3sg ‘And as for Deng himself, he is a man who knows how to look for fish or meat.’ [DengsFish 4.3-13.1]

The majority of cases in which we have observed the use of bǎa represent instances of proper inclusion. In addition, bǎa can also express equation (identification), that is, equating specific real-world referents. In this use, the noun phrase following bǎa references a single real-world entity, and the topic (before bǎa) is equated to it. This is shown in (67a,b). (67)

́ ̄ a.^ jí  bǎa mɛ̂n lʊ̂́ʊɲ ɪ ̄ɪ gɛ́n kɪ ́ nʌ̀ʌʌk-ɔ̀ pr.2sg nomp idp.sg take.turns:ov prp.pl pr.3pl prp kill-inf ‘You are the one they (will) kill in turn.’ [KeepTheSecret 75.2-77.2] b.^ kɪ ́ djɛ̄ɛr jí bǎa cjếeek̄-āa prp truth pr.2sg nomp wife:prt-1sg ‘Truly, you are my wife.’ [KeepTheSecret 223.4-225]

A particular type of identificational function is the ‘specificational’ use (Higgins 1979). In this usage, the noun phrase referring to the single real̄ references a world entity is topical. It is illustrated in (68a), where ɟấaak-ɔ̄ single real-world entity. To compare, the order of the arguments is reversed in (68b), and here the meaning is one of proper inclusion. (68)

a. ɟấaak̄-ɔ̄ bǎa twɔ́ɔŋ chief-sg nomp Twong ‘The chief is Twong.’

b. twɔ́ɔŋ  bǎa ɟấaak̄-ɔ̄ Twong nomp chief-sg ‘Twong is a chief.’ SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

42

Different from the other nonverbal predicate markers considered so far, bǎa does not inflect for TAM or voice at all. This is illustrated in (69), which clearly refers to an event in the past, even though this is not marked at all on the predicate marker. Similarly in relation to (67a) above, it is clear from the preceding part of the narrative that the speaker is discussing a situation in the future; but there is no specification for Future tense in the clause. (69)^

bǎa rûun mɪ ́ɪɪ pjʌ̀ʌʌrɔ̀ wɪ ̄c dâa á-biikjɛ̀ ̂́ l nomp year  100 ten head exsp:foc crd-sixth ‘It was the year 1690.’ [TheDescendantsOfWaang 542.5-546.5]

kɪ ́ pjʌ̄ʌr á-biiŋwɛ̀ ̂́ ɛn prp ten:pl crd-ninth

It is however possible to express the functions of TAM and voice in a construction involving the nominal predicate marker bǎa, with the support of the verb {bèeet} ‘stay, sit’. The lexical meaning of this verb is illustrated in (70). (70)^

ɛ́ɛ gɛ́ á-bèeet kɪ ̀  kɛ̂ɲ=ánɪ ́ interj pr.3pl pst:stay prp place:cs=def ‘Yes; they stayed in that place.’ [TheDescendantsOfWaang 184.5-186.1]

This intransitive verb is frequently used in a grammaticalized manner, in serializations with predicate heads such as bǎa, which cannot be inflected for TAM or voice themselves. In such serializations, {bèeet} contributes the specification for TAM and/or voice. This is illustrated in (71) and (72). In both of these examples, the topic is resumed pronominally ahead of the second predicate within the serialization, through the pronoun ɛ́; this resumption is characteristic of serializations. In (71), the serialization is á-bèeet ɛ́ bǎa […], whereby á-bèeet carries Past tense, and the nonverbal predicate construction with bǎa expresses proper inclusion. (71)^

kɪ ̀ jǐii gɛ́ bɔ̂ɔl-ɪ ̀ kɛ̂ɲ=ánɪ ́ ɲwɔ̂m-ɪ ́ wàaat ɪ ̄ɪ prp ad:pl pr.3pl front-prt time:cs=def marry:inf-prt relative.pl prp.pl ̄ ̄ kʌ̂́ʌl-ɪ à dɔ̀ɔŋ-ɔ̀ á-bèeet ɛ́ bǎa mìit ̪-ɪ ̀  rɛ̂m-ɪ ̀ kàl family:pl-cs mdf big-pl pst-stay pr.3sg nomp hold:inf-prt blood-prt family ‘For them at that time, the marrying of relatives by royal families was (a way) to preserve the blood of the family.’ [TheDescendantsOfWaang 398.6-402.9]

In (72), the serialization with {bèeet} serves to contribute voice. The time adverbial clause introduced by kàa is invariably positioned in the topic slot, displacing the subject to the position following the verb. This marked SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 43

argument structure is morphologically signposted on á-bēeet, which is in

Applicative voice.24 Illustration (72) below shows that it is the same with bǎa: here as well, time adverbial kàa conditions the use of the Applicative voice.

But because bǎa does not have an Applicative voice form itself, {bèeet} is used in a grammaticalized manner contributing Applicative voice, which is marked here through the Mid tone on the stem syllable of á-bēeet.25 (72)

kàa á-bēeet dɛ̄ɛŋ ɛ́ bǎa ɟấaak̄-ɔ̄, pâac á-bèeet áa ádūuul-ɔ̄ when pst:stay:xv Deng pr.3sg nomp chief-sg village pst-stay nomp circle-sg ‘When Deng was chief, the village was round (lit. a circle).’

There are two further uses of bǎa that warrant discussion. First, there is a

construction in which bǎa is followed by a patient nominalization derived

from a transitive verb root. The patient nominalisation is a productive and

regular morphological operation, yielding a noun referring to the semantic patient of the event referred to by the source transitive verb. For example, {màk} ‘catch’ yields à-mǎk ‘captive, prisoner’, and {ŋɔ̀l} ‘cut’ yields à-ŋɔ̌l

‘circumcised person’. Combined with such patient nominalisations, bǎa

conveys completive aspect. This is illustrated by the examples in (73). In

(73a), à-kàaan is the plural of à-kǎn ‘hidden one’, the patient nominalization of {kàn} ‘hide’; in (73b), à-jɛ̌p ‘opened one’ is the patient nominalization of {jɛ̀p} ‘open’.26 (73)

a.^ ʊ̀ lìiip-ɛ́ (k)ɪ ́ ɲwɔ̄ɔl-ɪ ́ pʌ̀ʌʌr conj sneak.up:atp-3s prp offspring.pl-prt hippo ́ ̄ à mɔ̂k bǎa  à-kàaan  ɪ ̄ɪ mēě ek gɛ́n idp.pl mdf nomp patn-hide:pl prp:pl mother.pl:prt pr.3pl ‘And he sneaked up on hippo calves that were hidden by their mothers.’ [TheDescendantsOfWaang 303.7-306.2]

