Navigation – Plan du site

AccueilNuméros69Morphological focus and its agree...

Morphological focus and its agreement features in Kakabe

La focalisation morphologique et ses caractéristiques d’accord
Александра Валентиновна Выдрина. Морфологический фокус и его согласовательные признаки
Alexandra Vydrina
p. 30-86

Résumés

L’article porte sur l’expression morphologique du focus au moyen de la particule focalisante (FP) en kakabé. Dans cette analyse, sont traitées des problématiques qui sont rarement étudiées en lien avec le focus morphologique :

i) la position de FP par rapport aux frontières du constituant focalisé, en particulier au groupe verbal (VP) et au groupe de flexion (IP) ;

ii) l’impact du statut du « donné » sur la position de FP ;

iii) FP par rapport aux types de CP et les limites de CP.

Je montre que, bien que FP ressemble à certains égards au focus prosodique, sa distribution ne s’intègre pas entièrement dans une explication prosodique du focus, contrairement aux affirmations de Büring (2010) et Féry (2013).

Plus précisément, je propose qu’en kakabé l’opérateur de focalisation est situé plus bas que CP. De la même façon que l’opérateur Q de Cable (2006, 2010), qui peut être impliqué dans une relation d’accord bipartite, l’opérateur de focalisation s’accorde simultanément avec C0, qui est positionné au-dessus, et avec DP, qui se situe au-dessous. Je montre comment le modèle de distribution de FP respecte les conditions de l’Accord (Chomsky 2000, 2001) comme la localité, la supériorité et l’activité.

Haut de page

Texte intégral

1. Introduction

1Up to this moment, there has been little research on the morphological expression of focus, as compared to focus marking through prosody or word order. For languages that signal focus morphologically, the accounts rarely go beyond stating that the focus marker is adjoined to the focused constituent. In particular, the question of the distribution of the morphological focus marker with respect to the boundaries of the focus constituent that can be a DP, a VP or an IP, is rarely addressed. There are some notable exceptions, for example, Hartmann and Zimmermann (2007; 2009) provide a detailed analysis of morphological focus marking in Chadic languages. But even they do not approach, for example, the issue of how givenness affects focus marking, a question that is, nevertheless, central for the research on prosodic focus.

2The issues concerning the position of sentence stress in relation to discourse-givenness and to the boundaries of focus constituents have been among the main concerns of Optimality Theory accounts of focus (e.g. Truckenbrodt 1995; Büring 2016). These accounts are mostly formulated based on data from languages that mark focus prosodically and/or syntactically rather than morphologically. At the same time, since the constraints and principles formulated in them are supposed to be universal, they are expected to apply to languages irrespective of the mode of expression of focus. Büring (2010) and Féry (2013) explicitly address the question of morphological focus and both suggest that morphological focus is subject to the same constraints as non-morphological focus. Following Büring’s (2010) FocusProminence constraint, that builds on Truckenbrodt (1995), focus is always realized as structural prominence that is defined on the prosodic structure.

3This paper explores the distribution of focus particle (FP) in Kakabe, a Mande language. The research on this little-known and understudied language is based on the data that I have been collecting in Guinea since 2008, and apart from elicitation it is supported by a large corpus of natural texts. By its distribution, on the first approach, FP appears to show striking similarities to sentence stress as the marker of focus, for example in English. Example (1) shows that FP follows the direct object in two cases: first, when only the object DP is in focus and, second, when the whole VP is in focus. The same way, sentence stress falls within the focus constituent on the direct object in the English sentence equivalents. Small capitals indicate the word that bears the sentence stress and the underlining indicates the extent of focus. As can be seen, sentence stress targets the same word as FP.

(1)

ka

kɔ̀rɛ́ɛ̀

lè

sàn.

3sg

pfv.tr

rice.art

fp

buy

4(What did he buy?) ‘He bought rice.’

5(Whad did he do?) ‘He bought rice.’

6Strikingly, Kakabe FP also appears to follow the pattern of the distribution of sentence stress with respect to givenness. In languages like English, when a discourse given constituent is part of focus, it is avoided by the sentence stress. The same is found for FP in Kakabe: thus, in (2) FP does not appear on the direct object that is part of VP focus because of its givenness.

(2)

ka

Sɛ́ku

gbàsi

lè.

3sg

pfv.tr

Seku

hit

fp

7(What did Seku’s mother do?) ‘She hit SekuG.’

8Therefore, the hypothesis by Büring (2010) and Féry (2013), according to which morphological focus markers are determined on the prosodic structure the same way as stress is rather compelling. However, under closer investigation, FP appears to manifest properties that cannot be easily fitted into the prosodic prominence account of focus. I show that FP does not behave like stress with respect to CP boundaries and CP types, and that FP is attracted to the DP as illustrated in (1) not because of its prosodically-prominent position but because of its syntactic categorization.

9After I explore the possibility of accounting of FP distribution in terms of prosodic hierarchy and show that it encounters some unresolvable difficulties, I proceed to the formulation of my proposal. I posit for Kakabe a Focus operator as it is done in the Cartographic approach (e.g. Rizzi 1997, Belletti 2001, Aboh 2006) but differing from it in that Foc in Kakabe can be in various locations in the syntactic tree (see Section 5.3). In my analysis, Foc operator enters Agree relation with two other elements in the structure: with the head of CP above it and with the DP below it. It is therefore structurally close to Q-operator in the Q-based theory (Cable 2006, 2010), an element that also agrees with CP and, in part of languages, with the wh-word in its c-command. Agreement is understood in the lines of the probe-goal approach (Chomsky 2000, 2001), with the representation of feature types as in feature valuation theory of Pesetsky and Torrego (2007). I demonstrate how the pattern of FP distribution follows from the conditions on Agree, such as locality, superiority and activity conditions. The feature type theory allows to provide an account of the interrelation between focus marking and distinct types of CPs.

10The paper proceeds as follows. Section 2 provides general information about Kakabe and its grammatical features that are relevant for the discussion. Section 3 outlines the main facts about the distribution of FP in Kakabe. Section 4 provides an attempt to apply the prosodic account of focus to the Kakabe focus particle and points out the difficulties that are encountered. Section 5 proposes the alternative, agreement-based, account for the presented Kakabe evidence, and outlines the comparison with the Q-based theory and focus in the Cartographic approach. The Foc/DP agreement part of the proposal is explored on the Kakabe data in Section 6, and Section 7 investigates the relationship between CP and Foc. Finally, Section 8 raises the question of how the agreement analysis of focus proposed in the paper can be applied to other languages with morphological focus marking.

2. Kakabe and the general information about the language

2.1. Sociolinguistic information and the data

11Kakabe is a little-known Mande language spoken by approximately 50 000 people in Guinea. It is subject to dialectal variation with Northern Kakabe (NK), Western Kakabe (WK) and Central Kakabe (CK) as main dialectal zones. Phenomena related to the expression of focus, which is the main object of this study, are mostly the same across these dialectal zones.

12The present research is based on first-hand data collected during my fieldwork in 2008-2020. It relies on a corpus of 20 thousand words with a large number of speakers and consisting of natural texts of various genres. The corpus contains about two thousand occurrences of FP. The most part of the corpus is available online.1 Apart from the corpus data, the investigation of the distribution of FP relies on elicitation and grammaticality judgement by at least two speakers. In the paper, examples originating from corpus are provided with a reference by which they can be found in the corpus.

2.2. Kakabe basics

13Kakabe is a tonal language, with H and L, floating L tones and downdrift (see the discussion about downdrift and prosodic boundaries in Section 4.2). The distance between the underlying (lexical) tones and their surface realization can be rather important, and multiple tonal processes are applied to lexical tone. The lack of tonal marker on certain syllables in the second line of transcription reflects the absence of any underlying tone. The transcription provided in the examples represents only tones before the application of tonal rules, except for some cases where they are due to grammatical tone.

  • 2 SOVX word order is typologically unusual but typical for Mande languages. It does not occur in lang (...)

14Kakabe shows typical for Mande languages S(aux)OVX word order, where X stands for any non-direct complement or adjunct.2 Adverbs most frequently appear rightmost, but can also be placed leftmost, and a limited number of them can appear between the subject and the auxiliary. Nothing can intervene between the auxiliary and the VP, nor between the object and the verb.

(3)

sí

nìngé

sàn

mànsáà

yen

sínàn.

3sg

pot

cow.art

buy

chief.art

bnf

tomorrow

15‘He will buy a cow for the chief.’

16Kakabe has a very rigid word order that cannot be modified by information structure. Wh-phrases are always in situ:

(4a)

Yɔ́n

(de)

kà

nìngéè

sàn?

who

fp

pfv.tr

cow.art

buy

17‘Who bought the cow?’

(4b)

ká

fɛ́n

dè

sàn?

3sg

pfv.tr

what

fp

buy

18‘What did he buy?’

4c.

kà

nìngéè

sàn

míntɔ?

3sg

pfv.tr

cow.art

buy

where

19‘Where did he buy the cow?’

  • 3 The transitivity split, with the differentiation of the perfective in transitive constructions and (...)

20TAM and polarity categories are expressed cumulatively by a paradigm of auxiliary markers which are referred to as predicative markers in the Mandeist tradition. They are presented in Table 1. The paradigm contains the existential copula bi and the negative existential copula béle. Perfective aspect in intransitive clauses is encoded by the suffix -ta on the verb, whereas in transitive clauses it is expressed by the post-subject auxiliary ka.3 The contrast between focused perfective báti and the non-focused ka/-ta perfective is discussed in Section 7.4.

Affirmative

Negative

perfective

báti (focused), ka tr./ -ta intr.

máa

stative-resultative

bi -len (NK, WK)

bi -nden (CK)

béle -len (NK, WK)

béle -nden (CK)

imperfective

bi -la

béle -la

potential

si

tée

subjunctive

ni

káni

imperative

Ø

temporal/conditional

máni

infinitive

kà

21Table 1. Kakabe inflectional paradigm

3. The observed facts about the placement of FP

22I use the term ‘focus’ as equivalent to the syntactic constituent that is F-marked and that can be identified through Question and Answer Congruence (Rochemont 2011).

23In general, FP lè/dè (the dè allomorph appears after nasals) is placed either at the right edge of focus or at the right edge of some constituent within focus. The only exception to this is that, when focus is part of an embedded CP, FP appears elsewhere; this case is dealt with in Section 4.5. Apart from this exception, FP is always at the right of focus if it is coextensive with a DP, cf. (5a) and (5b). Focus constituent is always in situ.

(5a)

Kàyéè

lè

ká

nìngéè

sàn.

Subject

focus

man.art

fp

pfv.tr

cow-art

buy

24(Did the man or the woman buy the cow?) ‘The man bought the cow.’

(5b)

Kàyéè

ká

nìngéè

lè

sàn.

Object

focus

man.art

pfv.tr

cow-art

fp

buy

25(Did the man buy the cow or the goat?) ‘The man bought the cow.’

26Next, when focus is broader than a DP, for example, when the whole VP is in focus as in (6), FP can appear inside focus. Taking for the moment only the cases where focus does not include discourse-given elements, in general, FP appears on a DP rather than on the verb.

(6)

ka

kɔ̀rɛ́ɛ̀

lè

tàbi.

VP focus

3sg

pfv.tr

rice.art

fp

prepare

27(What did he do?) ‘He prepared the rice.’

28The same is found for sentence focus where focus is the whole IP. Here again, FP appears after the DP (7).

(7)

Mónbilè

lè

bèe-ta.

IP focus

car.art

fp

crash-pfv.i

29(What happened?) ‘The car crashed’ [kkec_av_narr_150124_ak1_124].

30When a focused IP contains more than one DP, FP appears on the one that is linearly first. Thus, in (8a) FP follows the subject DP, whereas the construction with FP following the object in (8b) cannot have the IP-focus interpretation.

(8a)

Mùsa

lè

ka

Sɛ́ku

gbàsi.

Musa

fp

pfv.tr

Seku

hit

31(What happened?) ‘Musa has hit Seku.’

(8b)

Mùsa

ka

Sɛ́ku

lè

gbàsi.

Musa

pfv.tr

Seku

fp

hit

32(What did Musa do?/Who did Musa hit?) ‘Musa hit Seku.’

33#(What happened)? ‘Musa has hit Seku.’

34See also an example from natural text, with three DPs inside an IP-focus:

(9)

Túlâ-n

dè

ká

ɲàaréè-nù

súbɛ

tólóɲɛ̀

là.

mouse.art-pl

fp

pfv.tr

cat.art-pl

choose

game.art

obl

35(Beginning of a story:) The mice chose the cats to have a party [tula_SNKeita_2009_003].

36To summarize, in VP or IP focus, FP is attracted to a DP inside it. However, this attraction can be counteracted by the discourse status of the DP in question. There is an ever growing literature dedicated to the problem of givenness with respect to focus (e.g. Schwarzschild 1999; Féry and Samek-Lodovici 2006; Büring 2016; see Rochemont 2016 for an overview). In general, the authors agree that part of the focus phrase can be given. In what follows, givenness, ‘G-markedness’ is indicated following the principles of ‘F plus G notation’ model from Büring (2016). In this model, all that is not focus is G-marked. Thus, in (10), the subject is G-marked and out of Focus.

37(10) (What did Mary do?) SheG killed Bill.

38Crucially, focus constituent can contain a G-marked constituent (in the examples below, I omit the G-marking that signals givenness outside of focus for its obviousness and display it only on given constituent inside focus). Let us look at an example. In (11), the DP Bill is already activated due to the preceding context, whereas the lexical content of the verb is new (see the discussion of this example in Arregi 2016: 194). Therefore, a G-marked constituent is part of focus in (11).

39(11) (What did Bill’s mother do?) She killed BillG.

40The whole content of the focus phrase can be given as the focused but discourse-given object, as in (8) from Schwarzschild (1999: 145).

41(12) (Who did Bill’s mother praise?) She praised HIMG.

42Examples (13) and (14) illustrate givenness in focused IPs. In (13) from Féry and Samek-Lodovici (2006: 146), the whole sentence is focus and at the same all given (except for the word too). Therefore, the whole IP is F-marked and G-marked at the same time. Finally, (14) from Zubizarreta and Nava (2011: 659) is an instance of IP-focus that contains both G-marked and non-G marked subconstituents: covers is given, the rest is not (except for the locutor subject which is always given by default).

43(13) Jack said the American president drinks. What did Gilles say about the French president?

44He said [the French president drinks too]G.

45(14) Why are these notebooks missing their covers?

46Because I’m drawing pictures on the [covers]G

47As can be seen by comparing (10) and (11), givenness can affect the position of the sentence stress. By default, in English, the sentence stress falls on the rightmost element in the focus phrase, as in (10). However, when the rightmost element is G-marked, stress appears elsewhere, as in (11) and (14). As for (13), where sentence stress is on too, Féry and Samek-Lodovici (2006: 146) argue that the presence of this word is due to a repair mechanism: since all the other words are given, placement of the sentence stress on any of them would violate the Destress Given (DG) constraint, and this is avoided by the insertion of too which is not given and hosts the stress.

48Let us now turn to Kakabe. Analogously to sentence stress in English, FP in Kakabe does not appear on a given DP inside a focus. Thus, in (15) FP appears after the verb, since the object DP is given.

(15)

ka

Sɛ́ku

gbàsi

lè.

3sg

pfv.tr

Seku

hit

fp

49(What did Seku’s mother do?) ‘She hit SekuG.’

50See also (16) representing an excerpt from a natural narrative in which the speaker explains how he and his friends managed to take gasoline from tanks. The referent of rèzèrwárùnú ‘the tanks’ appears first in (16a). Therefore, when it appears again in (16b), it is already activated. The utterance in (16b) addresses the QUD ‘What did we do to get the gasoline?’ therefore, its focus is the VP rèzèrwárùnú sɔ̀gɔ́ ‘pierce the tanks’.

