Object left-dislocation, topicalization and the syntax-phonology mapping of intonation phrases in Bàsàá 1 Fatima Hamlaoui1 & Emmanuel-Moselly Makasso2 ZAS, Berlin; 2 Centre National d’Education-MINRESI, Yaoundé Talk given the 06/13/2013 Introduction Over the past few decades, a number of studies have argued that the speech flow is organized into a finite set of hierarchically organized phonological domains more or less reflecting the syntactic constituency (a.o. Selkirk, 1978, 1984; Nespor and Vogel, 1982, 1986). Above the word level, two prosodic levels are assumed : the phonological phrase and the intonation phrase. If it is rather widely accepted that the phonological phrase is formed in relation to syntactic domains, that is lexical XPs (a.o. Truckenbrodt, 1999, 2007; Selkirk, 2011; Ito and Mester, 2011) or spellout domains (a.o. Dobashi, 2003; Ishihara, 2007; Kratzer and Selkirk, 2007), defining (identifying the basis for) the intonation phrase has been a more problematic entreprise. Not only syntactic but semantic and pragmatic factors have been argued to play a role (e.g. Sense-Unit-Condition (Selkirk, 1984) and speech acts/commamarked constituents (Selkirk, 2005)). The notion of ‘syntactic clause’ has long been taken to be the basis of Intonation Phrases (a.o. Downing, 1970; Emonds, 1970, 1976). It has been brought back to the forefront in the recent literature, with Optimal Theoretic constraints (Prince and Smolensky, 2004) requiring Alignment/Wrapping of ‘vPs/IPs, TPs’/‘CPs’/‘ForcePs’ with intonation phrases (a.o. Truckenbrodt, 2005, 2007; Zerbian, 2007a; Cheng and Downing, 2009). The notion of clause is also central in the recent Match Theory (Selkirk, 2011), where M ATCH C LAUSE is defined as follows : (1) M ATCH C LAUSE A clause in syntactic constituent structure must be matched by a corresponding prosodic constituent, call it ι, in phonological representation. The main aim of this talk is to address what is ‘a central question for [...] any theory of the syntactic-prosodic constituency relation’ : ‘the choice of the constituents that figure in the correpondence constraints’ and particularly how to best 1 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD characterize the notion of ‘clause’ in A LIGN /M ATCH constraints related to the intonation phrase, as the one given in (1) (Selkirk, 2011, 17). In order to address this issue, we discuss a special type of left-dislocation (LD), the ‘pseudo-passive’ found in a number of Bantu and Nilo-Saharan languages. Bàsàa (Bantu A43 in Guthrie’s classification), the object of this talk, is one of them. 1 First, we show that in Bàsàa, this type of LD displays the type of prosody observed in a number of Romance, Germanic and Bantu languages : an intonation break separates the left-dislocated NP from the remainder of clause. Second we discuss a number of syntactic properties of this type of LD and general properties of the languges that attribute them a ‘pseudo-passive’ value. We argue that pseudo-passive LD is different from the oft-discussed ‘aboutness’ LD (also called ‘hanging topics’) found in Romance, in German and in many Bantu languages. [This is not yet explicitly shown here, but the idea is that analyses in terms of adjunction to the root-clause, location of the topic under a functinal projection in the C-domain or movement and ellipsis analyses of dislocation fail to capture the basic properties of pseudo-passive left-dislocation]. Finally, if the alternative syntactic analysis we propose is on the right-track, we need more flexibility than in the recent syntax/prosody mapping approaches to account for the phrasing found in Bàsàa. We argue that the needed flexibility is found in the approach offered by Szendrői (2001), where the size of the intonation phrase basically depends on the largest overt extended projection of the verb. 1 Part 1 : The prosody of Bàsàa left-dislocation Bàsàa is an SVO language. Like most Bantu languages, it is a tonal language which contrasts two level tones, H(igh) and L(ow) on lexical stems. 2 On the surface, Bàsàa displays five pitch levels. H, L, ŤH (downstepped High), HL and LH. Vowel length is phonemic. According to Hyman (2003), there are two major tone rules in this language : – High Tone Spread (HTS), by which a H - L sequence becomes H - H – Falling Tone Simplification (FTS), by which a HL - H sequence becomes H - ŤH In ongoing work with Gjersoe and Makasso, we argue that the domain of application of HTS is the phonological phrase (= lexical XPs). Under this view, HTS shows that in a simple SVO sentence, Bàsàa displays the following phonological phrasing : (S)φ (V O)φ . This is briefly illustrated in (2) (see Gjersoe, Hamlaoui, Makasso, forth). 1. The second author of this talk is a native speaker of Bàsàá and all the statements made here reflect his judgments. Bàsàá is not exempt of dialectal variation and other Bàsàá speakers might have different judgements. From Hyman (2003, footnote 37) it seems that in some dialects of Bàsàá it is acceptable to demote an agent and express it by means of the preposition ni in our ‘neutro-passive’ case. Here, this alternative is marginally acceptable with instruments. 2. We consider noun class prefixes to be underlyingly toneless here. 2 Bantu 5 Paris (2) Talk given the 06/13/2013 ǹ-tÉhÉ áÓÓNgÉ áÓ-áá-sô. sóGól à 1-grand-father SM1 pst1-see 2-children 2pro-2conn-all ‘The grand father saw all the children.’ Typically, a high tone on the last mora of the subject does not spread onto the first mora of the verb complex (the agreement marker) (sóGól à → *sóGól á). In contrast, HTS applies between the verb and the phrase that immediately follows it, here /áOONgÉ/ (→ [áÓÓNgÉ]). HTS fails to apply between the two objects of a ditransitive verb, indicating that Bàsàa does not wrap the VP into one phonological phrase. The following phonological phrasing is observed in (3) : (S)φ (V OI)φ (OD)φ . (3) sóGól à ǹ-sÓmb mááNgÉ Bìtámb. grand-father sm1 pst1-buy child shoes ‘The grand-father bought the child shoes.’ (→ *mááNgÉ Bítámb) The domain of application of FTS is more important to the present study. As shown in (4) the domain of application of this rule is larger than the phonological phrase. (4) máNgòlò. áÒÒNgÉ áÓ-áá-só Ťáá ḿ-áára 2-children 2pro-2conn-all sm2 pst1-pick.up 6-mangoes ‘All the children picked up the mangoes.’ HTS applies within the phrase /áÒÒNgÉ áÓ-áá-só/ creating a falling tone on the last mora of this phrase [áÒÒNgÉ áÓ-áá-sô]. This is visible in (2). Whenever this phrase is in subject position, FTS applies (áÒÒNgÉ áÓ-áá-sô áá → áÒÒNgÉ áÓ-áá-só Ťáá). In the prosodic hierarchy adopted in the present work (Intonation phrase (ι) > Phonological Phrase (φ) > Prosodic Word (ω)), the domain of application of FTS is taken to be the intonation phrase : [(S)φ (V O)φ ]ι . FTS also applies between the two objects of a ditransitive verb, as illustrated in (5), indicative of the phrasing [(S)φ (V OI)φ (OD)φ ]ι . (5) sóGól à ǹ-tí áÓÓNgÉ áÓ-áá-só Ťndáp. grand-father SM1 pst1-give 2-children 2pro-2conn-all 9-house ‘The grand-father gave all the children a house.’ Let us now turn to the prosody of LD. Left-dislocation of XPs is a common syntactic process by which a syntactic phrase is located – in purely descriptive sense – in the ‘left-periphery’ of the clause. More or less obligatorily, depending on the language and the LD-XP, the left-dislocated phrase can be resumed by a pronoun/marker within the clause (either in its canonical position or attached to the verb, depending on the language). It has often been observed that the left-dislocated phrase forms a separate prosodic unit from the rest of the clause. By way of illustration, this is the case in Romance languages (a.o., Frascarelli, 2000; Frascarelli and Hinterhölzl, 2007; de Cat, 2007; Feldhausen, 2008), in Germanic (a.o., Frey, 2004, 2005), as well as in many Bantu languages (see a.o. Downing, 2011, and re3 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD ferences therein). As illustrated in (6), Bàsàa also displays the pattern : additionally from forming its own phonological phrase (displayed by the absense of HTS between the LD-XP and the subject), it is separated from the subject by an intonation break preventing FTS to apply. (6) áÒÒNgÉ áÓ-áá-sô sóGól à ǹ-tÉhÉ áÓ. 2-children 2pro-2conn-all 1-grand-father sm1 pst1-see 2them Lit : ‘All the children, the grand-father saw them.’ We take this evidence to indicate the following phrasing of simple clauses with a left-dislocated object : [ (LD-XP)φ [(S)φ (V O)φ ]ι ]ι . It is worth noting that the intonation break separating the fronted-object from the verb is not obligatorily realized by a pause (statistics to come). Let us now turn to the syntax of LD in Bàsàa. 2 Part 2 : The syntax of Bàsàa LD In terms of basic word order, Bàsàa is no different from the common SVO pattern of Bantu languages, with recipient preceding patient in double object constructions. We will show here that it differs from many well-discussed (Eastern/Southern) Bantu languages in that it shows a connection between LD and passive voice. Many Bantu languages display passive morphology, as illustrated in the example (7) from Swahili : (7) barua hii i-li-andik-w-a na Juma 9-letter dem-9 sm9-pst-write-pass-fv by 1-Juma ‘This letter was written by Juma.’ Swahili (‘full’) passive As illustrated in (8)b, Bàsàa has a passive affix too. The passive morpheme however reduces the valency of the verb (suppression of the agent), as shown by the ungrammaticality of (8)c. (8) a. síNgá ì n-ÃÉ tólò 9-cat SM9 pst1-eat 1-mouse ‘The cat ate the mouse.’ Active b. tòlò à n-ÃéBâ. 1-mouse SM1 pst1-eat-pass. ‘The mouse was eaten.’ Neutro-Passive c. *tòlò à n-ÃéBâ nì síNá. 1-mouse SM1 pst1-eat-pass by cat. ‘The mouse was eaten by the cat.’ ‘full’-Passive The grammatical alternative to (8)c is given in (9). It consists in ‘foregrounding’ the patient by left-dislocating it. 4 Bantu 5 Paris (9) Talk given the 06/13/2013 tòlò síNgá ì n-ÃÉ ñÉ 1-mouse 9-cat SM9 pst1-eat it ‘The mouse was eaten by the cat.’ Pseudo-Passive Let us have a look again at the example (6) repeated below for convenience. LD is actually ambiguous between a passive-like reading and a contrastive reading. (6) áÒÒNgÉ áÓ-áá-sô sóGól à ǹ-tÉhÉ áÓ. 2-children 2pro-2conn-all 1-grand-father sm1 pst1-see 2them Lit : ‘All the children, the grand-father saw them.’ Equi : ‘All the children were seen by the grand-father.’ or ‘All the children, the grand-father saw (not all the adults).’ In order to disambiguate this sentence, a pronoun can be added between the LD-XP and the subject. This is illustrated in (10). In this configuration, only the contrastive reading is available : (10) áÒÒNgÉ áÓ-áá-sô áÓ sóGól à ǹ-tÉhÉ áÓ. 2-children 2pro-2conn-all them 1-grand-father sm1 pst1-see 2them Lit : ‘All the children them, the grand-father saw them.’ Equi : ‘All the children, the grand-father saw (not all the adults).’ This pseudo-passive strategy is not available with pronouns referring to speechact participants (1rst and 2nd person) : they can be the subject of a neutro-passive, but if they are left-dislocated in an active sentence, then only the contrastive reading obtains. The latter pattern is illustrated in (11). (11) mÈ mama à ǹlémbél mÉ ǹlámb (wÈ à lémbél wÉ tSÒBí). I mother sm1 pst-cook-appl me meat (you sm1 cook-appl you fish) Equi : ‘For me, Mom cooked meat (while for you, she cooked fish).’ / * ‘I was cooked meat by Mom (and you were cooked fish (by her)).’ The pseudo-passive use of LD is not specific to Bàsàa. A recent study by Bostoen and Mundeke (2011) shows that the Bantu language Mbuun (Bantu B87) also distinguishes two types of LDs, one of them being a pseudo-passive. Resumption is optional in pseudo-passive LD (except for speech act participant, for which it is obligatory), and obligatory in the other type of [contrastive ?] LD. (12) ba-án taar o-á-(bá-)bol 2-child father sm1-pres-(om2-)beat ‘The children are beaten by father.’ Mbuun passive OSV (13) a. b. taar o-á-pa nkáb i-kOOn. father sm1-pres-give beggar 5-banana ‘Father gives the beggar a banana.’ nkáb i-kOOn taar o-á-(lá-)mó-pa beggar 5-banana father sm1-pres-(om5-)om1-give 5 Hamlaoui & Makasso c. Prosody/Syntax Bàsàa LD ‘Father gives the beggar a banana.’ i-kOOn nkáb taar o-á-lá-(mó-)pa 5-banana beggar father sm1-pres-om5-(om1-)give ‘The banana, the beggar is given it by father.’ Mbuun non-pseudopassive LD From what we know so far, it seems that resumption is more obligatory in the Bàsàa pseudo-passive than in its Mbuun counterpart but more investigations, probably corpus-based, are needed. From a more general perspective, ‘pseudo-passives’ (= a passive meaning without actual passive morphology) seem to be a common phenomenon in Bantu languages. The other types of transitive pseudo-passives that have been reported – the impersonal passive and the subject-object reversal – are respectively illustrated in (14) from Kula and Marten (2010) and in (16) from Kimenyi (1980). To the best of our knowledge, the impersonal pseudo-passives have also been reported for Lunda by Kawasha (2007), who mentions that the same construction has been reported for Luvale, KiKaonde, Luchazi and Kimbundu (see references in Kawasha’s paper). The subject-object reversal is often discussed in connection to Dzamba, Kinande and Kirundi (which is part of the same language as Kinyarwanda) see e.g. Ura (1999) for references). (14) a. b. bá-alí-ly-a ífy-ákulya (ku mu-mbúlu) sm2-pst-eat-fv 7-food by 3-wild.dog. ‘The food was eaten by the wild dog.’ Ify-ákulya bá-alí-ly-a (ku mu-mbúlu) 7-food sm2-pst-eat-fv by 3-wild.dog. ‘The food was eaten by the wild dog. Bemba ba-(‘impersonal’-) passive In the impersonal pseudo-passive, the agent is demoted and appears as an oblique. The verb displays a default subject agreement in class 2 (somehow agentive compared to default/expletive agreement in locative classes found in a number of other impersonal sentences in some Bantu languages). The patient is realized as an object, either in situ or fronted (with or without object resumption depending on its class/on the language). In passing, Mbuun also has an impersonal passive, but a major difference with Bemba is that the agent is disallowed from being introduced as an oblique : (15) a. b. mo-íb o-á-pic í-lÓN O-ma-in 1-child sm1-pres-throw 5-dish loc-6-ground ‘The child throws the dish on the ground.’ i-lON ba-á-lá-pic O-ma-in 5-dish sm2-pres-om5-throw loc-6-ground ‘The dish is thrown on the ground (by someone).’ Mbuun ba-(‘impersonal’) passive 6 Bantu 5 Paris Talk given the 06/13/2013 In the subject-object reversal, the patient appears preverbally and subject-agrees with the verb, while the agent follows the verb, but is not marked as an oblique. (16) Igitabo ki-som-a umuhuûngu 7-book sm7-read-asp 1-boy ‘The book is being read by the boy.’ Kinyarwanda OVS Going back to Bàsàa, several questions arise : – Considering that Bàsàa has passive morphology, how come LD of the object is used to express a passive meaning instead of demoting the agent (making it an oblique) and realizing the object of the active as the subject of the passive verb (as seen in Swahili, French or English for instance) ? – Is the pseudo-passive left dislocation underlyingly the same structure as the aboutness LD (‘hanging-topic’ in Frey (2005)’s terminology) reported in Romance/Germanic languages ? The answer to these questions is, in our view, crucial in reaching a better understanding of the dimensions along which Bantu languages vary. 2.1 LD and Passive, some contextual syntax In our view, the neutro-passive and the pseudo-passive left-dislcation observed in Bàsàa are consitent with a general requirement of the grammar of this language that the grammatical subject encode the highest argument available on the thematic hierarchy. As discussed in Hamlaoui & Makasso (2012/forth) , Bàsàa displays little, if none, inversion structures as the ones often discussed for Bantu languages outside of the Northwest area. – General lack of inversion structures (no post-verbal logical subject, even right-dislocated) – No locative subject markers/no dummy subjects – Subject focusing in situ – Reverse clefts (i.e. Focus Copula Background / *Copula Focus Background) Verbs that tend to take an expletive pronoun/subject-marker in other languages display some sort of dummy (thematic-like) lexical subject. This is illustrated with the data in (17) to (20) from Hamlaoui & Makasso (2012/forth) : (17) nǑp à-ńŤnÓ. 1-rain SM1-PRES-rain Lit : ‘The rain rains.’ (= ‘It is raining’) (18) (yǑm) í-ńŤnÉnÉ lÉ (...) (7-thing) SM7-PRES-show that Lit : ‘The thing shows that (...)’ (= ‘It seems that’) (19) (yǑm) í-ńŤléNá lÉ (...) (7-thing) SM7-PRES-make that Lit : ‘The thing makes that (...)’ (=‘It happens that’) 7 Hamlaoui & Makasso (20) Prosody/Syntax Bàsàa LD (yǑm) í-ńŤtàGáÉ lÉ (...) (7-thing) SM7-PRES-make that Lit : ‘The thing makes that (...)’ (= ‘It happens that’) From what we know so far, Bàsàa is a rigid SVO language, with an even stricter syntax/semantics linking/mapping than configurational languages like English. In this sense, it is very similar to Lango, a Nilo-Saharan language spoken in Uganda described by Noonan (1977, 1992) as an ‘indirect role marking language’. Both languages use ‘grammatical roles as subject and direct object as part of [a] strategy for indicating semantic roles, such as agent and patient (...) Word order, which is fairly rigid, is the primary device used to indicate grammatical roles in simple sentences’ (Noonan, 1992, 119). Lango has no morphological passive and it happens that it also displays a functional-passive LD. This structure, called a ‘passiveanalog’ in the literature on Lango, is illustrated in (21) to (23). Resumption is only obligatory for first and second person pronouns (like in Mbuun) and complement of prepositions, as visible from (21) and (22). In contrast, non-human direct objects never display a resumptive pronoun, as illustrated in (23) (21) yín dákô òmìyí dyÈl. you woman 3s-give-perf-2s goat ‘you were given a goat by the woman.’ Lango pseudo-passive (22) án rwòt òmìrò dyàN bòt´@. I king 3s-give-perf cow to-1s ‘I was given the cow by the king.’ (23) àpwô àtîn ònÈnÒ. hare child 3s-see-perf ‘The hare was seen by the child.’ We propose that the strong requirement for the grammatical subject to encode the highest item on the thematic hierarchy is central in distinguishing Bàsàa from many other Bantu languages and understanding the very essence of the pseudopassive LD. This strategy encodes the lesser topical status of the agent compared to the patient while leaving the preferred semantics/syntax mapping untouched : the agent is encoded as the grammatical subject and the patient is encoded as the grammatical object. In this sense, this type of LD is a non-demotional, foregrounding pseudo-passive, in the sense of Foley and van Valin (1985); Keenan and Dryer (2007). We propose the mini-typology given in Table 1 to account for how the passive LD compares to the types of transitive (pseudo-)passives we have considered so far. Contrary to the pseudo-passive LD of Bàsàa and Mbuun, the types of (pseudo-)passive displayed by Swahili/English/French and Bemba involve a grammatical demotion of the agent (from subject of the verb to oblique). Bemba differs from 8 Bantu 5 Paris Talk given the 06/13/2013 F IGURE 1 – Mini-typology Transitive Passive Swahili in that it seems to disallow actual grammatical promotion of the patient. It optionaly foregrounds it and an expletive (agentive-like) subject-marking is found on the verb (these observations are made by Kula and Marten (2010)). If we are right in thinking that semantics/syntax mapping is not totally random, our account makes the prediction that in Bemba, it should be rare to find (if they exist at all) structures in which the subject of a transitive verb corresponds to a semantic role lower on the semantic/thematic hierarchy than the agent, while the agent is present and realizes a non-subject grammatical function. So far, this is corroborated by the data available in Kula and Marten (2010). [As suggested by the grammaticality of intransitive formal locative inversions like the one given below, even if we are on the right track, the picture is much more complex and more investigations are needed to make sense of the ban observed in the Bemba transitive passives, in which a theme cannot (simply) realize the grammatical subject. (24) a. mu-n-gânda mu-alí-sńdam-a umu-mbúulu. 18-9-house sm18-past-sleep-fv 3-wild.dog ‘In the house slept the dog.’ Bemba LI b. *mu-n-gânda mu-alí-séndam-w-a (na/ku umu-mbúulu) 18-9-house sm18-past-sleep-pass-fv by 3-wild.dog ‘In the house was slept (by the dog).’] As seen in the table, we propose that the famous Kinyarwanda OVS structures display the reverse pattern : they grammaticaly promote the patient without demoting the agent. We will come back to the case of Kinyarwanda after having discussed the syntactic structure of Bàsàa pseudo-passive LD in more detail. 2.2 One form and two different uses ? Nope... The question we wish to address in this part is whether the pseudo-passive LD is structurally similar to the very common ‘aboutness’ LD we know a.o. from 9 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD French (a.o. de Cat (2007) and references therein) and that seems to be attested in many languages (see a.o. Bresnan and Mchombo (1987) for Chichewa, Zeller (2009) for Zulu, Frey (2004, 2005) for ‘hanging-topics’ in German or Frascarelli (2000); Frascarelli and Hinterhölzl (2007) for Italian and references therein). Typically, the relation of the LD-XPs to the rest of the clause is so ‘loose’ that they have often been considered to be ‘free-floating’ discourse topics (Bresnan and Mchombo, 1987, 746), with the LD-XP being linked with the clitic/pronoun/object marker through anaphoric binding. Under such an approach, hanging-topics and pseudo-passive LD would be one form, with two different functions. It seems to us though, that several arguments speak against this view. Referentiality The most striking difference between the two types of LD is related to the referential status of the LD-XP. Here is a quote from de Cat (2007), which is consistent with what we know about LD in French and in many other languages (even, probably, some Bantu languages) : ‘Topics per se do not have quantificational properties (Rizzi, 1997). There is nonetheless the possibility that a topic might act as an operator due to the inherent properties of the type of XP that instantiates it. However, as is well known, quantifiers, non-generic indefinites and wh-elements (which are standardly regarded as operators) cannot be topics (and hence cannot be dislocated) : they do not meet the requirement that topic referents be readily identifiable in the context. This requirement is illustrated in (25) (examples from de Cat (2007)).’ (25) a. *Tout homme, il est mortel. any/every man he is mortal b. *Chaque potager, il a son robinet. each allotment it has its tap The pseudo-passive is similar to more prototypical passives (where the most topical item = the grammatical subject) in that non-referential expressions can be fronted. Bostoen and Mundeke (2011) show that in Mbuun, quantificational, indefinite NPs and even wh-phrases do not resist pseudo-passive LD : (25) a. b. (26) a. b. ntaam ká-dzú-í ndaam a-nsa lion sm1-kill-perf some 2-antelope ‘The lion has killed some antelopes.’ ndaam a-nsa ntaam ká-(bá-)dzú-í some 2-antelope lion sm1-kill-perf ‘Some antelopes have been killed by the lion.’ Mbuun passive OSV mpfúm o-á-sóól á-swéta á-mÉs o-á-tén á-saa president sm1-pres-choose 2-soldier 2-one sm1-pres-refuse 2-other ‘The president chooses certain soldiers, he refuses others.’ á-swéta á-mÉs mpfúm o-á-(bá-)sóól á-saa 2-soldier 2-one president sm1-pres-(om2)-choose 2-other 10 Bantu 5 Paris Talk given the 06/13/2013 o-á-(bá-)tén sm1-pres-(om2-)refuse ‘Certain soldiers are chosen by the president, others are refused.’ (27) a. b. mvám kábá-ker O-nkEr. merchant neg1-do 3-thing ‘The merchant does nothing.’ O-nkÉr mvám kábá-ker. 3-thing merchant neg1-do ‘Nothing is done by the merchant.’ Mbuun passive OSV The following examples also show that non-specific indefinite expressions can be fronted in Bàsàa : (28) híŤGí ǹ-tómbá nÃèé ì ǹnÓl jÓ. each 3-sheep 9-lion sm9 pst1-kill 3-it Lit : ‘Each sheep, the lion killed it.’ / ‘Each sheep was killed by the lion.’ (29) tÒ ǹ-tómbá wárá nÃèé ì ǹnÓl Ťáéé. even 3-sheep one 9-lion sm9 pst1-kill neg Lit : ‘no sheep, the lion killed.’ / ‘No sheep was killed by the lion.’ (more acurately : even a sheep, the lion didn’t kill) The French counterpart of both the above examples is ungrammatical : (30) *Chaque mouton, le lion l’a tué. each sheep the lion it-has killed. ‘The lion killed each sheep.’ (31) *Même un mouton, le lion n’a pas tué/ne l’a pas tué. Even one sheep the lion neg-has not killed/neg it-has not killed ‘The lion has not even killed one sheep.’ (32) *Aucun mouton, le lion a tué/l’a tué. no sheep the lion has killed/it-has killed ‘The lion killed no sheep.’ In Mbuun, even wh-phrases can be fronted in with a pseudo-passive reading : (33) a. b. O-káár nké ká-wó-kon ? 1-woman what sm1-pst-plant ‘What did the woman plant ?’ ńké O-káár ká-wó-kon ? what 1-woman sm1-pst-plant ‘What was planted by the woman ?’ Mbuun passive OSV A unified analysis of Romance/Germanic aboutness LD and pseudo-passive LD would have to account for this major difference between the two types. In passing : the fact that non-referential quantificational expressions can be the subject 11 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD of passive verbs and can be fronted in some Bantu and Nilo-Saharan languages seems to question the ‘well-known’ assumption that these expressions are ‘antitopic’ or that topicality is involved in diathesis... Anaphoricity In her paper on French LD, De Cat (p88) mentions that ‘even in instances where the dislocated element has quantifier-like properties, the resumptive pronoun is attributed a fixed reference. The relation between the resumptive and its antecedent is merely anaphoric and is no different to what it would be if the antecedent was omitted’. It is not yet clear to us how to best account for this fact, but in Bàsàa, the omission of the fronted NP does change the interpretation : no passive reading obtains if the object of a transitive verb is only expressed as a pronoun : the pronoun simply has an anaphoric value. Also, obviously, in the type of LD illustrated in (28), the pronoun has a non-referential quantified expression as an antecedent. There is thus no anaphoricity/coreference but quantifier-binding (the pronoun covaries with the quantifier). Discourse givenness Whereas the left-dislocated NP is in some sense normally discourse-given or ‘already established in the discourse’ in French [although it does not have to be for LDed-subjects], Bàsàa’s pseudo-passive LD-XP is part of an allfocus sentence. Pseudo-passive LDs can be used to answer all focus questions. This is illustrated in (34). (34) A : íñùúŤkíí áÒÒNgÉ áá jé màséé ? why 2-children sm2 be happy ‘Why are the children happy ?’ B : rìnùní rí NÓN múŤmbúl màláám í pùmá. 13-birds sm13 pst1-build 6-nests 6-beautiful loc orange.tree ‘The birds have built beautiful nests on the orange tree.’ 0 B : mùmbúl màláám rìnùní rí NÓN mÓ í 6-nests 6-beautiful 13-birds sm13 pst1-build 6-them loc pùmá. orange.tree Equi : ‘Beautiful nests have been built by the birds on the orange tree.’ Comparing again with [colloquial] French, it seems possible to have a leftdislocated subject in an answer to an all-focus question, but clearly not an object. This fact, in our view, also participates in bringing Bàsàa’s LD closer to genuine passive voice in the sense that it primarily involves a change in ‘view-point’ compared the active voice. Additionally, the fact that the expression of the LD-XP is considered optional in languages like French goes together with the idea that LD very often targets given/non-focused XPs. 12 Bantu 5 Paris Talk given the 06/13/2013 Iterativity Another difference between Germanic/Romance LD is related to the number of (non-contrastive) LD-XPs a sentence can contain. Typically, in Germanic/Romance LD, a sentence can contain several aboutness topics. This is illustrated for French in (35). (35) tu comprends Jacqueline, sa mère, la bonne, elle la lui you understand Jacqueline her mother the housekeeper she her to-her refile. gives(familiar) Approx Equi : ‘you understand, Jacqueline’s mother gives her housekeeper to her.’ (Gadet 1989 :171) In Rizzi (1997, 295–297)’s cartographic approach of the C-domain, TopP projections are assumed to be iterative : ‘there can be an indefinite number of topics [...] topic phrases can undergo free recursion. From what we know so far (at least for Bàsàa and Lango), pseudo-passive LD only allows one phrase to be fronted. Typically, in ditransitives, either object can be left-dislocated but only one at a time. (36) a. Paul à ǹlÓŤná Makasso ǹlámb. Paul sm1 pst1-bring(come.with) Makasso 3-meat ‘Paul brought meat to Makasso.’ b. ǹlámb Paul à ǹlÓŤná wÓ Makasso meat Paul sm1 pst1-bring(come.with) 3-it Makasso Equi : ‘The meat was brought to Makasso by Paul.’ c. Makasso Paul à ǹlÓŤná ñÉ ǹlámb. Makasso Paul sm1 pst1-bring(come.with) 1-him 3-meat Equi : ‘Makasso was brought meat by Paul.’ d. *Makasso ǹlámb Paul à ǹlÓŤná ñÉ ŤwÓ. Makasso meat Paul sm1 pst1-bring(come.with) 1-him 3-it Litt : ‘Makasso, the meat, Paul brought it to him.’ In passing : there is no mirror-image to the Bàsàa LD on the form of rightdislocation. Right-dislocation of a subject is ungrammatical in Bàsàa and rightdislocation of an object results in a sort of ‘intensive’ meaning. (37) síNgá ì n-ÃÉ ñNÉ tòlò 9-cat SM9 pst1-eat it 1-mouse Litt : ‘The cat ate it, the mouse.’ ‘The cat ate some especially big mouse’ (In colloquial French : ‘Il a mangé une de ces souris !’) The evidence gathered so far suggests that, just like in the analyses offered for Lango by Noonan and Bavin Woock (1978); Noonan (1992) and Woolford (1991), and more recently by Bostoen and Mundeke (2011) for Mbuun, the Bàsàa pseudopassive LD fronts an object within the clause, still (somehow) treating it as a core13 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD F IGURE 2 – ‘Pseudo-Passive’ OSV argument of the verb. Our proposal is that this position is a ‘Topic’ position within the inflectional domain, above the position hosting the grammatical subject (here Spec,TP). 