24 In Chapter 1, the use of Applicative voice in time adverbial clauses is illustrated for verbs in (42b) and (67b). 25 Illustration (72) also includes an instance of another nominal predicate marker, áa. Its properties will be described in Section 10. 26 In (72c), there is a mismatch in grammatical number between the patient nominalisaation à-gwɛ̌ɛt, which is grammatically singular, and kwáan-ɪ ́ ŋjēeɲ, its head in the superordinate clause, which is grammatically plural, as seen from the modifier mɔ́. We would expect the plural form of the patient nominalization, which is à-gwɛ̀ɛɛt. We speculate that the mismatch is due to the semantic interpretation of money as singular (constructio ad sensum). SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 44 b.^ kɛ̂ɲ à  á-wʌ̄t ̪ɪ ̄ gɛ̂n kàl time:cs mdf pst-reach:xvs pr.3pl:nom compound à-lɪɪɪt ̂́ ̄ ̪ gɛ̂n ɔ̄ɔt ɛ́ bǎa à-jɛ̌p seqpst-see:xv pr.3pl:nom house pr.3sg nomp patn-open ‘When they reached the compound, they saw that the house was open.’ [KeepTheSecret 470.1-473.2]

This use of bǎa with patient nominalisations fits within its distribution as a copula marking nominal predicates, in that these patient nominalisations are grammatically nouns. Semantically, however, the combination of bǎa with these nominalisations refers to the outcome of an event. Hence, this use of bǎa is akin to the use of an auxiliary combined with a non-finite verb form to express perfective aspect. In fact, it may be that this construction is being reinterpreted as a auxiliary verb construction, with the patient nominalization in an infinitival role. In support of this interpretation, we note that the patient nominalization may fail to agree for grammatical number. This is shown in (74). In (74a), à-gwɛ̌ɛt ‘written one’ is the grammatically singular patient nominalization of {gwɛ̂ɛt} ‘write’; in (74b), à-kǔm is the same derivation of {kùm} ‘cover’. Both are singular forms; the corresponding plural are à-gwɛ̀ɛɛt and à-kùm, respectively. As seen from (74a,b) both appear in the singular, rather than agreeing for number with the plural noun in the topic slot. The data are not consistent on this issue: in (73a), the patient nominalization à-kàaan does appear in the plural, in agreement with the grammatically plural noun in the topic. (74)

a.^ ɛ́ ʊ̀-rìp-ɔ̀  kɪ ́  kwáan-ɪ ́  ŋjēeɲ pr.3sg impf-add prp number.pl-prt money.pl mɔ́ bǎa à-gwɛ̌ɛt kɪ ̀ ɪ ̄ɪcμ  wʌ̀ʌɲ-ɔ̀ mdf.pl nomp patn-write prp inside:prt paper-sg ‘He is adding up the sums of money, that are written on the paper.’ [AchangVillagePart3 418.7-421.9] b.^ ʊ̀ gɛ́ bǎa à-kǔm  kɪ ́ ʊ́twɔ̌ɔc conj pr.3pl nomp patn-cover prp k.o.water.plant ‘And they (=fish) were covered with water plants.’ [AchangVillagePart3 885.8-887.9]

The other use of bǎa that requires further discussion is of an informationstructural nature. It is illustrated in the extracts from narratives in (75). In each case, the part of the utterance that follows bǎa represents a well-formed main clause by itself. This suggests that bǎa functions as a clause-level focus marker here. If bǎa were functioning here as a nominal predicate marker, then it would head the main clause, followed by a subordinate clause. In (75a), for example, SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

45

́ ʌɲ̄ -ɔ̄ would function as the common argument this would mean that bʌ̂ʌ

with a following relative clause. If this were the case, the à that follows ́ ʌɲ̄ -ɔ̄ would be the modification marker. While this works well in terms bʌ̂ʌ

of the idiomatic translation, it is incorrect as a grammatical analysis, because ́ ʌɲ̄ -ɔ̄ is in the base form. If the à that follows it were the relativizer, we bʌ̂ʌ ́ ̄ -ɪ̄, the construct state form. We conclude, then, that would instead find bʌ̂ɲ

bǎa functions here as a modifier to the clause as a whole, lending focus to the topic, in conjunction with the focus marker that is associated with the same constituent. We use the term focus here in line with the definition in Krifka

(2007:18), who writes that “[f]ocus indicates the presence of alternatives that are relevant for the interpretation of linguistic expressions.” In (75a), locusts are presented as the most serious risk to crops, as compared to drought; in

(75b), the spear is presented as a cause of death, as opposed to a buffalo; in (75c), the Sun, is presented as the strongest, as opposed to the North Wind. (75)

a.^ bǎa  bʌ̂́ʌʌɲ̄ -ɔ̄ à tɪ ̂ɪ  rʌ̀ʌʌc-ɔ̀ kɪ ́ djēer jɔ̂w nomp locust-sg foc surpass:nt bad:nmlz prp truth adv ‘It is the locust that is really the worst (factor that can damage the crop).’ [TheVillageAchangPart1 63.1-65.3] gɛ́ ʊ̀-lîŋ-ɔ̀ kɪ ̂nɪ ̀ bǎa tɔ́ŋ  à á-nʌ̀k ján b.^ kɛ̀-lɔ́k cond-turn pr.3pl impf-hear quot nomp spear foc pst-kill pr.1sg ‘If they hear that it was a spear that killed me, […].’ [KeepTheSecret 72.0-74.6] c. ̂ bǎa jín à tɛ̂ɛk nomp 2sg foc strong ‘It is you who is strongest.’ [TheNorthWindAndTheSun 53.3-54.6]

This analysis is corroborated by the elicited examples in (76), which contrast

the use of bǎa as a nominal predicate marker for proper inclusion (76a) with its information-structural use as a focus marking construction (76b). The ́ n̄-ɪ ̄ à in (76a) cannot be indistinguished from phonetic realization of mʌ̂ʌ

that of mʌ́ʌn à in (76b), because the suffix that marks the construct state is

elided before the relativizer. Crucially however, the initial pronoun gɛ́ ‘pr.3pl’ cannot be omitted when bǎa is used as a nominal predicate marker in (76a). This follows from the fact that 3rd person plural pronominal topics cannot be dropped. In conclusion, bǎa functions in a fundamentally different way in

(76b).

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 46 (76)

a. gɛ́ bǎa mʌ̂́ʌn̄ -ɪ ̄ à ŋʌ̂ʌɲ mák  kɪ ́ rɪ ́c pr.3pl nomp women-cs mdf know catch:infa prp fish.pl ‘They are the women who are skillful at catching fish.’ à  ŋʌ̂ʌɲ mák kɪ ́ rɪ ́c b. bǎa  mʌ́ʌn nomp women-cs foc know catch:infa prp fish.pl ‘It is the women who are skillful at catching fish.’