(16a)

Gàzwál

mín

b-áà

búutɔ̀,

rèzerwár-nù

búutɔ̀,

gasoline-art

rel

be-3sg

inside

tank.art-pl

inside

51‘The gasoline that was there, in the tanks,

(16b)

mà

ká

wò

fóo

wíide,

pàsɛ́

mà

kà

1pl

pfv.tr

that

univ

empty

how

1pl

pfv.tr

rèzerwárù-núG

sɔ̀gɔ́

lè

tank.art-pl

pierce

fp

52we took it all: we pierced the tanksG.’ [kkec_av_narr_150124_ak1_077].

53Finally, (17) illustrates the case where the whole focused constituent is given. In this case, since nothing else is available, FP is hosted by the G-marked pronoun.

(17)

Ǎndè

lèG

báà

sàn-na

kà

dí

bólo.

3pl.lg

fp

be-3sg

buy-ger

inf

3sg

give

1sg

hand

54(Do you buy your clothes yourself or do your parents buy you clothes?)

55It is them who buy me clothes. (kkec_av_conv_131207_talk01_228)

56The pattern of placement of FP as it is presented so far, shows striking similarities as compared to the distribution of prosodic markers of focus in languages like English. The next section provides an attempt to apply to the case of Kakabe the standard account of prosodic focus realization.

4. Focus as prosodic prominence

4.1. Background

57The prosodic hierarchy of Nespor and Vogel (1986) is widely accepted as the standard representation of phonological structure. In this representation, the levels relevant to focus realization are, bottom up, phonological word (ω), phonological phrase (φ), intonation phrase (ɩ) and utterance. The hierarchical structure can be viewed as a metrical tree (14), with prosodic units representing nodes at each corresponding level. Each node has exactly one strong daughter, which is the head at the lower level; a prosodic unit marked as strong can have only weak sister(s). In (18) John and Mary are heads at the level of ω, and Mary is also the head at the level of PP φ, in contrast to John. Each strong ω receives phrasal stress and each strong φ receives sentence stress.

58(18)

59The same can be represented as a bracketed metrical grid in (19) with “x” standing for the metrical beat at each hierarchical level. At ω level, each ω corresponds to the most prominent syllable from the level below, Ma- in Mary. Next, at φ level, each φ is headed by the most prominent ω: Mary in saw Mary. Finally, ɩ is headed by the most prominent φ. In English, the highest prosodic prominence, corresponding to the maximal number of the superposed heads ‘x’ is translated as the sentence stress on the corresponding syllable

60Next, the prosody-syntax interface is subject to the mapping constraints Wrap and StressXP (Truckenbrodt 1995, the formulation is taken from Samek-Lodovici 2005: 699):

61(20) Wrap: Each lexically headed XP is contained inside a phonological phrase.

62StressXP: Each lexically headed XP must contain a phrasal stress (where phrasal stress refers to the head of a phonological phrase).

63In accordance with Wrap, saw Mary in (19) is phrased as one φ with one phrasal head, which in English is translated into phrasal stress. This phrasal stress assignment respects, at the same time, StressXP, because Mary is included in the lexical projection of the verb.

64The designation of strong/weak elements among sisters is defined by the right vs. left-headedness parameter. Depending on the language, head is the leftmost or rightmost element within a constituent. As can be seen from the representation in (19), English is right headed both at the level of phonological phrase (rightmost ω is the head within the φ saw Mary) and at the level of ɩ phrase, with the rightmost φ saw Mary being the head as opposed to φ John. To take an example of a left-headed language, in Hungarian, both φ phrases and ɩ phrases are left headed (Szendroi 2003).

65Now let us turn to the focus-prosody interface. This mapping is regulated by the Stress-Focus constraint (Samek-Lodovici 2005: 697, based on Truckenbrodt 1995):

66(21) StressFocus: For any XPF and YP in the focus domain of XPF, XPF is prosodically more prominent than YP.

67In the case of IP focus as in (18), focus does not interfere with the default prosody as resulting from prosodic constraints. Since, by definition, the whole IP is F-marked, it will at any case include the constituent that is mapped to the prosodic unit with the sentence stress. When, on the other hand, focus is narrower than IP, some repair mechanism may be required for XPF to be aligned with the prosodic unit that receives the sentence stress. In English this is done through the operation of stress shifting (see, e.g. Reinhart 2006: 148), that due to Strict Layering prosodic constraint (Selkirk 1995) also implies the insertion of an additional ɩ boundary to the left of XPF. Thus, Subject focus in (22) results in prosodic structure with a sentence stress on John. Additionally, all phrasal heads after focus undergo deletion in English (Büring 2010)

68John saw Mary (as in an answer to ‘Who saw Mary?)

  • 4 The way this constraint functions can vary depending on how givenness and focus are represented in (...)

69Finally, another non-prosodic constraint interfering with default stress assignment concerns givenness already discussed in Section 3: DestressGiven or Givennes Realization (Büring 2016: 173). Roughly speaking, this constraint prohibits the G-marked constituent to carry the sentence stress4. Going back to the example (10) discussed earlier and reproduced in (23), stress appears not on the rightmost prosodic word Bill but on killed since Bill is given from the context.

70(23) (What did Bill’s mother do?) She killed BillG

4.2. Kakabe FP as prosodic prominence

71Going back to FP in Kakabe, let us assume that the pattern of its distribution is, underlyingly, due to the same constraints as the distribution of sentence stress. In other words, we can assume that FP is the spell-out of the prominence defined on the prosodic structure and, at the same time, subject to focus-related and givenness constraints.

72Let us look again at FP in an IP focus utterance where, as said above, the prosodic structure is default (24). Under the assumed logic, the position of FP in (24) indicates that ɩ phrase is left-headed in Kakabe since FP is found in the leftmost φ.

73Next, in (25) with VP focus, focus phrase is mapped onto the second φ and the prosodic head must be within this φ. Since this is not the case in the default prosodic structure (24), the latter needs to be amended. One can assume that this is done via the shift of the ɩ head from the left φ to the right φ, as represented in (25).

74The configuration in (25) allows us to define the default head orientation at the φ level. The φ phrase that has become the head of ɩ phrase due to stress shift consists of two words, as opposed to only one word in the head φ in (24). Therefore, assuming that FP spells out the right boundary of the most prominent word within ɩ, the fact that it appears after the first word within focus indicates that φ is left-headed the same way as ɩ phrase is. Thus, we end up with the left-headed default prosodic structure of Hungarian type as analysed by (Szendroi 2003: 44).

75(26) Possible parametrization for Kakabe:

76a. Align-Foc-ɩ-L: Align a focus with the left boundary of an intonation phrase.

77b. Align-Foc-φ-L: Align a focus with the left boundary of a prosodic phrase.

78Next, as already discussed, Kakabe FP displays the same sensitivity to givenness as sentence stress. Again, if we consider FP as a reflex of prominence, the interference of givenness with the placement of FP follows from DestressGiven. Thus, in (27) focus is coextensive with VP, therefore, the propsodic phrase (Sɛ́ku gbàsi)φ to which this VP is mapped, must be the more prominent φ in ɩ. Next, due to (26b) the leftmost word, i.e. Sɛ́ku must be the head within the phrase, yet, this word is G-marked and therefore cannot bear sentence stress. Accoridngly, a two-step derivation takes place. First, ɩ head shifts from Mùsa to the second φ Sɛ́ku gbàsi due focus on VP (StressFocus). Second, since φ in Kakabe is left-headed, the main prominence would be assigned to Sɛ́ku and not to gbàsi. Yet, Sɛ́ku is G-marked and this would violate DestressGiven, therefore, the main prominence is on gbàsi.

  • 5 The presented Kakabe evidence also complies with Féry's (2013) modification of the ‘Focus as Promin (...)

79To summarize, so far, Kakabe data fits well the claim that Prominence is the universal prosodic property of focus advanced Truckenbrodt (2005) and Büring (2010)5.

4.3. No conclusive phonological evidence in favour of the prosodic account

  • 6 Stress is understood here as such rhythmical structuring of the prosodic unit, whereby one syllable (...)

80It should be noted that Kakabe does not provide any conclusive evidence as to the existence of phonological correlates for FP placement (assuming as has been done above that it spells out prosodic prominence). To begin with, Kakabe has no stress, neither at the level of φ, nor at the level of ɩ6. Even though, in general, lexical tone and stress are not mutually exclusive, and there are languages that have both (see e.g. Yip 2002: 256), nevertheless, Kakabe belongs to the category of languages that have lexical tone only.

81Nevertheless, Kakabe phonology does have an indicator of ɩ boundary which is downdrift, or automatic downstep (for the discussion of terminology; see Connell & Ladd 1990; Yip 2002: 147ff): in a sequence of alternating H and L tones, every next H is pronounced lower than the preceding one. Crucially, the domain of its realization in Kakabe is ɩ, accordingly, the left boundary of ɩ is preceded by downdrift reset. Going back to FP, the phonological word to which FP is right-adjoined is expected to be preceded by downdrift reset considering the rule formulated in (26): since both ɩ and φ are left-headed, FP must be right-adjoined to the leftmost word within ɩ. This is, however, not what is found in the data. Thus, no downdrift reset is found before Sɛ́kù in (25) nor before gbàsí in (27). Even though, under the assumed analysis, they must be ɩ-leftmost, their respective H tones are pronounced lower than the first H tones of the intonation phrase.

  • 7 Kakabe has a partial tone-leveling operation of the type HLH = HꜜH where an L tone between two H to (...)

82Still, the absence of any correlation between downdrift reset and FP does not oblige us to dismiss the prosodic account. For example, one can suppose that the position of ɩ head as in (24) and (25) simply results from the ‘swap’ operation, whereby the head changes its position without the insertion of an additional boundary (see Büring 2010 for the proposal about the ‘swap’ strategy as opposed to boundary insertion strategy). Another way of rescuing the prosodic account is to assume, following Féry (2013), that focus alignment with prosodic phrases is parametrized with respect to the level of its application: it is aligned either with ɩ or φ depending on the language. Under this approach, one can suppose that Kakabe is an φ-alignment language. This means that only Align-Foc-φ-L (26b) and not Align-Foc-ɩ-L (26a) should apply to Kakabe. Yet, here again, no straightforward evidence for the existence of any phonological cues of the presence of a φ-boundary before the word with FP can be found7.

83Finally, the presence of prosodic cues could strengthen the hypothesis of the prosodic nature of Kakabe FP, yet, it is not required, because metrical structure can be understood in an abstract way. To summarize: I do not find any phonology-internal evidence of the involvement of prosody in focus marking, yet, this alone does not require us to abandon the prosodic account of FP.

4.4 Attraction to DP and the absence of prosodic alignment

84As demonstrated above, no supporting evidence for the prosodic analysis of FP can be found in the domain of phonological processes that exist in Kakabe. However, since prosodic prominence can be understood abstractly, it can still hold. What is more critical for this analysis is that the placement of FP inside a focus constituent appears to be due exclusively to the attraction to DP as a syntactic category, whereas the alignment with a prosodic boundary proves to be dispensable altogether under closer investigation.

85Let us consider first examples (28a) and (28b). Both represent VP-focus sentences with each focused VP containing a DP. Crucially, the DP is preverbal in (28a) and postverbal in (28b). From the alignment constraint, one would expect FP to right-adjoin to the leftmost word within the VP. However, in both cases FP is placed immediately after the DP. Hence, (28b) violates the alignment constraint.

(28a)

ka

nìngéè

lè

sàn.

3sg

pfv

cow.art

fp

buy

86(What did he do?) ‘He bought a cow.’

(28b)

dòn-ta

bóɲɛ̀

lè

tɔ.

3sg

enter-pfv.i

house.art

fp

in

87(What did he do?) ‘He entered the house.’

88This alone, however, does not refute the prosodic analysis for FP. The fact that arguments are more often stressed in prosodic focus languages has been known for a long time, see, for example. Gussenhoven (1992), Ladd (1996: 246ff). This property of arguments is particularly tangible in such languages as German that manifest VO and OV word order alternation. Thus, in (29a) and (29b) sentence stress is placed on DP rather than on verb independently of its position with respect to the verb. To account for this phenomenon, Schwarzschild (1999: 173) proposes HeadArg constraint, and (Büring 2001: 14) A/P (Argument-over-Predicate) constraint. This constraint, as can be seen in (29), is ranged higher than the alignment with a prosodic boundary.

89(29) German: VP focus and stress on argument

90a. Sie liest ein Buch. b. Sie hat ein Buch gelesen.

91She reads a book. She has a book read.

92What is crucial for our discussion is that for the account of sentence stress distribution both HeadArg and the alignment constraints are necessary. Prosodic alignment determines the word that hosts sentence stress when more than one DP is present inside a focus, or when focus has more than one word and neither of them is a DP. Since German is a prosodically right-headed language (Féry 2011), in (30), with IP focus including three DPs, it is the rightmost DP that receives sentence stress. Sentence in (31) has VP focus that includes two prosodic with neither of them being a DP, and, in line with the prosodic head alignment characterizing this language, main stress is on the rightmost word.

93(30) German: IP focus with three DPs (Féry 2011: 1906)

94Annemarie hat dem Nachbarn ihr neues Kombi gezeigt

95‘Annemarie showed the neighbour her new station wagon.’

96(31) German: VP focus including two non-DP words

97Er fing an zu singen

98‘He started to sing.’

99Differently from German, Kakabe does not align FP with respect to any prosodic boundary. When three DPs are present in an IP focus, as in (32), FP follows the first DP, which would suggest the leftmost alignment of the prosodic head. However, when no DP is present inside a focus, FP appears at its right edge, rather than after the first prosodic word.

(32)

Mùsa

lè

ka

Sɛ́ku

gbàsi

lúumɛ̀

.

Musa

fp

pfv.tr

Seku

hit

market.art

place

100(What happened?) ‘Musa hit Seku at the marketplace.’

(33)

fɔ́lɔ-ta

bòri-la

lè.

3sg

start-pfv.tr

3sg

run-ger

fp

101(What did he do?) ‘He started running.’

102Another piece of evidence against the relevance of prosodic alignment for FP comes from constructions with focus involving complex DPs. As shown in (34), when focus is coextensive with a complex DP, FP is placed at the end of it. Yet, if FP followed the alignment with the left boundary of intonation phrase, one would expect it to appear after the first DP. The contrast between (32) and (34) is rather suggestive of the irrelevance of prosody for the placement of FP: both start with two DPs mapped on two phonological phrases, but FP is placed after the first DP in (32) and after the second DP in (34).

(34)

Mànsáà

lá

mùséè

lè

gbàndiya-ta.

chief.art

poss

wife.art

fp

fall.il-pfv.i

103(Who fell ill?) ‘The chief’s wife fell ill.’

104The same placement of FP is found in IP focus with a complex (non-given) subject:

(35)

Mànsáà

lá

mùséè

lè

gbàndiya-ta.

chief.art

poss

wife.art

fp

fall.il-pfv.i

105(What happened?) ‘The chief’s wife fell ill.’

106Consider also (36) showing that, no matter how many DPs are included in the complex DP, FP always appears at the right edge of the whole DP.

(36)

Mànsáà

lá

mùséè

la

nìngéè

lè

gbàndiya-ta.

chief.art

poss

wife.art

poss

cow.art

fp

fall.ill-pfv.i

107(Who fell ill?) ‘The cow of the chief’s wife fell ill.’

108(What happened?)The cow of the chief’s wife fell ill.’