3 The structure proposed is given in Figure 2. This particular topic position is a reflex of the strong semantics/syntax restriction preventing the demotion of the agent and forcing a ‘divorce’ (Noonan, 1992) between grammatical subject and topic. As can be seen, for the time-being we remain agnostic as to whether the pseudo-passive involves movement or base-generation of the fronted-XP as well as to whether the resumptive-pronoun is a genuine pronoun or the spell out of a trace (a variable). More investigations are needed to provide an answer to these issues. Before turning to the syntax-phonology mapping of intonation phrases, let us say a word about the Kinyarwanda/Kirundi OVS pseudo-passive (more to come on this topic in a separate note. See FH’s website). Choose your own adventure. If you are interested in comparative Bantu and want to know more about pseudo-passives, read the next sub-section. If you are interested in prosody and in the syntax/phonology mapping of intonation phrases, jump directly to Section 4, the 3rd and last part of this transcript. 2.3 Connecting OSV to OVS As stated by Bostoen and Mundeke (2011), it is worth drawing a parallel between the OSV pseudo-passive and the OVS pseudo-passive found, among other languages, in Kinyarwanda (Kimenyi, 1980, 1988), Kirundi (Ndayiragije, 1999). OVS sentences have several properties that show that (i) the only subject property 3. The label ‘Topic’ is really used for convenience. The label ‘pivot’ might be more appropriate, as wh-phrases seem to be able to occupy this position too in Mbuun and it might be considered problematic that a wh-phrase topicalizes. In some sense, what we have is a (German-like) ‘prefield’ below C, within the inflectional domain, as shown by the fact that the Bantu languages discussed here are not V2 and that the verb can subject-agree with what’s in this particular prefield, as in Kinyarwanda. 14 Bantu 5 Paris Talk given the 06/13/2013 of the preverbal object lies in its agreeing with the verb and (ii) the postverbal subject is not an object and retains subject properties. As can be seen from the Kanyok example given in (38) (from Bostoen & Mundeke’s paper, reported to them by Timothee Mukash-Kalel), in the OVS pseudo-passive, the postverbal agent is not an oblique-NP. (38) a. b. ba-tùw bà-dyààdy mi-saany 2-fisherwoman sm2-eat 4-fish ‘The fisherwomen eat the fish.’ mi-saany yì-dyààdy ba-tùw 4-fish sm4-eat 2-fisherwoman ‘The fish is eaten by the fisherwomen.’ Kanyok passive OVS The postverbal agent can neither be left-out, nor substituted by an object marker on the verb. As for the preverbal object, it cannot be represented only by a subject marker on the verb. ‘NPs advanced to subject by the [S-O] reversal rule do not acquire the properties of basic subjects, such as raising, deletion under identity, and ha-insertion ; the only subject property they acquire is verb agreement’ (Kimenyi (1980, 145) from Morimoto (2006, 166)). Morimoto (2006) also argues that the postverbal agent can only be seen as a postverbal grammatical subject. In her analysis, subject-object reversal is a case of topic-agreement construction, instead of subject-agreement. From that perspective, and as stated by Bostoen & Mundeke, the OVS pseudo-passive is ‘quite the opposite’ of the pseudo-passive LD-XP in that, in the former the verb agrees with the topic whereas in the latter, it agrees with the subject. Our view of OVS is slightly different though. As already shown in Table 1, both pseudo-passives have in common that they do not demote the agent and that they foreground the patient (by making it clause-initial). Bàsàá and Kinyarwanda have (at least) two things in common here : (i) SpecTP attracts the highest thematic role rather than the most topical phrase and (ii) agreement with the verb strongly needs a surface Spec-Head configuration. The Kinyarwanda promotion of the patient, that is, the fact that it agrees with the verb, is a mere consequence of a strong application of the surface Spec-Head requirement in verb agreement [if we are on the right track, we generally expect no/very limited agreeing inversion in Kinyarwanda, just like in Bàsàá]. 4 The main difference between the two languages here resides in how high the verb moves. The fact that the verb moves to Topic0 in Kinyarwanda is what leads it to agree with the topic rather than with the grammatical subject (Configuration-based agreement wins over Role-based agreement). There is thus a split between subjecthood (being in SpecTP) and agreement with the verb. In other words, from what we know so far, we would not distinguish Salience-based 4. Languages with agreeing inversion (cf. Jenneke van der Val’s Bantu5 poster) would thus potentially be cases in which agreement is maintained with the item with the highest thematic role despite the fact that on the surface it is not in a Spec-Head relation with the verb. In case of conflict, Role-based agreement would then take priority over Configuration-based agreement in these languagues. 