Chapter 9*

FOCUS: First & Last

1. Introduction In the preceding chapter, we examined two languages from the perspective of FOCUS and its alliance with the Patient function and the morphosyntax of sentence-initial position. The unifying link was a reliance on Behagel’s First Law. One of the conclusions to that discussion was that word order typologies which rely on syntactic (meaningless) tokens do not reveal the relations among languages that our descriptions suggest to be present. Witness the actual connections between the FOCUS initial languages Warao & Urarina, Bella Coola & Yogad, and Haida where their word orders (OVS/ OSV, VSO, and no basic order, respectively) show no similarity. A semantic typology (Chapter 10) should be more illuminating. This chapter exists only because some languages have ended by invoking word order in the expression of FOCUS, and they are not random in that usage. Since not all languages turn to word order in signalling FOCUS, not all languages will find a place in this discussion. In that sense, this is not a true typology of FOCUS. But because FOCUS has impinged to shape the syntactic contour of clauses in at least some languages, the phenomenon touches upon the matter of word order typology more generally. Before proceeding to further discussion of FOCUS in relation to biases in its morphosyntactic expression, we will look a bit more at the notion of typing languages with syntactic word order. Recognizing failures in attempts to type languages using orders of the three tokens S, O, and V, Dryer (1997) proposes an alternative based on the couplets OV, VO, SV, and VS. Dryer relies on frequency to identify a basic order and he requires that an order appear at least twice as frequently as its competitors in order to be basic (Dryer 1997.74). ______* It finally occurred to me that, in writing this book, there will be long periods of time in which nothing seems to happen. That is not entirely true. thought that it might be interesting to make a chapter available in progress so that one could watch it being built. This began as Chapter 9, but as it approached 100 pages without behing half finished, I broke the chapter in two. Now, Chapters 9 & 10 are in progress. In showing the work in progress, I exercise the right to misspell, to make mistakes (and maybe correct them), to try an idea (and maybe abandon it), to change my mind (maybe without reason), to be irritated with the literature (and maybe vent), to not know (and maybe learn), etc. In the meantime, Pay no attention to that man behind the curtain. 2 SYNTAX & SEMANTICS

Languages seem in this way always to select a basic order from among the couplets OV, VO, SV, and VS, while it is not uncommon for a language to fail to reveal one of the triads (SOV, SVO, etc.) as basic, especially when the 2X criterion is used. This system produces four types: VS&VO, SV&VO, SV&OV, and VS&OV. Dryer adduces eight arguments in support of it. First, languages which are indeterminately VSO or VOS (Fijian) now have a typological home in VS&VO. Second, with respect to language types providing “the basis for predicting other typological characteristics” (Dryer 1997.75), collapsing VOS and VSO into the VS&VO type is an advantage since “there are no known differences between VSO and VOS languages.”1 Third, languages that are VSO or VOS “can easily change from one order to the other” (Dryer 1997.76). Fourth, tactically, the observed frequency with which languages elide A and O makes typing them using all three terms of the S, O, and V triumvirate difficult to execute. Using VS&VO and so forth will yield a type while using SVO and so forth will not (again using the 2X frequency criterion) (Dryer 1997.79). Fifth, some languages exploit the morphosyntax of word order so effectively that no single one stands out when using the 2X frequency criterion. Such languages will continue to exemplify one of the four types based on the doublet orders. Sixth, some languages will fail to reveal a choice between VO or OV or between SV or VS, but never both. “It is not clear, for most of the languages cited here, that they can be classified by the traditional typology. The proposed typology allows it to be clear exactly where one can assign a basic order” (Dryer 1997.86). Seventh, the doublet typology recognizes that “the order of the subject and verb is much less important than the order of the object and verb. By distinguishing these two parameters, as the typology proposed here, we explicitly separate out the more important parameter from the less important one” (Dryer 1997.66).2 VO/OV is the “more important parameter” because the VO languages which combine VSO and SVO appear more similar to each other than either does to the OV type, which combines SOV and OSV.3 Eighth, the

1 Except that they are not the same.

2 This is one expression of the pattern of Verb-Object Bonding (Tomlin 1986.73-101).

3 The similarities within this syntactic type are expressed other syntactic properties. For example (Dryer 1997.86),

SVO languages are like verb-initial languages in being predominately prepositional ...[and] prenominal relative clauses are ... rare in both SVO languages (1%) and verb-initial languages (0%). For various other characteristics too, SVO languages pattern much like verb-initial languages. This demonstrates FOCUS: First and Last 3 typology which Dryer proposes easily incorporates intransitive expressions, whereas the alternative S, O, and V typology “is based entirely on the word order of transitive clauses and ignores intransitive clauses” (Dryer 1997.87). Although both Tomlin (1986), as the representative of the traditional triad typology, and Dryer yearn for some semantic component to their types, it remains by and large elusive.4four Their types are formal ones created with

the validity of the claim that there is a fundamental distinction between VO and OV languages, a distinction which is isolated in the typology proposed here.

The claim of similarity betweenVSO and SVO to the exclusion of OSV/OVS contradicts the fact that VSO, OSV, and OVS languages are that way because they have melded FOCUS with either EVENT/Verb or PATIENT and then associated the result with sentence-initial position.

4 Tomlin (1986.138-139) There seem to be two paths that typological research must take. In its initial stages, both for the sub-discipline as a whole and for specific studies, attention must be devoted to the systematic observation of structural regularities and variation exhibited cross-linguistically ... But in its second stage typological research turns the order of inquiry around. It begins with the functional principles and inquires how they are manifested structurally in the languages of the world ... By turning the order of inquiry around a new sort of opportunity arises. no structural presuppositions are made, and the resulting typological observations can be more fine grained ... The vocabulary of this new typology continues, however, to be one of how functional “information is manifested cross-linguistically” and not one in terms of how the functional/semantic contents are interrelated independently of its morphosyntax. It continues to be a syntactic typology. It is more motivated, but it is also uncertain that the same syntactic types would emerge. Dryer (1997.82), while proposing a purely syntactic typology based on frequency, nevertheless asks: Do we not also want to say something about what conditions the choice of VS and VO as opposed to the less frequent orders SV and OV [in Hanis Coos and in Papago]? The answer of course is yes ... And clearly what determines the choice among the different orders is more important than the relative frequencies of the different orders. He then points to his answer in this specific case: “Definite nominals generally follow the verb, while indefinite nominals generally precede.” If “what determines the choice” is more important, why is the typology not in terms of that, whatever it may be. Both Tomlin and Dryer end by trying to explain or justify their types. Tomlin 1986 is devoted to explaining the relative frequencies of the types with reference to three functional principles: the Theme First Principle, Verb-Object Bonding, and the Animated First Principle. Except that they are not all functional. Verb-Object Bonding is the name given to a collection of syntactic observations: (i) noun incorporation on the verb prefers objects (Tomlin 1986.79), (ii) sentence qualifies “occur on the side of the verb opposite the object” (Tomlin 1986.81), (iii) “sentence adverbials are not permitted to inervene between the verb and object” (Tomlin 1986.84), (iv) “the distribution of modal elements is such that in general the object is not separated from the verb” (Tomlin 1986.87), 4 SYNTAX & SEMANTICS reference to formal criteria. In bypassing semantics and semantic regularities, any such typology is ultimately arbitrary.5 Either way constructs a Procrustean Bed which many languages find uncomfortable. Haida, for example, has no use for ordering in terms of S’s, O’s, and V’s whether arrayed as triplets or doublets. That is simply irrelevant for Haida. Palauan, while it seems to be VOS, has its a ... a morphosyntax, a strong complement to the VOS grammar. The Palauan a ... a grammar exists without reference to S’s, O’s, and V’s and it resists attempts to reduce it to such. Like Palauan, Yogad appears to be a typical VSO language, yet it pairs its VSO grammar with an alternative X ay Y syntax in the expression of TOPIC (Chapter 14). Like the Palauan a ... a, Yogad X ay Y does does not decompose into S’s, O’s, and V’s. Wolof complements its SVO morphosyntax with the addition of non-SVO grammar in giving expression to degrees of FOCUS. Wolof has a collection of four morphosyntactic expressions.

2. Syntagmatic FOCUS, Bipartite FOCUS, Paradigmatic FOCUS & Singularity of FOCUS Although the primary purpose of this chapter is the discussion of word order as an expression of FOCUS, along the way, we will encounter languages which exemplify some combination of a four-way distinction in the semantic organization of FOCUS ... of a particular sort. These dimensions to the organization of FOCUS will provide one of the primary ways of creating a typology of FOCUS. The differences among the four can be preliminarily sketched as follows. If FOCUS is, roughly, the semantic enhancement of some one component of an and so forth. Verb-Object Bonding is not a“functional principle” (Tomlin 1986.73). As depicted by Tomlin, it is not a principle that is prior to grammar; it follows from observed grammar. In no way can it “shape[...] the grammars of natural languages.” Dryer 1997 only sporadically, as noted above, refers his types to another explanation, and it is not clear that all SV and OV languages would have the same explanation.

5 Dryer (1997.75) suggests that there is a way in which the arbitrariness can be diminished and the type justified: “A typological distinction is more significant if it provides the basis for predicting other typological characteristics.” However, Haspelmath’s (2007.126) conclusion appears more accurate (By “categories,” Haspelmath intends “formal categories such as affix, clitic, and compound” (124)) : The most important consequence of the non-existence of pre-established categories for language typology is that crosslinguistic comparison cannot be category-based, but must be substance-based, because substance (unlike catetories) is universal. Hear, hear. FOCUS: First and Last 5 utterance, that enhancement must occur against a background. “Enhance- ment” (or whatever other word is chosen) is a relationship.6 It implies that the enhanced has “more” of something than the other content that is not enhanced, and the nature of the “other” will produce, ultimately, at least four variations of FOCUS. First, the “other” can be the other component functions of the PROPOSITION (e.g., the Agent is focused, but not the Patient nor the EVENT) within the reach of the semantics of ROLE and VOICE. Second, the enhancement may occur against a background of competitors to manifest a single propositional function (e.g., the one individual that is the Agent, and not an alternative one). The first enhancement has a syntagmatic cast to it, and the second, a paradigmatic one. In the Syntagmatic FOCUS, those components of the PROPOSITION which lack FOCUS continue to be present and expressed.7 Somali provides us an example of Syntagmatic FOCUS (Saeed 1984.125):

(1) (a) Yaa Ø yeelay? [who it did] ‘Who did it?’

(b) Axmed baa Ø yeelay [Ahmed FOCUS it did] ‘Ahmed did it’

(c) *Axmed baa [Ahmed FOCUS]

“If the verb does not appear, then neither can baa ...” (Saeed 1984.125). The Syntagmatic FOCUS is primus inter pares. Northern Sotho provides an example of Paradigmatic FOCUS. The language contains an element that marks ASSERTION (Lowrens 1991.71):

(2) Phahlane ke [Phahlane ASSERT teacher] ‘Phahlane is a teacher’ which is also involved in Paradigmatic FOCUS (Zerbian 2006b.397 & 2007b.326):

6 The substantive character of “enhancement” will occupy us more centrally in Chapter 11.

7 This is the “both-and” that makes it syntagmatic. 6 SYNTAX & SEMANTICS

(3) (a) Ké mang (yo) a nyaka-ng [ASSERT who RPRN.CL1 CL1 look.for-REL nga:ka? CL9.doctor] ‘Who is looking for a doctor?’

(b) Ké a nyaka-ng ngaka [ASSERT SC1-old.man SC1 look.for-REL CL9.doctor] ‘The old man is looking for a doctor’

(c) Ké [ASSERT SC1-old.man] ‘The old man’

Unlike the Somali in (1c), the Northern Sotho in (3c) can respond to the question of (3a) without mention of the remaining functions of the question. Mokgalabje ‘the old man’ stands in an exclusionary relation with other possible answers (e.g., Ke moruti ‘the teacher’) and not in a combinatory relation with the other components of its PROPOSITION. In the paradigmatic organization of FOCUS, possible FOCUSES (e.g. the EVENT, another PARTICIPANT, Place, etc.) are absent from the PROPOSITION.8 The third dimension of FOCUS has properties of both the Syntagmatic and the Paradigmatic FOCUSES, and probably stands between them. This utterance from Modern Greek provides an example (Georgiafentis 2005.165):9

(4) O Janis tin efaje tin turta [the-nom John it ate-3s the-acc cake] ‘What John did was eat the cake’

The English gloss of (4) is misleading. It suggests the morphosyntax of a free relative clause, What John did, but the Modern Greek which the free relative clause translates is completely other. The Verb efaje is the FOCUS of the utterance, and will have a distinguishing focal stress. The material which precedes is minus any focal stress and simply names two of the PARTICIPANTS

8 This is the “either-or” that makes it paradigmatic. Commonly, after the PARTICULAR that bears this FOCUS is named, the unfocused alternative may appear in some expression of ‘not Y’.

9 For discussion of Modern Greek, see section 3.1.1 below and especially Chapter 10, section 3.2. FOCUS: First and Last 7

of the PROPOSITION: O Janis ‘John’ and tin ‘it’. The morphology of o Janis identifies it as Agent, and the shape of the pronoun identifies it as Patient. Preverbal, sentence-initial position without focal stress is one of the grammatical marks of that is termed TOPIC in Modern Greek. There is in (4) no morpology that might be taken as express ASSERTION, although ASSERTION must be present or (4) would not be an declarative utterance. It might be thought that the EVENT/Verb in (4) — because it is the EVENT — is signalling ASSERTION. But ... (5) is parallel to (4) in its bipartite FOCUS, yet the FOCUS (and ASSERTION?) is o Janis ‘John’ and not the EVENT/Verb (Keller & Alexopoulou 2001.306):

(5) Tin apelis-e o Janis ti Maria [her fired-3s the-nom John the-acc Mary] ‘John fired Mary’

By and large, (4) and (5) seem to consist of two parts, a TOPIC and a FOCUS.10 Hence, the label “bipartite” for this FOCUS. Syntagmatic FOCUS operates within the semantics of the PROPOSITION, meaning the EVENT, its PARTICIPANTS, ROLE, and VOICE. The element that is focused is contained in the PROPOSITION along with the non-focused ones. Bipartite FOCUS works without the semantics of EVENT, ROLE, and VOICE. The elements that are candidates for FOCUS are arrayed with the non-focused ones assembled together (initially in Modern Greek) and marked off in some manner11 from the focussed.12 In Paradigmatic FOCUS, the elements against which the focused term contrasts are not mentioned at all. If present, they exist in the context in which paradigmatic FOCUS is used.

10 The tin turta ‘the cake’ that completes (4) is an unnecessary part that can be ignored. Its function is discussed in Chapter 10, section 3.2.

11 In Modern Greek, it is the focal stress.

12 Traditionally, the initial material has been identified with TOPIC, and it has been asserted that languages always order TOPIC before FOCUS (or “comment”). It seems intuitively appropriate that the background against which FOCUS is set forth should be known at the time the FOCUS is announced. When speaking of bipartite FOCUS, it may be true that TOPIC precedes FOCUS, but more broadly, this cannot be correct. Bella Coola is clearly FOCUS initial, followed by TOPIC (Cf. Chapters 3 & 13). But the Bella Coola FOCUS and TOPIC are implemented within a semantic frame of the PROPOSITION, created of EVENT, PARTICIPANT, ROLE, and VOICE, unlike the bipartite FOCUS, which has none of that. At the same time, it also seems clear that the TOPIC of bipartite FOCUS is not “topic” in precisely the same way that TOPIC is otherwise. Cf. Chapters 12 - 19. 8 SYNTAX & SEMANTICS

This scheme of FOCUS is differentiated not by the “scope” of FOCUS. In all three, Tom, for example, may be syntagmatically, bipartitely, or paradigmatically focussed. It is the frame within which the FOCUS is accomplished that differentiates them. As we move from Syntagmatic FOCUS through Bipartite FOCUS to Paradigmatic FOCUS, the background against which the FOCUS occurs seems to be successively diminished in each stage. This diminution can continue until there is no background frame at all. There is only pure, unalloyed FOCUS, a Singularity of FOCUS. Somali provides an example of this (Saeed 1987.211):

(6) (a) Waa rún-táa [INDPART truth-FEM.your] [Lit. ASSERTION/FOCUS your truth’ or ‘Your truth is’] ‘You are right’

(b) Wàa kúwan [INDPART these.ones] ‘Here they are’

The Indicator Particle waa is one of a set of Somali morphemes that signals ASSERTION and FOCUS simultaneously. When propositional organization is present, the Indicator Particle waa will exhibit subject agreement, and participate in a Syntagmatic FOCUS with ASSERTION/FOCUS is on the EVENT (Saeed 1987.207):

(7) W-uu i Ø siiyey [INDPART-he me it gave] ‘He gave it to me’

When there is no content in the PROPOSITION other than the EVENT, agreement is absent, and the ASSERTION/FOCUS of waa is compacted with the sole remaining content. (Cf. section 3.2.1.4 for discussion) Paradigmatic FOCUS relies heavily (but not solely) on the semantics of ASSERTION for its realization.13 Like Bipartite FOCUS, it consists of two components, an ASSERTION and the context in which the ASSERTION is made. Bipartite FOCUS appears semantically to place FOCUS in the fore, letting it imply ASSERTION, while the paradigmatic FOCUS places ASSERTION in the

13 Cf. Miya, Chapter 10, section 3.4. FOCUS: First and Last 9

fore and letting it mark FOCUS. Bipartite and Paradigmatic FOCUS also differ on the issue of exclusion. In Bipartite FOCUS, there seems to be no hidden paradigm of unmentioned, unfocussed alternatives. In this way, Bipartite FOCUS is like Syntagmatic FOCUS. In both Bipartite FOCUS and Syntagmatic FOCUS, ASSERTION is an accessory. Syntagmatic FOCUS may employ ASSERTION (cf. Chapter 10), but does not depend crucially on it. No language will shape FOCUS exclusively as Bipartite FOCUS or Paradigmatic FOCUS. All languages will have some syntagmatic composition of FOCUS, but they may lack either Bipartite FOCUS or Paradigmatic FOCUS. The contrast between the four types of FOCUS will constitute one of the major typological dimensions to its semantic organization.14 In the remainder of this chapter and the next, we will develop the contrasts between Syntagmatic, Bipartite, Paradigmatic FOCUS, and Singularity of FOCUS as we encounter them.15 In section 3, we will continue our consideration of the use of word order to express FOCUS beginning with a review of FOCUS in FOCUS initial languages. Some FOCUS initial languages will be unmarked/basic/neutral verb-initial. Some FOCUS initial languages will be unmarked/basic/neutral verb-final, and some will be unmarked/basic/ neutral subject-initial. In section 4, we will examine FOCUS final languages, and in section 5 we will introduce languages that employ a “crossover” marking. In Chapter 10, we will look at what it means to have in situ FOCUS.