109It should be noted that there is no ban on the placement of FP inside of a complex DP, meaning that the position of FP as in (34) and (35) cannot be due to the impossibility for FP to appear DP-internally. FP is found inside a complex DP in the context when the focus is narrowed to the first DP which is the possessor as in (37). Importantly, such a construction cannot have the interpretation with focus on the complex DP as a whole.

(37)

Mànsáà

lè

la

mùséè

gbàndiya-ta.

chief.art

fp

poss

wife.art

fall.ill-pfv.i

110Whose wife fell ill?) ‘The chief’s wife fell ill.’

111#(Who fell ill?) ‘The chief’s wife fell ill.’

112The same as what has been said about possessor DPs is true for coordinate structures with DPs. If focus includes the whole coordinate DP structure, FP is placed at its right edge (38a), whereas the construction with lè after the first DP is possible but allows only the interpretation with the narrow focus DP as in (38b).

(38a)

ka

támaatè

nin

bàntárà

lè

sàn.

1sg

pfv.tr

tomato.art

and

cassava.art

fp

buy

113(What did you buy?) ‘I bought tomatoes and cassava.’

(38b)

ka

támaatè

lè

nin

bàntárà

sàn.

1sg

pfv.tr

tomato.art

fp

and

cassava.art

buy

114(You bought cassava and what?) I bought tomatoes and cassava.

115#(What did you buy?) ‘I bought tomatoes and cassava.’

116To sum up, the evidence presented above strongly suggests that FP is attracted to DP as a syntactic category, and otherwise, i.e. when no DP is present in a focus, is placed at the end of focus phrase, rather than being governed by prosodic constraints in any context.

4.5. Ban on FP in embedded CPs

117The next context in which FP shows different behaviour as compared to sentence stress focus involves embedded clauses. Strikingly, when a focus phrase is inside an adverbial purpose clause as in (39), FP appears not to the right of focus embedded in this clause, but instead at the end of the matrix clause. The structure with FP inside an adverbial clause as in (39b) is considered inadmissible or at least less preferable as compared to (39a). Besides, structures with FP inside and adverbial clause never occur in texts.

(39a)

ka

mániɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

rice.art

buy

118(What did he ask you to buy?) ‘He asked me to buy rice.’

119(Litt. “He asked me that I should by rice”).

(39b)

??

ka

mániɲinka

[cp

ni

kɔ̀rɛ́ɛ̀

lè

sàn].

3sg

pfv.tr

1sg

ask

1sg

sbjv

rice.art

fp

buy

120FP is not allowed either inside relative clauses or conditional antecedents, for illustrations see Section 7.6.

121Note that the construction with FP at the end of a matrix clause can also have interpretations with focus extending over larger constituents:

(40)

ka

mániɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

ka

mániɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

ka

mániɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

ka

mániɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

rice.art

buy

122(What did he ask you to buy?/What did he ask you to do?/What did he do/What happened?) ‘He asked me to buy rice.’

  • 8 A detailed discussion of focus embedded within a subordinate clause in German can be found, for exa (...)

123The prosodic account of focus does not predict any restrictions on focus marker related to clause subordination. And indeed, there appear to be none in languages signaling focus with sentence stress. In language like English and German focus on a subconstituent is itself in situ and is marked in situ by stress. 8Thus, in (41a) a noun embedded in a subordinate clause is in focus, and, accordingly, this noun hosts the sentence stress (see the discussion of this example in Gussenhoven 1994: 90). The stress patterns in (41b) and (41c) show that, in contrast to Kakabe (40), sentence stress can also appear on a word embedded in a subordinate clause in the case of broader focus including a subordinate clause.

124(41) a. (What ex-convict wearing something red was he warned to look out for?)

125He was warned to look out for an ex-convict in a red shirt.

126b. (What was going on?) He was warned to look out for an ex-convict in a red shirt.

127c. (What did he do ?) He was warned to look out for an ex-convict in a red shirt.

128Let us summarize. We have seen that there are similarities between the assignment of sentence and the placement of FP in Kakabe, namely, therefore, the hypothesis advanced in Büring (2010) and Féry (2013) according to which prosody is the universal underlying principle of focus realizations seems at first very compelling. However, apart from givenness sensitivity and the preference of arguments over predicates, FP displays properties that do not fit into this account unless numerous stipulations are made. For this reason, I propose a different analysis in the following sections.

5. Focus projection

5.1. The proposal

129In the core of the current proposal lies the assumption that focus is represented in the syntactic structure by Foc operator with its own functional projection. As argued in more detail later (Section 5.3), it differs from the Foc projection of the Cartographic approach (Rizzi 1997) in that it can be in multiple locations in the syntactic tree. Nevertheless, I share the basic insight of that approach that focus is essentially related to CP.

130The next central component of the proposal is that Foc operator enters into an agreement relation with two other elements in the syntactic structure: (i) with CP above it and (ii) with DP below it. These agreement relations are conditioned by the presence of two features on FocP head. The feature that it shares with C is Foc and the feature that it shares with DP will be referred to as D-feature. Focus projection FocP can dominate any non-terminal element of the clause in Kakabe, taking the XP that corresponds to the logical focus as its complement, its head is aligned to the right.

131Assuming that in Kakabe CP head is on the left (judging by the more frequent leftmost position of complementizers) and FocP head is on the right (judging by the placement of FP), Foc projection and its relations with CP and DP can be represented as in (42).

132(42) Focus operator in Kakabe

133[CP C [Foc] … [[DP[D] … Foc, [D, Foc] FocP]]

134As already mentioned, Foc projection can take as complement not only DPs but other XPs as well such as IPs, VPs, and PPs. Thus, in examples below, Foc operator dominates an IP, a VP and a PP, respectively. As can be seen, FP appears in FocP head only in (44), whereas in (43) and (45) it appears on the DP insider the FocP. As argued in detail in Section 6, the placement of FP in the head of FocP is default, whereas its appearance on DP is due to successful D-agreement.

(43)

[[Mónbilè

lè

bèe-ta

IP]FocP].

IP

focus

car.art

fp

crash-pfv.i

135(What happened?) The car crashed.’

(44)

ka

[[

Sɛ́ku

gbàsiVP]

lè

FocP]

VP

focus

3sg

pfv.tr

Seku

hit

fp

136(What did Seku’s mother do?) ‘She hit Seku.’

(45)

bòyi-ta

[[kɔ́lɔɲɛ̀

lè

PP]FocP]

PP

focus

3sg

fall-pfv.i

well.art

fp

in

137(Where did he fall?) ‘He fell in a well.’

138The fact that FocP does not take as complements terminal elements is seen, for example, from the impossibility to place FP after the auxiliary or complementizer, occupying I0 and C0, respectively. Thus, FP is not possible after the potential auxiliary si in (46a), nor after the complementizer in (46b), neither is it admitted after the head of coordinate construction, as shown in (46c).

(46a)

*À

si

lè

tága

sáarè

tɔ.

3sg

pot

fp

go

city.art

in

139Intended meaning: ‘He will go to the city.’

(46b)

*Kàtáa

lè

fɔ́lɔ-ta

báara-la

mà

béle

ɲɔgɔnyén-na.

since

fp

2sg

start-pfv.tr

work-ger

1pl

be.neg

see.each.other-ger

140Intended meaning: ‘Since you go to the city, do no not see each other.’

(46c)

*À

ka

kɔ̀rɛ́ɛ̀

ànín

dè

bàntárà

sàn.

3sg

pfv.tr

rice.art

and

fp

cassava.art

buy

141‘He bought rice and cassava.’

5.2. Bi-partite Agreement chain in Q-based theory (Cable 2010)

142The bipartite architecture of focus represented in the syntactic structure with FocP as a vantage point as in (42) is parallel to the representation of the Q operator phrase in Q-based theory (Cable 2006; 2010; see also Hagstrom 1998; Kishimoto 2005). Q operator appears in interrogative utterances and it can be lexified by an overt Question (Q) particle. Cable notes that Q-particle is not restricted to questions, and it should not be confounded with the interrogative complementizer head, which also has a tradition of being referred to as ‘Q’. Crucially for our discussion, at least in some languages, this operator agrees both with the C head above it and with a wh-word under it.

143An example of a language that uses an overt Q particle is Tlingit (Na-Dene) where Q-particle is used in interrogative sentences:

144(47) Q-particles in Tlingit (Cable 2010: 7,8)

a.

Daa

i

éesh

al’óon?

what

Q

your

father

he.hunts.it

145‘What is your father hunting?’

b.

Aadóo

yaagú

ysiteen?

who

boat

Q

you.saw.it

146Whose boat did you see?’

147Q-particle accompanies the wh-word, but it is not necessarily adjacent to it, cf. (47a) and (47b). This indicates that the Q-projection with the particle in its head dominates the phrase containing wh-word and not just the wh-word. In Cable’s analysis of Tlingit Q-particle takes the phrase containing the wh-operator as complement, thus projecting the phrase minimally containing the Q-particle and its sister (Cable 2010: 38). Next, QP is the target of movement to CP through agreement with C. The interrogative C head agrees with the Q particle , as a result, the QP headed moves to CP, hence its leftmost position in (47).

148Q operator is present in all languages but is overt only in some. Thus, English sentence (48a) has the structure (48b) with Q that is not spelled-out. One of the main advantages of Q-theory, is that it reduced pied-piping to simply phrasal movement.

149(48) The Pied-Piping Structures of English Under the Q-Based Theory (Cable 2010: 143)

150a. Whose father’s cousin’s uncle did you meet at the party?

151b. [ QP [ [ [ [ whose ] father’s ] cousin’s ] uncle ] Q ] did you meet at the party?

152What is particularly relevant in Q-based theory in relation to the current analysis is that Q particle can agree not only with C but also with the wh-word c-commanded by it. The idea of Q/wh-agreement is taken by Cable from the analysis of interrogatives in Japanese and German by Kratzer and Shimoyama (2002). But differently from the agreement between C and Q, the agreement between Q and wh-operator is subject to parametric variation. For example, wh-word agree with Q particle in German and English but not in Japanese and Tlingit.

153Q/wh-agreement has important consequences in the form of restrictions on the possible position of the wh-word for the languages that have it. Thus, in English, a Q/wh-agreement language, no pied-piping past island is possible, as can be seen in (49). In contrast to that, Q does not agree with wh-operators in Tlingit and, as predicted, pied-piping past islands is possible, as shown in (50). The ungrammaticality of the construction in English is expected because, under the standard assumption, the agreement cannot apply across islands. In Tlingit, on the other hand, wh-word does not bear Q feature, and therefore is not required to agree with Q.

154(49) No pied-piping past islands in English (Cable 2010: 144)

155a. * [DP A fish [CP that is how big ]] do you want?

156b. * [DP A book [CP that who wrote ]] did you buy?

157(50) Pied-piping past islands in Tlingit (Cable 2010: 143)

a.

[[Wáa

kwligeyi

CP]

xáat

NP]

i

tuwáa

sigóo?

how

it.is.big.rel

fish

q

your

spirit.at

it.is.happy

158Literally: ‘A fish that is how big do you want?’

159To conclude, the relationship between Foc operator and DP in its c-command in Kakabe is parallel to the Q/wh-agreement posited in the Q-based theory. As is demonstrated in the remained of the paper, representing focus as an agreeing operator allows us to provide a unified account of the properties of FP and the restrictions that govern its distribution. But before proceeding to the further exploration of the expression of focus in Kakabe, some words need to be said about the analysis postulated here and the Cartographic approach, which is done in the following section.

5.3. Foc operator in the Cartographic approach

160The other dominant account of focus apart from the prosody-syntax account discussed in Section 4 is the one where it is represented as a separate location in the syntactic structure within the cartography (e.g. Rizzi 1997; 2004; Belletti 2001; Aboh 2004; 2016; Haegeman 2012). Below is provided the outline of the main points of the cartographic representation of focus in comparison to the current proposal.

161Rizzi (1997) and his followers posit a focus projection, FocP, in the left periphery, which is the CP decomposed into a number of projections as in (51). Some constituent lower in the structure can bear a focus feature due to which it enters into a relationship with the focus projection.

162(51) Structure of clausal left periphery (following Rizzi 1997):

163[ForceP ... [TopP ... [FocP ... [FinP ... [IP ... XPF ]]]]]

164The main motivation for this representation of focus comes from languages where focus constituents always appear in the left periphery and are accompanied by a focus particle. The relationship between the focus phrase and focus projection therefore appears to be straightforward: focus phrase moves to its specifier, whereas the focus particle is a lexification of the head of the focus projection. This type of focus expression is found, for example, in Kwa languages. Thus, in Gungbe (Kwa < Niger-Congo), the focused XP appears in the left periphery of the clause, cf. (52a) and (52b) with focused subject and focused object are leftmost, correspondingly. Next, there is a focus particle wɛ̀ that appears to the right of the focused XP, which, according to the standard cartographic analysis, appears in the head of the focus projection.

165(52) Gungbe (Aboh 2004: 237)

a.

Sɛ́ná

wɛ̀

xìá

wémà.

Sena

fp

read-prf

book

166Sena read a book.’

b.

Wémà

wɛ̀

Sɛ́ná

xìá.

book

fp

Sena

read-prf

167‘Sena read a book.’

168Belletti (2001; 2002) proposes that Foc projection can also be found lower than CP, namely in the vP periphery, as in (53).

169(53) vP < FocusP < (TopicP) < (TopicP) < VP.

170Her proposal is motivated by the evidence from Italian where focus in question-answer pairs appears to the right of the verb. Consider the position of the subject focus constituent in the answer in (54b).

(54)

a.

Chi

ha

parlato?

b.

ha

parlato

Gianni

who

has

spoken

has

spoken

Gianni

171Who has spoken? Gianni has spoken.

172Aboh (2007) adds to the discussion evidence from Aghem, a Grassfield Bantu language, where focus is placed in the position immediately after verb. Building on Belletti’s analysis, he proposes that the position of focus projection is subject to parametric variation: depening on the language, it can be in the articulated complementizer periphery or, lower, in the vP periphery. Accordingly, languages where focus is in the complementizer layer can be referred to as high-focus languages, and the ones with focus in the verbal layer instantiate the low-focus type.

173Turning now to Kakabe, let us suppose that there is a focus projection in the CP, and a focus feature on the focus constituent similarly to Gungbe. As for the in situ position of focus in Kakabe, it can be dealt with through positing EPP on C. In line with Miyagawa's (2001) influential analysis, the wh-in-situ phenomenon is due to the absence of EPP feature on C. In languages like English, C has EPP feature that requires wh-phrases to move to its specifier. In languages like Japanese, on the other hand, C lacks EPP, and no overt movement of wh-phrase takes place. The same can be assumed for the focus XP and Focus projection in the left periphery: Gungbe has EPP on Foc projection whereas Kakabe does not have it. As for the low-focus phrase option posited for Aghem, it is ruled out for Kakabe because of the focused subject consturction, where focus appears before auxiliary and is therefore higher than vP.

174The main point that the current analysis shares with the cartographic respresentation (in the version of Rizzi 1997) is that CP plays a role in the expression of focus. The existence of a relationship between CP and focus allows to account for the asymmetry between root and embedded clauses with respect to the possibility of focus expression that is found in Kakabe. For a particular type of embedded clauses, such as adverbial subjunctive CPs illustrated in (55), the focalization a subconstituent can not be expressed in situ. FP appears not next to such a subconstituent but instead in the matrix clause, cf. (55a) with an ‘ex situ’ FP as opposed to (55b) with in situ focus in a root clause:

175(55) a. Focus embedded in adverbial CP => FP raised to matrix clause

ka

manìɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

rice.art

buy

176(What did he ask you to buy?) ‘He asked me to buy rice.’

177(Litt. “He asked me that I should by rice”).

178b. Focus in a root clause => FP in situ

ka

kɔ̀rɛ́ɛ̀

lè

sàn.