8. Nominal predications formed using jɪ ́ŋ

A second nominal predicate marker is jɪ ́ŋ. Its use is illustrated by the examples from narratives in (77). Different from bǎa, jɪ ́ŋ is not used to express proper inclusion, but only to express the relation of equation (identification) between noun phrase referents. For example, in the contexts where bǎa is used to describe entities as they are introduced in discourse – as in (66): ‘Achang is a big village’; ‘Deng is a hunter’ – bǎa cannot be substituted with jɪ ́ŋ. This is because the complement that follows jɪ ́ŋ is invariably definite. Presumably for this reason, the lexical frequency of jɪ ́ŋ in texts is considerably lower than that of bǎa. (77)

a.^ ʊ́ɟwɔ́k  à jɪ ́ŋ  wʌ̂́ʌʌtμ̄  ɟúur Ojwok foc nomp:nt son:prt Jur ‘Ojwok is the son of Jur.’ [TheDescendantsOfWaang 571.7-573.8] b.^ gɛ́ cwɔ̂lɪ ̀ gɛ́=áa àkǒool běeen pr.3pl call-iter:ov pr.3pl=nomp leader.pl quant à jɪ ́ŋ ɲɪ ́ŋ-ɪ ́ d̪wʌ̌ŋ mɔ́ɔɔ gɛ́n mdf nomp:nt eyes.pl-prt status  idp.pl:prt pr.3pl ‘They were all called leaders, which are their titles (lit.: their names of status).’ [TheDescendantsOfWaang 226.9-230.3] c.^ cè jɪŋ́̂ a djēer interj nomp:nevp foc truth ‘So apparently it was the truth!’ [TheNorthWindAndTheSun 52.1-53.1]

The nominal predicate marker jɪ ́ŋ is segmentally identical to the locative predicate marker jɪ ̂ŋ (Section 5), and with certain specifications of TAM and voice they are homophonous, i.e., identical not just in segmental but also in suprasegmental terms. Because of this, we will juxtapose the two across specifications for TAM and voice. In No Tense and in Future tense, the two are distinguished by tone: the nominal predicate marker has a High tone (78a,c), SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

47

which distinguishes it from the locative marker, which has a Low Fall instead (78b,d). (78)

a. kùl à jɪ ́ŋ bɔ̀ɔɔt ̪ɔ̀ Kul foc nomp craftsman-s ‘Kul is the craftsman.’

b.

kùl à  jɪ ̂ŋ kàl Kul foc locp compound ‘Kul is in the compound.’

c.

d.

kùl ʊ́-jɪ ̂ŋ kàl Kul fut-locp compound ‘Kul will be in the compound.’

kùl ʊ́-jɪ ́ŋ bɔ̀ɔɔt ̪ɔ̀ Kul fut-nomp craftsman-s ‘Kul will be the craftsman.’

The rest of the paradigm of nominal predicate marker jɪ ́ŋ is identical to that

of the locative predicate marker jɪ ̂ŋ. In (79a,b) this is illustrated for the NonEvidential Past, where both have the High Fall, and in (79c,d) for the Past tense, which has the Low Fall. (79)

a. kùl à jɪŋ́̂ bɔ̀ɔɔt ̪ɔ̀ Kul foc nomp:nevp craftsman-s ‘Kul was apparently the craftsman.’

b.

kùl à jɪŋ́̂ kàl Kul foc locp:nevp compound ‘Kul was apparently in the compound.’

c.

d.

kùl à á-jɪ ̂ŋ kàl Kul foc pst-locp compound ‘Kul was in the compound.’

kùl à á-jɪ ̂ŋ  bɔ̀ɔɔt ̪ɔ̀ Kul foc pst-nomp craftsman-s ‘Kul was the craftsman.’

In the applicative as well, the two paradigms are indistinguishable. This is shown in (80). (80)

a. ɛ́n à jɪŋ̄́̂ kùl pr3s foc nomp:xv:nt Kul bɔ̀ɔɔt ̪ɔ̀ craftsman-s ‘This is why Kul is the craftsman.’

b. ɛ́n à  jɪŋ̄̂́ kùl kàl Kul foc ocp:xv:nt Kul compound ‘This is why Kul is in the compound.’

In addition to voice and TAM, jɪ ́ŋ inflects for subject marking. Subject marking

on the nominal predicate marker jɪ ́ŋ is identical to subject marking on the

locative predicate marker jɪ ̂ŋ both in terms of distribution (1ast, 2nd, and 3rd

singular only, and not in the Non-Evidential Past) and in terms of phonological form. The subject marked forms of the nominal predicate marker jɪ ́ŋ are

shown in (81a), and the corresponding plural forms are show in (81b). Both can be compared with those the locative predicate marker jɪ ̂ŋ in (61a) and (61b), respectively.

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk (81)

48

a. ɛ́n à jɪ ́ŋ-áa / jɪ ́ŋ /  jɪ ́ŋ-ɛ́ bɔ̀ɔɔt ̪ɔ̀ pr.3sg foc nomp:xv-1sg / nomp:xv:2sg / nomp:xv-3sg craftsman-sg ‘Because of this I / you / they (sg.) are a craftsman.’ b. ɛ́n à  jɪŋ̄́̂ wāa / wɔ̂n / pr.3sg foc nomp:xv pr.1pl.inc:nom / pr.1pl.exc:nom wûn / gɛ̂n bòot ̪-ɪ ̀ pr.2pl:nom / pr.3pl:nom craftsman-pl ‘Because of this we (incl.) / we excl. / you (pl.) / are craftspeople.’

Finally, also like the locative predicate marker, the nominal predicate marker can contract to /jaa/ when followed by a focus marker, e.g. jɪŋ̂́ + a yields jấa. Table 7 presents its complete paradigm, with the paradigm of locative jɪ ̂ŋ alongside for the sake of comparison. The fact that they have an Applicative voice parallels the paradigm of verbs: both transitive and intransitive verbs also have an Applicative voice. Table 7. The forms of the nominal predicate marker jɪ ́ŋ with different specifications for TAM and voice. The corresponding forms of the locative predicate marker jɪ ̂ŋ are presented alongside. Those forms where the two forms diverge are underlined. jɪ ́ŋ Nominal

jɪ ̂ŋ Locative

Past / No Tense / Fut

Non-Ev. Past Past / No Tense / Fut

Non-Ev. Past

Base

á-jɪ ̂ŋ / jɪ ́ŋ / ʊ́-jɪ ́ŋ

jɪŋ̂́

á-jɪ ̂ŋ / jɪ ̂ŋ / ʊ́-jɪ ̂ŋ

jɪŋ̂́

Applicative

jɪ̂́ŋ̄ / á-jɪ̂́ŋ̄ / ʊ́-jɪ̂́ŋ̄

jɪŋ̂́

jɪ̂́ŋ̄ / á-jɪ̂́ŋ̄ / ʊ́-jɪ̂́ŋ̄

jɪŋ̂́

At the same time, the voice systems of these nonverbal predicate markers are more restricted than those of verbs in that they do not present Object voice forms. That is, transitive verbs can shift the internal argument to the topic slot and mark this on the verb, through inflection for Object voice (cf. Chapter 1, Section 3.2.1). In contrast, the nonverbal predicate markers jɪ ́ŋ and jɪ ̂ŋ do not allow for this morphosyntactic operation, even though they have two noun phrase slots in their structure, just as transitive verbs do.27 The tonal marking of TAM on the stem syllable of jɪ ́ŋ – with the stem form /jɪ ̂ŋ/ in Past and /jɪ ́ŋ/ in No Tense and Future – has parallels with the verb system. Both in the Subject voice of transitive verbs and in intransitives, we 27 However, the lack of Object voice in relation to jɪ ́ŋ could also be attributed to its functional range, which is limited to equalition / identification. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