3. FOCUS Initial languages When a language uses sentence-initial position in the expression of FOCUS, certain content may associate preferentially with that FOCUS and to appear “unmarked” in sentence-initial position. Such appeared to be the case with the O-initial languages. We will examine several V-initial languages, and our conclusion will be similar to the one concerning languages with O-initial order. Where a language appears to be V-initial, in the neutral case, it will have associated FOCUS preferentially with the semantics of the EVENT/Verb

14 Some questions that come immediately to mind are: “Are the distinctions valid ones?” “Is there a language that does have all four types?”, “Why can a language not have only a Paradigmatic (or Bipartite) configuration of FOCUS?”, “Why is Syntagmatic FOCUS universal?” etc. Lambrecht (1994.221-238), using a more “abstract” (210) idea of FOCUS, proposes three “Types of focus structure”: predicate-focus structure, argument-focus structure, and sentence-focus structure. The typological dimension proposed in this section and illustrated below in this chapter and the following two is clearly not the same as Lambrecht’s.

15 At the same time I will try to justify the claims of this paragraph, which for the moment are just that, claims. 10 SYNTAX & SEMANTICS and then exploited initial position for its expression. Our claim will be the extreme one that all VSO and all VOS languages are FOCUS initial.16 FOCUS initial languages may also exist without FOCUS being associated with other semantics. Such was the case with Haida, and such will be the case for some, but not all, SVO and for some SOV languages languages as well. In a subject-initial language which also uses the sentence-initial position for FOCUS, the subject-initial clause (SVO or SOV) will itself always express a FOCUS that never falls on the S. The subject, when focused, will require other expression than being placed sentence-initially. Cf. Wolof (Chapter 5) or Kinyarwanda (Chapter 3).

3.1 FOCUS Initial Languages that Are VSO In this section, we will review briefly a number of languages that have been reported to have an unmarked, neutral or basic word order that is VSO. In the past 20 years or so, verb-initial languages have attracted much attention: The Pragmatics of Word Order: Typological dimensions of verb initial languages (Payne 1990), The Syntax of Verb Initial Languages (Carnie & Guilfoyle 2000), Verb First: On the syntax of verb-initial languages (Carnie, Harlie & Dooley 2005). And portions of others: Basic Word Order: Functional principles (Tomlin 1986), Word Order Rules (Siewierska 1988), and Pragmatics of Word Order Flexibility (Payne 1992). While most of the work is informed by some version of formal theory, in which shape exists on its own,17 some nevertheless approach the conclusions put forth here:

16 Dryer (1997.77) cites 7 VSO languages “With Opposite Property” with respect to “Initial Q.” There are 5 VSO languages and 2 VOS languages “With Opposite Property” with respect to “Initial wh.” “Initial Q” refers to “Sentence initial question particle in yes/no questions.” Our primary caution is then with “Initial wh”: “wh-expression obligatorily in sentence-initial position.” Assuming that responses in these languages match the initial wh-syntax, this is what we expect. FOCUS is sentence-initial, but the alternative case, in which “wh-expression [is] not obligatorily in sentence-initial position (typically in situ),” raises our interest. Our claim here is not that all FOCUS in unmarked verb-initial languages is sentence-initial, but that at least some expression of FOCUS in response to wh-questions will be. Such languages, e.g. Modern Greek below, may have a second expression of FOCUS that employs in situ morphosyntax. Modern Greek could then plausibly be one of the 5 VSO languages cited by Dryer, and Palauan could then be one of the 2 VOS languages. The issue is not that FOCUS in these languages is not sentence-initial, but that there exists at least a second implementation of FOCUS, and its grammar — because it is a second, distinctly constituted FOCUS — cannot exploit also sentence-initial position.

17 In this way of thinking about things (Carnie, Harlie & Dooley 2005.2): One important topic ... [is] the role of methodology and data sets in determining how V-initial order is derived. FOCUS: First and Last 11

We aim to show how the unmarked (or default) verb-initial, or — more generally — predicate-initial feature of Wanyi finite clauses is actually an instance of the obligatory focus-initial property of these clauses in an informationally ‘neutral’ clause (Laughren, Pensalfini & Mylne 2005.367)18

In languages commonly identified as VI [Verb Initial], predicates occur initially in temporally sequenced clauses. Pragmatically marked information is also initial. In etic terms there are different degrees of pragmatic markedness. In more rigid VI languages only the most marked situations (e.g. focus of contrast) will trigger preverbal position, while information question words, answers to information questions, and other less highly marked information may follow the verb. (Payne 1995.479)19

It was pointed out in the previous chapter (footnote 76) that the implementation of “neutral”, “unmarked” or “basic” was various. Different researchers often appealed to different indicators. While declining to accept the attribution of neutral, purely syntactic order as a useful typological device,

“Derivation” and “movement” are central in this mode (Davis 2005.31-32): As work on the syntax of predicate-initial languages has progressed, it has become clear that even though they constitute a small minority of the world’s languages, their grammars exhibit considerable internal diversity. For example, whereas a V(head)-raising analysis has become fairly standard for VSO languages such as Irish ..., a VP(predicate)-raising analysis has met with greater success in accounting for VOS systems such as Zapatec ... and Niuean .... These analyses are based on underlying SVO order, with subsequent movement of the verb or one of its projections ..., but it has also been argued that some V-initial systems have base-generated VOS order ... with VSO order derived by movement of the subject.

18 That sounds promising, but then the authors conclude (392): In attempting to account for the data presented, we have argued that both verbal and nominal predicates are XPs, which can occupy specifier (and complement) positions, but which cannot be phrasal heads. We argue that the specifier of CP is where focus is assigned to an XP through its relationship with a feature [+focus] in the head of CP. The head of CP may also be the locus of the features [+wh] or [+negative]. These three features, [+wh], [+negative] and [+focus], contrast in C; the presence of one precludes the presence of the other. One of the features must be present in C, with [+focus] being the default feature.

19 Since, for better or worse, we have assumed that the morphosyntax of answers to wh- questions is also the morphosyntax of FOCUS, the proposal made in this section with respect to verb-initial languages must differ a bit from Payne’s. Namely, if a language is correctly identified as verb-initial, then at least one expression of FOCUS will employ initial position as well (and it will answer a wh-question). This, of course, does not preclude other implementations of FOCUS, semantically and morphosyntactically. 12 SYNTAX & SEMANTICS we will assume that the perception of “neutral” is nevertheless real, but that neutral/basic VSO order is itself something that requires explanation. In all cases, they will also be FOCUS initial, and in all cases they have matched the semantics of FOCUS with the semantics of EVENT. We will accept a researcher’s declaration of a neutral order at face value unless another researcher has challenged the conclusion. Cf., for example, the disagreement about Apuriña in the preceding chapter and Modern Greek in the following section.

3.1.1 Modern Greek 20

20 In this section, when I refer to “the literature,” I intend the following sources: Alexiadou 1996 & 1999b, Alexiadou & Anagnastopolou 2000, Georgiafentis 2001, Georgiafentis 2005, Georgiafentis & Sfakianaki 2002 & 2004, Georgiafentis & Lascaratou 2007, Haidou 2000, Keller & Alexopoulou 2001, Holt, Mackridge & Philippaki-Warburton 1997, Horrocks 1983, Joseph & Philippaki-Warburton 1987, Lascaratou 1998, Le Gac & Yoo 2002, Mennen & Den Os 1994, Philippaki-Warburton 1985 & 1994, Roussou & Tsimpli 1994 & 2006, Tsimpli 1990, 1995 & 1998, and Tzanidaki 1998. With the exception of Joseph & Philappaki-Warburton 1987, all these studies reference some avatar of Generative Grammar, GPSG or the like in trying to explain Modern Greek syntax. They, nevertheless, often contain useful examples and comments about them. I have recently been made (re)aware of the subgenre of passive-aggressive gammars. Holt, Mackridge & Philippaki-Warburton’s 1997 grammar of Modern Greek is by and large useless to a nonspeaker of MG. All the examples are transcribed using the Greek alphabet. But that is not the issue. Linguists should certainly be able to make out the Greek alphabet. The problem is that the authors fail to include a line of grammatical glosses in their examples, and they do not segment the Greek. This means that if one wants to learn from an example, it is necessary to refer to a glossary (should there have been one) or to go through the grammar hunting for other examples that have shared lexicon. Unless one is obsessed, that will not happen. Similarly, Elsaid Bardawi, Michael G. Carter & Adrian Gully’s 2004 Modern Written Arabic: A comprehensive grammar is 812 pages intellectual effort to no purpose. Like Holt, Mackridge & Philippaki-Warburton’s MG grammar, there is no line of grammatical glosses to any of the Arabic examples, and to make things impossible, the Arabic utterances are given in the Arabic writing system. There must be at least 4 people in the world with both the desire and the training to use that book. The Greek grammar was published in the old Croom Helm descriptive grammar series, and the Arabic one is published in the successor Routledge series. One suspects that it could be the publisher, but Rudolf P. G. de Rijk’s 1,370 page Standard Basque: A progressive grammar, published in 2008, is of the same sort. Even though the transcription uses latin characters, it is just as useless as the others, but now the publisher is The MIT Press. Newman’s (2000) 760 page grammar, The Hausa Language (Yale University Press), is of this sort, and Jagger’s (2001) Hausa (John Benjamins), has grammatical glosses accompanying perhaps 5% of the examples. To add to the difficulty of Newman’s The Hausa Language, the 80 chapters are presented according to the alphabetical order of their titles so that, for example, Focus is Chapter 28 after Chapter 27, Expressions of Contempt, and before Chapter 29, Frequentives. Newman (2000.4) explains: The grammar is designed as a reference work. The eighty topics covered are each treated in separate units and presented in alphabetical order. The expectation is that the grammar will be consulted like an encyclopedia or a FOCUS: First and Last 13

Modern Greek is spoken natively by some 12,000,000 to 13,000,000 people. The vast majority of these — some 9,000,000 — are found in the country of Greece itself, where 95% of the population speaks Greek. Some 5,000,000 Greek speakers are to be found in Cyprus (where Greek is one of two official languages) with the remainder scattered in small enclaves in Southern Italy, the Ukraine, and Egypt, and in the numerically more significant communities in Australia, Canada, Great Britain, and the United States that make up most of the “Hellenic diaspora”. (Joseph & Philippaki-Warburton 1987.1)

Modern Greek may seem a strange language with which to initiate a section on verb-initial languages because there is no agreement on what its basic word order in fact is, or whether it has one. Philippaki-Warburton (1985.138 et passim) was one of the first to argue “that V-S-O is the basic pattern” in Modern Greek. Most of those now working on Modern Greek appear to share her opinion. Tsimpli (1995.177) concurs: “... the basic order in MG is VSO.” Haidou (2000.165) confirms the earlier judgments: “VSO is the unmarked focus information string. VSO strings are the ones that can be uttered out of the blue and the whole sentence is the new information.” Thus:21

(1) Vrike o Janis dhulya (Philippaki-Warburton [found-3s the-nom John work] 1985.124) ‘John found work’

(2) Estile o Janis to dhema (Tsimpli 1995.177) [sent-3s the-nom John the-acc parcel] ‘Yanis sent the parcel’

Similarly, intransitives are VS:

dictionary rather than being read from the beginning to end like a novel ... The hope is that once the reader becomes accustomed to the approach adopted here, he or she will find this to be a user-friendly manner of presentation ....

21 There is variation in the transliteration of the Modern Greek examples. Sometimes Tsimpli will employ i (1998.216) and sometimes dh- (1995.199) for see. Haidou (2000.175) writes diavasi ‘read’ where Alexiadou (1999b.58) writes dhjavasa. Alexiadou (1999.59) writes aghorase ‘bought’. Tsimpli (1995.193) writes agorase, and she (1998.207) also writes . I have arbitrarily written the fricatives with h and chosen j over i. In the choice between Yanis and Janis ‘John’, I have generalized the j, replacing y here and elsewhere. In the matter of grammatical glosses, I have regularized them to Tsimpli’s notation. I have also occasionally regularized the transliteration of lexical items, i.e. turta ‘cake’ is always that, and not tourta. Efage ‘ate’ and not efaje, etc. 14 SYNTAX & SEMANTICS

(3) Eklapse o Janis (Roussou & Tsimpli 2006.318) [cried-3s the-nom John] ‘John cried’

These patterns are the ones used to respond to the equivalent of ‘What happened?’ (Philippaki-Warburton 1985.123, Alexiadou 1996.41 and 1999b.54):22

(4) (a) Ti ejine? [what happened-3s] ‘What happened?’

(b) Panreftike o Petros tin Ilektra [married-3s the-nom Peter the-acc Ilektra] ‘Peter married Ilektra’

(c) %O Petros pandreftike tin Ilektra [the-nom Peter married-3s the-acc Ilektra] ‘Peter married Ilektra’

Sentence (4c) is inappropriate as a response to (4a) because “SVO orders ... are unacceptable as thetic statements” (Alexiadou 1996.41).23 The VSO order of (4b) is the appropriate answer to (4a):

... V-S-O is not thematic, and as such it is very appropriate as an answer to

22 While the SVO utterance of (4c) fails in response to (4a), there is this interesting wrinkle to the pattern (Keller & Alexopoulou 2001.309): (i) Kana neo? [any new] ‘Any news?’

(ii) O Janis pulis-e TO AFTOKINITO. [the-nom John sold-3s the-acc car] ‘John sold the car’ While SVO (4c) fails, SVO (ii) succeeds. The difference is focal stress on to aftokinito ‘the car’. Sentence (ii) has the form of TOPIC FOCUS, a strategy which we will examine further in Chapter 10, section 3.2, in the discussion of in situ FOCUS in Modern Greek.

23 Alexiadou (1999b.54) and Alexiadou & Anagostopolou (2000.176) repeat this assertion; “SVO orders ... are unacceptable in these contexts....” SVO utterances are acceptable in other contexts. See (5), (6a) and the discussion below. FOCUS: First and Last 15

questions requiring only new information, such as ... Ti ejine ‘What happened’? (Philappaki-Warburton 1985.122)

...VSO orders in MGr are best understood as answers to the question ‘what happened’. (Alexiadou 1996.41)

The most natural answer to the question ‘What happened?’, i.e. when all information is new, is the VSO order, which is morphologically, intonationally and pragmatically least marked .... (Lascaratou 1998.156)

... VSO may also converge with a neutral intonation ... [and be] a natural answer to a wide-focus question of the ‘What happened?’ type (Roussou & Tsimpli 2006.318)

“Neutral intonation” seems to be “the generally falling intonation of statements” that Joseph & Philappaki-Warburton (1987.4) describe. In Figure 1, (a) is the statement intonation, which contrasts with (b), the yes-no question intonation, which “is a rising one, with a slight fall at the end of the utterance ....” Tsimpli (1998.204) observes further that there is a “‘default’ sentential

(a) O tákis píje s to nosokomío [the-nom Takis went-3s to the-acc hospital] ‘Takis went to the hjospital’

(b) O tákis píje s to nosokomío? [the-nom Takis went-3s to the-acc hospital] ‘Did Takis go to the hospital?’