1sg

pfv.tr

rice.art

fp

buy

179(What did you buy?) ‘I bought rice.’

180The embedded vs. root clause asymmetry with respect to focus is addressed in the so-called CP truncation analysis developed by Haegeman (2003; 2010; 2012). In this line of research, whereas root clauses have a full CP structure, certain types of embedded clauses have truncated CP, where Topic, Focus and Force projections are missing:

181(56) Root vs. Adverbial clauses according to Haegeman (2010: 632)

182a. Root clauses: Top Focus Force Mod Fin IP

183b. Adverbial clauses: Sub Mod Fin IP

184Apart from the subordinate vs. root CP asymmetry, the focus-in-CP approach is also well-suited to deal with the ban on multiple foci, that as shown in Section 7.3 is present in Kakabe. However, some questions are difficult to tackle with the version of cartographic representation as described above. First, it does not provide any instruments to deal with broad focus. Under this approach, focus is a phrase with Foc feature, but nothing is said about the relationship between the elements within focus phrase (or rather, it is implicitly assumed that focus particle will always appear either at the left or at the right of the whole focus constituent). Second, returning to example (55), though the truncation analysis does predict that focus should be treated differently two types of CP, the exact way in which focus is dealt with in the subordinate CPs is not applicable to Kakabe. If there is no focus projection, no focus can be present inside the CP that lacks this projection. But if there were no focus in (55a) then nothing would motivate the presence of FP in the matrix clause: as follows from the context, the focused constituent is in the root clause rather than in the matrix clause. To conclude, these are the reasons why Foc in the proposed analysis is closer to Cable’s (2006, 2010) Q-particle than to Foc projection in Cartography. In what follows, I will first outline the implementation of D-agreement that takes place between Foc and a DP in its domain (Section 6). After that follows the analysis of Foc-feature agreement that holds between C head and Foc projection (Section 7).

6. Foc/D-agreement

6.1. FP assignment and D-agreement

185The standard approach to structural case in the recent literature is that it is assigned to a DP by a nearby functional category when the Agree relation in the sense of Chomsky (2000, 2001) holds between a DP and this category (for recent overviews, see e.g. Baker 2015; Bárány 2018). T0 assigns Nominative case and v0 assigns Accusative case to respective DPs that they agree with.

186Parallel to case assignment by T0 and v0, in Kakabe Foc0 assigns FP to a DP thorough D-agreement relation that holds between them. In what follows I show that D‑Agreement, and, accordingly, FP assignment is subject to conditions that commonly hold for Agreement. First, applied to the case of Kakabe, DP must be in the с-command domain of Foc. Second, no other DP can intervene between the DP candidate for FP assignment and Foc (intervention condition). Third, Foc and DP must be clausemates due to locality condition: the candidate DP cannot be separated from Foc by any CP boundary. Fourth, the assignment of FP is subject to Activeness condition. As will be argued, givenness condition that disallows FP to appear on a G-marked DP is parallel to the activeness condition in T agreement when the verb fails to agree with the nearby DP if the latter is assigned a non-structural case.

187Agreement is understood here in the sense of Agree operation in the Minimalist Program as a probe-goal relationship:

188(57) Agree (following Chomsky 2000; 2001)

189(i) An unvalued feature F (a probe) on a head H scans its c-command domain for another instance of F (a goal) with which to agree.

190(ii) If the goal has a value, its value is assigned as the value of the probe.

191Based on this definition (58) formulates the operations of D-Agreement and of the FP-assignment resulting from it:

192(58) Kakabe Foc/D-Agreement and FP assignment

193(a) D-Agreement:

194Foc has unvalued D-feature and acts as probe.

195DP has valued D-feature and act as goals in the domain of Foc.

196(b) FP-assignment:

197When D-feature is valued in Foc projection, it assigns FP to the DP with the goal feature.

198Somewhat anticipating the discussion, when one of the conditions on Agree between Foc operator and DP is not met or, simply, when no DP is present in the domain of Foc operator, the valuation process fails, and FP is not assigned to a DP. In this case, Default agreement in the sense of Preminger (2009, 2014) takes place. Default agreement is discussed in 6.4, but before that we will first consider how the conditions on Agreement determine the distribution of FP in Kakabe.

6.2. Intervention condition

199To begin with, Foc/D-agreement is subject to intervention condition, according to which probe has to be valued by the closest goal. According to (58b), the assignment of FP is spelled out as enclitization to the corresponding DP. In the case of VP focus, as in (59), FP appears on the object DP due to the successful D-feature valuation as presented below (the Foc features on FocP and CP are left out for the moment).

200(59) VP focus and FP assignment

Músà

ka

Sɛ́ku

lè

gbàsi.

Musa

pfv.tr

Seku

fp

hit

201(What did Musa do?) ‘Musa hit Seku.

202FP cannot appear in (59) on the subject DP Musa because the latter is not in the c-command domain of Foc. In contrast to that, in (60) below, Foc takes the whole IP as its complement and, therefore, D-feature on the subject DP is the closest goal that it agrees with. It is also the only one due to intervention condition.

203(61) Agreement and FP assignment with Foc over a complex DP

a.

ka

[[mànsáà

la

nìngéè

lè

DP

]FocP]

màyita.

1sg

pfv

chief.art

poss

cow.art

fp

sell

204(What did you sell?) ‘I sold the chief’s cow.’

b.

ka

[[[mànsáà

la

nìngéè

lè

DP

]

màyita

VP]FocP].

1sg

pfv

chief.art

poss

cow.art

fp

sell

205(What did you do?) ‘I sold the chief’s cow.’

206FP can appear after the first DP only when Foc operator takes as complement exclusively the possessor DP, as in (62).

(62)

ka

[[mànsáà

lè

DP]FocP]

la

nìngéè

màyita.

1sg

pfv

chief.art

fp

poss

cow.art

sell

207(Whose cow did you sell?) ‘I sold the chiefs cow.’

208Example (63) illustrates the placement of FP in a postpositional phrase (PP). Here, Foc takes as complement a PP and FP is right-adjoined to the DP inside the PP.

(63)

Wò

bi

táa-la

[[mót

lè

PP]FocP]

2pl

be

go-ger

motorcycle.art

fp

in

káa

[[mɔ́nbilè

lè

PP]FocP]?

or

car.art

fp

in

209‘Do you go on motorbike or by car? [kkec_av_conv_131207_talk01_162]

6.3. Givenness and the Activity condition

  • 9 It should be noted that the givenness effect on the placement of FP is represented in a simplified (...)

210As demonstrated earlier, FP fails to appear on a G-marked DP if it can appear elsewhere within the focus constituent. For this reason, FP appears on a non-given object DP in (64a) as opposed to (64b) where the DP is given.9

(64a)

ka

Sɛ́kù

lè

gbàsi.

VP focus and non-given Object

3sg

pfv.tr

Seku

fp

hit

211(What did Musa do?) He hit Seku.

(64b)

ka

Sɛ́kùG

gbàsi

lè.

VP focus and given Object

3sg

pfv.tr

Seku

hit

fp

212(What did Seku’s mother do?) ‘She hit SekuG.’

213Under the agreement account of FP assignment proposed here, this givenness effect can be attributed to activity condition on agreement or something structurally analogous to it: G-markedness makes a DP inactive for agreement.

214Activity condition as it is formulated for φ-feature agreement (Chomsky 2001) has it that agreement is blocked by a non-structural case assigned to the DP. Consider (65) from Icelandic discussed by Preminger (2014: 130) and originally from Sigurðsson (1996). The DP with dative case is unable to agree with the finite verb, and therefore no agreement between the subject and the verb takes place in (65a). This contrast with the case where the same verb stem, used in the meaning ‘walk hand in hand’ takes nominative instead of dative argument, and the φ-features of the argument appear on the verb form (65b).

215(65) Agreement in Icelandic (Sigurðsson 1996)

a.

Strákunum

leiddist/*leiddust.

Dative => Agreement blocked

boy.the.pl.dat

were.bored.3sg/*3pl

216‘The boys were bored.’

b.

Strákarnir

leiddust/*leiddist

Nominative => Agreement

boy.the.pl.nom

walked.hand.in.hand.3pl/*3sg

217‘The boys walked hand in hand.’

218Another clear example of such a differential agreement can be found in Hindi-Urdu (Bhatt 2005). Hindi is a split ergative-accusative language: depending on the construction, agents can be ergative or non-marked (nominative), whereas objects can be marked accusative or non-marked (absolutive). One sentence can combine two marked as well as two unmarked arguments or have only one of them with a marked case. The verb agrees with the structurally most prominent argument that is not case-marked overtly. Accordingly, when both are marked, a default agreement marker, namely, that of masculine singular, appears on the verb. Thus, (66a) combines two non-overtly case marked arguments, and since nominative argument is more salient, it triggers agreement on the verb. In (66b), on the other hand, the agent is ergative, hence overtly marked, and the object is non-marked and transmits the values of its φ-features to the verb. Finally, in (66c) both arguments are overtly case-marked, so none of them is accessible for agreement and default agreement marker appears on the verb.

219(66) Hindi-Urdu agreement (Bhatt 2005: 759, 768)

220a. Nominative subject, Accusative object, both non-overtly case-marked

Rahul

kitaab

par:h-taa

thaa.

Rahul.M

book.F

read-hab.m.sg

be.pst.m.sg

221‘Rahul used to read (a/the) book.’

222b. Ergative subject, Accusative object, only object is non-overtly case-marked

Rahul-ne

kitaab

par:h-ii

thii.

Rahul-erg

book.f

read-pfv.f

be.pst.f.sg

223‘Rahul had read the book.’

224c. Ergative Subject, overtly marked accusative object

Mona-ne

is

kitaab-ko

par:h-aa

thaa.

Mona.f-erg

this.obl

book-acc

read-pfv.m.sg

be.pst.m.sg

225‘Mona had read this book.’

226Preminger (2014: 160) analyses such systems where case forms are distributed into those that can trigger agreement as opposed to those that cannot do it as case discrimination. Thus, in Icelandic, unmarked case is accessible for agreement, whereas dependent case is not. See also the discussion of case discrimination in (Bárány 2018). We will return later to the question of default agreement that takes place when no DP can be accessed for agreement.

227The discrimination between DPs with respect to FP assignment as in (64) is also reminiscent of Differential Object marking (DOM) phenomenon, known since (Bossong 1991); for a detailed description see (Dalrymple and Nikolaeva 2011). The parallel is even closer considering that DOM is concerned with the discourse status of the object. In languages with DOM, topical/specific objects display overt marking whereas non-specific and non-topical are unmarked. This can involve the dependent marking i.e. Accusative case as opposed to none on the DP, or head marking, i.e. presence or absence of object agreement marker on the verb. To illustrate the latter, in Palauan (Austronesian) verbs agree with objects, but only if they are both singular and specific, cf. (67a) and (67b).

228(67) Palauan differential object agreement (Woolford 2000: 218)

a.

Te-‘illebed-ii

a

bilis

a

rengalek.

subj.3pl-prf.hit-obj.3sg

the

dog

the

children

229‘The kids hit the dog.’

b.

Te-‘illebed

a

bilis

a

rengalek.

subj.3pl-prf.hit

the

dog

the

children

230‘The kids hit a dog/the dogs/some dog.’

231To summarize, the dependence between discourse status of a DP and the possibility to assign FP to it enters into a wider range of phenomena that can be described as discrimination between DPs with respect to their ability to enter into agreement.

6.4. Locality condition

232Finally, agreement is subject to locality condition which means that it cannot cross the CP boundary (Chomsky 2000, 2001)

233Thus, in (68), wótè ‘money’ is the DP that is expected to host FP, considering the above discussion. It is inside a focused VP, therefore c-commanded by Foc operator, and it is the first non-given DP in the domain of operator. Nevertheless, as can be seen, FP cannot appear on it. This restriction is due to the fact that the D feature of Foc operator cannot probe for its valued counterpart on DP across the CP boundary.

(68)

maɲìninka

lè

[CP

ni

wótè

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

money.art

dí

bólo].

give

3sg

to

234(What did he do?) ‘He asked me to give him money.’

*À

maɲìninka

[CP

ni

wótè

lè

3sg

pfv.tr

1sg

ask

1sg

sbjv

money.art

fp

dí

bólo].

give

3sg

to

235Compare it to (69), where a non-given DP also follows the verb and is under Foc operator, but, in contrast to (68), it is not separated by a CP boundary from Foc. As expected, FP appears on this postverbal DP inside the postpositional phrase.

(69)

[ǹ

maɲìninka

[dɛ̀ɛmárè

lè

laPP]

VP].

3sg

pfv.tr

1sg

ask

help.art

fp

obl

236(What did he do?) ‘He asked me for help.’

237At the end of the section we will return to the question why in (68) FP appears after the verb in the matrix clause rather than at the end of the embedded CP.

6.5. Default Agreement

238As demonstrated above, D-feature of Foc probes for a valued D-feature on DP, and in certain situations the valuation is successful which leads to the placement of FP on DP. However, as has been shown, there are also numerous cases when the DP does not meet the required conditions or when there is simply no DP in the c-command of Foc. Chomsky’s (2000, 2001) version of Agree needs to be successful, otherwise the derivation does not converge. Preminger's (2009; 2014) approach to agreement minimally differs from Chomsky’s (2000, 2001) Agree in that, for him, it is a fallible operation, meaning that it is allowed to fail. In his model, when a probe attempts to value its features and fails to do so, the derivation, nevertheless, does not crash since probe can get a default value. This is what is found in the examples from Icelandic and Hindi-Urdu cited earlier. When no DP with an unmarked (structural) case is available, the verb form is assigned a default agreement value (65a) for Icelanding and (66c) for Hindi-Urdu. The example with DOM in (67) from Palauan, where no value is assigned to the verb, could also be interpreted along these lines.

239Going back to Kakabe, as can be seen in the examples given earlier, when the D-probing of Foc fails to find a valid goal and therefore, FP is not placed on a DP, it is placed elsewhere. The default outcome is the merger of FP with the head of Foc. On the model of the findφ(f) operation in Preminger (2014: 159), the D-Agreement Foc in Kakabe can be defined as follows:

240(70) D-Agreement in Kakabe:

241Given an unvalued feature D on Foc, look for an XP bearing valued instance of D. Upon finding such an XP, check whether its discourse status is acceptable with respect to activity condition.

242yes => assign FP to the DP with the goal feature

243no => merge FP with the Foc0

  • 10 It should be noted that, as already mentioned, under the current analysis, Foc operator can take as (...)

244Examples (71) and (72) illustrate the default outcome of Foc/D Agreement in Kakabe. In both cases, FP is linearized sentence finally, but this results from two different positions of Foc operator. In (71) Foc operator takes as complement VP and since no DP is present in it, FP is in Foc0. In (72), the complement of Foc is IP, and the DP in subject position is G-marked, therefore, agreement is blocked and, again FP appears in Foc0.10

245(71) VP focus (72) IP focus

[

[sɔnkɔ-taVP]

lèFocP]

[IP[

bòyi-ta]

lèFocP].

3sg

cry-pfv.tr

fp

1sg

fall.down-pfv.tr

fp

246(What did he do?) ‘He cried.’ (What happened?) ‘I fell down.

247The default agreement is also the outcome in (64b) where VP focus contains a G-marked object DP, with which Foc does agree because of its givenness, and therefore FP appears after the verb.

248A special case of default agreement is represented by a situation when Foc immediately dominate a DP that is G-marked as in (73). FP linearly appears after such DP, since nothing else is present under Foc operator, but, since agreement is blocked, it appears not on DP in the syntactic structure but in the default Foc0 position.

(73)

[Ňdè

G

lè

FocP]

ka

kɔ̀rɛ́ɛ̀

sàn.