15 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD F IGURE 3 – ‘Pseudo-Passive’ OVS agreement languages (= agreement with the topic) from Role-based agreement languages (= agreement with the highest thematic role), as proposed by Morimoto (2006). For us, a determining argument in favor of the existence of Salience-based agreement in Bantu would be a language in which the phrase that ‘subject’-agrees with the verb is (i) not the highest thematic role available and (ii) topical and (iii) has at no point in the derivation been in a Spec-Head configuration with the verb [this could be an SVO pattern with ‘inverted’-verb-agreement with a genuine topical non-subject or agreement with left-dislocated non-subject]. From what we know, no such Bantu language has been reported so far. The structure we assume for the Kinyarwanda/Kirundi OVS is shown in Figure 3 [again, for the time-being we stay agnostic as to the details of the derivation of the OVS (movement/base-generation etc, which are important but not central to the present discussion]. Presently, we can only speculate as to the motivation of the verb movement to Topic. As will be discussed in more detail somewhere else, we suspect that Kinyarwanda displays a strong affinity between focus and the right-edge of the clause (similar to colloquial French (Hamlaoui, 2009), and to Sotho/Tswana (Zerbian, 2007b)). OVS has indeed been discussed in connection to focus on the agent (a.o. Ndayiragije (1999)). The verb movement could be seen as a sort of altruistic movement ensuring the alignment of the agent with the right-edge of the clause [a frequent phenomenon in Bantu/Romance dislocation, see a.o. Cheng and Downing (2009) on this topic), ultimately making the pseudo-passive OVS optimal to satisfy interface requirements related to the expression of focus. As kindly reminded to us by Denis Creissels, the passive is known for being a strategy to focus agents in Sotho/Tswana. OVS would fullfil a similar function, without the notion of focus being directly encoded in its syntactic structure. Bàsàá (and we suspect Mbuun too) do not display an affinity between focus and the right-edge of the clause (see Hamlaoui & Makasso, forth, on focusing in Bàsàá). 16 Bantu 5 Paris Talk given the 06/13/2013 Informally, here is a brief list of a few simple constraints we have considered so far, that, ranked in different ways yield several of the different word order/agreement patterns discussed in the Bantu literature : – Subject-Verb agreement : surface Spec-Head (Configuration-based) and/or Role-based (with the Highest Thematic Role available) [≈ Baker (2008)] – Choice of Grammatical Subject (i.e. phrase that goes to Spec,TP, agreement with the verb is not a diagnostic) : Role-based (The Highest Thematic Role available) and/or Salience-based (the most Topical item) – Expressing focus : rejection-based (i.e. avoid focused-subjects) and/or attractionbased (i.e. align focus to the right of VP/TP) – Expressing topic : topic first ? – Availability of expletives : yes or no (e.g. Bàsàá has no expletive) This section ends here for now. To be continued. 3 Part 3 : The syntax-phonology mapping of intonation phrases So far, we have discussed the prosody of left-dislocation in Bàsàá and we have shown that the LD-XP is phrased separately from the rest of the clause. We have then turned to the syntax of left-dislocation in Bàsàá and argued that its pseudopassive value is related to a type of dislocation that is different from the one extensively discussed in Romance/Germanic languages. We have proposed that the fronted-XP occupies a Topic position within the inflectional domain, above TP, accounting for its intermediate status between a left-dislocated topic and the topical subject of a passive verb. As mentioned in our introduction, the syntactic clause is taken to be the basis of intonation phrase formation. What syntactic projection corresponds to the root clause is thus not a trivial issue. Although people usually seem to intuitively know what a clause is, what is inside or outside of it, there is little consensus over how to define a clause. Within the generative framework, a widespread view is that a clause is a CP (Complementizer Phrase) or one of its Cartographic avatars, even a root clause. According to this view, items that have been argued to adjoin to the the root clause – typically left and right-dislocated phrases in Romance and German – are considered to adjoin to CP. If there is evidence for this approach in V2 languages like German, there is little empirical evidence for that in English, French or in many Bantu languages. 5 In the Bantu syntax/phonology interface literature, this issue is often acknowledged by a.o. Zerbian, who assumes that the highest projection in the Bantu languages she discusses is IP (Inflection Phrase). We share Zerbian’s view on this 5. By way of illustration, French allows left-dislocation in embedded clauses. The LD-XP can only follow the complementizer, indicating that the target position of the LD-XP is actually below C rather than above it. Considering that a root-clause is a CP forces a non-uniform syntactic analysis of LD-XP in declarative clauses. 17 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD F IGURE 4 – Match Syntax/Prosody Mapping matter. In the trees provided here, the CP level is only represented to indicate that the type of clause we discuss can potentially be embedded. Recent approaches to the syntax/prosody mapping of intonation phrases have however integrated the CP projection into their mapping algorithm of intonation phrases. As can be seen in (39), in Selkirk (2011)’s approach, the lowest intonation phrase in a sentence corresponds to the egdes of the complement of C. On the assumption that LD-XPs are either located in the CP domain or adjoined to the root clause/CP, this approach does account for the fact that left-dislocated items can be separated from the ‘standard clause’ by an intonation break. This is illustrated in Figure 4. [We refer the interested reader to Rizzi (1997) and Selkirk (2011) for details related to the ‘illocutionary clause’] (39) ‘Standard clause’ vs. ‘Illocutionary clause’ 1.‘The complement of the functional head Complementizer0 ’ → CP [Comp0 [Standard clause]]CP 2.