Figure 1: Modern Greek Statement Intonation and Yes-No Question Intonation. stress which falls on the sentence-final constituent in languages like Greek.” It’s presence goes unmarked in the examples cited in the literature on Greek syntax. Nevertheless, there are those who are not convinced that Modern Greek is VSO.24 Alexiadou (1996.43) seems less certain: “MGr has an SVO order in

24 Still others follow another tact. Tzanidaki (1998) discusses Modern Greek word order within a theoretical frame of Word Grammar and concludes (349) “that the exhibited patterns 16 SYNTAX & SEMANTICS the VP, other orders being derived via movements,” and (1999.46) says, “Gr permits VSO orders.” If we take frequency as the criterion for basic word order, then the language is clearly not VSO. Lascaratou (1998.157) provides these percentages of transitive utterances in a textual corpus of 2,530 sentences:25

Order Percentage

SVO 49.2 SOV 0.7 VSO 1.1 VOS 0.7 OSV 0.4 OVS 8.7 VO 34.7 OV 4.3

Figure 2: Percentages of Transitive Utterances in a Textual Corpus.

Intransitive sentences are SV, 49.7% of the time (Lascaratou 1998.159). She (1998.153) then observes, “SVO is the main clause order par excellence, which justifies the widely view that M. Greek is an SVO language,” but then concludes “M. Greek could plausibly be classified as having free word order with SVO as its dominant active transitive order.”26 My suggestion here is that it is the typological frame which forces one response to “What is the ‘basic’ order?” that is unproductive. We will see that Modern Greek has at least two major morphosyntactic configurations. One implements a proposition which expresses a continuing TOPIC, and the second is used when the proposition lacks TOPIC continuity, and does not meld its content with the context, but stands alone, in isolation. The first gives expression to the TOPIC in initial position and then follows it with a FOCUS.

in Greek word order variation may be accounted for by reference to two grammatical principles, the V →O and ← V → principles.”

25 Lascaratou does not, however, record the use of strong stress. So we do not know which percentage of the SVO utterances are implementing a FOCUSED S and how many are a TOPIC S.

26 In a footnote, Lascaratou (1998.168) describes some of the “controversial theoretical discussion.” FOCUS: First and Last 17

Because Agents are frequent TOPICS,27 the syntax of SVO is a common manifestation, but others are possible expressions of the TOPIC initial formula. See (9c), (14c), and (17c) below. The second morphosyntactic configuration employs a grammar which appears to be VSO, and which places FOCUS in initial position. Modern Greek has both a TOPIC initial syntax and also a FOCUS initial syntax, used when there is no continuing TOPIC. The two can intersect in such a way that sentence stress is the sole distinguishing mark. In the absence of a continuing TOPIC, when a piece of propositional content assumes the semantics of FOCUS, it will occur initially and assume a stronger stress (Roussou & Tsimpli 2006.341):28

(5) SINADELFI dhjamartirithikan s-ton pritani [colleagues protested-3p to-the-acc rector] ‘It was colleagues who protested to the Rector’

“Emphasis [FOCUS, PWD] is marked by placing stronger than normal stress on the emphatic [focused, PWD] element” (Joseph & Philappaki-Warburton 1987.96).29 When there is content which reaches into the context and connects with it as TOPIC, that content will be initial as well, but it will contrast with (5) in not having a stronger stress (Roussou & Tsimpli 2006.341):30

27 Cf., for example, the expression of TOPIC in Bella Coola and the reliance on the S position in the VSO formula. Chapter 12.

28 There is some inconsistency in the notation. Tsimpli (1990) does not mark the presence of the stress of FOCUS at all in her examples, but she does recognize its presence in her discussion of the examples. Holt, Mackridge & Philippaki-Warburton (1997.434) employ bold italics. Georgiafentis & Sfakianaki (2004.939) use underlining to note “the most prominent constituent in the order” and CAPITALIZATION to indicate “contrastive focus” (943). Roussou & Tsimpli (2006.336) decide to “use capital letters to indicate focus.” That would probably be equivalent to Georgiafentis & Sfakianaki’s use of underlining. Tsimpli 1995 employs UPPER CASE. For Philippaki-Warburton (1985.118), “capitals indicate prominent stress.” However, later in the same paper (131), she uses the same notation for content that “is emphatically stressed,” and it notes “an emphatic focus.” Joseph & Philippaki-Warburton (1987.97) use “italics in the Greek and boldface in the translation to indicate emphasis.”

29 Tsimpli (1995.178) notes the presence of “heavy/focal stress, typical of focus phrases in general”. Haidou (2000.166) further confirms, “the focus of the sentence must also carry the main stress in the sentence.”

30 Alexiadou & Anagnostopolou (2000.175, 208) have these examples:

(i) (a) O JANIS aghorase to aftokinito [the-nom John bought-3s the-acc car] ‘John bought the car’ 18 SYNTAX & SEMANTICS

(6) (a) I sinadelfi dhjamartirithikan s-ton pritani [the-nom colleagues protested-3p to-the-acc rector] ‘The colleagues protested to the Rector’

(b) *Sinadelfi dhjamartirithikan s-ton pritani [colleagues protested-3p to-the-acc rector] ‘Colleagues protested to the Rector’

The contrasting behaviors of (5) and (6) illustrate the absence and the presence of TOPIC, respectively. Because TOPIC extends into the context to

(b) O Janis ekapse ti supa [the-nom John burnt-3s the-acc soup] ‘John burned the soup’ These examples are from Tsimpli (1990.228-229 and 1995.177): (ii) (a) To vivlio edhose i Maria sto [the-acc book gave-3s the-nom Maria to.the.acc Jani John] ‘Maria gave the book to John’ (b) To vivlio to-edhose i Maria sto [the-acc book it-gave-3s the-nom Maria to.the.acc Jani John] ‘Maria gave the book to John’ (iii) (a) Sto Jani edhose i Maria to [to.the-acc John gave-3s the-nom Maria the-acc vivlio book] Maria gave the book to John’ (b) Sto Jani tu-edhose i Maria to [to.the-acc John him-gave-3s the-nom Maria the-acc vivlio book] ‘Maria gave the book to John’ Tsimpli’s (1990.238-239) explanation of them is this:

...the preposed object receives heavy stress, typical of focussed phrases, and is not related to a resumptive pronoun, while in [sic] the other preposed object is not heavily stressed and is related to a resumptive pronoun ... I will refer to the preposed object in ... [(iia)] as Focus, while the preposed object in ... (iib)] will be referred to as Topic. FOCUS: First and Last 19 make a connection with some portion of it, sinadelfi in (6b) fails. The absence of a determiner — present in (6a) — asserts that sinadelfi is being introduced into the c ontext with this utterance and therefore cannot constitute the link with content already present in the context. Hence, the content of sinadelfi contradicts the grammar of TOPIC (sentence-initial position & absence of strong stress), which asserts the presence of that connection. Placing sinadelfi after the V in a VSO order avoids that contradiction because there is no continuing TOPIC asserted (Roussou & Tsimpli (2006.349):

(7) Dhjamartirithikan sinadelfi s-ton pritani [protest-3p colleagues to-the-acc rector] ‘Colleagues protested to the Rector’

The Patient function may show the same contrast of being sentence-initial and then carrying strong stress or not, and thereby carrying FOCUS or expressing a continuing TOPIC (Tsimpli 1995.177):31

(8) (a) To VIVLIO edhose i Maria [the-acc book gave-3s the-nom Maria s-to Jani to-the.acc Jani] ‘It is the book that Maria gave to Jani’

(b) To vivlio to-edhose i Maria [the-acc book it-gave-3s the-nom Maria s-to Jani to-the.acc Jani] ‘The book, Maria gave it to Jani’

Several observations amplify the use of an unstressed sentence-initial constituent as TOPIC. First, we have seen above that a VSO expression is the response to a question such as Ti ejine ‘What happened?’, but if the question is a bit more specific such as (Philappaki-Warburton (1985.121):

(9) (a) Ti ekane o Janis? [what did-3s the-nom John] ‘What did John do?’

31 Note that TOPIC Patients differ from TOPIC Agents in that the former also occur with an agreement prefix attached to the verb. 20 SYNTAX & SEMANTICS

(b) %Filise o Janis ti Maria [kiss-3s the-nom John the-acc Maria] ‘John kissed Maria’

(c) O Janis filise TI MARIA [the-nom John kiss-3s the-acc Maria] ‘John kissed Maria’ then the VSO sentence of (9b) is not the answer: “This sentence is not at all appropriate in contexts which require a theme [TOPIC, PWD]” (Philappaki- Warburton 1985.122). Sentence (9c) is. Although o Janis is not itself in sentence-initial position in (9a), its very mention acts to establish it in the context, and sentence (9c) recognizes that. Alexiadou (2000.119) offers examples that clearly demonstrate the presence of TOPIC:

(10) (a) I Maria mu estile ena grama. [the-nom Maria me wrote-3s a letter. To grama irthe simera. the-nom letter arrived-3s today] ‘Maria wrote me a teller. The letter arrived today’

(b) %I Maria mu estile ena grama. [the-nom Maria me wrote-3s a letter. Irthe to grama simera. arrived-3s the-nom letter today] ‘Maria wrote me a teller. The letter arrived today’

Ena grama ‘a letter’ is introduced in the first sentence, and continued reference to it, i.e. TOPIC, has to be effected by sentence-initial position in the following utterance. The VS order in (10b) — Irthe to grama simera — is acceptable Greek, but it is not appropriate to its context. “It is clear that first position, whatever else it might be, is the characteristic position for thematic NPs [In the absence of strong stress, of course, PWD]” (Philappaki-Warburton 1985.122).32

32 Kanis ‘no one’ can appear initially only when it is focused (Philappaki-Warburton 1985.131):

(i) KANIS dhen irthe [no.one not come-3s] ‘No one came’ FOCUS: First and Last 21

Having a TOPIC in sentence-initial position, then allows FOCUS to occur with the sentence-final constituent when the default sentence stress is on that constituent, as in (9c), or to occur on other constituents following the TOPIC (Tsimpli 1990.247):

(11) (a) O Janis edhose to vivlio STI MARIA [the-nom gave-3s the-acc book to-the-acc Maria] ‘John gave the book TO MARIA’

(b) O Janis edhose TO VIVLIO sti Maria [the-nom gave-3s the-acc book to-the-acc Maria] ‘John gave the BOOK to Maria’

As one might expect now, since any degree of semantic FOCUS that selects some portion of a PROPOSITION, in place of applying to the entirety of the PROPOSITION, by that very fact of discriminating and selecting, it is natural that only one portion of a Modern Greek utterance may carry strong stress and be FOCUS. Otherwise, it would not be FOCUS. Thus, two FOCUSES in the same utterance will fail (Alexiadou & Anagnostopolou 2000.175, Tsimpli 1995.181, 191):33

Semantically, kanis “is the indefinite nonspecific item par excellence in MG, and as such it is very difficult for it to be used as theme [TOPIC, PWD].”

33 “... sentences with multiple focalization are ungrammatical” (Horrocks 1983.107); “... the sentence can have only one focused element” (Alexiadou & Anagnostopolou 2000.175); and “... only one focused argument can be preposed” (Tsimpli 1995.181). This is also true of questions (Tsimpli 1990.250): (i) *Pjos ti efage? [who-NOM what-ACC ate-3s] ‘Who ate what? (ii) *Ti pjos efage? [what-ACC who-NOM ate-3s] ‘Who ate what?

But if the multiple FOCUSES are joined by ke ‘and’, then two or more (i.e. actually still one) are acceptable (Holton, Mackridge & Philippaki-Warburton 1997.418 and Lascaratou 1998.162):

(iii) Me pjo-n ke jati malón-es? [with whom and why quarrel-2s] ‘With whom and why are you quarreling?’ (iv) Pj-os pu, póte, jati ke pos xtipis-e tin Elen-i? 22 SYNTAX & SEMANTICS

(12) (a) *STO JANI TA VIVLIA dhosame [to-the-gen John the-acc books gave-3s] ‘It is John we gave the books to’

(b) *TIS MARIAS TA VIVLIA estile [the-gen Maria the-acc books sent-3s o Janis the-nom John] ‘John gave the BOOKS to MARIA’

In contast to FOCUS, TOPIC is not so constrained. An utterance may make multiple connections to its context. Thus, we find the sentences of (13) each with two TOPICS (Tsimpli 1995.180-181):34

[who-NOM where when why and how hit-3s the-ACC Helen-ACC] ‘Who, where, when, why and how hit Helen?]

More generally, more than one FOCUS may appear in a single utterance, so long as there is still but a single FOCUS sentence-initially (Tsimpli 1998.222):

(v) KANENAS dhen ipe TIPOTA se KANENA [no one not said-3s nothing to no one] ‘Nobody said anything to anyone’

The restriction on the number of FOCUSES applies to the sentence-initial FOCUS, and not the in situ FOCUS(ES). The semantic differences between them permit this contrast in behavior. See Chaper 10, section 3.2.

34 We may note that both the Patient and the Recipient, as TOPICS, are followed by a pause intonation (noted by the commas), and both are reflected on the verb by clitics: (i) #To grama egrapse o Petros [the-acc letter wrote-3s the-nom Peter] ‘Peter wrote the letter’

Example (i), with no stress of FOCUS on the Patient to grama, is from Roussou & Tsimpli (2006.338). They comment on it as follows (2006.343): “This construction is marked as odd [Their “#,” PWD], as the object bears no focus, and [it] is not a topic either, since there is no associated clitic present.” Notice that this comment appears to imply that the use of a clitic will mark the constituent which it reflects as a TOPIC. The Agent is neither followed by a pause, nor is it reflected on the verb by clitics, except in a “highly restricted” context (Joseph & Philippaki-Warburton 1987.212, 216), i.e. occurrence with ná ... ‘Here [he is]!’ and pú n ‘Where is [he]?’ The order of Patient and Recipient as preverbal TOPICS can vary (Alexiadou 1999b.55):

(ii) Tis Maria to vivlio tis to edhose [the-gen Mary the-acc book cl-gen cl-acc gave-3s] ‘Sh/e gave the book to mary’ FOCUS: First and Last 23

(13) (a) Tus fitites, oli i kathigites [the-acc students all the-nom lecturers tus-ispostiriz-un them-support-3p] ‘All the lecturers support the students’

(b) Tis Maria, ta vivlia, tis-ta-estile [the-gen Maria the-acc books her-them-sent.3g o Janis the-nom John] ‘John sent the books to Maria’

Since both FOCUS and TOPIC may appear before the EVENT/Verb, and since they do not conflict semantically, there is the possibility of an utterance containing both (Tsimpli 1990.244-245):

(14) (a) Tis Maria TA VIVLIA tis-edhose [the-gen Maria the-acc books her-gave.3s o Janis the.nom John] ‘To Maria, John gave the BOOKS’

(b) Afto to grama TIS MARIAS [this-acc the-acc letter the-gen Maria to-estile o Janis it-sent.3s the-nom John] ‘This letter, John sent to MARIA’

The possible grammatical conflict — the competition for sentence-initial position — is resolved by placing the TOPIC sentence-initially and letting the FOCUS follow.35 They are additionally distinguished by the presence or

(iii) To vivlio tis Maria tis to edhose [the-acc book the-gen Mary cl-gen cl-acc gave-3s] ‘Sh/e gave the book to Mary’ No attempt is made to describe the semantic contrast between these usages. Alexiadou says that it is a “free ordering.” Alexiadou records no pauses in (ii) and (iii) to match those in (13).

35 The reverse sequence of FOCUS - TOPIC fails (Tsimpli 1995.183):

(i) (a) *TO PETRO i Maria simbathi 24 SYNTAX & SEMANTICS absence of strong stress. Furthermore, it seems that an utterance may be complete if both a TOPIC and a FOCUS is present. Consider this question and answer (Haidou 2000.180):36

(15) (a) Pjos efage tin turta? [who ate-3s the-acc cake] ‘Who ate the cake?’

(b) Tin efage tin turta o Janis [it-acc ate-3s the-acc cake the-nom John] ‘John ate the cake’

[the-acc Peter the-nom Maria likes-3s] ‘It is Peter that Maria likes’

(b) TO PETRO simbathi i Maria [the-acc Peter likes-3s the-nom Maria] ‘It is Peter that Maria likes’ Cf. also Tsimpli 1998.199. Russou & Tsimpli (2006.334-335) add these examples: (ii) O Petros to grama to estile [the-nom Peter the-acc letter it sent-3s] ‘Peter sent the letter’ (iii) O Petros to estile to grama [the-nom Peter it sent-3s the-acc letter] ‘Peter sent the letter’

Because the presence of the clitic conflicts with the semantics of FOCUS, in (ii), to grama ‘the letter’ cannot bear the stress of FOCUS and is therefore signaled as TOPIC. And because the potential conflict in signaling preverbal TOPIC and FOCUS has been solved by placing TOPIC first in the sequence, o Petros in (ii) must also be TOPIC. Sentence (ii) is TOPIC + TOPIC + FOCUS, where the EVENT estile bears the stronger stress. In (iii), the EVENT can be focused as it was in (ii) (but not to grama), or o Petros can be FOCUS since making it such does not conflict with a TOPIC as in (ii).