1sg.lg

fp

pfv.tr

rice.art

buy

249(Who bought the rice?) ‘It is me who bought the rice.’

250Finally, it is necessary to comment on the position of FP in the case of a subordinate CP included in focus, where FP appears to be before the CP discussed in relation to locality condition in (68). The seemingly internal position of FP with respect to the focus phrase that is found there, is due to the fact that CP undergoes extraposition, and (68) therefore had the representation as in (74). In line with this analysis, the subordinate CP is base-generated in VP and then undergoes extraposition to an IP-adjoined position before spell-out. Therefore, FP does appear at the end of focus phrase and not inside it, if the base structure is concerned. Accordingly, probing for D-feature on a DP and the licencing of FP on this DP takes place after the extraposition, otherwise the CP would be extraposed with the FP.

251(74) Prosodic extraposition of subordinate CP

[VP

maɲìninka

ti

lè]

[CPi

ni

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

3sg

ask

fp

1sg

sbjv

rice.art

buy

252(What did he do?) He asked me to buy rice.

253Extraposition of sentential complements is triggered by a conflict between the mapping constraints and prosodic well-formedness (Büring 2013: 870). CPs require to be mapped onto a separate intonation phrase, which is problematic when the CPs is inside another CP. In order to improve this situation, the embedded CP is extraposed to the right of the utterance. Evidence that supports the existence of this motivation for extraposition in Kakabe is that, in this language, the extraposed CPs are preceded by downdrift reset. And, as already argued in Section 4.3, downdrift reset is an indication of intonation phrase boundary.

7. C/Foc agreement

254This section analyses the restrictions on the distribution of FP that, as I argue, follow from the agreement between C and Foc projections. These restrictions can be grouped into three types: (i) a ban on FP inside adverbial and relative CPs as opposed to root clauses and speech verb CP complements; (ii) a ban on multiple FP instances within one CP; (iii) incompatibility within one CP of FP and the focused perfective auxiliary.

255The distinction between feature types in feature valuation theory by Pesetsky and Torrego (2007) are crucial for the analysis of the relationship between FocP and CP. Therefore, before proceeding to the analysis of these three groups of cases, the following section provides the outline of feature valuation theory that will be used in the analysis.

7.1. Feature valuation theory (Pesetsky and Torrego 2007)

256The core element of feature theory of Pesetsky and Torrego (2007) is that interpretability and valuation are feature properties independent from one another. It therefore differs from Chomsky’s (2000, 2001) representation of features and agreement that posits a biconditional relation between feature values: interpretable features are valued, whereas uninterpretable features are unvalued: “A feature Foc is uninterpretable iff Foc is unvalued”.

257The interpretability property defines whether a feature makes a contribution to semantics or not (only interpretable instances of F do), and this property cannot be inspected by syntax. The valuation property, on the other hand, is visible to syntax. For example, the number and gender feature is valued in the lexical entries of nouns, whereas determiners and adjectives are lexically unvalued and receive their values from nouns through agreement. Pesetsky and Torrego (2007) dispense with Chomsky’s biconditional that ties together interpretability and valuation and assume instead that these two feature properties are independent from one another. Therefore, in their model, the following feature types are possible:

258(75) Features types (Pesetsky and Torrego 2007: 269)

259uFval uninterpretable, valued iFval interpretable, valued

260uF [ ] uninterpretable, unvalued iF [ ] interpretable, unvalued

261To give an example, the interpretable instance of tense feature, iT[] on the Tns head ‘learns its value’ from the uninterpretable but valued instance of tense on v, uTval.

262The second basic component of their theory is that Agreement is viewed as ‘feature sharing’, an approach current in Minimalist and HPSG frameworks. Pesetsky and Torrego (2007) make a distinction between occurrences of F and instances of F. Occurrences of F are distinct Fs present in the syntactic structure that might undergo agreement but have not done so yet. Instances of F, on the other hand, are the output of Agree. The application of Agree results in the replacement of the unvalued instance of F that acts as probe by another instance of F as formulated in (76).

263(76) Agree as feature sharing (Pesetsky and Torrego 2007: 268)

264(i) An unvalued feature F(a probe) on a head H at syntactic location α (Fα) scans its c-command domain for another instance of F (a goal) at location β (Fβ) with which to agree.

265(ii) Replace Fα with Fβ, so that the same feature is present in both locations.

266If the goal F has a value, the outcome of Agree, following (56), is that this value is present both in the location of the goal and in the location of the probe: the unvalued probe is replaced with the valued instance of the goal. Going back to the example of tense feature, before Agree, Tns has unvalued T whereas after Agree it has a valued instance of T, which is a copy of the valued T located on v.

267Agree is constrained by an LF-interface restriction based in the Radical Interpretability principle adopted from (Brody 1997: 143-144). According to this restriction, each feature must receive a semantic interpretation in some syntactic location. Within the feature sharing approach to agreement as formulated in (76), this means that an uninterpretable valued F must enter into Agree relation with an interpretable instance of F (Pesetsky and Torrego 2007: 272). Thus, tense feature on v must agree with tense feature on T head, because it is in this location that it can obtain an interpretation. On the other hand, valuation is a precondition for interpretability. This means that interpretable but unvalued features are compelled to act as probes and, accordingly, it rules out structures where an interpretable unvalued feature can find no valued counterpart in its domain. From this follows that for every occurrence of a valued uninterpretable F there must be an interpretable unvalued occurrence of F that can probe for it. Resuming the above said, the interpretability condition can be formulated as in (77).

268(77) Interpretability condition on Agreement (based on Pesetsky and Torrego 2007)

269Every occurrence of uFval must be probed by iF[ ]

270From (77) follows, for example, that uFval is licensed in a CP only if this CP contains an occurrence of F that is interpretable and not yet valued. This will be crucial for the analysis of Kakabe that follows.

271The raising of the subject in raising infinitival constructions as in (78) is triggered by the fact that otherwise Mary cannot be interpreted. The proposal of Pesetsky and Torrego (2007) is that v in infinitival clauses has an uninterpretable instance of T, as opposed to v of a finite clause. Accordingly, the interpretation of the T feature on Mary can occur only when it raises to the matrix clause.

272(78) Mary seemed to like the play.

7.2. CP types and FP licensing

273Let us now turn to Kakabe. Adverbial CPs display obligatory raising of FP to the matrix clause, whereas CPs of the type of speech verb complements allow focus to be realized in situ, the same way as it is possible in a root clause. In what follows I propose an analysis of this phenomenon in terms of Agreement and feature valuation.

274Example (79) shows that FP cannot appear inside and embedded adverbial clause when focus is on an argument inside this embedded CP. As can be seen, FP appears instead before the embedded clause. Interpretations with broader focus, e.g. VP focus are also possible, as shown in (79a).

275(79a) Object focus in an adverbial CP => FP raised to matrix clause

ka

manìɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

ka

manìɲinka

lè

[cp

ni

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

rice.art

buy

276(What did he ask you to buy?) He asked me to buy rice.

277(What did he ask you to do?) He asked me to buy rice.

278(Litt. “He asked me that I should by rice”).

279(79b) *à ka ǹ manìɲinka [cp ǹ ni kɔ̀rɛ́ɛ̀ lè sàn]

2803sg pfv.tr 1sg ask 1sg sbjv rice.art fp buy

281Apart from adverbial subjunctive clauses, FP are infelicitous inside relativized CPs, conditional antecedents as well as preposed temporal-adverbial CPs. The illustrations are given in Section 7.6 which analyses the exact position of FP resulting from raising.

282Importantly, as already mentioned, not all subordinate clauses disallow the internal placement of FP. Thus, in (80) FP appears in situ, i.e. adjacent to the DP in focus inside the subordinate speech report.

283(80) Object focus inside a Speech report CP => in situ FP

ka

[VP

fɔ́

[CP

(kó)

Músà

ka

[kɔ̀rɛ́ɛ̀

lè]FocP]

sàn].

3sg

pfv.tr

3sg

say

qu

Musa

pfv.tr

rice.art

fp

buy

284‘He said that it was rice that Musa bought.’

285Note that the case illustrated in (79) is different from the case that we were dealing with in Section 6.3, reproduced in (81) for convenience.

(81)

[[ǹ

maɲìninka

lè

[CP

ni

wótè

dí

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

money.art

give

bólo]

VP]FocP].

3sg

to

286(What did he do?) ‘He asked me to give him money.’

  • 11 As discussed in Section 6.5, the adverbial CP is extraposed before spell-out, but is nevertheless b (...)

287In (81), FP is assigned by Foc located in the matrix clause, and FP cannot be licensed on the DP inside the adverbial clause because of the CP boundary (locality restriction).11 In cases like (79), on the other hand, Foc operator originates inside the imbedded CP taking the focused DP inside it as complement, therefore there is no CP boundary to separate the two and Foc/DP agreement is not blocked. Therefore, the reason FP appears not adjacently to the DP inside the adverbial clause is different.

288The existence of this contrast is confirmed by the fact that speech verb complements that, as has been shown in (80), can focus its subconstituents with a FP in situ, at the same time, cannot have a FP inside if it is assigned from Foc operator of the matrix clause, as shown in (82). In other words, as it is expected, locality condition is violated just the same by the presence of clause boundary, irrespectively if this clause is a speech verb complement or a subjunctive clause.

289(82) Speech report CP embedded under Foc ≠ (80)

ka

fɔ́

lè

[CP(kó)

Músà

ka

kɔ̀rɛ́ɛ̀

sàn].

3sg

pfv.tr

3sg

say

fp

qu

Musa

pfv.tr

rice.art

buy

290(What did he do?) ‘He said that it was rice that Musa bought.’

291All the above said indicates that there are two types of subordinate CPs with respect to the acceptability of licensing FP inside it: (i) CPs like speech verb complements that are not subject to focus-related restriction and (ii) CPs like adverbial or relative clauses that do not allow FP inside. Feature valuation theory by (Pesetsky & Torrego 2007) outlined earlier is well-suited to treat this contrast.

292I propose for Kakabe that the head of CP has an unvalued instance of Foc feature that can be of two types: interpretable unvalued iFoc[ ] and uninterpretable unvalued uFoc[ ]. As for Foc operator, it has the valued but uninterpretable counterpart of the feature, uFoc+. Therefore, in the syntactic structure, focus receives lexical valuation on the Foc operator but is interpreted on C. And since, as just said, not all CPs contain a C head with an interpretable instance Foc, therefore, interpretation is not possible in all of them. Next, when the interpretation is not possible, in order to avoid violating the interpretability condition (77), Foc operator moves to the matrix clause (or, alternatively, to the edge of the embedded CP as argued in 7.5). This final position is where OpF licenses FP.

293In line with this analysis, no restrictions are attested on FP inside a gerund constituent, as can be seen in (83). Gerunds only have VP structure and do not introduce an additional CP. Therefore, Foc operator and its features inside a gerund VP remain visible for the CP features of the root clause.

(83)

fɔ́lɔ-ta

[sòbéè

lè

dàmu-la

VP].

3sg

start-pfv.I

meat.art

fp

eat-ger

294(What did he start eating?) ‘He started eating meat.’

295My analysis of the contrast between CPs follows the logic of Pesetsky and Torrego’s (2007) Agree account of how CP are typed in relation to wh-phrases. Examples from their paper reproduced below show that a wh-phrase have to match with a specific type of CP:

296(84) Wh-phrases and clause types (Pesetsky and Torrego 2007: 271)

297a. I wonder [what Mary bought __ ]. (interrogative)

298b. *the book [what Mary bought __ ] (relative)

299(85) a. I wonder [why she left]. (interrogative)

300b. the reason [why she left] (relative)

301c. *John left [why Mary left]. (free relative)

302In their analysis, this matching between CP type and wh-type is explained by agreement through shared Q feature. Interrogative CPs have an interpretable unvalued Q feature, CiQ[ ] , whereas declarative CPs bear an uninterpretable unvalued Q feature, CuQ[ ]. What is a wh-phrase that bears a valued Q probed by its unvalued counterpart on C. Next, C with uninterpretable Q, even though it cannot interpret the valued Q of the wh-phrase, can serve as an intermediate landing site in the successive cyclic movement in cases like (86).

303(86) What do you think [ __ we should do __ ]?

304Going back to Kakabe, it follows from the interpretability condition on agreement (77) that the valued instance of Foc on Foc operator must agree with its interpretable counterpart. However, when Foc is under a CP with unvalued Foc on C, the required agreement is impossible. The iFoc[ ] on C of the main clause, obviously, cannot agree directly with uFoc+ of the operator in the embedded clause for the locality reason. The impossibility of interpretation through Agree within a subordinate CP is, therefore, a trigger for the raising of Foc operator to the matrix CP.

  • 12 It should be noted that, even though subordinate clauses show no restriction on the prosodic expres (...)

305It should be noted that the ban on in situ focus marking in embedded clauses attested in Kakabe is part of a more general cross-linguistic tendency. Restrictions on the expression of information-structure categories within subordinate clauses have been reported for a number of unrelated languages, including Tundra Yukagir (Matić 2014), Aghem (Hyman and Polinsky 2009) and Somali (Saeed 2004: 269-270). In languages where overt marking of information-structure categories such as topic and focus is not allowed inside a subordinate clause, the corresponding meanings are either left unspecified or are expressed indirectly (Matić et al. 2014: 14). Quite tellingly, the languages that are reported to have restrictions on the expression of focus in an embedded clause all mark focus morphologically. In contrast to that, there is no such restriction on the prosodic focus marker. As already mentioned in Section 4.5, languages like English put the sentence stress on the focus constituent irrespectively of whether focus is in the root or in a subordinate clause (illustrated in example 41). This is in line with the general argument developed in the present paper, namely, that morphological focus is an element determined on the syntactic structure, as opposed to sentence stress signalling focus in languages like English and related to syntax only via prosody. Accordingly, since sentence stress is not an element directly determined by syntax, Foc feature on C does not affect it.12

306In other words, in Kakabe, but possibly in languages with morphological focus more generally, focus marking is a root clause phenomenon. It has been known at least since Hooper and Thompson (1973) that root clause phenomena are allowed in some embedded clauses, that are therefore not ‘truly subordinate’; see for overview Heycock (2006). Hooper and Thompson's (1973) proposal associates root clause phenomena with assertion and with their own illocutionary force. Sentential complements of verbs of saying are commonly recognized to be asserted, cf. the acceptability of focus marking inside complements of ‘say’ in Kakabe as opposed to adverbial and relative clause.

307To conclude, in the framework of feature valuation theory, the specific behaviour of subordinate clauses with respect to focus translates as the uninterpretable character of Foc feature on C in the periphery of a subordinate clause. One of the advantages of this representation, for the case of Kakabe, is that it explains the raising of FP to the matrix clause as motivated by the need to be probed by an interpretable feature of the root clause.

7.3. Ban on multiple foci

308The next piece of evidence in favour of positing Foc feature on C and its valued counterpart on Foc operator is the ban on multiple foci that exists in Kakabe. Consider (87), where, following from context, both direct and indirect object DPs should be focused. However, FP can appear only on the direct object (87a) and FP on both direct and indirect object is not allowed as shown in (87b). Finally, (87c) demonstrates that FP can be placed only on the higher DP of the two, therefore, the structure with FP on the postverbal DP is infelicitous in the context where focus is on a higher DP as well.

(87a)

Ò’óyè

ka

bàâ

lè

dí

Músà

bólo.

no

1sg

pfv.tr

goat.art

fp

give

Musa

to

309(Did you give a cow to Alseniy?) ‘No, I gave a goat to Musa.’