49

find alternations whereby No Tense and Future have a specification for tone involving higher F0 than the Past tense. And in various sets of intransitive verbs, the alternation is between Low Fall in Past vs. High in No Tense and Future, just as it is in the paradigm of jɪ ́ŋ. For example, ‘disappear’ has Past Tense á-lʌ̂ʌʌɲ and No Tense and Future lʌ́ʌʌɲ and ʊ́-lʌ́ʌʌɲ, respectively; and ‘run’ has Past Tense á-rɛ̂ŋɪ ̀ and No Tense and Future rɛ́ŋɪ ́ and ʊ́-rɛ́ŋɪ ́, respectively. The High Fall in the Non-Evidential Past of jɪ ́ŋ reminds of the same specification for tone in this inflection for dɪ ̂ (Section 2.2). In all of the above examples, jɪ ́ŋ is accompanied by the focus marker. This is not a coincidence: jɪ ́ŋ requires licensing by a. This sets it apart from bǎa, which can be used with or without focus marking (82a,b). As for jɪ ́ŋ, even when inflected for tense, as in (82c,d), the focus marker is required. Without focus marking, the clause is ungrammatical: *kùl á-jɪ ̂ŋ bɔ̀ɔɔt ̪-ɔ̀. In this sense, jɪ ́ŋ is notably different from predications headed by lexical verbs, where the presence of a TAM prefix satisifies syntactic licensing. (82)

a. kùl bǎa bɔ̀ɔɔt ̪-ɔ̀ Kul nomp craftsman-s ‘Kul is a craftsman.’ c.

kùl à á-jɪ ̂ŋ  bɔ̀ɔɔt ̪-ɔ̀ Kul foc pst-nomp craftsman-s ‘Kul was the craftsman.’

b. kùl à bǎa bɔ̀ɔɔt ̪ɔ̀ Kul foc nomp craftsman-s ‘Kul is a craftsman.’ d.

kùl á-jɪ ̂ŋ  à bɔ̀ɔɔt ̪-ɔ̀ Kul pst-nomp foc craftsman-s ‘Kul was the craftsman.’

In addition, if the focus marker follows the No Tense form jɪ ́ŋ, the clause is still incomplete (83a). It is only through the addition of a subordinate clause that the whole is grammatical (83b). Illustration (83a) is to be contrasted with (77a), where the focus marker preceeds jɪ ́ŋ, and the sentence is well-formed. (83)

a. *ʊ́ɟwɔ́k jɪ ́ŋ a  wʌ̂́ʌʌtμ̄ ɟúur Ojwok nomp:nt foc son:prt Jur ‘Ojwok is the son of Jur.’ a wʌ̂́ʌʌtμ̄ ɟúur, kɛ̀ ɲwɔ̂ɔ́ ɔm̄ mɛ̂j́ ̄-ɛ̄ ɪ ̀ɪ ɟúur b. ʊ́ɟwɔ́k jɪ ́ŋ Ojwok nomp:nt foc son:prt Jur sub marry:xv mother:prt-3sg prp Jur ‘Ojwok will be the son of Jur, if Jur marries his mother.’

It is clear the use of jɪ ́ŋ places additional requirements in terms of focus marking and syntactic licensing, beyond those familiar from transitive verbs (Chapter 1). We do not have an explanatory analysis for this.

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

50

9. Nominal predications formed using áa

The next nonverbal predicate marker taking a noun phrase complement is áa. It is found in serializations only: áa never constitutes the sole predicate head within the clause. Its use is illustrated by the narrative examples in (84). Note that, in each case, there is another predicate head earlier in the clause. The fact that these are serializations is also evident from the resumption as a pronoun of the topic in some of the examples. Hence, the 3rd plural topic is resumed before áa in (84b), and the 1st plural exclusive topic in (84c). It is only 3rd singular topics that are not resumed pronominally before áa; this may be for phonological reasons, i.e., as it is vocalic (ɛ́), just as áa itself.28 The generalization is that, if áa is not preceded by a independent pronoun topic, its topic is to be interpreted as 3rd singular, as in (84a) and (84d). (84)

́ ̄ à a.^ ʊ́kwʌ̂ʌ á-kwàɲ à  dɪ ̀wʌ̀ʌʌt áa  ɟấaak̄-ɔ̄ mɛ̂n ɟʌ́kɪ ́   Okwa  pst-take foc Diwaat nomp chief-sg idp.s mdf rule:amb:nt kɪ ̀  bấaaŋ̄-ɛ̄ prp after.prt-3sg ‘Okwa picked Diwaat to be the chief to rule after him.’ [TheDescendantsOfWaang 15.2-18.9] b.^ gɛ́ cwɔ̂lɪ ̀ gɛ́=áa àkǒool běeen pr.3pl call:iter:ov pr.3pl=nomp leader:pl quant ‘They are all called leaders.’ [TheDescendantsOfWaang 226.9-228.3] c.^ wɔ́ nū̌uut wɔ́  mīit ̪-ɪ ́ rɪ ́ɪ wɔ́n wɔ́n áa  kàl pr.1plex aux:nt pr.1plex hold:nt refl pr.1plex pr.1plex nomp family ‘We still hold together as a family.’ [TheDescendantsOfWaang 796.4-798.6] d.^ kấā ʊ̀-ŋùt-ɔ̀ kí t ̪jêp conj impf-cut:atp prp k.o.acacia kấā  gɔ̀ tjɛ̂ɛɛŋ-ɛ̀ áa jʌ̂ʌj mɛ́ mjéeer̀ conj pr.3sg.obl craft:3sg nomp vehicle mdf.sg beautiful:ctg ‘and he cut an acacia, and crafted it to a beautiful boat.’ [AchangVillagePart3 132.5-137.6] e. ̂ kɪ ̀  cʌ́n à-jʊ́ʊʊt-ɛ́  gɔ̀n áa ábóoop-ɔ̀ prp end seqpst-find:xv-3s pr.3sg.obl nomp ambatch-sg ‘In the end, he realized it was a piece of ambatch wood.’ [TheDescendantsOfWaang 110.7-112.6]