36 Haidou uses underlining to mark “the most prominent stress in the sentence.” As is not infrequent in the literature, there is some disagreement about the data. Keller & Alexopoulou (2001.306) add these examples: (i) Pjos apelis-e ti Maria? [who fired-3s the-acc Mary] ‘Who fired Mary?’

(ii) ??Tin apelis-e ti Maria O JANIS [her fired-3s the-acc Mary the-nom John] ‘John fired Mary’ where (ii) is exactly parallel to (15b), but judged questionable. FOCUS: First and Last 25

Notice the presence of the clitic tin reflecting the Patient tin turta ‘the cake’, a pattern repeated from (8b) and (11) where a non-Agent functions as TOPIC. Given the question of (15a), it would appear that ‘ate the cake’ is the TOPIC continuing from the question into the response, and that the FOCUS in the response is ‘John’. The utterance of (15b) is then composed entirely of a TOPIC, Tin efage tin turta, followed by a FOCUS, o Janis. We turn our attention now to the FOCUS initial, V-initial morphosyntax. Wh-questions “Are those interrogative sentences formed with one of a set of question words, many of which share as a formal element the initial element p- ... The question word is typically fronted to the beginning of the main clause” (Joseph & Philippaki-Warburton 1987.5).37 Thus (Tsimpli 1998.197 and Roussou & Tsimpli 2006.318):38

(16) (a) Pjon idhe o Stefanos? [whom saw-3s the-nom Stefanos] ‘Whom did Stefanos see?’

(b) TI MARIA idhe o Stefanos [the-acc Maria saw-3s the-nom Stefanos] ‘It was Maria that Stefanos saw’

(17) (a) Pjos episkevase ton ipolojisti su? [who repaired-3s the-acc computer your] ‘Who repaired your computer?’

37 “There is no special intonation associated with such [question-word] questions, though the yes-no question rising-falling intonation can occur in question-word questions” (Joseph & Philippaki-Warburton 1987.6).

38 Tsimpli actually has these utterances paired: (i) (a) Pjon idhe o Stefanos? [whom saw-3s the-nom Stefanos] ‘Whom did Stefanos see?’

(b) TON STEFANO idhe i Maria [the-acc Stefano saw-3s the-nom Maria] ‘It was Stefano that Maria saw’ I have taken the liberty of swapping Maria and Stefano(s) so that the answer is appropriate to the question. 26 SYNTAX & SEMANTICS

(b) O Janis episkevase ton ipolojisti mu [the-nom John repaired-3s the-acc computer mine] ‘John repaired my computer’

Georgiafentis & Sfaniaki (2004.951) offer this as one of the conclusions to their experimental investigation into word order in Modern Greek. It pertains to how wh-questions are answered :39

The second major finding of our experiments is that the order used by the subjects in answering questions designed to elicit subject focus ... was not VOS, but rather SVO or OclVS ....

Thus (Georgiafentis & Sfakianaki 2004.951):

(18) (a) Pjos dhjavaz-i to vivlio? [who read-3s the-acc book] ‘Who’s reading the book?’

(b) O Janis dhjavaz-i to vivlio [the-nom John read-3s the-acc book] ‘John is reading the book’

(c) To vivlio to-dhjavaz-i o Janis [the-acc book it-read-3s the-nom John] ‘John is reading the book’

We have two ways to respond to a question. The first exploits the FOCUS initial, or VSO, resource. That is the answer found in (18b). When an O or S co-occurs with FOCUS in sentence-initial position it will bear the stress of FOCUS (and is sometimes notationally distinguished with upper case or the like). The second way to respond is to placte a TOPIC initially — To vivlio todhjavazi — and then follow that with a FOCUS — o Janis. Tsimpli (1998.215) provides this contrast that adds some detail to the semantic composition of Modern Greek FOCUS:40

39 Others concur that in response to a question such as (18a), that an Agent initial utterance with strong, focal stress is the “most acceptable pattern” (Georgiafentis & Sfakianaki 2004.951).

40 We noted above that Tsimpli (in this same article) has described a “‘default’ sentence stress” on the final constituent. In (19a), that would be o Jani. She (1998.204) also says “If this is correct, then the possibility of the contrastive interpretation is also available here due FOCUS: First and Last 27

(19) (a) Idhes TO JANI? [see-3s the-acc John] ‘Did you see John?’

(b) TO JANI idhes? [the-acc John see-3s] ‘Is it John that you saw?’

“... whereas in ... [(19a), with the default sentence-final stress, PWD] the yes/no question ranges over the whole proposition, in ... [(19b)] the presupposition is that you saw someone and the question operator refers to the individual variable identified by the focus operator, namely Jani” (Tsimpli 1998.215). This is clearly the now familiar semantics of FOCUS.41 In addition to the default intonation pattern associated with the in-situ object.” That is, the upper case O JANI in (19a) is not phonologically distinct from the default sentence stress. Tsimpli (1998.215) reports in a footnote that Philippaki-Warburton has said that (19a) can have an alternative intonation “starting low and rising on JANI without any significant fall at the end.” In this case, (19a) “can also have the reading of” (19b). My guess is that (19a) should probably be written without upper case TO JANI if the notation is used consistently. Georgiafentis (2001.138-139) cites this contrasting pair: (i) Efage tin turta o Janis [ate-3s the-acc cake the-nom John] ‘John ate the cake’

(ii) Efage tin turta O JANIS [ate-3s the-acc cake the-nom John] ‘JOHN ate the cake (not Thanassis)’ ‘It was JOHN that ate the cake’ Although these are VOS utterances, the facts of prosody may carry over. In sentence (i), o Janis “receives the main prominence” by virtue of the default sentence-final stress, an “information focus interpretation” (Georgiafentis 2001.152). There is no pause before the S as there can be (cf. below). In (19a), Georgiafentis’ “information focus” “ranges over the whole proposition,” and so does not single out to Jani. In (ii), o Janis is “contrastively focused.” “Contrastive focus” appears to exist only in sentence-final position where the default stress would otherwise be. I have found no indication in the literature of a contrast between focal stress and contrastive stress in any other position, although occasionally sentence-initial FOCUS will be called “contrastive”, semantically, not phonologially. If this distribution is true, then Modern Greek maintains a consistent two-way contrast between the presence of FOCUS and its absence. “Contrastive focus” — meaning the pronunciation transcribed by the UPPER CASE in (ii) — is just how Modern Greek distinguishes between default stress and something more.

41 Of (i), with focal stress on ton Petro: 28 SYNTAX & SEMANTICS

to combining the stress of FOCUS with one of the actants in sentence-initial position, it is also possible to place the stress of FOCUS on the EVENT/Verb itself (Joseph & Philappaki-Warburton 1987.98):

(20) Léne pos o Jánis dhen théli [say-3pl comp the-nom John neg want-3s na fígi prt leave-3s] ‘They say that John does not want to leave’

Sentence (20) “has several different implications depending on which constituent is emphatically stressed. With emphatic stress on léne, the implication is that this is only a rumour ....”42 The second response to (18) exploits a TOPIC-FOCUS organization in the manner of (15b) above. That is the answer of (18c).43 Georgiafentis (2005.165) provides another example which supports interpreting Tin efage tin turta in (15b) and To vivlio to-dhjavazi in (18c) as TOPIC:

(i) Ton Petro idh-a [the-acc Peter saw-1s] ‘I saw Peter’

Tzanidaki (1998.243) reports these two “readings” of FOCUS: (ii) It was not John I saw; it was Peter. (iii) I did indeed see Peter (though you did not expect me to).

42 Alexiadou (1999b.55) has this example:

(i) KSERI Ispaniki i Meropi [know-3s Spanish the-nom Meropi] in which “the verb is focalized.” In Alexiadou 2000, the example is (128):

(ii) KSERI o Janis Germanika [know.3s the-nom John German]

43 Notice that the Patient in (18c) precedes the verb whereas in (15b), it follows. Haidou calls tin a “clitic” (as do Joseph & Philippaki-Warburton 1987.211-215) and separates it from the verb in his notation. Philippaki-Warburton et al. (2004.963) argue that “clitics are not affixes but full words.” Others adjoin clitics to the verb with a hyphen. I have not regularized this difference. The formal contrast is that to is 3rd singular accusative neuter, and tin is 3rd singular accusative feminine. FOCUS: First and Last 29

(21) O Janis tin efage tin turta [the-nom John it ate-3s the-acc cake] ‘What John did was eat the cake’

In (21), “a weak pronoun (a clitic), i.e. a grammatical element, has been added. The result is that the object tin turta ‘the cake’ can no longer be accented and thus focused ... The verb efage ‘ate’ will receive the main accent [the default, sentence-final stress, PWD] and thus be the focus ... The object tin turta ‘the cake’ can only be old information ....”44 One must assume that on the model of (13a), that (22) is also possible:

(22) Tin turta, o Janis tin efage [the-acc cake the-nom John it ate-3s] ‘What John did was eat the cake’

Sentences (21) and (22) appear to be composed enitrely of TOPIC and FOCUS: TOPIC + FOCUS + TOPIC & TOPIC + TOPIC + FOCUS. It is unknown what the semantic difference might be. They are variations of the TOPIC + FOCUS utterances (9c) and (15b) above.45 While the sentence-initial contents that are focused are often written with uppercase, no one ever seems to write PJOS ‘Who?’, and no one comments on it.46 The clear implication is that the wh-question words do not carry the stress of FOCUS. Since Joseph & Philippaki-Warburton (1987.98) assert “an emphatic [i.e. focused, PWD] element is always accompanied witht emphatic stress,” we must wonder whether the wh-question words themselves express FOCUS, or to what degree. Tsimpli (1995.192) observes that “single clause sentences cannot contain both a wh-phrase and a focus phrase, irrespective of whether any of them is preposed or in situ.” A “focus phrase” here is one that bears a nondefault stress (or if sentence-final, then a stress that is “emphatic”). Wh-words appear occupy a middle ground on a scale of FOCUS. They do not themselves occur with the stress of FOCUS, but they occur in the FOCUS

44 FOCUS is generally incompatible with the preverbal clitic (Tsimpli 1995.179):

(i) *TO JANI ton sinantisa xtes [the-acc John him met-1s yesterday] ‘It is John I met yesterday’

45 We speculate on this in Chapter 10, section 3.2.

46 Recall the earlier observation that the presence of non-default stress is marked in various ways and, seemingly, not consistently. 30 SYNTAX & SEMANTICS

position and do not permit FOCUS to appear elsewhere. (Recall from above that utterances do not allow more than one FOCUS.) Modern Greek has two partially overlapping exploitations of syntax. The first is based on a syntactic verb-initial order that is semantically FOCUS initial. This syntax places non-EVENT content before the verb and marks it with strong stress in order to signal FOCUS. The second morphosyntactic pattern places TOPIC sentence-initially and signals it as TOPIC by not accompanying it with strong stress. The presence of two distinct constructions one marking TOPIC and a second that is appropriate to the absence of a continuing content and which marks FOCUS is repeated in Yogad (Cf. Chapter 14). Like Modern Greek, Yogad appears to be VSO and to use its sentence- initial position to signal FOCUS, and like Modern Greek, Yogad has a second construction X ay Y, which is used to track content which continues from the preceding contenxt.47 It is TOPIC. In the implementation of FOCUS and TOPIC, languages may ensconce both within a single morphosyntactic complex (like Bella Coola. Cf. Chapters 2 & 12), or languages may factor them into two or more syntaxes. As do Modern Greek and Yogad. Our initial assertion that a verb-initial language will also be FOCUS initial appears to be supported by Modern Greek. In semantic terms, the VSO syntax is FOCUS initial. The semantic association between sentence-initial FOCUS and the EVENT in Modern Greek, which produces the impression of VSO, is further amplified by usages which disallow a VSO or a VS expression (Alexiadou 1996.36, 42)

(23) (a) *Agapai o Janis ti Maria [loves-3s the-nom John the-acc Maria] ‘John loves Mary’

(b) O Janis agapai ti Maria [the-nom John loves-3s the-acc Maria] ‘John loves Mary’

(c) Agapai o Janis ti Maria [loves-3s the-nom John the-acc Maria]

47 Modern Greek differs from Yogad in that it has a morphosyntactic possibility that is VOS. On the surface, Modern Greek and Palauan would appear to be similarly organized. The VOS order has elicited some special attention from those working on Modern Greek (Alexiadou 1999b, Haidou 2000, Georgiafentis 2001, Georgiafentis 2005, Georgiafentis & Sfakianaki 2004). Because FOCUS in the VOS pattern is exclusively in situ, it will be discussed Chapter 10, section 3.2 FOCUS: First and Last 31

‘John fell in love with Mary’

(24) (a) *Epeze o Janis [play-imp-3s the-nom John] ‘John was playing’

(b) O Janis epeze48 [the-nom John play-pst-imp-3s] ‘John was playing’

(c) Epekse o Janis [played-perf-3s the-nom John] ‘John played’

(25) (a) *Kinigai i gata pondikja [chases-3s the-nom cat mice-acc] ‘Cats chase mice’

(b) I gata kinigai pondikja [the-nom cat chases-3s mice] ‘Cats chase mice’

(c) Kinigai i gata pondikja [chases-3s the-nom cat mice-acc] ‘The cat chases mice’

What we are seeing here is the semantics of FOCUS refracted through the prism of EVENT semantics. The following is an explication for how that is and why the (a)-sentences of (23) - (25) fail. We begin with a preliminary rationale for why EVENTS/Verbs, more than any other semantics, should have an affiliation with FOCUS. When human intelligence encounters human experience, the interaction will not be homogeneous. Some experience will inevitably be more abrupt, different, exciting than other experience. Differentiation in experience is mirrored by differentiation in eliciting the orienting reflex.49 The variety of the encounter will be unending, but when it

48 This expression is not contained in Alexiadou 1996, but I assume that — given the way things are going — that it is acceptable.

49 The orienting (startle) reflex is a “response to novelty” (Hinde 1970.131.ff). The response 32 SYNTAX & SEMANTICS is present and encoded into language (as it is and will be),50 the human response is to direct the listener, “Heed!” In some languages, the EVENT is, by that fact alone, able to satisfy the animus of FOCUS. In others, namely Modern Greek, some EVENTS will, in themselves, fail to achieve the “novelty” required to support FOCUS. There is not enough happening to justify ‘Heed!’ Thus in (23a), agapai ‘loves’ is so ongoing and without noticeable change that it fails as FOCUS. But if the sense is ‘fell in love’ as in (23c), then a change has been effected that is sufficiently abrupt to support FOCUS ... and then the sentence-initial position of FOCUS is available to it. Alternatively, (23b), which avoids the syntax of sentence-initial FOCUS will suffice. In (24a), there is more kinesis, but it is undirected and unshaped, like a swarm of gnats. But when the EVENT is formed and made abrupt by a conclusion (the perfective aspect), the congealed act stands out sufficiently against its background to satisfy the noticeability requirement of FOCUS. The same bounding shaping can be achieved by introducing a qualifying space (Alexiadou 1996.45):51

(26) (d) Edo pezun pedja [here play-3pl children] ‘Children are playing here’

In (25a), the lack of formation is manifest in the generic sense, i.e. cats, attributed to it. But it the sense is more specific, i.e. the cat, as in (25c), then the EVENT is more precise and noticeable, and the utterance succeeds. Reversing the order in (25b), allows the generic meaning because sentence- initial FOCUS no longer places any demands on kinigai. Finally, we find (Alexiadou 1996.45):

is generally, and for this reason this class of behavior has been also called fear (Hinde 1970.349): “Fear ... is elicited by stimuli that are strange, novel, or surprising, and is associated with fixation on the source.”

50 The ontogenesis is described in Chapter 10.

51 But if the space is itself too general to bound the event, it will continue to fail: (i) *se spilies pezun pedja [in caves play children] ‘Children are playing in caves’ The vague “in caves” fails. One must assume that if the phrase were ‘in those caves’, then (i) would pass muster. FOCUS: First and Last 33

(27) (a) *Kseri o Janis ti Maria [klnows-3s the-nom John the-acc Maria] ‘John knows Mary’

(b) KSERI o Janis ti Maria [klnows-3s the-nom John the-acc Maria] ‘John knows Mary’ in which FOCUS reveals itself directly as the facilitator. If the EVENT is by itself so languid that it does not support sentence-intial position, just pump it up with strong stress. In complementary fashion, Holt, Mackridge & Philippaki-Warburton (1997.427) identify “various factors which contribute to the choice of sentences where the verb is placed before the subject and object ... [among which is] The verb is one of appearing, entering, being, coming, beginning, etc.” These are EVENTS which exhibit transitions. A change is effected, which in turn supports the novelty that prompts FOCUS, i.e. sentence- initial position.52 One can, of course, never predict precisely how the nature of EVENTS will play out in ascending to sentence-initial FOCUS. But we should always be able after the fact to recognize the patterns. Returning to our motivation for examining Modern Greek, the conclusion is that yes, to the degree that it makes sense to say that the language is VSO, it is FOCUS initial.