(87b)

*Ò’óyè

ka

bàâ

lè

dí

Músà

lè

bólo.

no

1sg

pfv.tr

goat.art

fp

give

Musa

fp

to

(87c)

Ò’óyè

ka

bàâ

dí

Músà

lè

bólo.

no

1sg

pfv.tr

goat.art

give

Musa

fp

to

310#(Did you give a cow to Alseniy?) ‘No, I gave a goat to Musa.’

311OK(Did you give a cow to Alseniy?) ‘No, I gave a goat to Musa.’

312See also (88) from a natural text, where the context imposes a discontinuous focus on subject and then on the locative object, whereas FP appears only on the subject DP in the first clause:

(88)

Dóo-n

dè

nà-la

ɲín-na

mà

yen,

other-pl

fp

come-ger

3sg

look.for-ger

1pl

for

mà

béle

táa-la

ɲín-na

kàn

do.

1pl

be.neg

go-ger

3sg

look.for-ger

place

some

313‘Others come to our place to look for them [for the shoes], we do not go to look for them in a different place’ [kkec_av_conv_131220_mosquee6_102].

  • 13 The licensing of FP is parallel in some respects to wh-phrase movement in languages like English. M (...)

314The ban on multiple FPs demonstrated in (87b) is expected within the C/Foc agreement account. If more than one Foc feature is present in the domain of one CP, only the goal feature closest to the probe feature on C can be interpreted through valuation. This is so because, first, the closest occurrence of the feature is probed first and, second, once an unvalued instance of a feature is valued, it stops probing. The other valued Foc instances would therefore be left with their Foc features uninterpreted, which goes against the interpretability requirement (58). Hence the restriction in (89).13

315(89) Maximum one Foc operator in the domain of CiFoc[ ]

316From (89) follows that the language simply does not have a way to express multiple foci inside one CP. In line with that, the double foci interpretation for the structure in (87a) with FP after the object DP, is one of the interpretations along with, whereas in languages like English these interpretations are mapped onto two structures distinct by their stress pattern.

(87’)

’óyè

ka

bàâ

lè

dí

Músà

bólo.

no

1sg

pfv.tr

goat.art

fp

give

Musa

to

317(Did you give a cow to Musa?) ‘No, I gave a goat to Musa.’

318Going back to the generalization in (89), the specification of the CP type in as CiFoc[ ], namely, that its head bears an interpretable unvalued instance of Foc, is necessary since clauses with uninterpretable instances of Foc are not able to satisfy the interpretability requirement. In what follows, it will be shown that this restriction bears a more general character since it involves not only Foc occurrences but also concerns Foc feature that can be present on an auxiliary.

7.4. FP and the perfective focus split14

  • 14 Editor’s note: Alexandra also published a lengthy text dealing with this topic: Vydrina, Alexandra. (...)

319The marking of the perfective aspect in Kakabe is characterized by what can be called ‘focus split’ that is instantiated in the contrast between the inflectional markers báti and ka/-ta. Their aspectual value is identical and they differ by their relationship to focus (as for the ka vs. -ta distinction, ka is used with transitive predicates, whereas the suffix -ta appears on intransitive predicate, see Section 2). The utterance with the auxiliary báti in (90a) has the interpretation with the focus on the aspectual value of accomplishment or on the truth value. (90a) is appropriate, for example, in a situation where the listener expects the meat to be prepared, therefore the accomplishment of this event and not the content of the event is at issue. By contrast, when the auxiliary ka is used, as in (90b), the aspectual value of perfectivity is not in focus. Crucially, báti cannot co-occur with FP, as shown in (90c).

(90a)

báti

sòbéè

tàbi.

Aspect/polarity focus

3sg

pfv.f

meat.art

prepare

320(Has he prepared the meat or is he still doing it?)

321(Has he prepared the meat or hasn’t he?)

322‘He has prepared the meat.’

(90b)

ká

sòbéè

lè

tàbi.

3sg

pfv.tr

meat.art

fp

prepare

323(What did he prepare?) ‘He prepared the meat.’

(90c)

*À

báti

sòbéè

lè

tàbi.

3sg

pfv.f

meat.art

fp

prepare

324Apart from the restriction to co-occur in the same CP with FP, báti is also banned from interrogatives with wh-words (91a), as opposed to polarity questions (91b) where báti and not ka/-ta is used:

(91a)

kà

(*báti)

fɛ́n

sàn?

wh-interrogative

2sg

pfv.tr

pfv.f

what

buy

325‘What did you buy?’

(91b)

báti

nìngéè

sàn?

polarity interrogative

2sg

pfv.tr

cow.art

buy

326‘So, have you bought the cow (as expected)?’

327Next, báti is excluded from relativized clauses (92a) and from conditional antecedents (92b): only ka can be used in such environments.

328(92a) Relativized clause

[REL

ká

(/*báti)

nìngéè

mín

sàn]

ànu

báti

wò

dàmu.

3sg

pfv.tr

pfv.f

cow.art

rel

buy

3pl

pfv.f

that

eat

329‘They have eaten the cow that he bought.’

330(92b) Conditional antecedent

[COND

S-àa

ká

(/*báti)

nìngéè

sàn]

àn

si

wò

dàmu.

if-3sg

pfv.tr

pfv.f

cow.art

buy

3pl

pot

that

eat

331‘If he buys a cow, they will eat it.’

332The pattern of distribution of the two Kakabe perfectives indicates that, within the framework adopted in the present paper, báti should be analysed as lexically specified for valued Foc feature, uFoc+. The ka/-ta perfective, on the other hand, bears an uninterpretable and unvalued Foc. From this analysis naturally follows that I0 heads with [báti, uFoc+] cannot appear in CPs that have another instance of uFoc+, and Foc operator does have such a valued instance of Foc. In other words, the ungrammaticality of (90c) is explained by the fact that there are two uninterpreted but valued Foc, one on I0, and on Foc0, whereas there is only one interpretable counterpart on C0 for feature valuation, as presented in (93).

333(93) * CiFoc[ ] … IuFoc+ … Foc uFoc+

334Therefore, the restriction of in (89) can be reformulated in a more general way as in (94) which can be considered as a subconstraint following from the interpretability constraint (77).

335(94) Maximum one X0 of the type ­[uFoc+] in the domain of CiFoc[ ]

336Let us now look again at (91a). As Beck (2006) and Cable (2010), inter alia, claim, wh-words are obligatorily focused when functioning as interrogative operators in wh-questions. Therefore, the incompatibility of báti with wh-words in (91a) is due to the same reason as the incompatibility between báti and FP or between two instances of FPs under one CP. All these cases lead to a situation where one of the valued Foc features remains uninterpreted.

337Finally, the ban on báti in a conditional protasis or a relative clause as shown in (92) is due to the same reason as the raising of Foc operator from embedded clauses to the matrix clause: the impossibility for uFoc+ to receive valuation under CuFoc[ ]. However, the outcome for the focused perfective báti is different as compared to Foc operator. For Foc operator, the problem of the impossibility to receive an interpretation of its feature within a subordinate CP is resolved through raising to the matrix clause. As for báti, it cannot raise to the matrix clause for the obvious reason that the I0 of the matrix clause is already occupied by a lexical head, therefore, it is simply unacceptable in this environment.

338There is one last point that should be added about the distribution of the two perfectives. There is a specific type of conditional where báti is allowed in the antecedent CP; it is illustrated in (95a) and (95b).

(95a)

Sì

báti

nà

háray

bì

díya-la

lè

dóndèn.

if

2sg

pfv.f

come

so

3sg

be

please-ger

fp

little

339‘If you have come here, it is because you like it’ [sajoya_SNKeita_2009_031].

(95b)

S-àa

báti

ɔ̀ya,

àrɛti!

if-3sg

pfv.f

finish

3sg

stop

340‘If (as you say) it is finished, stop now!’ [tale_mammadu-boyi-konde2_23dec2011_140]

341Haegeman (2003) refers to the conditional of the type (95) as premise-conditionals as opposed to the event-conditional corresponding to (92b). An event-conditional, according to her definition, contributes to event structure, it modifies the main clause event. A premise-conditional, on the other hand, structures the discourse: “it makes manifest a proposition that is the privileged context for the processing of the associated clause” (Haegeman 2003: 319). See also Bhatt and Pancheva (2006: 671) for a similar analysis of two types of conditionals. Crucially, according to Haegeman (2003), the respective antecedents of these two types of conditionals differ by their internal syntax, namely, by the type of their left periphery. As already discussed in Section 5.1, Haegeman (2003, 2006, 2010) proposes that CP can be complete or truncated. Event-conditional antecedents belong to the second type and have a reduced CP that does not contain Force, Topic nor Focus projections, whereas the premise-conditionals belong to the first type and have a complete left periphery, including Focus (Haegeman 2003: 335). As already argued, the truncation analysis whereby Focus is absent altogether from the clause does not fit Kakabe evidence, therefore, instead the opposition between CP types is assumed to follow from the distinctions between the types of Foc features as outlined above. But despite the difference in the exact analysis of CP types, the opposition between the two types of conditionals points into the same direction, namely that Foc expression is affected in true subordinates as opposed to other types of embedded CPs.

342In relation to the more general question of focus marked jointly with TAM and polarity, it should be noted that this phenomenon is found in grammars of many African languages. Waters (1979) and Hyman and Watters (1984) were the first to attract attention to the fact that focus can target inflectional categories, mainly on the data of Bantu languages. The exact way in which focus is involved in the inflectional paradigm differs across languages. There are languages where the focus split appears across multiple TAM categories. For example, in Hausa (Chadic), both perfective and imperfective have different focused and non-focused forms. In KiRundi (Bantu), the focus split is present in three TAM categories: Immediate Past, Recent Past and Future. In Efik (Bantu), it is present in past, present and future, but not in progressive (examples of these three languages are cited from Hyman & Watters 1984: 249). It is now commonly accepted that the conjoint/disjoint alternation, which is typical of Eastern Bantu languages, has to do with differences in information structure, correlating directly or indirectly with constituency (Hyman and Watters 1984; (Güldemann 2003; Watters 2010; Van der Wal 2017). In all the reported cases the same distributional restrictions are attested: focused TAM or polarity forms are excluded from conditionals antecedents, wh-interrogatives, relativized clauses and they are incompatible with argument focus.

7.5. Landing site for Foc

343The last question that will be addressed in this section is that of the landing site for Foc operator after it raises to the matrix clause for the reasons discussed above. We have already seen a case where an adverbial clause appears to the right of the matrix clause, an example is reproduced in (96). It has been argued in Section 4.5, that adverbial clauses undergo a phonologically conditioned extraposition to the right. Therefore, the position of FP in (96) can be interpreted in the following way. The necessity to escape from the CP with the uninterpretable Foc feature triggers the raising of Foc operator to the matrix CP where the feature becomes interpretable. One can assume that the operator raises to the position where it adjoins to the closest XP node that dominates this CP. In the case of (96) this XP is VP since the adverbial clause is base-generated in the position of postverbal argument inside a VP. After that, the adverbial CP is extraposed to the right.

344(96) Object focus in an adverbial CP => FP raised to matrix clause

ka

[ǹ

manìɲinka

ti

lè]

[cp

ni

kɔ̀rɛ́ɛ̀

sàn]i.

3sg

pfv.tr

1sg

ask

fp

1sg

sbjv

rice.art

buy

345(What did he ask you to buy?) ‘He asked me to buy rice.’

346Let us now look at relativization. The most common strategy of relativization in Kakabe is correlativization, as illustrated in (97a). A head-internal relative clause precedes the root clause, and a resumptive determiner wò appears in the position of the relativized argument in the root clause. When focus is on a DP inside such a preposed relative clause, FP is spelled out on the resumptive pronoun in the main clause (97a). Again, FP is not possible within the relative clause, as shown in (97b).

(97a)

[RELSàâ

kà

kàyéè

mín

kìn]

ànu

ka

wò

lè

lakɛ̀ndɛya.

snake.art

pfv.tr

man.art

rel

bite

3pl

pfv.tr

that

fp

heal

347‘They healed the man who the snake bit’ (not the man bitten by the scorpion), Litt.: “The snake bit that man, they healed him”.

(97b)

??[RELSàâ

lè

kà

kàyéè

mín

kìn]

ànu

ka

wò

snake.art

fp

pfv.tr

man.art

rel

bite

3pl

pfv.tr

that

(lè)

lakɛ̀ndɛya.

fp

heal

348The placement of FP appears on the resumptive demonstrative can be easily explained within Bhatt's (2003) analysis of correlatives. According to this analysis, the relative clause is base-generated in the position adjoined to the DP that it modifies and is then fronted to an IP-adjoined position. Transposing this analysis to Foc operator and relativization in Kakabe, first, Foc operator raises to escape from the relativized CP where its Foc feature cannot be interpreted, after which it is adjoined to the closest XP above this CP which is the relativized DP. Finally, the relative clause is fronted, leaving behind Foc operator adjoined to the DP in the matrix clause.

349Next, (98) illustrates the case of a focus DP inside an adverbial clause that is preposed to the matrix clause. As can be seen, FP in this case is placed at the end of the adverbial clause, just before the matrix clause. This can be interpreted in a way that Foc operator raises to the specifier of the embedded CP (providing that right-branching specifiers are possible, cf. the Spec-Rt constraint in (Grimshaw 1997). And since, according to Chomsky (2001), the specifier position is the edge of a phase that is, therefore, visible to the higher CP, this renders the moved element visible to the agreement features of the matrix C.

350(98a) VP focus in an adverbial CP => FP raised to matrix clause

’óyè

àn

ka

bùntálè

yén

nɛ́ɛ̀

lè

ànu

kúle-ta.

no

3pl

pfv.tr

scorpion.art

seen

when

fp

3pl

cry-pfv.i

351(Is it when they saw a snake that they cried?)

352‘No, it is when they saw a scorpion that they cried.’

(98b)

??Ò’óyè

àn

ka

bùntálè

lè

yén

nɛ́ɛ̀

ànu

kúle-ta

no

3pl

pfv.tr

scorpion.art

fp

see

when

3pl

cry-pfv.i

353The same position of FP, i.e. linearly at the end of embedded CP and, supposedly, in its Specifier, is found for conditional antecedents. Consider (99) where the conditional antecedent contains a focused object and the FP appearing at the end of the dependent clause.

(99)

Ò’óyè,

s

ka

nɛ̀tɛkólè

dàmu

lè,

si

kɛ́ndɛya.

no

if

2sg

pfv.tr

locust.beans.art

eat

fp

2sg

pot

heal

354(If I eat mangoes, I will get better?) ‘No, If you eat locust beans, you will get better.’

??Ò’óyè,

s

ka

nɛ̀tɛkólè

lè

dàmu,

si

kɛ́ndɛya.

no

if

2sg

pfv.tr

locust.beans.art

fp

eat

2sg

pot

heal

355This raises the question why, in the cases of extraposed adverbial CPs (96) and relative CPs (98), Foc raises to the XP immediately dominating the embedded CP, whereas in (98) and in (99) it moves to the specifier of this CP. Supposedly, this is related to the extraposition requirement imposed in Kakabe CPs base-generated in VP or a DP. There may be a restriction according to which CPs cannot be extraposed together with the Foc operator, therefore, it needs to move higher than the extraposed CP.

356To conclude, further research may clarify the nature of the difference in the placement of Foc in preposed adverbials as compared to the case of correlatives and the adverbial CP extraposed to the right. At any case, the analysis according to which Foc needs to raise for the reason of feature visibility appears to be valid for the both cases.

8. Conclusion and extensions of the analysis

357The agreement analysis proposed for Kakabe may be extended to other languages. As already discussed in Section 7.2, several languages with morphological focus are reported to have restrictions on the expression of focus in embedded clauses. It has been argued in the paper that these restrictions follow from the agreement between C and Foc. Apart from that, this agreement relationship is also appropriate to treat the incompatibility that holds between inflectional markers and DP focus found in numerous languages in Africa (Section 7.4).