28 The lack of a 3rd singular pronoun before áa cannot be attributed to topic drop – the zero realization of 3rd singular pronominal topics that are not marked for focus, because topic drop is specific to the first or only predicate head only. In a serialization, a 3rd singular topic before a non-initial predicate head is not regularly dropped, as seen from e.g. twɔ́ɔŋ á-bèeet ɛ́ bǎa ̄ ‘Twong was a chief.’ ɟấaak-ɔ̄ SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

51

With respect to its function, áa allows for the addition of an argument to the clause, as illustrated by the examples in (84). Its use includes both proper inclusion, as in (84c), and equation / identification, as in (84a). áa does not inflect for tense-aspect-modality (TAM) or voice; this function is marked on a preceding predicate head within the serialization. And in fact, the other predicate head can serve merely to provide a specification for TAM or voice. For this purpose {bèeet} ‘stay’ is used in a grammticalised manner – just as it is elsewhere in Shilluk grammar – see e.g. Section 7, and also Chapter 4, page 13. This use of {bèeet} in conjunction with áa is illustrated in (85), where the inclusion of á-bèeet merely serves to contribute Past tense. The lack of marking for tense-aspect on áa fits with its distribution: verbal predicates in the non-initial position in a serialization are also restricted in the range of TAM forms they can appear in.29 (85)^ cʊ̄ʊl gɛ́ kɪ ́  bʊ̌ʊl á-bèeet gɛ́=áa mʌ̂́ʌʌtμ̪ Chol pr.3pl prp Bol pst-stay pr.3pl=nomp friend:pl ́̂ ̄ ̪ kéeer à á-piiit gɛ̂n since mdf pst-grow:xvs pr.3pl:nom ‘Chol and Bol were friends from childhood (lit. from when they grew up).’ [KeepTheSecret 0.0-2.5]

There is one context in which áa is confusable with the existential predicate marker dɪ ̂, specifically when the latter is in a juncture with the focus marker a, yielding dâa. The scenario in question is illustrated in (86). The first word of these two sentences is realized [dáa]. The interpretation that this would be a form of the existential marker does not fit with the meaning of the sentence. Instead, we are dealing here with the nominal predicate marker áa, as it appears here in a juncture with the conjunction dɛ́ ‘but’: the vowels in dɛ́ and áa are in hiatus, and hence they coalesce to dáa. (86)

a.^ dɛ́=áa cʊ́ʊt ̪-ɪ ́ mɔ́k-áa ànàn conj=nomp end-prt.pl idp.pl:prt-1sg:pl dxs.sg ‘But that is the end of the things I had to say.’ (lit. ‘That is the end of my things’) [SomethingStolenCannotBeSacrificed 159.7-160.7] gʌ̀ʌt à câaat ̪-ɛ̀ b.^ dɛ́=áa jùuuμ conj=nomp road-prt riverbank foc walk-3sg:pl ‘But (instead) he walks the road of the riverbank.’ [AchangVillagePart3 251.7-253.4]

29 The morphosyntactic properties of serializations in Shilluk will be described in detail in a following chapter dedicated to complex clauses. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

52

In line with the above descriptive analysis, both (86a) and (86b) are serializations, each with two predicate heads, the first one being the nominal predication headed by áa. In (86a), the second predicate is headed by the demonstrative (cf. Section 14 and also Chapter 5, Section 6.2.4). This can be seen from the fact the construction without ànàn is ungrammatical: *dáa cʊ́ʊt ̪-ɪ ́ mɔ́k-áa is ungrammatical. Similarly, dáa lʊ̀ʊt ̪ ànàn ‘But that is a stick.’ is grammatical, but *dáa lʊ̀ʊt ̪ ‘But it is a stick.’ is not. And in (86b), the second predicate is a lexical verb.

10. Nominal predications formed using bɛ̄ɛnɛ́

The final nominal predicate marker to be covered is bɛ̄ɛnɛ́. Its function can be characterised as identificational / equative (Higgins 1979, Mikkelsen 2011); by implication, nominal predications marked by bɛ̄ɛnɛ́ are invariably referential. Its use is illustrated by the examples in (87). Both are drawn from narratives in which a group has been identified. Note, how, in the first part of (85a), a particular boat is identified in detail; then bɛ̄ɛnɛ́ identifies this boat again. Similarly, (85b), references a village that is unique in not losing people to migration to the town. (87)

́ ̄ a.^ kấā ʊ́-kấaan̄ -ɔ̄ ʊ̀-dɔ̀ɔɔk-ɔ̀ áa mɛ̂n  ɲɪ ́ túuuk̀ conj nmlz-k.o.palm-sg impf-return nomp idp.sg hab play-xv ́ ̄ à ɪ ̄ɪ jấaɲ̄ t ̪ɔ̀n̪-ɔ̀ mɔ̂k pwōɲɪ ́ rɪ ́ɪ gɛ́n kɪ ́ kéew prp.pl men:cs small-pl idp.pl mdf teach:nt refl pr.3pl prp row:infa ‘After that, the palm boat was turned into something used to to play with by the young boys, those who were teaching themselves to row,’ ́ ̄ à gɛ́ kɪ ́ mɔ̂k jìi gɛ́n  dɪ ̂ bɔ̂ɔj-ɪ ̀ à t ̪ɔ̀n̪-ɔ̀ pr.3pl prp idp.pl mdf ad pr.3pl exsp net:pl-cs mdf small-pl ‘and by those who have small fishing nets.’ bɛ̄ɛnɛ́  gɔ̀n à ɲɪ ́ mʌ́ʌʌj ̀  ɪ ̄ɪ gɛ́n nomp pr.3sg.obl mdf hab fish:atp:xv prp:pl pr.3pl ‘It was the one used for fishing by them.’ [AchangVillagePart3 138.8-148.6] ̄ ̄ b.^ ácʌ̂ŋ pāaar-ɛ́ɛ bɛ̄ɛnɛ́  à bʊ̄ŋ kʌ̂́ʌl-ɪ Achang village:prt-1pl.inc nomp foc neg.exsp family-pl ́ +à) á-d̪wòt ̪ kɪ ̀ ɪ ̄ɪc-ɛ̄ mấā (=mɔ̂k mdf.pl cs past migrate prp inside:prt-3sg ‘Achang our village is the only one that no families migrated away from.’ [AchangVillagePart3 787.7-790]

Comparing bɛ̄ɛnɛ́ with bǎa, we find that bɛ̄ɛnɛ́ is consistently referential, that is, referring to single real-world entity, whereas bǎa is not consistently so. This SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

53

is shown in (88). The sentence in (88a) is felicitous in a context when the topic references a man who is one of the speaker’s teachers, i.e., ‘he is a teacher of

mine’. The sentence in (88b), in contrast, implies that the topic is referential, referencing the sole teacher or main teacher of the speaker. (88)

a. bǎa ɟâal-pwòɲ-āa nomp man-teach-1sg ‘He is my teacher.’

b. bɛ̄ɛnɛ́ ɟâal-pwòɲ-āa nomp man-teach-1sg ‘He is my teacher.’