3.1.2 Hdi

3.2 FOCUS Initial Languages that are SOV

3.2.1 Somali53

Somali (af soomaali) is spoken by more than four million people in the Horn of Africa. This includes all the citizens of the Somali Democratic Republic (Somalia), probably a majority of Djibouti citizens, and substantial minorities in Ethiopia and Kenya. It is the official language of Somalia and is the sole medium throughout Somali society, including schools, local and national administration ... Genetically, Somali is an East Cushitic language ... Somali has three basic dialect

52 There is no example of “being.” It is the odd one on the list and is probably not ινε.

53 The comments here depend on the following work: Andrzejewski 1975, Antinucci 1980, Gebert 1986, Hetzron 1965, Lecarme 1991, 1994 & 1999, Livnat 1983 & 1984, Saeed 1984, 1987 & 1999, and Svolacchia, Mereu & Puglielli 1995. 34 SYNTAX & SEMANTICS

groups: Common (Northern), Central, and Benaadir (Coastal) ... Common Somali is the most widespread and prestigious of the dialects and was a lingua franca among dialects in earlier times. Now, with some minor influences from the Benaadir dialect of the capital Mogadishu, it has become the national standard in Somalia. (Saeed 1984.11-13)

Somali is probably an SOV language:

Somali is usually said to have a word order of Subject-Object-Verb (S-O-V) ... However, accepting this basic order of S-O-V, a survey of the examples given in this study [Saeed 1984] will clearly reveal that in a great many cases this is not the surface word order; and this fact would be true of any corpus of Somali sentences ... much of the divergence away from S-O-V word order is caused by focus and topic structures.” (Saeed 1984.35-36)

The verbal piece is the basic element of a sentence and consists obligatorily of a verb and, optionally, or verbal pronouns, locative, and deictic prepositions The order of these elements within the verbal piece is fixed regardless of the order of any elements, noun phrases etc., outside the verbal piece; and is as follows:

subject — object — locatives — deictics — verb pronoun pronoun

(Saeed 1987.200)

Somali is a free word order language, though the preferred, unmarked order is probably SOV. (Livnat 1983.90 & 1984.6)

Le Somali est une langue OV qui se charactérise, parmi les autres langue cuchitiques, par l’extrême liberté de l’ordre de mots (en ce qui concerne tout au moins les NPs argumentaux). (Lecarme 1994.26)

Word order in clauses seems to be remarkably free. (Lecarme 1999.278)

3.2.1.1 ASSERTION While grammatically correct, the SOV (a)-members of (1) - (3) (Saeed 1984.23, 84, 91, 97, 126, 171 & 178, Saeed 1987.206, 212, and Saeed 1999.170) fail as complete ASSERTIONS:54

54 These orthographic choices are standard for Somali (Saeed 1987.15): c for a voiced pharyngeal fricative, dh for a voiced post-alveolar retroflex stop, and kh for a voiceless velar or uvular fricative. Somali is a tone language with “three basic tones, using two levels of pitch: HIGH tone ... e.g. á ... LOW tone ... left unmarked e.g. a ... FALLING tone e.g. àa” (Saeed 1987.20-21). Not every author marks tone in their examples. I have cited them as I found them. FOCUS: First and Last 35

(1) (a) Niman-ka baabuur-ka wata55 [men-the car-the drive] ‘The men who drive the truck’

(b) Nín-kii bàa gaadhígíi cusbáa watáy [man-the INDPART car.the new drove] ‘The man drove the new car’

(2) (a) Niman-kii lacag-tii xaday [men-the money-the stole] ‘The men who stole the money

(b) Cali lacag-tii wuu keenay [Ali money-the INDPART brought] ‘Ali brought the money’

(3) (a) Baabuur-kii i dhaafay [truck-the me passed] ‘The truck that passed me’

(b) Baabuur-kii baa i dhaafay [truck-the INDPART me passed] ‘The truck passed me’

(c) Baabuur-kii wuu i dhaafay [truck-the INDPART me passed] ‘The truck passed me’

While the (a)-utterances fail as ASSERTIONS, they succeed as NAMES. To achieve the status of a free standing ASSERTIONS, the (a)-utterances require additional morphosyntax, namely the material in bold italics. The same is true of these intransitives (Saeed 1999.170, 1984.69, 105 and Saeed 1984.101- 102):

55 Saeed 1987.63 has this version: (i) Nimánka baabùurka wadá [the men the truck drive] ‘The men who drive the truck’ 36 SYNTAX & SEMANTICS

(4) (a) Nin wanaagaan [man good] ‘A man who is good’

(b) Nin-ka ayaa wanaagsán [man.the INDPART good] ‘The man is good’

(c) W-uu wanaagsanyay [INDPART-he good] ‘He is good’

(5) (a) Nin-kii yimi [man-the came] ‘The man who came’

(b) Isaga yimi [he came] ‘He who came’

(c) Nimán-kii bàa yimi [men-the INDPART came] ‘The men came’

(d) Niman-kii ayàa yimi [men-the INDPART came] ‘The men came’

(e) Isaga wuu yimi [he INDPART came] ‘He came’

The three forms represented in (1) - (5) are baa, ayaa, and w-uu. In the tradition of Somali linguistics, these are called “Indicator Particles” (Andrzejewski 1975), and I have used “INDPART” as their grammatical gloss in the examples. Comparison of the respective (a)-sentences with the (b)- and (c)-sentences indicates that these morphemes are the carriers of the force of semantic ASSERTION. FOCUS: First and Last 37

En somali, tout énoncé déclaratif est caracterisé par l’usage complémentaire et obligatoire de baa/ayaa ou waa. (Lecarme 1991.34)

... either baa or waa ... must obligatorily occur in declarative main clauses. (Lecarme 1999.276)

EVENTS in isolation, e.g., yimi ‘came’ in (5a), by themselves fail to carry ASSERTION sufficient to complete a PROPOSITION.56 Before proceeding, we need to make three explanatory comments on the forms and their morphophonemics. First, baa and ayaa appear to be free alternates:

... in the present study baa and ayaa will continue to be treated as optional variants. (Saeed 1984.79)

The difference between baa and ayaa is mainly stylistic. In different environments one focus marker may be the speaker’s first choice, but the other is almost invariably accepted when suggested. The focus marker ayaa is slightly preferred in writing and in slow, deliberate and emphatic speech. When a speaker of Somali is asked to repeat a sentence with baa he has just uttered, very often he would repeat it using ayaa in stead of baa. (Livnat 1984.85)

J’admettrai ici que baa et ayaa sont en variation libre, et que ayaa inclut l’élément baa. (Lecarme 1991.34)

Baa and ayaa are held to be wholly equivalent apart from some stylisic differences, ayaa being a more formal, slow speech variant of baa. (Svolacchia, Mereu & Puglielli 1995.93)

Second, the Indicator Particle baa takes different shapes when inflected. There are two sets of Somali pronouns (Independent & Verbal) with one of them having a distinction between Subject and Object forms (Saeed 1984.158, 187.162-163). When a nominal (or Independent Pronoun) Subject is followed by an Indicator Particle, it will be sentence initial and the shape of the Particle will be baa.57 When a non-Subject is followed by baa, the complex will also

56 “Every indicative, affirmative sentence in Somali must contain one and only one indicator particle” (Livnat 1983.91). “... Somali uses the obligatory focus markers (F) [i.e., baa, ayaa, and waa, PWD] in every independent declarative sentence” (Gebert 1986.43). “Somali requires that in every main sentence a focus particle be present” (Svolacchia, Mereu & Puglielli 1995.66).

57 “When the indicator particle baa marks a subject in a sentence, it always occurs in its unconjugated form baa” (Livnat 1983.96). 38 SYNTAX & SEMANTICS

Independent Verbal Subject Object

‘I’ aniga aan i ‘you’ adiga aad ku ‘he’ isága uu — ‘she’ iyáda ay — ‘we (Incl)’ innága aynu ina ‘we (Excl)’ annága annu58 na ‘you (Pl)’ idínka aydín59 idín ‘they’ iyága ay —

Figure 3: Somali Pronouns. normally be sentence initial, and the Subject will follow (Gebert 1986.46):60

(6) Maryam b-uu Cali dilay [Maryam INDPART-he Ali beat] ‘Ali beat Maryam’

In that circumstance, baa is inflected, reflecting the person and number of the Subject. The form buu in (6) is the morphophonemic product of the third person singular verbal pronoun of Figure 3 following baa. When the Subject is not overtly expressed, that pronoun inflection is still present (Saeed 1984.79):

(7) Nin-k-uu dilayaa [man-the-INDPART.he is.beating] ‘He is beating the man’

Here, the sequence nin-ka + baa + uu reduces to ninkuu. The combination of baa plus the Verbal Subject Pronouns in Figure 3 produces the shapes of Figure 4 (Heztron 1965.118):

58 Saeed 1984.99 has aannu. Livnat 1983.96 has aanu.

59 Livnat 1983.96 has aydu.

60 “NPs focused by baa or ayaa occur leftmost within their sentence ...” (Saeed 1984.24). FOCUS: First and Last 39

Indicator Verbal Inflected Particle Subject Pronoun Indicator Particles

‘I’ baa aan ⇒ baan ‘you’ baa aad ⇒ baad ‘he’ baa uu ⇒ buu ‘she’ baa ay ⇒ bay ‘we (Incl)’ baa aynu ⇒ baynu ‘we (Excl)’ baa annu61 ⇒ baannu62 ‘you (Pl)’ baa aydín63 ⇒ baydin64 ‘they’ baa ay ⇒ bay

Figure 4: Somali Indicator Particle Plus Pronouns.

Third, the form wuu in (2b), (3c), and (4c) is morphophonemically parallel to buu in Figure 4. In utterances with a Verb as the EVENT, the Verb will be preceded by a Verbal Pronoun for the Subject, and for the Object, if present. Cf. Figure 3. They will appear between the Indicator Particle waa and the Verb. The Subject Verbal Pronoun will then combine with waa to produce forms parallel to the Inflected Indicator Particles in Figure 4 (Saeed 1984.101, 141 and Gebert 1986.44):

(8) (a) Cali w-uu Ø tiriyey [Ali INDPART-he it counted] ‘Ali counted it’

(b) Cali w-uu ku arkay [Ali INDPART-he you saw] ‘Ali saw you’

(c) Cabdi w-aan aqaan [Abdi INDPART-I know] ‘I know Abdi’

61 Saeed 1984.99 has aannu. Livnat 1983.96 has aanu.

62 Livnat (1983.96) has baanu.

63 Livnat 1983.96 has aydu.

64 Livnat 1983.96 has baydu. 40 SYNTAX & SEMANTICS

(d) Isagu w-uu dhintay [he INDPART-he died] ‘Him, he died’

The necessity of using one of the Indicator Particles in order to form a free standing declarative utterance suggests their absence is equally exploited. And it is, as we saw above in the expression of relative clauses. But the non-use of Indicator Particles goes beyond relative clauses to effect a variety of semantics where absence of ASSERTION is appropriate. In the formation of relative clauses, the initial position is the prominent one in establishing the component which the clause qualifies, i.e. the “head” (Saeed 1987.63)

(9) (a) Baabùur-ka nimán-ku wadàan [truck-the men-the drive] ‘the truck which the men drive’

(b) Nimán-ku baabùur-ka wadá [men-the truck-the drive] ‘the men who drive the truck’

Initial position continues to be important in the construction of other non- assertive expressions. The following illustrate some of those uses (Saeed 1999.224):

(10) (a) Mark-aan arkay (w-uu Ø kéenayay) (Saeed [time-I saw INDPART-he it bringing] 1987.65) ‘When I saw him (he was bringing it)’

(b) Islá márkíi uu yimíd (shàqùu bilaabay) (Saeed [ time he arrived began work] 1987.240) ‘As soon as he arrived (he began working)’

(c) Íntíi ay chashéynaysáy (saxìib-káy bàa sóo [while she dining friend-my INDPART here galay) (Saeed 1987.241) came] ‘While she was dining (my friend came in)’ FOCUS: First and Last 41

(d) Hadd-àad tagtó, (is jir) (Saeed 1987.243) [occasion-you go take care] ‘If you go (take care!)’

(e) Wax kast-ùu is-ku65 dayó (lacág b-ùu [thing each-he self-with tries money INDPART-he ká sameeya) (Saeed 1987.244) with makes] ‘Whatever he tries (he makes money at it)’

(f) Ín-uu imánayó (ay-áan ógahay) (Saeed [that-he is.coming INDPART-I know] 1999.234) ‘(I know) that he is coming’

(g) Xaggee-d tagtay? (Saeed 1999.202) [where-you went] ‘Where did you go?’

(h) Xaggee b-aad tagtay? (Saeed 1999.201) [where INDPART-you went] ‘Where did you go?’

(i) Kú shúb! (Saeed 1987.202) [in pour] ‘Pour it in!’

(j) Há tago!66 (Saeed 1987.203) [CLASSIFIER go] ‘May he go!’ ‘Let him go!’

65 “The combination of is and kú does not form a cluster, though the two words together are often written as if joined ... Together they convey a meaning of ‘joining, together, likeness’ ...” (Saeed 1987.195).

66 “... the minimum optative sentence is simply the verbal piece, or há with the verbal piece for the third person” (Saeed 1987.203). 42 SYNTAX & SEMANTICS

(k) Shòw kúu keenee67 (Saeed 1999.208) [CLASSIFIER for.you bring] ‘Perhaps he will bring it for you’

Another sentence-initial form wáxa(a) will play a role in the formation of FOCUS (cf. below). In (10a) - (10f), the initial element — like the sentence- initial head of the relative clause — deterines the character of the non- ASSERTION, which amplifies the co-occurring clauses (in parentheses). “They [i.e. the Indicator Particles] act thus as signals of completeness of the sentence and their absence, in certain structures, acts as a signal of dependence of a verbal form on the preceding noun or its equivalent” (Andrzejewski 1975.125). Sentences (10g) - (10i) show that absence of Indicator Particles are possible in independent utterances if they are not assertive. The Imperative (10h),68 Optative (10i), and Potential (10j) are marked by a “special verb form” (Saeed 1987.202, 203 & 204).

3.2.1.2 FOCUS We have seen that the Indicator Particles carry the semantic force of ASSERTION. We now discover that they simultaneously carry the semantics of FOCUS.69 Consider the following (Gebert 1986.48, Saeed 1984.24, 25, 28 & 125, 177-178, and Saeed 1999.192, 231):

(11) (a) Max-aa dhacay? [what-INDPART happened] ‘What happened?’

67 “Potential sentences ... usually occur with the word shòw, which we shall treat as the classifier for this sentence type. The minimal potential sentence consists of shòw with the verbal piece, which for these sentences does not contain subject verbal pronouns” (Saeed 1987.204).

68 “Note that, possibly as a result of the egalitarian nature of traditional Somali society, imperatives do not have the same associations of power and impoliteness as in English and are consequently much more commonly used” (Saeed 1987.73). This suggests diminished ASSERTION for the Imperatives as indicated by the missing Indicator particles.

69 Antinucci & Puglielli (1980.90) call this “the focus of assertion”. FOCUS: First and Last 43

(b) Cali baa Maryam dilay70 [Ali INDPART Maryam beat] ‘Ali beat Maryam’

(12) (a) Kuma ayaa kalluun-kii cunay? [who INDPART fish-the ate] ‘Who ate the fish?’

(b) Cali ayaa kalluun-kii cunay [who INDPART fish-the ate] ‘Ali ate the fish’

(13) (a) Y-aa Ø yeelay? [who-INDPART it did] ‘Who did it?’

(b) Axmed baa Ø yelay [Ahmed INDPART it did] ‘Ahmed did it’

(14) (a) Y-áa yimi? [who-INDPART came] ‘Who came?’

(b) Cali bàa yimi [Ali INDPART came] ‘Ali came’

(15) (a) Kuma ay-aad aragtay? [who INDPART-you see] ‘Who did you see?’

70 Confusingly, Svolacchia, Mereu & Puglielli (1995.91) have this utterance parallel to (11b) (i) Maryan baa Cali dilay [M. FM C. beat] ‘Ali beat MARYAM’ but with an odd gloss. Following the grammar of (11b), (i) should mean ‘MARYAM beat Ali’, and not the reverse. Recall that the uninflected Indicator Particle baa follows the Subject. 44 SYNTAX & SEMANTICS

(b) Amina ay-aan arkay [Amina INDPART-I saw] ‘I saw Amina’

(c) %Amina w-aan arkay [Amina INDPART-I saw] ‘I saw Amina’

(16) (a) Cali baa lacag-tii keenay, sow ma aha? [Ali INDPART money-the brought QM not be.NEG] ‘It was Ali who brought the money, wasn’t it?’