358Let us now look at whether the second part of the proposed analysis, namely, postulating a D-feature on Foc, can be extended to other languages. Positing the D-feature on Focus operator is indeed in line with certain general tendencies observed for languages with morphological focus marking. Thus, if a language shows the pattern of DP/VP and DP/IP focus syncretism (where DP is part of VP and IP respectively), then it is the DP inside the constituent that hosts the marker. Moreover, in the case of DP/IP focus syncretism, it is the subject, the structurally higher DP that hosts the focus marker. This pattern is discussed as ‘anti-pied-piping’ by Branan and Erlewine (2019).

359The syncretism between object DP/VP syncretism and, at the same time, between subject DP and IP focus found, for example, in Miyara Yaeyaman (Ryukyuan), that marks focus with the particle =du (Davis 2013):

360(100) Miyara Yaeyaman (Ryukyuan) =du focus marker (Davis 2013: 33)

a.

Hajasi-san=du

ziroo=ba

bari.

subject/IP focus syncretism

Hayashi-san=du

Jiro=acc

hit

361(Who hit Jiro?) Hayashi-san hit Jiro.

362(What happened?) Hayashi-san hit Jiro.

b.

Kunu

midun-pïto=o

izï=ba=du

fai

object/VP focus syncretism

this

female-person=top

fish=acc=du

eat

363(What did that woman eat?) This woman ate fish.

364(What did that woman do?) This woman ate fish.

365To cite another example, Konkonba uses two particles to mark focus. The particle is right adjoined to the object DP in object and VP focus (Schwarz 2007: 126). In the cases of IP focus and subject focus, a particle follows the subject (Schwarz 2007: 131). See Assman et al. (2019) and Branan and Erlewine (2019) for more examples of languages that have the pattern of morphological focus marking characterized by object/VP and subject/IP syncretism. This pattern find in our analysis is attributed to the existence of the D-feature on the focus operator.

366An interesting pattern is found, Gùrùntùm, a Chadic language. This language marks focus with the FP á that precedes the focus constituent, see the subject focus in (101). Strikingly, in difference to the languages discussed so far, a structure where FP á precedes the object as in (101b), has not only the object focus and VP-focus interpretations, but also the interpretation with the focus on verb only. The same type of syncretism, whereby no distinction is made between V, VP and object focus is found in Tangale, another Chadic language (Hartmann & Zimmermann 2007).

367(101) Gùrùntùm (Hartmann and Zimmermann 2009: 1347, 1355)

a.

rèená

wum

kwálíngálá-ì.

fp

king

chew

cola.nut-def

368(Who is chewing the cola nut?) ‘The king chewed the cola nut.’

b.

Tí

bà

wúm

kwálíngálá.

3sg

prog

gather

fp

cola.nut

369(What is he chewing?) He is chewing cola nut.

370(What is he doing with the cola nut?) He is chewing cola nut.

371(What is he doing?) He is chewing cola nut.

372This pattern can be rather easily accounted for within the framework proposed in the present paper. As has been argued for Kakabe, Foc operator can take as its complement only an XP and not terminal a node such as I0 or V0. One can assume that the same is true for Gùrùntùm and Tangale. In other words, for these languages narrow V focus is a case of V plus G-marked object focus. Gùrùntùm and in Tangale are then different as compared to Kakabe in that their focus operator has a ‘stronger’ D-feature that is not blocked by the givenness of a DP, and therefore FP is licenced before the DP even if it is G-marked.

373Another piece of evidence supporting the analysis where focus bears a D-feature comes from languages where VP or V focus is possible only through nominalization. This is the pattern used in languages that analyses Manfredi (1993): Vata (Kru), Igbo (Igboid), Yoruba (Yoruboid) and Haitian creole (French creole). In all the four languages the focalization of a verb required the verb to be nominalized. A nominalized copy of the verb appears in the sentence along with the non-nominalized instance of the verb. In Igbo this nominalized copy stays in situ (102). In the other three languages discussed in the paper, the nominalized copy of the verb moves to CP, as it is illustrated for Haitian Creole (103).

374Verb focus in Igbo (Manfredi 1993: 10)

(102)

Ézè

bu-ru

íbu

è-bú.

Eze

carry-Øasp

load

nom-carry

375‘Eze carried a load.’

376Verb focus in Haitian Creole (Manfredi 1993: 16)

(103)

Se

manje

li

mange

pen.

cop

eat-nom

3sg

eat

bread

377S/he ate bread.’

378There are languages that show a focus marking strategy that can be seen as the reverse in comparison to the strategy employed in Gùrùntùm and Tangale, namely, that of signaling focus on the auxiliary or the verb and never on a DP. This is typical, for example, of Atlantic languages; see for overview and discussion (Robert 2010). Thus, in Wolof focus is always marked jointly with the person-aspect marker. There is a three-way contrast between subject focus (104a), object focus (104b), and verb focus (104c) encoded in the person-aspect marker.

379Focus marking in Wolof (McLaugnlin 2004: 247)

(104a)

Maa-y

lekk

jën

1sg.sfoc-ipfv

eat

fish

380I eat fish.’

(104b)

Jën

laa-y

lekk.

fish

1sg.ofoc-ipfv

eat

381‘I eat fish.’

(104c)

Dama-y

lekk

jën.

1sg.vfoc-ifpv

eat

fish

382‘I eat fish.’

383For our analysis, this means that the presence of D-feature on Foc is subject to parametric variation. In cases like that of Wolof, one can assume that the focus operator has a fixed location in I0. It may also agree with the focused DP, but in a distinct way, as compared to DP/Foc agreement in Kakabe.

384It should be admitted that for the moment it is difficult to judge about how most of the criteria discussed on the Kakabe data are applied to other languages with morphological focus simply because detailed descriptions are very few. In particular, givenness is almost never discussed in this context when morphological focus languages are concerned. The same concerns focus in embedded clauses which is crucial in order to be able to judge about the effect of the CP boundary. Still, considering what has been said in this section, it appears that at least for some languages with morphological focus, the proposed agreement account might be helpful to explain the pattern of the distribution of their focus markers. And the investigation of Kakabe focus marking could be a starting point for the exploration of these parameters in relation to morphological focus.

Abbreviations

385Øasp – zero aspect

386acc – accusative case

387art – referential article

388bnf – benefactive

389cop – copula

390dat – dative case

391def – definite article

392dim – diminutive

393erg – ergative case

394f – (a) focus, (b) feminine

395fd – focus domain

396fp – focus particle

397g – given

398ger – gerund

399ident – identificational copula

400incl – inclusive

401inf ‑ infinitive

402ipfv ‑ imperfective

403lg – long form of pronouns

404m ‑ masculine

405neg – negation

406nom – (a) nominative case, (b) nominalization marker

407obj – direct object index

408obl – oblique

409of – operator focus

410ofoc – object focus

411pass – passive

412pc.st – stative participle

413pfv.of – perfective with operator focus

414pfv.i – intransitive perfective (without operator focus)

415pfv.tr – transitive perfective (without operator focus)

416pl – plural

417poss – possessive

418pot – potential

419prf ‑ perfect

420prog ‑ progressive

421pst – past

422q – interrogative

423qu – quotative marker

424qud – question under discussion

425rel – relativizer

426sf – sentence focus

427sbjv – subjunctive

428sg – singular

429sbjv – subjunctive

430sfoc – subject focus

431subj – subject index

432top – topic

433univ – universal quantor (‘all’)

434vfoc – verbal focus

Haut de page

Bibliographie

References

Aboh, Enoch Oladé. 2004. The morphosyntax of complement-head sequences: clause structure and word order patterns in Kwa (Oxford Studies in Comparative Syntax). Oxford: Oxford University Press.

Aboh, Enoch Oladé. 2007. Leftward focus versus rightward focus: The Kwa-Bantu conspiracy. In Nancy Kula & Marten Lutz (eds.), SOAS Working Papers in Linguistics, vol. 15, 81–104. London: SOAS.

Aboh, Enoch Oladé. 2016. Information structure. A cartographic perspective. In Caroline Féry & Shinichiro Ishihara (eds.), The Oxford handbook of information structure, 147–164. Oxford: Oxford University Press.

Arregi, Carlos. 2016. Focus projection theories. In Caroline Féry & Shinichiro Ishihara (eds.), The Oxford handbook of information structure (Oxford Handbooks in Linguistics), 185–202. First edition. Oxford, United Kingdom: Oxford University Press.

Assman, Muriel, Daniel Büring, Isabela Jordanoska & Max Prüller. 2019. Towards a theory of morphosyntactic focus marking. https://homepage.univie.ac.at/daniel.buring/locker/Assmannetal2019.MorphFoc.

Baker, Mark C. 2015. Case: its principles and its parameters (Cambridge Studies in Linguistics 146). Cambridge, United Kingdom: Cambridge University Press.

Bárány, András. 2018. Person, case, and agreement: the morphosyntax of inverse agreement and global case splits. Oxford: Oxford University Press.

Beck, Sigrid. 2006. Intervention effects follow from focus interpretation. Natural Language Semantics 14(1). 1–56.

Belletti, Adriana. 2001. Inversion as focalization. In Aafke C. J. Hulk & Jean-Yves Pollock (eds.), Subject inversion in Romand and the theory of universal grammar, 60–90. New York: Oxford University Press.

Belletti, Adriana. 2002. Aspects of the low IP area. In Luigi Rizzi (ed.), The structure of CP and IP. The cartography of syntactic structures, 16–51. Oxford: Oxford University Press.

Berg, Rob Van Den, Carlos Gussenhoven & Toni Rietveld. 1992. Downstep in Dutch: implications for a model. In Gerard J. Docherty & D. Robert Ladd (eds.), Papers in Laboratory Phonology II. Gesture, Segment, Prosody, 335–367. Cambridge: Cambridge University Press. (5 May, 2020).

Bhatt, Rajesh. 2003. Locality in correlatives. Natural Language & Linguistic Theory 21(3). 485–541.

Bhatt, Rajesh. 2005. Long distance agreement in Hindi-Urdu. Natural Language & Linguistic Theory 23(4). 757–807.

Bhatt, Rajesh & Roumyana Pancheva. 2006. Conditionals. In Martin Everaert & Henk van Riemsdijk (eds.), The Blackwell companion to syntax, 638–687. Malden, MA, USA: Blackwell Publishing.

Bocci, Giuliano. 2008. Criterial positions and left periphery in Italian. Evidence for the syntactic encoding of contrastive focus. Nanzan Linguistics Special Issue. 3. 35–70.

Bošković, Željko. 2002. On multiple Wh-fronting. Linguistic Inquiry 33(3). 351–383.

Bossong, Georg. 1991. Differential object marking in Romance and beyond. In Dieter Wanner & Douglas Kibbe (eds.), New analyses in romance linguistics, 143–167. Amsterdam: John Benjamins.

Branan, Kenyon & Michael Yoshitaka Erlewine. 2019. Anti-pied-piping. https://mitcho.com/research/handout-anti-pied-piping2019.pdf.

Brody, Michael. 1997. Perfect chains. In Liliane Haegeman (ed.), Elements of grammar: Handbook in generative syntax. 139–167. Dordrecht: Springer Netherlands. (12 May, 2020).

Büring, Daniel. 2001. Let’s phrase it! Focus, word order, and prosodic phrasing in German. In Gereon Müller & Wolfgang Sternefeld (eds.), Competition in syntax, 69–105. Berlin: M. de Gruyter.

Büring, Daniel. 2010. Towards a typology of focus realization. In Malte Zimmermann & Caroline Féry (eds.), Information structure. Theoretical, typological and experimental perspectives, 177–205. Oxford University Press. Oxford: Oxford University Press.

Büring, Daniel. 2013. Syntax, information structure, and prosody. In Marcel den Dikken (ed.), Cambridge handbook of generative syntax, 860–896. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511804571.029.

Büring, Daniel. 2016. Intonation and meaning (Oxford Surveys in Semantics and Pragmatics 3). Oxford: Oxford University Press.

Cable, Seth. 2006. The grammar of Q: Q-particles and the nature of wh-fronting, as revealed by the wh questions of Tlingit. Cambride, MA: MIT Ph.D. dissertation.

Cable, Seth. 2010. The grammar of Q: Q-particles, Wh-movement, and pied-piping (Oxford Studies in Comparative Syntax). New York: Oxford University Press.

Chomsky, Noam. 2000. Minimalist inquiries: the framework. In Roger Martin & David Michael (eds.), Step by step: essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge MA: MIT Press.

Chomsky, Noam. 2001. Derivation by phase. In Kenneth L. Hale & Noam Chomsky (eds.), Ken Hale: a life in language (Current Studies in Linguistics 36), 1–52. Cambridge, Mass: MIT Press.

Connell, Bruce & D. Robert Ladd. 1990. Aspects of pitch realisation in Yoruba. Phonology 7. 1–29.

Creissels, Denis. 1997. Postpositions as a possible origin of certain predicative markers in Mande. Afrikanistische Arbeitspapiere 50. 5–17.

Dalrymple, Mary & Irina Nikolaeva. 2011. Objects and information structure (Cambridge Studies in Linguistics 131). Cambridge ; New York: Cambridge University Press.

Davis, Christopher. 2013. Surface position and focus domain of the Ryukyuan focus particle du : Evidence from Miyara Yaeyaman. International journal of Okinawan studies 4(1). 29–49.

Drubig, Hans B. 1994. Island constraints and the syntactic nature of focus and association with focus. Arbeitspapiere des Sonderforschungsbereichs 340: Sprachtheoretische Grundlagen der Computerlinguistik 51.

Drubig, Hans B. 2003. Toward a typology of focus and focus constructions. Linguistics 41(1). 1–50.

Féry, Caroline. 2011. German sentence accents and embedded prosodic phrases. Lingua 121(13). 1906–1922.

Féry, Caroline. 2013. Focus as prosodic alignment. Natural Language & Linguistic Theory 31(3). 683–734.

Féry, Caroline & Vieri Samek-Lodovici. 2006. Focus projection and prosodic prominence in nested foci. Language 82(1). 131–150.

Grimshaw, Jane. 1997. Projection, heads, and optimality. Linguistic Inquiry 28. 373–422.

Güldemann, Tom. 2003. Present progressive vis-à-vis predication focus in Bantu. Studies in Language 27(2). 323–360.

Gussenhoven, Carlos. 1992. Sentence accents and argument structure. In Iggy M. Roca (ed.), Thematic structure: its role in grammar, 79–106. Berlin; New York: Foris Publications.

Gussenhoven, Carlos. 1994. Focus and Sentence accents in English. In Peter Bosch & Rob A. VanDerSandt (eds.), Focus: linguistic, cognitive, and computational perspectives (Studies in Natural Language Processing), 83–92. Cambridge: Cambridge Univ. Press.

Haegeman, Liliane. 2003. Conditional clauses: External and internal syntax. Mind and language 18(4). 317–339.

Haegeman, Liliane. 2010. The internal syntax of adverbial clauses. Lingua 120(3). 628–648.

Haegeman, Liliane. 2012. Adverbial clauses, main clause phenomena, and composition of the left periphery (Oxford Studies in Comparative Syntax). Oxford ; New York: Oxford University Press.

Hagstrom, Paul. 1998. Decomposing questions. Cambride, MA: MIT Ph.D. dissertation.

Hartmann, Katharina & Malte Zimmermann. 2007. In place ‑ out of place? Focus strategies in Hausa. In Kerstin Schwabe & Susanne Winkler (eds.), On information structure, meaning and form generalizations across languages, 365–403. Amsterdam: John Bendjamins. (15 May, 2014).

Hartmann, Katharina & Malte Zimmermann. 2009. Morphological focus marking in Gùrùntùm (West Chadic). Lingua 119(9). 1340–1365.