The nominal predicate marker bɛ̄ɛnɛ́ ends in /ɛ́/. Anywhere else in Shilluk morphophonology, a word-final vowel /ɛ/ marks 3rd singular – be it as a

possessive marker on nouns or a subject marker on verbs. In constructions with bɛ̄ɛnɛ́, however, the argument does not need to be singular. This is

shown in (89), where it is followed by ɟɪ ̀ɪ ‘people’. Nonetheless, this example

presents evidence in support of the analysis that the 3rd person form of bɛ̄ɛnɛ́

has become fully grammaticalized and ‘frozen’, as a non-inflecting copula. This can be seen from the fact that the deictic predication marker ànàn, which is grammatically singular – the corresponding plural is àgàk.

́ ̄ à (89)^ bɛ̄ɛnɛ́ ɟɪ ̀ɪ mɔ̂k á-dôoŋ ànàn nomp people idp.pl mdf pst-stay.behind dxs.sg ‘Those are the people who stayed behind.’ [TheDescendantsOfWaang 650.7-651.8]

11. Headless nominal predication In the preceding subsections, we have described the functional and formal properties of several nominal predicate markers. In addition, nominal

predication may also be headless, i.e., lacking any nominal predicate marker. This is restricted to relative clauses. Illustration (90) presents an example

of this, in the relative clause, where bɔ̀ɔɔt ̪-ɔ̀ represent a headless nominal predication. Crucially, there is no copula in the relative clause.30

30 The fact that à is the relativizer, i.e., the modification marker, and not the focus marker in this sentence can be seen from the fact that the preceding noun is the construct state: twɔ̂ɔ́ ɔŋ̄ is the construct state form of twɔ́ɔŋ ‘Twong’. This signposts that it is being modified, in this case by the relative clause. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk (90)

54

twɔ̂ɔ́ ɔŋ̄ à bɔ̀ɔɔt ̪-ɔ̀ bǎa ɟấaak̄-ɔ̄ Twong:cs mdf craftsman-sg nomp chief-sg ‘The Twong who is a craftsman is a chief.’

It is worthwhile to note that the distribution of headless nominal predications is more restricted than that of headless locative predications. Just as nominal predications, locative predications are grammatical in relative clauses; but, as described in Section 6, the latter are additionally grammatical as a main clause if there is focus marking; this is shown in (89a). In contrast, nominal predications cannot be used in this way, as seen from (89b). In the absence of a copula, here bǎa, the sentence is ungrammatical. (91)

a. twɔ́ɔŋ à kâl Twong foc compound ‘Twong is in the compound.’

b. twɔ́ɔŋ à *(bǎa) bɔ̀ɔɔt ̪-ɔ̀ Twong foc nomp craftsman-sg ‘Twong is a craftsman.’

12. Negative nominal predication The way negation is expressed in nominal predication is illustrated by the examples in (92). In (92a), it is predicated that the subject is not a member of the set of thieves (proper inclusion); in (92b), it is predicated that the referent of the omitted 3rd singular topic – a dish to be brought to the center of the village – is not equal to the dish people had expected (equation / identification). Both are expressed using pât ̪, followed by the preposition kɪ ́, which heads a prepositional phrase holding the predicated noun phrase. (92)

a.^ já pât ̪ kɪ ̀  kùur pr.1sg neg.nomp prp thief ‘I am not a thief.’ [SomethingStolenCannotBeSacrificed 213.6-214.7] ́ ̄ b.^ dɛ́ pât ̪ kɪ ́ mɛ̂n  à  á-rɔ́mɪ ̀  ɪ ̄ɪ gɛ́n conj neg.nomp prp idp.sg mdf pst-think:ov prp:pl pr.3pl ‘But it (=the dish that was brought) was not the one they expected.’ [DengsFish 93.8-95.6]

These examples show that negative nominal predication is expressed through a single morpheme, pất ̪, rather than through the combination of morphemes expressing negation and nominal predication separately. In this respect, the phenomena run parallel with those observed for existential predications (Section 3), where bʊ̄ŋ represents an antonym to the positive existential predicate marker dɪ ̂ (Section 2). SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

55

The morpheme in question is also used in the expression of negation in response to a yes/no question. This is shown in (93), which shows both affirmative and negative answers in response to a yes/no-question. (93)

̀ twɔ́ɔŋ cấm̄ a kwʌ̄n Twong eat:nt foc porridge ynq ‘Is Twong eating porridge?’

ŋɛ̀ɛɛ / pât interj neg.nomp ‘Yes. / No.’

Interestingly, while the affirmation ŋɛ̀ɛɛ is the same irrespective of TAM, the negative answer is sensitive to it. This is shown in (94); the forms are to be compared with the one in (93). Whereas pât ̪ is used when the verb in the yes/ no-question is in the No Tense form, nùutɪ ́ is used when the verb is in the Past (94a); nùutɔ̀ is used when it is in Non-Evidential Past (94b). Finally, the future tense negation marker is only used in a juncture with a lexical predicate head, as in (94c). It cannot be used as an utterance by itself: the answer *kʊ̄ʊ is ungrammatical. (94)

a. kwʌ̄n á-cấm ɪ ̀ɪ twɔ́ɔŋ ̀ porridge pst-eat:ov prp Twong ynq ‘Did Twong eat the porridge?’

ŋɛ̀ɛɛ / nūutɪ ́ interj neg.pst ‘Yes. / No.’

b.

̀ twɔ́ɔŋ càaam a kwʌ̄n Twong eat:nevp foc porridge ynq ‘Did Twong apparently eat the porridge?’

 ŋɛ̀ɛɛ / nūutɔ̀ interj neg.nevp ‘Yes. / No.’

c.

kwʌ̄n ʊ́-cấm ɪ ̀ɪ twɔ́ɔŋ ̀ porridge fut-eat:ov prp Twong ynq ‘Will Twong eat the porridge?’

 ŋɛ̀ɛɛ / kʊ̄ʊ cấm ɛ́n interj neg.fut eat:ov pr.3sg ‘Yes. / He will not eat it.’