(b) Haa, Cali baa lacag-tii keenay [yes Ali INDPART money-the brought] ‘Yes, Ali brought the money’

(c) Haa, w-uu Ø keenay [yes INDPART-he it brought] ‘Yes, he brought it’

(d) %Haa, lacag-tii b-uu keenay [yes money-the INDPART-he brought] ‘Yes, he brought the money’

(17) (a) Cali lacag-tii b-uu keenay, sow ma aha? [Ali money-the INDPART-he brought QM not be.NEG] ‘Ali brought the money, didn’t he?’

(b) Haa, Cali lacag-tii b-uu keenay [yes Ali money-the INDPART-he brought] ‘Yes, Ali brought the money’

(c) % Haa, Cali baa lacag-tii keenay [yes Ali INDPART money-the brought] ‘Yes, Ali brought the money’

(18) (a) Wìi-kii má maragay? [boy-the QM vomit] ‘Did the boy vomit? FOCUS: First and Last 45

(b) Haa, w-ùu matagay [yes INDPART-he vomited] ‘Yes, he vomited’

(b) Haa, wìi-kii w-ùu matagay [yes boy-the INDPART-he vomited] ‘Yes, the boy vomited’

(d) %Haa, wìi-kii ayàa matagay [yes boy-the INDPART vomited] ‘Yes, the boy vomited’

Relying on the heuristic that responses to questions will point us to the morphosyntax of FOCUS, we find that baa, ayaa, and waa are the indicators of FOCUS as well as the bearers of ASSERTION. Although Somali appears to distribute ASSERTION/FOCUS either to the end of a PROPOSITION (if the EVENT is asserted/focused) or to the beginning (if some other content is asserted/focused),71 it shares significantly with Mupun (Chapter 10, section 3.3) and with Bella Coola (Chapter 3). Each language manifests ASSERTION and FOCUS as a portmanteau. They differ in fairly trivial ways, i.e., how the grammar of ASSERTION and FOCUS is expressed. Somali is as we have seen. Mupun likewise employs morphology as the carrier of ASSERTION/FOCUS, but then places ASSERTION/FOCUS at the desired location in a language that is SVO. Bella Coola restricts ASSERTION/FOCUS to sentence initial position in a language that is VSO. The content that is to be asserted/focused is then cycled through sentence initial position. Like Rwanda (Chapter 3), in Somali, content that is being introduced into the conversation requires FOCUS (Saeed 1984.170):

(19) (a) Baabuur baa i dhaffay [truck INDPART me passed] ‘A truck passed me’

(b) *Baabuur w-uu i dhaafay [truck INDPART-it me passed] ‘A truck passed me’

71 Somali will later reveal itself to be more generally cast as a TOPIC + FOCUS language. Cf. below. 46 SYNTAX & SEMANTICS

(a) Baabuur-kii baa i dhaffay [truck-the INDPART me passed] ‘The truck passed me’

(b) Baabuur-kii w-uu i dhaafay [truck INDPART-it me passed] ‘The truck passed me’

(20) Waxa jiray nin (Saeed 1984.30) [what existed man] ‘There was a man’

While the sentences of (19) use Syntagmatic FOCUS, sentence (20) exemplifies a Bipartite FOCUS expressed by a TOPIC + FOCUS grammar. Cf. below. Either is satisfactory to introduce new content.

3.2.1.3 TOPIC Although the combination of ASSERTION/FOCUS baa with a non-EVENT is said to appear in sentence-initial position,72 there is at least one consistent pattern in which that is not so. Some element(s) will precede the term qualified by baa/ayaa. Cf., for example, (17a) above in which Cali preceded the focused/asserted lacagtii buu.73 We begin to make sense of this by considering the question of (21a) and possible answers and non-answers (Gebert 1986.47):

(21) (a) Mux-uu Cali sameyay? [what-INDPART.he Ali do] ‘What did Ali do?’

(b) Cali Maryam b-uu dilay [Ali Maryam INDPART-he beat] ‘Ali beat Maryam’

72 “NPs focused by baa or ayaa occur leftmost within their sentence ...” (Saeed 1984.24).

73 Contrast the shape of the Indicator Particle b-uu in (17a) with the shape baa in (16a). In (17a), Cali itself is outside ROLE and VOICE semantics of the PROPOSITION, and the propositional AGENT is present formally as the Verbal Prouon uu. In (16a), the AGENT Cali is within ROLE and VOICE system of the PROPOSITION, but it is the FOCUS as well; hence, no Verbal Pronoun is present, and the shape of the Indicator Particle is baa. FOCUS: First and Last 47

(c) %Maryam b-uu Cali dilay [Maryam INDPART-he Ali beat] ‘Ali beat Maryam’

Sentences (21b) and (21c) are both acceptable Somali utterances, yet (21b) — with Cali before the focused/asserted Maryam — properly answers (21a), while (21c) does not. Both (21b) and (21c) place ASSERTION/FOCUS on the Patient Maryam so they must differ in some other semantics ... which appears to be TOPIC. Question (21a) mentions Cali, who continues into the answer as TOPIC. And to signal TOPIC, position to the left of FOCUS is employed.74 Position to the right of ASSERTION/FOCUS as in (21c) signals a lack of continuity that is not appropriate in response to (21a). The following from Gebert (1986.47), with a context analogous to (21), confirms the use of the pre-ASSERTION/FOCUS position to mark TOPIC:

(22) (a) Mar-kii Cali guriga u soo nogday, Maryam [time-the Ali house-the to here came Maryam jikadayku jirtay. kitchenin was] ‘When Ali came home, Maryam, was in the kitchen.

(b) Dabadeed Cali Maryam b-uu dilay [then Ali Maryam INDPART-he beat] ‘Then Ali beat Maryam’

(c) %Dabadeed Maryam b-uu Cali dilay [then Maryam INDPART-he Ali beat] ‘Then Ali beat Maryam’

Beginning with (22a), only the conclusion of (22b) succeeds. The conclusion in (22c) has Cali following the ASSERTION/FOCUS in a position that is not semantically compatible with continuing content. Conversely, if content is placed initially in the position of TOPIC, as Cali is in (23b), and if the context appropriate to TOPIC is missing, for example, the question (23a), then there can be no continuity, and TOPIC is misused. Neither (21b)/(23b) nor (23c) is an appropriate response to (23a) (Gebert 1986.48, 51):

74 As in Modern Greek. 48 SYNTAX & SEMANTICS

(23) (a) Max-aa dhacay? [what-INDPART happened] ‘What happened?’

(b) %Cali Maryam b-uu dilay [Ali Maryam INDPART-he beat] ‘Ali beat Maryam’

(c) %Maryam Cali baa dilay [Maryam Ali INDPART beat] ‘Ali beat Maryam’

It is a sentence like (23d) that answers (23a):

(23) (d) Cali baa Maryam dilay [Ali INDPART Maryam beat] ‘Ali beat Maryam’

The candidate answers (23b) and (23d) share an SOV order, and the difference between them lies in the absence of pre-ASSERTION/FOCUS content in (23d), i.e., the Cali of (23b). Although I do not find this in the literature, I assume that (23e) would also be acceptable in answer to (23a):

(23) (e) Maryam b-uu Cali dilay [Maryam INDPART-he Ali beat] ‘Ali beat Maryam’

The essential thing is that there is no TOPIC content expressed where none is justified. Pre-ASSERTION/FOCUS position can be used to establish TOPICS as well as to note continuing ones. Context such as Time can be a self-establishing TOPIC (Gebert 1986.56):

(24) (a) Max-aa dhacay? [what-INDPART happened] ‘What happened?’

(b) Shaley Cali baa dhintay [yesterday Ali INDPART died] FOCUS: First and Last 49

‘Ali died yesterday’

And Saeed (1984.31) adds examples of this sort:75

(25) Suuqa, hilib geelku aad b-uu [market-the meat camel very INDPART-it qaalisan yahay expensive is] ‘The market, camel’s meat is very expensive’

There is one other usage in which non-asserted/focused terms precede ASSERTION/FOCUS, namely, waa. Appropriate to the pre-ASSERTION/FOCUS position as a marker of TOPIC, no content which is not established in the context may appear before waa (Saeed 1984.34):

(26) (a) Lacag-tii, w-aan helay [money-the INDPART-I found] ‘The money, I found it’

(b) *Lacag, w-aan helay [money INDPART-I found] ‘Some money, I found it’

Gebert (1986.50) adds this example:

(27) Cali Maryam w-uu dilay [Ali Maryam INDPART-he beat]

75 The comma following a sentence-initial TOPIC is an inconsistent mark: (i) Nin-kii, w-uu yimi (Saeed 1984.34) [man-the INDPART-he came] ‘The man, he came’ (ii) Wii-kii w-uu matagay (Saeed 1999.231) [boy-the INDPART-he vomited] ‘The boy vomited’

Both nan-kii ‘the man’ and wii-kii ‘the boy’ are TOPICS in (i) and (ii). If they are not TOPICS, then the Verbal Pronoun is absent from the Indicator Particle waa (Saeed 1984.172): (iii) Baabuur-kii waa yimi [truck-the INDPART came] ‘The truck came’ 50 SYNTAX & SEMANTICS

‘Ali beat Maryam’

Gebert (1986.50) evaluates (27) as follows:

... [(27)], with the focus on the verb, requires a context in which both NPs stand for given information ... Therefore ... [(27)] is to be considered a marked pattern with quite a limited distribution, used in specific contexts ... In fact, Somali informants are rather reluctant to accept patterns represented for ... [(27)] since it is difficult to create an approriate context for them.

Saeed (1984.177-178) provides such a context:

(28) (a) Cali lacag-tii b-uu keenay, sow ma aha? [Ali money-the INDPART-he brought QM not be.NEG] ‘Ali brought the money, didn’t he?’

(b) Haa, Cali lacag-tii w-uu keenay [yes Ali money-the INDPART-he brought] ‘Yes, Ali brought the money’

Sentence (28b) is also a response to (29):

(29) Cali lacag-tii w-uu keenay, sow ma aha? [Ali money-the INDPART-he brought QM not be.NEG] ‘Ali brought the money, didn’t he?’

In (28b), both Cali and lacagtii are repeated from the preceding utterance, and like Cali in (21b), both function as TOPIC.76 This meshes with Gebert’s statement that “both NPs stand for given information”. Post-verbal position has a function similar to the sentence-initial, pre- ASSERTION/FOCUS TOPIC position (Gebert 1986.54):

(30) Y-aa Maryam dilay? [who-INDPART Maryam beat]

76 Saeed (1984.170) confirms this: “... this means that waa does not occur with indefinite NPs.” Sentence (29), however, is left without an explanation. How are both Cali and lacagtii the TOPICS the grammar says they are? Saeed does not remark on this, but I suspect the explanation lies in sow ma aha, that is glossed as a tag question ‘Is it not so?’ (Saeed 1999.205). In (29), sow ma aha suggests a context in which Cali and lacagtii are already present and continue into the tag query of (29). FOCUS: First and Last 51

‘Who beat Maryam?’

(31) Cali baa Maryam dilay [Ali INDPART Maryam beat] ‘Ali beat Maryam’

(32) Maryam Cali baa dilay [Maryam Ali INDPART beat] ‘Ali beat Maryam’

(33) Cali baa dilay Maryam [Ali INDPART beat Maryam] ‘Ali beat Maryam’

When content has been mentioned in the preceding utterance, it has an option in addition to appearing in initial TOPIC position as in (31). Sentence (33) shows that it may also appear after the EVENT. As in the case of TOPIC, this option is denied when the content is not continuing from the preceding utterance (Gebert 1986.53):

(34) Max-aa dhacay? [what-INDPART happened] ‘What happened?’

(35) %Cali baa dilay Maryam [Ali INDPART beat Maryam] ‘Ali beat Maryam’

Now, (33)/(35) is not a possible answer, nor is (32); only (31) will do. Saeed (1984.33) recognizes the post-Verbal expressions as “‘afterthought’ topics”:

These ‘afterthought’ topics ... share many of the characteristics of topics proper. They may, for example, be separated by a pause from the sentence. They also play no grammatical role in the sentence; and thus like topics may always be deleted leaving a grammatical sentence.

“Afterthought” topics and TOPICS have an unexplored interplay (Saeed 1984.26, 41): 52 SYNTAX & SEMANTICS

(36) Cali mux-uu ku sameeyey kalluun-kii? [Ali what-INDPART.he with did fish-the] ‘What did Ali do with the fish?’

(37) Kalluun-kii Cali waa cunay [fish-the Ali INDPART ate] ‘Ali ate the fish’

Kalluunkii is “afterthought” in (36), but TOPIC in (37).

3.2.1.4 FOCUS of Singularity There is still more to waa. Consider this exchange (Gebert 1986.60)

(38) (a) Cali ma yimid? [Ali Q come] ‘Did Ali come?

(b) Haa, w-uu yimid [yes INDPART-he came] ‘Yes, he did’

(c) Ma yimid? [Q come] ‘Did he?’

(d) Haa, waa yimid [yes INDPART came] ‘Yes, [he] did’

(e) %Haa, w-uu yimid [yes INDPART-he came] ‘Yes, he did’

The form waa can appear inflected for a subject as in (38b) or without inflection.77 The response in (38b) is itself questioned in (38c), asking for

77 The inflected shapes of waa are (Saeed 1984.55): 1st w-aan keen-ay ‘I brought it’ 2ndSg w-aad keen-tay FOCUS: First and Last 53 confirmation. Now, the first answer of (38b) will not suffice. The only response is (38d), in which there is no constituent save the EVENT, which bears the ASSERTION/FOCUS. There is no ROLE and VOICE, nor is there a TOPIC. The EVENT with waa in (38d) is a Singularity of ASSERTION/FOCUS, while the EVENT with w-uu in (38b) serves in a pattern of Syntagmatic FOCUS.78 Gebert (1986.60) comments:

... the truth value of the event expressed by the verb in ... [(38d)] is asserted with more intensity than in ... [(38b)]. In other words, ... [(38d)] appears as more marked for the focus of assertion on the verb than ... [(38b)].

Contrasts such as (38b) & (38d) and the following illustrate the extreme in the compaction of FOCUS (Gebert 1986.61):

(39) (a) Cali w-uu yimid [Ali INDPART-he came] ‘Ali came’

(b) Cali waa yimid [Ali INDPART came] ‘Ali came’

Gebert again comments on the contrast between the two:

It seems in that that ... [(39a)] can be uttered in a context announcing a new state of affairs; it is considered thus an unmarked sentence .... On the other hand ... [(39b)] can apparently be used as an announcement (confirmation)of an expected event [Emph mine, PWD]. This means, again that sentences such as ... [(39b)] are marked for assertion on the verb.

Even though there is a TOPIC in (39b), the isolation of the EVENT in combination with ASSERTION/FOCUS maintains its intensity. And of (40a) & (40b):

3rdSgMsc w-uu keen-ay 3rdSgFem w-ay keen-tay 1Pl w-aannu keen-nay 2Pl w-aydin keen-teen 3rdPl w-ay keen-een

The verb as well shows agreement with the subject.

78 Cf. section 2 above. 54 SYNTAX & SEMANTICS

(40) (a) Nin-kii w-uu yimid [man-the INDPART-he came] ‘The man came’

(b) Nin-kii waa yimid [man-the INDPART came] ‘The man came’

... a sentence like ... [(40b)] cannot be uttered in isolation, but has to be followed by some other sentence. It functions, then, as the beginning of a story ... where as ... [(40a)] can be uttered as an autonomous statement ... Obviously, these differences are sometimes very subtle and difficult to isolate. Nevertheless, what is interesting about these data is the fact that the loss of the privileged status of the subject is expressed by the absence of the subject pronoun on the F-marker following either a NP different from S, or the verb.

The “privileged status” is the presence of ninkii, i.e. -uu, as a ROLE in the ROLE/VOICE organization of the PROPOSITION in (40a). Sentence (40b) has no ROLE/VOICE organization. It seems that the Indicator Particle waa, uninflected, exceeds the inflected Indicator Particle w-uu in some dimension. Gebert’s term “announcement” is suggestive of the variety of these waa’s. In addition to describing the contrast between (39a) & (40b), “announcement” — a presenting or a setting forth — is adequate to describe the introduction of content that is (40b). The act of announcing, presenting, or setting forth is an isolated one ... Here! Announcing that “The man came” differs from telling/relating that the man came in that it invites explanation.79 Something is to follow in order to motivate the announcement.80 Given the property of

79 Conversational implicature?

80 A contrast between an inflected and an uninflected Indicator Particle b-uu and baa is possible with a semantic difference that compares with the difference between w-uu and waa (Hetztron 1965.122-123): ... the alternation between bàa and bùu after a non-subjectal element is possible only when an explicit nominal subject is present — in its absence, only the conjugated form can occur ... From the sometimes vague indications of the informants, the following tentative conclusions can be drawn: when buu is used, the connection of the emphasized element with both the preceding and the following sentences is closer, it serves as a link between them. On the other hand, the use of bàa implies an anticipated element, which, in the case of temporal and modal expressions, serves as a general ‘back-ground’ for the following sentence by disconnecting it from the preceding one. It marks a change of scene, a shift of the time-axis, a discontinuity. FOCUS: First and Last 55 announcing, it is consistent that waa “occurs only in sentences in the present tense” (Livnat 1984.98). Other examples of the FOCUS of Singularity are these (Saeed 1987.210- 212, 1999.188-189, 239-240)

(41) Waa nabád [INDPART peace] ‘Peace!’ [Reply to a greeting]

(42) Waa yahay [INDPART be] ‘Okay’

(43) Waa in-aad tagtó81 [INDPART that-you go] ‘You must go’

(44) Wàa dhakhtar [INDPART doctor] ‘He is a doctor’82

(45) Wàa gabdhó [INDPART girls] ‘They are girls’

(46) Waa kúma? [INDPART who] ‘Who is it?’