Hayes, Bruce. 1995. Metrical stress theory: principles and case studies. Chicago: University of Chicago Press.

Heycock, Caroline. 2006. Embedded root phenomena. In Martin Everaert & Henk C. van Riemsdijk (eds.), The Blackwell companion to syntax (Blackwell Handbooks in Linguistics 19), vol. II, 174–209. Malden, MA, USA: Blackwell Pub.

Hooper, Joan B. & Sandra A. Thompson. 1973. On the applicability of root transformations. Linguistic Inquiry 4. 465–497.

Hovarth, Julia. 1986. Focus in the theory of grammar and the syntax of Hungarian. Dordrecht: Foris.

Hyman, Larry & Maria Polinsky. 2009. Focus in Aghem. In Malte Zimmermann & Caroline Féry (eds.), Information structure, 206–233. Oxford University Press. (29 December, 2018).

Hyman, Larry & John Watters. 1984. Auxiliary focus. Studies in African Linguistics 15(3). 233–273.

Kishimoto, Hideki. 2005. Wh-in-situ and movement in Sinhala questions. Natural Language & Linguistic Theory 23(1). 1–51.

Kratzer, Angelika & Junko Shimoyama. 2002. Indeterminate pronouns: The view from Japanese. In Yukio Otsu (ed.), Proceedings of the 3rd Tokyo conference on psycholinguistics, 1–25. Tokyo: Hituzi Syobo.

Ladd, D. Robert. 1996. Intonational phonology. Cambridge; New York: Cambridge University Press.

Ludwig, Reiner, Fabienne Salfner & Mathias Schenner. 2012. Adverbials in German: More on embedding and focus. Belgian Journal of Linguistics 26. 84–115.

Manfredi, Victor. 1993. Verb focus in the typology of Kwa/Kru and Haitian. In Francis Byrne & Donald Winford (eds.), Creole language library, vol. 12, 3–52. Amsterdam: John Benjamins Publishing Company. (19 May, 2020).

Matić, Dejan. 2014. Questions and syntactic islands in Tundra Yukaghir. In Rik van Gijn, Jeremy Hammond, Dejan Matić, Saskia van Putten & Ana Vilacy Galucio (eds.), Information structure and reference tracking in complex sentences, vol. 105, 127–162. Amsterdam: John Benjamins. https://benjamins.com/catalog/tsl.105.01mat. (22 September, 2014).

Matić, Dejan, Rik van Gijn & Robert D. van Valin Jr. 2014. Information structure and reference tracking in complex sentences: An overview. In Rik van Gijn, Jeremy Hammond, Dejan Matić, Saskia van Putten & Ana Vilacy Galucio (eds.), Typological Studies in Language, vol. 105, 1–42. Amsterdam: John Benjamins Publishing Company. https://benjamins.com/catalog/tsl.105.01mat. (22 September, 2014).

McLaugnlin, Fiona. 2004. Is there an adjective class in Wolof. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.), Adjective classes: a cross-linguistic typology. Oxford: Oxford Univ. Press.

Miyagawa, Shigeru. 2001. EPP, scrambling, and wh-in-situ. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 293–338. Cambride, MA: MIT Press.

Nespor, Marina & Irene Vogel. 1986. Prosodic phonology. Dordrecht: Foris Publications. (3 June, 2016).

Nikitina, Tatiana. 2011. Categorial reanalysis and the origin of the S-O-V-X word order in Mande. Journal of African Languages and Linguistics 32(2).

Pesetsky, David & Esther Torrego. 2007. The syntax of valuation and the interpretability of features. In Simin Karimi, Vida Samiian & Wendy K. Wilkins (eds.), Phrasal and clausal architecture syntactic derivation and interpretation, 262–294. Amsterdam: John Benjamins Publishing Company. (28 April, 2020).

Preminger, Omer. 2009. Breaking agreements: Distinguishing agreement and clitic doubling by their failures. Linguistic Inquiry 40(4). 619–666.

Preminger, Omer. 2014. Agreement and its failures (Linguistic Inquiry Monographs). Cambridge MA: The MIT Press.

Reinhart, Tanya. 2006. Interface strategies: optimal and costly computations (Linguistic Inquiry Monographs 45). Cambridge, Mass.: MIT Press.

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar, 281–337. Dordrecht: Springer Netherlands. (25 December, 2018).

Rizzi, Luigi. 2004. Locality and the left periphery. In Adriana Belletti (ed.), Structures and beyond. New York: Oxford University Press.

Robert, Stéphane. 2010. Focus in Atlantic languages. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, vol. 91, 233–260. Amsterdam: John Benjamins Publishing Company. (13 October, 2014).

Rochemont, Michael. 2011. Question answer congruence and focus phrase. Manuscript. University of British Columbia, ms.

Rochemont, Michael. 2016. Givenness. In Caroline Féry & Shinichiro Ishihara (eds.), The Oxford handbook of information structure (Oxford Handbooks in Linguistics), 41–63. First edition. Oxford, United Kingdom: Oxford University Press.

Saeed, John. 2004. In Brian Nolan (ed.), RRG 2004 books of proceedings - Linguistic theory and practice: Description, implementation, and processing, 258–279. Dublin: Institute of Technology Blanchards town.

Samek-Lodovici, Vieri. 2005. Prosody–syntax interaction in the expression of focus. Natural Language & Linguistic Theory 23(3). 687–755.

Schwarz, Anne. 2007. The particles lé and lá in the grammar of Konkomba. In Shin Ishihara, Stefanie Jannedy & Anne Schwarz (eds.), Interdisciplinary studies on information structure 8, 115–139. Potsdam: Universitätsverlag.

Schwarzschild, Roger. 1999. GIVENness, AvoidF and other constraints on the placement of accent. Natural Language Semantics 7(2). 141–177.

Selkirk, Elisabeth. 1984. Phonology and syntax: the relation between sound and structure (Current Studies in Linguistics Series 10). Cambridge, MA: MIT Press.

Selkirk, Elisabeth. 1995. Sentence prosody: Intonation, stress and phrasing. In John A. Glodsmith (ed.), The handbook of phonological theory, 550–569. Oxford, UK: Blackwell Publishing Ltd.

Sigurðsson, Halldór Ármann. 1996. Icelandic finite verb agreement. Working papers in Scandinavian syntax 57. 1–46.

Szendroi, Kriszta. 2003. A stress-based approach to the syntax of Hungarian focus. The Linguistic Review 20(1). https://doi.org/10.1515/tlir.2003.002.

Truckenbrodt, Hubert. 1995. Phonological phrases: their relation to syntax, focus and prominence. MIT PhD thesis.

Truckenbrodt, Hubert. 2002. Upstep and embedded register levels. Phonology 19(01). 77–120.

Truckenbrodt, Hubert. 2005. A short report on intonation phrase boundaries in German. Linguistische Berichte 203. 273–296.

Van der Wal, Jenneke. 2017. What is the conjoint/disjoint alternation? Parameters of crosslinguistic variation. In Larry M. Hyman & Jenneke Van der Wal (eds.), The conjoint/disjoint alternation in Bantu, 14–60. Berlin: Walter de Gruyter.

Watters, John R. 1979. Focus in Aghem. A study of its formal correlates and typology. In Larry Hyman (ed.), Aghem Grammatical Structure, 137–197. Los Angeles: University of Southern California.

Watters, John R. 2010. Focus and the Ejagham verb system. In Ines Fiedler & Anne Schwarz (eds.), Typological studies in language, vol. 91, 349–376. Amsterdam: John Benjamins.

Woolford, Ellen. 2000. Object agreement in Palauan: Specificity, humanness, economy and optimality. In Ileana Paul, Vivianne Phillips & Lisa deMena Travis (eds.), Formal issues in Austronesian linguistics (Studies in Natural Language and Linguistic Theory 49), 215–245. Dordrecht: Kluwer Academic Publishers.

Yip, Moira J. W. 2002. Tone (Cambridge Textbooks in Linguistics). Cambridge ; New York: Cambridge University Press.

Haut de page

Notes

1 The corpus is available at the ELAR archive: https://elar.soas.ac.uk/Collection/MPI43300, as well as at the Pangloss collection: https://lacito.vjf.cnrs.fr/pangloss/corpus/search.php?keywords=kakabe and the Corporan collection: https://corporan.huma-num.fr/Archives/corpus.php

2 SOVX word order is typologically unusual but typical for Mande languages. It does not occur in languages unrelated to Mande languages, except for the languages of the Senufo and Gur family, as well as in several other languages which are in contact with Mande (Nikitina 2011).

3 The transitivity split, with the differentiation of the perfective in transitive constructions and the perfective in intransitive constructions is a feature shared by Mokole languages as well as all Manding languages (a group within Mande, which is closest to Mokole). A diachronic account of the transitivity split is suggested in Creissels (1997).

4 The way this constraint functions can vary depending on how givenness and focus are represented in syntactic structure, see for discussion Büring (2016: Ch7).

5 The presented Kakabe evidence also complies with Féry's (2013) modification of the ‘Focus as Prominence’ proposal. She proposes that focus is rather Prosodic Alignment than Prosodic Prominence. At any case, the difference between the two versions is not essential for the present discussion. Characteristically, in her discussion of focus particles, Féry (2013) mentions only cases of narrow DP focus.

6 Stress is understood here as such rhythmical structuring of the prosodic unit, whereby one syllable is singled out as the strongest (Hayes 1995). Therefore, the presence of a phrasal stress means having one culminative syllable that per phonological phrase, and the presence of a sentence stress implies having one such syllable intonation phrase. This strength can have varying phonetic correlates, such as length, pitch, intensity. Stress often serves as the attractor of tone in languages that have both stress and lexical tone. None of these are found in Kakabe.

7 Kakabe has a partial tone-leveling operation of the type HLH => HꜜH where an L tone between two H tones becomes a downstepped H. Its domain is φ, accordingly, tones are cannot be levelled across a φ boundary. However, a word with FP to its right does not create an obstacle for this operation. Neither is it necessarily preceded by a partial downdrift reset which is another φ-boundary indicator (partial downdrift reset is opposed to full downdrift reset preceding ɩ phrases, as discussed above; on partial reset see e.g. Berg et al. 1992 and Truckenbrodt 2002).

8 A detailed discussion of focus embedded within a subordinate clause in German can be found, for example in Ludwig, Salfner, and Schenner (2012), see also (Drubig 1994; Drubig 2003).

9 It should be noted that the givenness effect on the placement of FP is represented in a simplified form in the present paper for the reason of space limitations. As I argue at length elsewhere, rather than being a binary feature, givenness on DPs is instead a system of hierarchically structured features, such as [±locutor], the recency of appearance of the referent in the discourse, disrupted vs. continuous referent. Accordingly, when more than one DPs are in agreement domain, and the DP highest in the structure is given, FP can appear on the lower DP that has less ‘givenness points’. Thus, in (i) with IP-focus, the FP on the subject is ruled out, because, being a locutor, it is located topmost on the givenness scale. Therefore, FP is assigned to the object DP instead a demonstrative pronoun, which is a non-locutor even though activated. In (ii), on the other hand, both the subject and the object DPs are locutors, therefore, FP appears on neither of them and is instead placed after the verb.

(i)

ká

kɛ̀

lè

tàrán

kɔ̀nkɛ́ɛ̀

kùn

tɛ́ɛ̀mà.

1sg

pfv.tr

this

fp

find

mountain.art

head

between

(The hare asked: ‘What has happened?’ He boy said:) ‘I found it on the top of the mountain’ [tale_Mammadou-Boyi-Konde_23dec2011_059].

(ii)

Àn

bí

wò

kɔ́n-dén

dè

tún.

3pl

be

2pl

dislike-pc.st

fp

only

(They didn’t give it to you.) ‘It is just because they didn't like you.’

10 It should be noted that, as already mentioned, under the current analysis, Foc operator can take as complement only non-terminal elements. Therefore, the system does not distinguish between narrow V-focus and a VP-focus with a G-marked direct object. This analysis for VP is supported by the evidence from incomplete complements of ‘or’ as well as be incomplete answers with FP. As is shown in examples below, the pronominal object cannot be omitted in either of the elliptical contexts.

(i)

ka

tɛ̀gɛ

lè

káa

fàra

lè?

fàra

lè.

3sg

pfv.tr

3sg

cut

fp

or

3sg

tear

fp

3sg

tear

fp

‘Did he cut it or tear it? (He) cut it.’

(ii)

*À

ka

tɛ̀gɛ

lè

káa

fàra

lè?

*Fàra

lè.

3sg

pfv.tr

3sg

cut

fp

or

tear

fp

tear

fp

11 As discussed in Section 6.5, the adverbial CP is extraposed before spell-out, but is nevertheless base-generated in the VP that is the complement of Foc operator.

12 It should be noted that, even though subordinate clauses show no restriction on the prosodic expression of focus, languages with prosody as the main correlate of focus expression can also have syntactic focus-related phenomena that are subject to these restrictions. For example, Bocci (2008) shows that Italian has contrastive focus left dislocation operation which is not available in subordinate clauses as opposed to root clauses.

13 The licensing of FP is parallel in some respects to wh-phrase movement in languages like English. Multiple wh-phrases can be present in a sentence but only the one closest to C moves to Spec,CP. According to the standard account, wh-movement is triggered by Q feature (or wh-feature, depending on the approach) on C, and the wh-phrase closest to C moves to Spec,CP, whereas the lower wh-phrase cannot move there.

(i) a. Who bought what?

b. *What who bought?

However, there is a significant difference between the two phenomena. In the case of wh-phrases more than one instance of them is possible, but only among these wh-phrases can be attracted to Spec,CP (as opposed, for example to Bulgarian, where multiple wh-phrases appear in Spec,CP, see Bošković 2002). In Kakabe, on the other hand, only one occurrence of Foc is possible following the restriction formulated in (89). Alternatively, one could suppose that instead of the restriction on the number of Foc operators within one CP as in (89), there is a restriction on the number of FP licensed by feature valuation. This would imply that multiple occurrences of Foc operator are admitted in a CP, but only the Foc feature of the higher Foc operator is successfully valuated, and this successful valuation licenses FP. However, this would violate the interpretability restriction (77).

It should be noted that the case of multiple wh-phrases may pose problems to the valuation theory of Pesetsky and Torrego (2007). As has been mentioned in Section 7.3, they posit valued but uninterpreted Q feature on wh- and an interpretable but unvalued Q on C in interrogative sentences. They then take it as granted that the uninterpreted valued Q feature cannot remain uninterpreted: “uninterpretable Q on a wh-phrase must enter an Agree relation with interpretable Q on C” (Pesetsky and Torrego 2007: 272). However, in cases like (i) only one of the two Q features receives interpretation through Q-probing by C.

14 Editor’s note: Alexandra also published a lengthy text dealing with this topic: Vydrina, Alexandra. 2020b. Operator focus in discourse and grammar: the two perfectives in Kakabe. Journal of African Languages and Linguistics 41(1). 99–145.

Haut de page

Pour citer cet article

Référence papier

Alexandra Vydrina, « Morphological focus and its agreement features in Kakabe »Mandenkan, 69 | 2023, 30-86.

Référence électronique

Alexandra Vydrina, « Morphological focus and its agreement features in Kakabe »Mandenkan [En ligne], 69 | 2023, mis en ligne le 18 mai 2023, consulté le 14 février 2025. URL : http://journals.openedition.org/mandenkan/3006 ; DOI : https://doi.org/10.4000/mandenkan.3006

Haut de page

Auteur

Alexandra Vydrina

Laboratoire de Phonétique et Phonologie, UMR-7018 CNRS

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-SA-4.0

Le texte seul est utilisable sous licence CC BY-NC-SA 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search