The examples in (93, 94) show that the expression of negation as a statement by itself, i.e., the equivalent of English ‘no’ is specific to TAM. This is also the case in the expression of negation on predications headed by verbs and by adjectives.31 In contrast, negation in nominal predications is marked solely by means of pât ̪. And when negative nominal predication is expressed in Past, Non-Evidential Past or Future, then a serialization construction is used, with the now familiar grammaticalized verb {bèeet} ‘stay’, which carries TAM. This is shown in (95a), which is to be compared with (92a). In contrast, the use of the other TAM specific negation markers is ungrammatical here (95b).

31 Note for example, the use of the past-tense negation marker nūutɪ ́ in já nūutɪ ́ kwʌ̀ʌʌtɔ̀ pr.1sg neg.pst steal:nevp ‘I was not stealing’, which is headed by an intransitive verb form; and jāat ̪ nūutɪ ́ bʌ̂ʌr tree neg.past tall ‘The tree was not tall’, which is headed by an adjective. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk (95)

a.

56

já á-bèeet /  bèeet-ɔ̀ / ʊ́-bèeet  já pât ̪ kɪ ́ kùur pr.1sg pst-stay stay-nevp fut-stay pr.1sg neg.nomp prp thief ‘I was not / apparently was not / will not be a thief.’

b. *já nūutɪ ́ / nūutɔ̀ / kʊ̄ʊ kɪ ́ kùur pr.1sg neg.pst neg.nevp / neg.fut prp thief ‘I was not / apparently was not / will not be a thief.’

13. Nominal predications of similiarity formed using nɪ ̄

The nominal predicate constructions discussed in the preceding sections express proper inclusion and/or equation – or the negation of these functional relations. In addition, predications that take a noun complement can be headed by nɪ ̄, which expresses similarity rather than equation. In most occurrences in our corpus, nɪ ̄ appears as a noun phrase modifier. This is shown ́ ̄=ánɪ ́ modifies the adjective dɔ́ɔɔc̀; and in (96b), nɪ ̄ in (96). In (96a), nɪ ̄ mɛ̂n ́ modifies the noun phrase gîn. mɛ̂n (96)

a.^ nū̌u rɔ̂mɪ ̀ gɛ̂n kɪ ̂nɪ ̀ gɛ́ ʊ́-jîtɪ ̀ kɪ ́ pɪ ́ɲ neg.pst think pr.3pl.nom quot pr.3pl fut-find:amb prp land ́ ̄ =ánɪ́ mɛ́ dɔ́ɔɔc̀ nɪ ̄ mɛ̂n mdf.sg good:ctg sim idp.sg=def ‘They had not thought that they would find a land as good as that one.’ [TheDescendantsOfWaang 169.0-171.3] ́ b.^ t ̪ùrɔ̀ cɛ̂t ̪  à gôook-ɔ̀ gîn nɪ ̄ mɛ̂n Thuro adv foc make:atp:nevp something:cs sim idp.sg:dxs ‘It must have been Thuro who did something like this.’ [TheDescendantsOfWaang 115-116.4]

However, this morpheme can also head a predication, with a noun phrase complement that refers to the entity to which the topic is predicated to be similar. This use is illustrated in (97). This example illustrates that for nɪ ̄ to head a predicate, it needs to be licensed syntactically by the focus marker. (97)^ ɛ́ d̪ʌʌ ̀ j nāa rêeeɟ-ɔ́ pr.3sg adv sim=foc fish-sg ‘It is seemingly like a fish.’ [TheDescendantsOfWaang 105.1-106.2]

As a predicate marker, nɪ ̄ can take TAM marking, in particular for past and future tense, through prefixation. These forms are illustrated in (98a) and SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

57

(98b), respectively. There is no Non-Evidential Past, perhaps because the meaning ‘to be similar to something’ comes with its own mirative stance. (98)

a. gìn-ánɪ ́ á-nāa rêeeɟ-ɔ́ something:cs-def pst-sim=foc fish-sg ‘The thing was like a fish.’ b. gìn-ánɪ ́  ʊ́-nɪ ̄ rêeeɟ-ɔ́ something:cs-def fut-sim fish-sg ‘It will be like a fish.’

Just like dɪ ̂ and the locative predication marker jɪ ̂ŋ, nɪ ̄ requires syntactic licensing in the No Tense, in Past tense, but not in Future tense. The requirement of syntactic licensing is fulfilled most often by a focus marker clitic following nɪ ̄; this juncture is shown in (97) in relation to the No Tense form and in (98a) in relation to the Past tense form. This analysis can explain why the stem form nɪ ̄ rather than nāa is found in No Tense and Past if the focus marker is associated with another constituent within the clause, because the focus marker can appear no more than once in a clause. This is shown in (99). (99)

gînμ à á-nɪ ̄ rêeeɟ-ɔ́ something:cs-dxs foc pst-sim fish-sg ‘This thing was like a fish.’

14. Adjectival and deictic predications The grammatical properties of both adjectival and deictic predications are described in earlier chapters: adjectival predications in Chapter 4, Section 3.1; deictic predications in Chapter 5, Section 6.2.4. For a detailed discussion of these phenomena, the reader is referred there. Here we merely spell out their key characteristic: both are expressed morphosyntactically as headless predicates. This is illustrated by the examples in (100). In (100a), the predicate is the adjective léeet;̪̀ and in (100b) the predicate is the plural proximal demonstrative pronoun ɛ̀k. Note that, in these examples, neither subject nor predicate are morphologically marked for their syntactic status, and there is no copula.

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk (100)

58

a.^ dèeel-āa léeet ̪̀ jɛ̂t ̪-ɪ ́ cjʌ̂́ʌŋ běeen body:prt-1sg hot:ctg inside.pl-prt day:pl quant ‘My body was hot constantly.’ [DownWithIllness 109-110.7] ́ áwàc-ɪ ́ɪ ɛ̀k, mɔ̂k  ʊ́-mùuuc-á mɛ̂j́ ̄-āa b.^32 wìij-ɔ́, father-sg k.o.fruit:pl-2sg dxs.pl idp.pl:dxs fut-give:spat-1sg mother-1sg ‘Father, your awajo fruits are these, and those I will go give to my mother.’