(47) Waa xaggée? [INDPART where] ‘Where is it?’ ‘Where are they?’

81 Sentence (35) asserts/focuses the nominalized ‘that you go’, and the implication of that content being set forth is that it is so, hence the ‘must’. Andrzejewski’s (1975.146) gloss for a similar example, ‘It is necessary (or obligatory) ...’ suggests that this is a deontic ‘must’.

82 “If the first NP in a ‘A waa B’ sentences [sic] is omitted, a third person subject, ‘it’ or ‘he, she, they’, is understood” (Saeed 1987.210). 56 SYNTAX & SEMANTICS

This expression can signal a literal setting forth (Saeed 1987.211):

(48) Wàa i kan [INDPART me this] ‘Here I am’

3.2.1.5 Waxa(a) and Paradigmatic FOCUS In summary to this point, it appears that Somali ASSERTION/FOCUS is initial in any utterance in which it appears, and it fails to be initial only when it is preceded by content that is TOPIC.83 That applies to ASSERTION/FOCUS when it combines with non-EVENTS, as well as when FOCUS/ASSERTION occurs with EVENTS. Somali is then structured as

TOPIC + ASSERTION/FOCUS84

83 Saeed (1999.229) states that the pre-FOCUS/ASSERTION material is unordered: “full NPs which are known or given ... occur freely in any order.” (i) Cali warqád-dii w-uu li dhiibay [Ali letter-the INDPART-he me.to passed] ‘Ali passed the letter to me’ (ii) Warqád-dii Cali w-uu li dhiibay [letter-the Ali INDPART-he me.to passed] ‘Ali passed the letter to me’

84 Matters in Somali are, however, a bit more complex than presented above. Consider, for example, this question (Gebert 1986.50): (i) Cali y-uu dilay? [Ali who-INDPART.he beat] ‘Who did Ali beat?’

We would expect that, because Cali precedes the grammar of ASSERTION/FOCUS, it would be TOPIC and in a following utterance, if Cali were present, it would continue to precede the grammar of FOCUS/ASSERTION. That is, Cali would continue as TOPIC, and (ii) would be the appropriate answer: (ii) Cali Maryam b-uu dilay [Ali Maryam INDPART-he beat] ‘Ali beat Maryam’ The response (ii), that we would expect, “is supported by the data” (Gebert 1986.51), but “In fact ... [(iii)] is a perfect answer to ... [(i)]” as well (Gebert 1986.51): (iii) Maryam b-uu Cali dilay [Maryam INDPART-he Ali beat] ‘Ali beat Maryam’ FOCUS: First and Last 57

We now encounter difficulty with Somali. Recall from above that the sense of a relative clause is affected by composing a normal utterance, but omitting the mark of ASSERTION/FOCUS, e.g.,

(49) (a) dukaan-kii niman-ku doonayaan (Saeed 1984.107) [shop-the men-the looking.for] ‘the shop the men are looking for’

(b) Akhri buugag-ga aan kuu keenay (Antinucci & [read books-the I you.to brought] Puglielli 1980.89) ‘Read the books that I brought to you!’

A mark of definiteness appears on dukaan ‘shop’, niman ‘men’, and buugag ‘books’ in (49), and on wax ‘thing’ in (50) & (51):

(50) Wix-ii baa jebay (Saeed 1984.59) [thing-the INDPART broke] ‘The thing broke’

(51) Wáx-a la íi sheegay ín-uu (Saeed 1987.237) [thing-the one me told that-he

For some reason, the question (i) is behaving as a context different from (21a). The context of (i) allows us to ignore the presence of Cali in the question when the answer is formed. The context of (21a) does not permit us to ignore Cali in the context. Sentence (i) questions the Patient, but if we question the Agent, we discover the same possibilities (Gebert 1986.51): (iv) Y-aa Maryam dilay? [who-INDPART Maryam beat] ‘Who beat Maryam?’ (v) Maryam Cali baa dilay [Maryam Ali INDPART beat] ‘Ali beat Maryam’ (vi) Cali baa Maryam dilay [Ali INDPART Maryam beat] ‘Ali beat Maryam’

A non-TOPIC looking Maryam is possible in (vi), as the non-TOPIC looking Ali was acceptable in (iii). It may be that dabadeed ‘then’ in (22b) and (22c) is operative in enforcing TOPIC continuity. As matters now stand, there are contexts in which TOPIC continuity must be maintained and those in which it may not be. 58 SYNTAX & SEMANTICS

tagó go] ‘I was told that he goes] [Lit. ‘The thing one told me is that he goes’]

(52) (a) Wáx b-aan akhrinayaa (Saeed 1999.238) [thing INDPART-I am.reading] ‘I am reading something’

(b) Wax ma dheceen? (Saeed 1984.59) [thing QUESTION happen] ‘Did anything happen?’

About the morphology of ‘definiteness’ Saeed (1987.150, 152) explains:

There are two definite articles, meaning ‘the’, and each occurs in a masculine and a feminine form, agreeing with the noun to which it is attached:

the first is ka (masc.), ta (feminine) the second is kíi (masc.), tíi (feminine)

The difference between the two types is one of remoteness in space and time from the speaker: if the noun is distant or in a sentence referring to the past, the article kíi/tíi will be used. If the noun is physically close to the speaker, or, in an expression of time, refers to the present or future, then ka/ta will be used .... there is no indefinite article ....

The shape -ku & -ga (41), -ii (50), and -a in (51) are normal morphonemic variants.85 The absence of a suffix in (52) expresses indefiniteness. We have also seen that in addition to the relative clause, there is a second kind of clause that lacks ASSERTION/FOCUS, but which does not have the sense of a relative clause. I repeat a couple of them here:

(53) Hadd-àad tagtó, is jir! (Saeed 1987.243) [occasion-you go take care] ‘If you go, take care!’

85 Among the morphophonemic rules are these: “k becomes g after g, w, y, or i ... k is dropped after h, x, q, c, kh ... t becomes d after all vowels and d, x ....” (Saeed 1987.151). FOCUS: First and Last 59

(54) Wax kast-ùu is-ku dayó lacág b-ùu [thing each-he self-with tries money INDPART-he ká sameeya (Saeed 1987.244) with makes] ‘Whatever he tries he makes money at it’

These clause initial terms appear to carry a mark of definiteness. For example, in (53) the initial word is “haddíi ‘the time, occasion’ ... which convey[s] the meaning ‘if’” (Saeed 1987.243). Haddíi seems composed of had ‘time’ and díi (< tíi Fem. Definite). In (54), the sequence is wax kastá plus the verbal pronoun uu. Kasta is composed of kas plus the Feminine Definite ta. The combination wax kas-ta-uu is realized as wax kastuu. The relative clauses of (49) are distinguished from those in (53) & (54) in two ways. First, the initial terms in the relative clause occupy a syntactic position which marks them as filling a ROLE in the PROPOSITION, e.g., Patient in (41). The clause initial terms of the second type do not fill a ROLE function. Second, the relative clauses themselves usually occupy a syntactic position within a larger clause that identifies the term they name as functioning as some ROLE in that more inclusive PROPOSITION. The clauses of the second sort accompany another PROPOSITION, but do not have a function within it. This leads us finally to our point:

(55) Wux-uu yidhi w-aan ka haqay (Saeed 1984.76) [thing-he said INDPART-I at feared] ‘I was afraid of the thing/what he said’

(56) Wúx-uu sheegay inaanu garánéyn (Saeed 1999.225) [what-he reported that.not.he understand.NEG] ‘He said that he didn’t understand’ [Lit. ‘What he said (was) that he did not understand’]

There seems to be a difference here in the two wax-clauses. In the first, wax identifies more concrete, specific content, whereas in the second, wax is more a vague semantic holder, to be made whole only when the second part of the utterance provides the substance, i.e. inaanu garaánéyn ‘that he didn’t understand’. The contrast is recognized in the literature by accepting an element waxa in Somali, distinct fom wax. Waxa has a range of glosses ‘a concrete thing, an object,’ ‘something, anything,’ ‘a person, people’ (Saeed 1984.58) and which is a lexicalised version of wax & ka: “The status of waxa 60 SYNTAX & SEMANTICS is clear ... [It] is the lexical item wax ‘thing’ plus the definite article -a ...” (Saeed 1984.42).86,87 The waxa-clauses appear only in the morphosyntax of the TOPIC + FOCUS sort,88 and are further confined to expressing the TOPIC. In Andrzejewski’s (1975.135) view:89

... I propose to introduce a new term and call them “heralding sentences.” The choice of the term is suggested by the fact that the indicator wáxa places emphasis on the noun or its equivalent which somes later in the sentence, thus heralding as it were, its arrival.

Andrzejewski’s “heralding” appears to be a fair recognition of the semantics of TOPIC. The following question and answer pairs demonstrate the FOCUS part of the TOPIC + FOCUS morphosyntax:

86 Saeed (1984.77) comments further: The translation of the lexical item wax-a (where -a is the definite article ‘the’) into English presents difficulties since its range of meaning corresponds to no one English word ...The closest approximation is ‘who, what’, but while ‘What brought it was a train.’ is possible, ‘Who brought it was me.’ is a little strange. For this reason two translations are usually given: waxa as ‘the one(s)’ in ‘The one(s) who VP be NP’; and a cleft version ‘It be NP who VP.’ (where the underlined NP is in focus).

87 There is also a contrast between waxa and waxaa when (at least) the waxa-clause specifies an intransitive subject: (i) Wáxa yimi nimán-kii (Saeed 1984.213) [what came men-the] ‘What came was the men’ (ii) Waxaa tagay Cali (Livnat 1984.87) [F left Ali] ‘Ali left’

The difference is unexplained, except for Livnat’s (1984.89) postulation that waxaa is actually wax+baa. Saeed (1999.194) notes “the word waxa (waxaa)” and then cites it (et passim) as “waxa(a)”.

88 I have found no examples that contradict this.

89 In Andrzejewski’s 1975 description, waxa is an Indicator particle as are baa, ayaa, and waa. Note that waxa is not always an Indicator Particle. Compare sentences (49) & (50). It is only in the morphosyntactic context of TOPIC + FOCUS that waxa might be an “Indicator”. Note also that if waxa is an Indicator Particle, this will produce utterances with two Indicators, whereas otherwise, the limit has been one per utterance. Cf. below. FOCUS: First and Last 61

(57) (a) Kuma ay-aad aragtay? (Saeed 1984.24) [who INDPART-you see] ‘Who do you see?’

(b) Wax-aan arkay Amina [who-I saw Amina] ‘Who I saw was Amina’

(c) Amina ay-aan arkay [Amina INDPART-I saw] ‘I saw Amina’

(58) (a) Kuma ayaa kalluun-kii cunay? (Saeed 1984.25) [who INDPART fish-the eat] ‘Who ate the fish?’

(b) Waxa kalluunn-kii cunay Cali [who fish-the ate Ali] ‘Who ate the fish was Ali’

(c) Cali baa kalluun-kii cunay [Ali INDPART fish-the ate] “Ali ate the fish’

The (b)-utterances of (57) & (58) express FOCUS as do the (c)-expressions, with the contrast between them being that the former employ Paradigmatic FOCUS and the latter, Syntagmatic FOCUS. Consistent with this Paradigmatic formation of FOCUS, Saeed (1987.213) observes. “wáxa only focuses noun phrases which follow its verb.”90

90 Cf. also Saeed 1999.194. Livnat’s (1984.89) solution to waxa syntax of TOPIC + FOCUS is this: I suggest that waxaa is not a different focus marker from baa but rather an instantiation of a structure containing baa which results from the application of syntactic and phonological rules. According to this hypothesis, in sentences with waxaa (or one of its conjugated forms) the focus marker baa follows a dummy NP, waxa (‘the thing’ from wax ‘thing’), and the focused NP is extraposed to sentence final position ... The form waxaa is phonologically derived from the combination of the word waxa and the focus marker baa. If baa is followed by a subject clitic, the result is one of the conjugated forms waxaan, waxaad, wuxuu, etc. 62 SYNTAX & SEMANTICS

The TOPIC + FOCUS expressions of (57b) - (58b) are unusual in two ways: (i) They appear to require a TOPIC built exclusively with wax (or waxa?) and (ii) The grammar of those utterances is odd in that they lack any morphological mark of ASSERTION/FOCUS. There is no baa/ayaa or waa in (57b) or (58b), yet they pass as acceptable independent utterances. It is not always unambiguous that we are looking at an example that has wax or one that is built on waxa, or whether there is a distinction:

(59) (a) Waxa yimi tareen (Saeed 1984.42) [thing came train] ‘What came was a train’

(b) Waxa yimi waa tareen [thing came INDPART train] ‘The thing that came was a train’

If this is so, then an underlying form such as wax+baa+aan contradicts the otherwise universal Somali pattern of allowing one ASSERTION/FOCUS per sentence. Cf. (52b). It violates Livnat’s earlier pronouncement: “Every indicative, affirmative sentence in Somali must contain one and only one indicator particle” (Livnat 1983.91). Another problem with this solution is that it relies on an abstract morphonemics that completely hides the element baa. It never reveals itself phonologically. There is no way to factor waxaad nonarbitrarily into wax + baa + aad or wax + aad. Saeed (1984.151-152) faces a similar problem with question words. For example: (i) Maxay b-aad sameynayssa? [what indpart-you do] ‘What are you doing?’ (ii) Max-aad sameynayssa? [what-you do] ‘What are you doing?’

In place of seeing the possibility of wh-questions with or without FOCUS as yes-no questions occur (Saeed 1987.219-220): (iii) Ma Ali bàa yimi? [QUESTION Ali INDPART come] ‘Did Ali come?’ (iv) Cali má yimi? [Ali QUESTION come] ‘Did Ali come?’

Saeed (1984.152) proposes an ad hoc looking solution: Wh-NPs not ending in a long vowel generally allow a phonological coallescense [sic] rule to combine the focus particle baa (and any pronoun) with the NP .... FOCUS: First and Last 63

(60) (a) Wáx-aan dóonayaa sháah (Saeed 1999.196) [thing-I want.PROG tea] ‘What I want is tea’

(b) Wáx-aan dóonayaa waa sháah [thing-I want.PROG INDPART tea] ‘The thing that I want is tea’

Saeed (1984.43, 50-60) develops an argument that the (a) & (b) pairs are related by a “waxa-cleft reduction rule,” so that (59a) is derived from (59b). In Saeed 1999, the derivational relation “is probably best seen as a historical process” (196). The TOPIC + FOCUS expressions with waa are “verbless waa sentences containing a relative clause” and the TOPIC + FOCUS without waa are are “clefts” (196). This would explain Saeed’s contrasting glosses for the pairs in (59) & (60), but it unfortunately moves the contrast between the members of the pairs from where it is (e.g., tareen vs. waa tareen) to where it is not (e.g., Waxa yimi).91 In spite of the homophony of the waxa yimi clauses in (59) and the like, Saeed persists detecting a contrast between them in which “the NP wáx-a in ... [(59b)] is reanalysed as a unit wáxa in ... [(59a)] and its head no longer participates in nominal morphosyntax” (Saeed 1999.196). I.e., (59a) no longer contains a relative clause like (59b) — a clause like those in (49) — but has an initial clause like the clauses in (53) & (54). You just cannot hear the difference.92

91 Saeed (1984.43) notes the problem of “the waxa’s being merely homophones,” which confirms that the contrast between the (a)- & (b)-pairs lies in the FOCUS portion of the utterance.

92 This may not be entirely accurate. Andrzejewski (1975.135) provides one example of a tonal contrast between the waxa clause and the relative clause, writing “Even in cases which give the impression of similarity[,] heralding sentences are always differentiated formally from other sentences”: (i) Wáx-ay tidhi anígu garán máayó [what-she said I understanding not.am] ‘What she said was ‘I don’t understand’” (ii) Wáx-ay tidhí anígu garán máayó [thing-she said I understanding not.am] ‘I don’t understand what [i.e. the thing that] she said’

Sentence (ii) contains a relative clause built upon wax ‘thing’, and (i), a waxa clause. Andrzejewski makes no more of the contrast than to note it, “It must be observed that the noun w˘ax (+ the def. art. gen.) frequently occurs as the headword of a nominal clause containing a dependent clause which may outwardly [!] resemble the first part of a heralding 64 SYNTAX & SEMANTICS

The helpful, missing datum would be to know more precisely what it means to say tareen rather than waa tareen in (59). There is a semantic contrast. One indication of the difference is that a waa FOCUS in a TOPIC + FOCUS expression is permitted when the TOPIC is other than one built on waxa:

(61) Wax-ani waa hub (Saeed 1984.43) [thing-this INDPART weapon] ‘This thing is a weapon’

(62) Meesh-aad tegin waa Marka (Saeed 1984.112) [place-you went INDPART Merca] ‘The place you went to is Merca’

(63) Lacag waa wax-aan doonayaa (Saeed 1984.136) [money INDPART what-I want] ‘Money is what I want’93

(64) Cali waa cidda uu raadinayo (Saeed 1984.141) [Ali INDPART person he looking.for] ‘Ali is the person he is looking for’

(65) Nimán-ka Ø kéenayaa waa askár (Saeed 1987.96) [men-the it bringing INDPART soldiers] ‘The men who are bringing it are soldiers’ sentence.” Saeed (1984.60), citing Andrzejewski, mentions the possibility of an audible contrast: “It has been claimed that these two items are ‘formally differentiated’ by tone patterns.” Other than to provide and to cite the one contrasting pair above, neither Andrzejewski nor Saeed pursues the issue. It is curious that a pattern so central to the grammar of Somali would not receive more attention. There is one additional mark of the contrast. In the waxa clause, there is agreement for gender (but not number [Saeed 1984.76] nor person-number [Saeed 1984.56]) between the TOPIC and the FOCUS (Saeed 1984.53): (ii) Waxa timi gabadh [who came.FEM girl] ‘Who came was a girl’ (iii) *Waxa yimi gabadh [who came.MASC girl] ‘Who came was a girl’

93 Saeed glosses (54) as ‘MONEY is what I want’, but that seems almost certainly a mistake. FOCUS: First and Last 65

TOPICS formed without waxa (e.g., those in [61] - [65]) may not appear without the ASSERTION/FOCUS marker waa. I.e.,

(66) *Wax-ani hub

This asymmetry indicates that the choice between sháah and waa sháah in (60), and the like, is a meaningful one.94 The constraint between the kinds of TOPIC that are compatible with the kinds of FOCUS suggests that the more specific ones of (61) - (65) require waa-FOCUS. Then, the waa-less FOCUS matches the less specific waxa clause TOPIC. Compare these two questions:

(67) Wáxa Ø keenáy áyo? (Saeed 1999.238) [ it brought who] ‘Who brought it?’ [Lit. ‘The one who brought it (was) who?’]

(68) Nin-kii ay raaceen wàa ayó? (Saeed [man-the they accompanied INDPART who] 1999.212) ‘Who is the man they travelled with?’

The open, unlimited Wáxa keenáy ‘Whoever bought it’ appears with ayo ‘Who?’ alone. The more limited, defined Ninkii ay raaceen ‘The man they travelled with’ needs waa to accompany ayo. Additional examples of waxa clauses as TOPIC are these:95

94 An additional observation that suggests a contrast is “the semantic restriction in ... [TOPIC + FOCUS expressions with waa] that the complement [FOCUS] be inanimate (to match wáx ‘thing’)” (Saeed 1999.196) This appears, however, to be contradicted by sentences such as (65) and also (Saeed 1984.51): (i) Cali waa askari [Ali INDPART soldier] ‘Ali is a soldier’ Unless (iii) is not, for some reason, a “verbless sentence”.

95 The waxa-clauses appear in the TOPIC + FOCUS morphosyntax of (69) - (75), but they also occur in other uses as naming expressions. “Note that waxa relative clauses are genuinely free relatives, and are not limited to equational sentences” (Saeed 1984.76): (i) Cali wúx-uu cabbó síi! (Saeed 1999.143) [Ali thing-he drinks give] ‘Give Ali something to drink!’ 66 SYNTAX & SEMANTICS

(69) Wax-aan aqaan in-uu imanayo (Saeed 1984.320) [what-I know that-he come] ‘What I know is that he is coming’

(70) Wáxa la íi sheegay ín-uu tagó (Saeed 1987.239) [what one me told that-he went] ‘I was told that he went’ [Lit. ‘What one told me (was) that he went’]

(71) Wáx-aan joojiyey in-aan mùus cunó (Saeed 1987.238) [what-I stop that-I bananas eat] ‘I stopped eating bananas’ [Lit. ‘What I stopped (was) that I eat bananas’]

(72) Wáx-aan joojiyey shaneemáda tagisteeda (Saeed 1987. 239) [what-I stop cinema going.its] ‘I stopped going to the cinema’ [Lit. ‘What I stopped (was) its going to the cinema’]

(73) Wáx-aan doonayaa lacag (Saeed 1984.42) [what-I want money] ‘What I want is some money’

(74) Wáx-aan raadinayaa niman-ka Ø kéenayá (Saeed [what-I searching.for men-the it bringing] 1999.96) ‘What I am searching for are the men who are bringing it’

(75) Waxaa la doortay goor-tii Cismaan (Saeed 1987.239) [what one chose time-the Cismaan la dilay one killed] ‘He was chosen when Osman was killed’ [Lit. ‘What one chose him the time one killed Osman’]

Sentences (69) - (72) show a nominalized form, e.g.,‘that he is coming’, as FOCUS. And the asymmetry suggests that the waxa clause as TOPIC in a TOPIC + FOCUS utterance encodes an expression of Paradigmatic FOCUS. FOCUS: First and Last 67

3.2.1.6 Conclusion Somali appears to fairly illustrate the typology suggested in section 2. Syntagmatic FOCUS is found in (76) (Saeed 1987.191):

(76) Nín-kíi b-àan cèel-ka xádhig ka-gá [man-the INDPART-I well-the rope with-from sóo saaray there pulled] ‘I pulled the man out of the well with a rope’

There is no TOPIC in (76), and there are other terms co-occurring with nankii within the frame of the ROLE & VOICE semantics of the PROPOSITION that might have been, but were not selected for ASSERTION/FOCUS. Bit by bit, the content that occurs between the initial FOCUS and the EVENT can be placed outside semantics of ROLE & VOICE, thus reducing the background against which FOCUS is made (Gebert 1986.59):

(77) (a) Cali w-uu ka shaqeeyaa warshad-da [Ali INDPART-he in works factory-the] ‘Ali works in the factory’

(b) Cali waa ka shaqeeyaa warshad-da [Ali INDPART in works factory-the] ‘Ali works in the factory’

In the (a)-sentence Cali and ka remain within the ROLE & VOICE organization of the PROPOSITION. Warshadda is outside that structure. In the (b)-sentence, Cali is also outside that structure. The successive removal from of content from eligibility for ASSERTION/FOCUS continues until there is no semantics of ROLE & VOICE, just TOPIC and ASSERTION/FOCUS:

(78) Nin-kii shanda-dii uu keenay waa tan (Saeed [man-the suitcase-the he brought INDPART this] 1984.195) ‘The man, the suitcase he brought is this one’

(79) Cali wux-uu doonayaa waa shaah (Saeed [Ali thing-he want INDPART tea] 1984.208) ‘Ali, the thing he wants is tea’ 68 SYNTAX & SEMANTICS

(80) In dhakhtar-kii yimi, war-kaas waa run (Saeed [that doctor-the came news-that INDPART truth 1984.213) ‘That the doctor came, that news is true’

And finally there is no TOPIC, just ASSERTION/FOCUS, a Singularity of FOCUS:

(81) Waa Júmce (Saeed 1999.188) [INDPART Friday] ‘It’s Friday’

(82) Waa hagàag (Saeed 1999.189) [INDPART straightness] ‘Okay’ [Lit. ‘It is straightness’]

Then, the mark of ASSERTION/FOCUS is removed from the PROPOSITION creating a symbiotic TOPIC-FOCUS, that expresses Paradigmatic FOCUS (Saeed 1999.196):

(83) Wáxa timi gabádh dhèer oo qurúx badán [what came.FEM girl tall and beauty much] ‘There came a tall, very beautiful girl’ [Lit. ‘What came was a tall, very beautiful girl]

Outside this morphosyntactic expression, Wáxa timi ‘What came’ is not possible, nor is gabádh dhèer oo qurúx badán ‘a tall, very beautiful girl’ possible. Here, Wáxa timi and gabádh dhèer oo qurúx badán are exist only in their co-occurrence, TOPIC-FOCUS. We must conclude finally that Somali is, as several have claimed, only an uncertain SOV language. In this context, our concern with SOV was/is the prediction that no S should be the preferred FOCUS. That is, in an SOV language, FOCUS should not be, like the S, placed in neutral fashion in initial position. There must be some other morphosyntactic expression for it. Somali seems not to challenge that prediction.

3.2.2 Boni

3.3 FOCUS Initial Languages that are SVO FOCUS: First and Last 69

3.4 FOCUS Initial Languages that are VOS

4. FOCUS Final Languages

5. FOCUS Crossover Languages There is a kind of language which uses word order to signal FOCUS, but which requires two positions to signal it exhaustively. Languages in this group are similar to Telugu in that they seem rely on the contrast between the semantics of the position and the semantics of of the content occupying that position. They use the opposite of Behagel’s First Law. In place of positioning like-with-like, they put like-with-unlike. The startle of the juxtaposition achieves the effect that is FOCUS, i.e. “Look here first!”

5.1 Kanakuru The published information on the grammar of Kanakuru is contained in Newman 1974:96 The Kanakuru language is spoken in north-eastern Nigeria along the Rivers Hawal and Gongola ... The peoples own name for themselves is ‘Dera’, but the term ‘Kanakuru’ has established itself as the standard designation for the group both informally throughout Nigeria and in the scientific literature abroad Kanakuru belongs to the Bole-Tangale cluster of the Plateau-Sahel branch of the Chadic language family. (Newman 1974.ix)

Green (2007.203) suggests that Kanakuru has “a focus position after the direct object ....” This appears to be so, but Kanakuru also has a FOCUS position before the Agent. Kanakuru is a SVO language (Newman 1974.16, 22, 23):97

(1) Ngoje a ko-no kom [Ngoje TP98 catch-me rat] ‘Ngoje caught me a rat’

96 There are other works on Kanakuru, but they draw on Newman 1974 as does this chapter: Samek-Lodivici 1998 & Tuller 1992.

97 Underlining indicates an implosive.

98 ”Tenses in Kanakuru are partially marked by ... pronouns that incorporate the underlying tense constituent as a feature. Most tenses have no overt marker apart from these ‘tense- pronouns’ (tp’s); a few do use a tense marker in addition to the tp ... The tp’s agree in person, number, and gender with the underlying subject. If the underlying subject is a pronoun [+pn], it is obligatorily deleted when the tp is added” (Newman 1974.16-17). 70 SYNTAX & SEMANTICS

(2) Guwari a tade kilei [stone TP break pot] ‘The stone broke the pot’

(3) Nà jobe jokoi Ngoje [I wash cap PREP Ngoje] ‘I washed the cap for Ngoje’

(4) Amboi à lewo-wu [boys TP tire-ICP99] ‘The boys have tired’

Wh-questions of the Agent place the question after the EVENT/Verb, and after the Patient, if the verb is transitive (Newman 1974.64, 65):

(5) (a) Share karami mandai? [take.away crocodile who] ‘Who took away the crocodile?’

(b) Share karami mandai wún koa? [take.away crocodile who REL they catch] ‘Who took away the crocodile that they caught?’

(6) Kur mandai? [refuse who] ‘Who refused?’

There are no utterances cited as answers to wh-questions, but it appears certain that the utterances which express “emphasis”100 are the way to do it, and in those, the Agent again appears in the position following the Patient (Newman 1974.63):

(7) (a) Nai meni shi [drink beer he]

99 “In Kanakuru, intransitive sentences ... are formally marked by a pronominal suffix attached to the verb. This pronoun, which I am calling ‘intransitive-copy-pronoun’ (icp) copies the person, number and gender of the subject onto the verb” (Newman 1974.23).

100 “Question words are inherently [+e] ...” (Newman 1974.66). FOCUS: First and Last 71

‘He drank the beer’

(b) Nai meni shi ane [drink beer he up] ‘He drank the beer up’

(8) Job-ro nani Basha [wash-it I PREP Basha] ‘I washed it for Basha’

Sentences such as (5b) in which the clause which modifies the Patient follows the Agent, thus allowing the Agent direct access to position immediately after the Patient, provide evidence that it is specifically the O __ position that marks FOCUS for the Agent. Compare also the TOTALITY MARKER ane, that follows the Agent in (7b), and the Beneficiary in (8) that similarly follows the Agent. Sentence (9) combines some of the above (Newman 1974.64):

(9) Ade shiruwoi shé wura ane [eat fish cat REL she fry up] ‘The cat ate up the fish that she fried’

When constituents other than the Agent are questioned, the wh-word precedes the Agent (Newman 1974.66):

(10) wún kapa? [what they plant] ‘What did they plant?’ as do non-Agent constituents that bear “emphasis” (Newman 1974.66, 67):

(11) Ngoje wún bela [Ngoje they choose] ‘They chose Ngoje’

(12) Lowoi1 náa na shi1 [boy1 I catch-him him1] ‘I caught him for the boy’

“The original i.o. slot must be filled by a pronoun remnant” (Newman 72 SYNTAX & SEMANTICS

1974.67), and the pronoun in (12) identifies lowoi ‘the boy’. Instruments behave similarly:

(13) Guwat1 náa mukwa kwara yiki1 [stone1 they throw.at goat with.it1] ‘They threw a stone at the goat’

Newman (1974.66) combines the two expressions of FOCUS in this way:

The converse of the rule ... that moves emphasized subjects into the predicate is ... [the] rule ... that shifts emphasized nouns from the predicate to the front of the sentence. The result in both cases is deviation from neutral word order.

Matters of FOCUS in Kanakuru are not, however, so simple as the above description makes them appear. There are elements of Kanakuru FOCUS that are reminiscent of Hausa, which we take up in Chapter 10, section 3.2.

5.2 Pero Like Kanakuru, Pero is a Nigerian Chadic language:

Pero, or pók pìpérò (lit. Language of Pero) is a Chadic language spoken in Northeastern Nigeria, within an area of about 140 square miles bounded by 9.4 degrees latitude north and 11.3 degrees longitude east ... The current number of speakers of Pero is not known, but it probably exceeds twenty thousand. (Frajzyngier 1989.1)

Pero is a SVO language101 (Frajzyngier 1989.160, 161, 162, 165):

(1) Wúji dáklò-kò mínà-i [fire destroy-COMPLETIVE house-DEFINITE] ‘Fire destroyed the house’

(2) Tà-pálù píijì nín cínú [FUTURE-pack drum 3pl ] ‘They will pack drums’

101 “In a sentence with two arguments and no focused elements, the semantic roles of the arguments are indicated by the position relative to the verb. The agent is indicated by the position preceding the verb, and the patient is indicated by the position following the verb. If the agent is a 3p. specific pronoun, i.e. it is introduced by the particle nin-, the order of elements is Verb-Patient-Agent ...” (Frajzyngier 1989.161). FOCUS: First and Last 73

(4) Nì-mójó-kò cìgó-tò [1SG-embrace-COMPLETIVE body-3F] ‘I embraced her’

(3) Míjíbà wáat-nà [stranger come-COMPLETIVE] ‘A stranger came’

Compare these (Frajzyngier 1989.227):

(4) (a) Díllà cébí-nà cándè [Dilla plant-COMPLETIVE yam] ‘Dilla planted the yam’

(b) Cébí-nà cándè Díllà [plant-COMPLETIVE yam Dilla] ‘Dilla planted the yam’

Frajzingier (1989.226) notes:

The main device for putting an element of a sentence in focus is to place it in a position different from the one it occupies in the unmarked sentence. For elements that occupy the non-initial position in the sentence, the initial position becomes the position of focus. For the elements that occupy the initial position, the clause final position becomes thge position of focus.

All the examples of Agent FOCUS have only a patient and an Agent, so we cannot determine whether the relevant position is sentence-final or following the Patient as in Kanakuru. And there are no clear examples of a focussed Patient. For example, (Frajzyngier 1989.229):

(5) Mínà-ì díg-kò táttà [house-DEFINITE build-COMPLETIVE father] ‘As for the house (his) father built it’

There is a possibility of a pause following mínà-ì, which is supported by its TOPIC-looking, and not FOCUS-looking gloss. The postverbal Agent táttà suggests, rather, that it, and not mínà-ì is the FOCUS. Although the data are not clear, it appears that Pero exemplifies the same pattern of FOCUS grammar that Kanakuru does. 74 SYNTAX & SEMANTICS

6. Conclusion

[Version: June 14, 2009]