15. Conclusion Shilluk presents an extensive range of predicate structures that are not headed by a verb. The nonverbal predicate marker with the widest functional range is dɪ ̂. Beyond its core function of existential predication, the scope of existence / availability can be restricted in three ways. First, it can express possessive predication, whereby existence / availability is restricted to a Possessor. Second, it can express deictic predication, whereby existence / availability is restricted to a Location or a Time. Third, it can express a predication of affliction, whereby existence / availability is restricted to an Experiencer. With respect to the formal properties of nonverbal predications in Shilluk, we can distinguish three types: headless predicates, predicates headed by morphologically invariant copulas, and predicates headed by a function morpheme that presents a range of inflections. The first type, headless predication, is used to express adjectival predications, deictic predications, locative predications and nominal predications (the latter are only found in relative clauses). The second type includes the morphologically invariant copulas bǎa, áa, and bɛ̄ɛnɛ́. Third, there are the predicate markers that inflect: dɪ ̂, bʊ̄ŋ, nû́t, jɪ ̂ŋ, jɪ ́ŋ, and nɪ ̄. All of these inflect for TAM, albeit in the case of nû́t and nɪ ̄ only in some of the values. With the exception of nɪ ̄, they all present an Applicative voice forms. Different from verbs, none of them inflect for Object voice. Subject marking is only found on the locative predicate marker jɪ ̂ŋ and the nominal predicate marker jɪ ́ŋ. The nonverbal predicate markers that take TAM prefixes on a par with verbs – dɪ ̂, bʊ̄ŋ, jɪ ̂ŋ, jɪ ́ŋ, and nɪ ̄ – also pattern with verbs in that they are subject to syntactic licensing. In verbs, syntactic licensing is required if the verb is marked for a TAM distinction that that does not involve a prefix, unless the verb is lexically specified for progressive aspect (cf. Section 6 of 32 There is no reference to a text here – it is a recollection of a conversation by the second author. SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

59

Chapter 1; Section 2.2.1.1 of the current chapter). In relation to the nonverbal predicate markers dɪ ̂, jɪ ̂ŋ, jɪ ́ŋ, and nɪ ̄, the requirement for syntactic licensing is more extensive, in that they require focus marking also if the verb is in the Past tense form, even though this inflection involves a TAM prefix. And in the case of jɪ ́ŋ, syntactic licensing is more constraining still, in the sense that the focus marker satisfies it in certain positions only. While we have expressed the generalisations that appear to hold for syntactic licensing in relation to each of the nonverbal predication marker, it is not always clear what the explanation is, i.e., why a particular nonverbal predicate marker would need to be licensed. This is a direction for further research.

Abbreviations used in glosses The following abbreviations are used in the glosses. adv amb apl atp aux card conj cs ctg def dxs excl exsp foc fug fut hab idp impf indir inf

Adverb Ambitransitive Associative plural Antipassive Auxiliary verb Cardinal Conjunction Construct state Contingent form (of adjective) Definite Deixis marking Exclamation Existential predicate marker Focus Centrifugal deixis Future Habitual Independent pronoun Imperfective Indirect speech marker Infinitive

neg nevp nexp nmlz nomp nt obl ov patn pet pl pr pres prp prt pst quant quot refl sg seqp

Negation Non-evidential past Negative existl. pred. marker Nominalizer Copula for nominal predicates No tense Oblique pronoun Object voice Patient nominalization Centripetal deixis Plural Independent pronoun Present Preposition Pertensive Past Quantifier Quotative Reflexive Singular Sequential past

SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk

infa interj iter locp mdf nom

Agent-oriented infinitive Interjection Iterative Locative predicate marker Modification marker Nominative

60 sim sub whq xv xvs ynq

Similarity Subordination marker wh-question marker Applicative voice Applicative voice w. inversion Yes/no question

Acknowledgements We gratefully acknowledge the Leverhulme Trust, which supported this work through the research grants “A descriptive analysis of the Shilluk language” (RPG-2015-055) and “Suprasegmentals in three West Nilotic languages” (RPG-2020-040). Spontaneously uttered examples are drawn from narratives by Akwaci Otto Gwado, Gabriel Basha Oyo, Maria Bocay Onak, Nyathom Thabo Odhong, Otto Gwado Ayoker, Teresa Akic Awanh, and Viviana Nyacyan James. We gratefully acknowledge their contributions. In the final stages of the preparation of this chapter, Akil Ismael and an anonymous reviewer for Language Documentation & Conservation provided detailed feedback, which led to substantial improvements; and then Brenda Thornley typeset the paper in journal layout. We are very grateful to them for their engagement and effort.

Bibliography Andersen, Torben (1988). Ergativity in Päri, a Nilotic OVS language. Lingua 75, 289-324. Creissels, Denis (2014). Existential predication in typological perspective. Revised version of a paper presented at the 46th Annual Meeting of the Societas Linguistica Europaea (Split, 18 - 21 September 2013). [http:// www.deniscreissels.fr/public/Creissels-Exist.Pred.pdf] Dixon, R.M.W. (2009). Basic linguistic theory / Volume 2: Grammatical topics. OUP. Dryer, Matthew S. (2007). Clause types. In Timothy Shopen (ed.) Language typology and syntactic description, volume 1: Clause Structure (2nd edition). SPECIAL PUBLICATION NO. 14 , 2023

Nonverbal predication in Shilluk 61

CUP, 224–275. Higgins, Roger Francis (1979). The pseudo-cleft construction in English. New York: Garland. Hyman, Larrry M. & John R. Watters (1984). Auxiliary focus. Studies in African linguistics 15(3), 234-273. Krifka, Manfred. (2008). Basic notions of information structure. Acta Linguistica Hungarica 55, 243–276 Mikkelsen, Line (2011). Copular clauses. In Claudia Maienborn, Klaus von Heusinger, and Paul Portner (eds.) Semantics: An International Handbook of Natural Language Meaning, volume 2. Berlin: Mouton de Gruyter, 18051829. Overall, Simon, Rosa Vallejos, and Spike Gildea (2018) Nonverbal predication in Amazonia. In Overall, Rosa Vallejos, and Spike Gildea (eds.) Nonverbal predication in Amazonian languages. John Benjamins, 1-49. Payne, Thomas E. (1997). Describing morphosyntax: a guide for field linguists. CUP. Remijsen, Bert, Otto G. Ayoker & Timothy Mills (2011). Shilluk. Journal of the International Phonetic Association 41(1), 111-125. Remijsen, Bert & Otto G. Ayoker (2014). Contrastive tonal alignment in falling contours in Shilluk. Phonology 31(3), 435-462. Remijsen, Bert & Otto G. Ayoker (2020). Floating quantity in Shilluk. Language 96(3), e135-e156. Remijsen, Bert, Otto G. Ayoker & Signe Jørgensen (2019). Ternary vowel length in Shilluk. Phonology 36, 91-125. Remijsen, Bert & Otto G. Ayoker (no date). Collection of Shilluk narratives and songs. Edinburgh DataShare. [dataset, online at https://datashare.is.ed. ac.uk/handle/10283/425] Sapién, Racquel-Maria (2018). Nonverbal predication in Kari’nja (Cariban, Suriname). In Simon Overall, Rosa Vallejos, and Spike Gildea (eds.) Nonverbal predication in Amazonian languages. John Benjamins, 103-134. Stassen, Leon (2009). Predicative possession. Oxford University Press. Vydrin, Valentin (2020). Non-verbal predication and copulas in three Mande languages Journal of West African Languages 47, 77-105. Wade, Terence (2000). A comprehensive Russian grammar. Blackwell Publishers.

SPECIAL PUBLICATION NO. 14 , 2023