‘The higest syntactic projection of the sentence’ → ForceP [illocutionary clause]ForceP Cheng and Downing (2009) and Downing (2011) also base intonation phrases on CPs. In their approach, non-selected (i.e. root, parantheticals, non-restrictive relative clauses) CPs introduce an intonation phrase break. The lack of prosodic integration of LD-XPs is related to Phase (here CP) adjunction. As illustrated in the Figure 5, the intonation phrase is expected to embody everything that is contained within CP. [We here ignore the break introduced by vP (the other phase) in that, in most Bantu languages, such a break would make both the grammatical subject and a topic be (intonation-)phrased separately from what follows, i.e. the verb] 18 Bantu 5 Paris Talk given the 06/13/2013 F IGURE 5 – Phase-based Syntax/Prosody Mapping In both approaches, the presence of an intonation break within the inflectional domain, as found in the Bàsàá pseudo-passive, is unexpected. Interestingly, this is independent of ones prefered (purely theoretical) view of whether a root clause is a CP or an IP. On the assumption that information structure can influence prosodic phrasing, it has been proprosed that Topics introduce intonation phrase breaks. Combined with the above-approaches, a constraint such as the one offered by Frascarelli (2000, 23) in (40) or more recently Feldhausen (2008) in (41), would indeed capture the correct phrasing for both LDs. (40) Topic Domain Hypothesis for Italian : A Topic is mapped into its own I. (41) Align-Topic, Right Align the right edge of a topic constituent to the right edge of a prosodic [FH : intonation] phrase. This approach is however problematic in that in all the languages discussed in this paper, there are topical items for which the high ranking of this constraint would wrongly predict the presence of an intonation phrase break. To give only a few examples, this constraint would predict the presence of an intonation break : Between a subject in SpecTP and the verb in Romance, Bantu and in English : –In SVO sentences, in which the subject is typically topical –In a number of inversion structures in Bantu languages that seem to allow topical non-agents to occupy Spec,TP –In ‘standard’ promotional/demotional passives –In English wh-questions containing a contrastive topic (e.g. ‘I know why 19 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD you came by, but what is John doing here ?’) Between the phrase in SpecCP (‘the prefield’) and the verb in German root clauses : –In root clauses in which a non-subject immediately precedes the verb (that is, located in SpecCP) and probably in any SVO clauses in which the subject is assumed to be in Spec,CP and the verb in C Being aligned with an intonation phrase break does not seem to be a gereral property of topics. Rather, it seems to be a common property of certain types of ‘fronted’, outside of their canonical position, dislocated phrases. 6 It is thus related to the syntactic structure rather than the information structure. A flexible approach to the syntax-prosody mapping of intonation phases that offers the chance to capture the difference between the topics that normally phrase separately from the ones that do not is the one offered by Szendrői (2001). In a nutshell, in this approach, the basis for intonation phrase formation is the largest overt extended projection of the verb [the discussion is for the time-being limited to root clauses]. According to Grimshaw (1997, 376) : ‘An extended projection is a unit consisting of a lexical head and its projections plus all the functional projections erected over the lexical projection. The smallest verbal projection is VP, but IP and CP are both extended projections of V.’ Typically, left-dislocated phrases do not belong to the largest overt extended projection of the verb, be they within or outside the inflectional domain, adjoined or in a specifier position. On the contrary, the above-cited types of topics that tend not to be aligned with an intonation phrase break sit in the Specifier position of the largest overt extented projection of the verb (TP in English, French and many Bantu languages declaratives, standardly CP in German declaratives) or lower in the structure. As shown in Figure 6, this approach captures the prosody of Bàsàa pseudo-passive LD. Figure 7 shows that the same proposal captures the intonation phrasing of a basic declarative/interrogative clause in German. Finally, this approach makes the prediction that in OVS languages, there should normally be no intonation break betwen O and V as shown in 8. Acknowledgment This research was initiated in the ANR/DFG Bantu Psyn project of Annie Rialland and Laura Downing. We are very grateful to them and to the audience at the Bantupsyn 2012 workshop on dislocation and metatony. Our heartful thanks go to Corneille Marius Bias, Seunghun Lee, Prof. Jacob Mabé, Yukiko Morimoto, Juvénal Ndayiragije, Laurent Roussarie, Hubert Truckenbrodt and Robert van Valin. Both authors were generously funded by the Alexander von Humboldt foundation. 6. This is somehow acknowledged by Feldhausen, who states : ‘A LIGN -T OP ( IC ), R, a constraint which accounts for the obligatory right boundary of dislocations (i.e. topics)’. 20 Bantu 5 Paris Talk given the 06/13/2013 F IGURE 6 – Bàsàá (declarative) TP = ι F IGURE 7 – German (declarative) CP = ι Références Baker, M. (2008). The syntax of agreement and concord. Cambridge University Press. Bostoen, K. and Mundeke, L. (2011). Passiveness and inversion in Mbuun (Bantu b87, DRC). Studies in Language, 35(1) :72–111. Bresnan, J. and Mchombo, S. (1987). Topic, pronoun, and agreement in Chicheŵa. Language, 63(4). 21 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD F IGURE 8 – Prediction : Kinyarwanda OVS = ι Cheng, L. L.-S. and Downing, L. J. (2009). Against FocusP : arguments from Zulu. In Kucerova, I. and Neeleman, A., editors, Proceedings of the UCL Workshop on Information Structure. Cambridge University Press. de Cat, C. (2007). French dislocation : Interpretation, Syntax, Acquisition. Oxford University Press. Dobashi, Y. (2003). Phonological phrasing and syntactic derivation. PhD thesis, Cornell University. Downing, B. T. (1970). Syntactic structure and phonological phrasing in English. PhD thesis, University of Texas at Austin. Downing, L. (2011). The prosody of ‘dislocation’ in selected bantu languages. Lingua, 121(5) :772–786. Emonds, J. (1970). Root and Structure-Preserving Transformations. PhD thesis, MIT. Emonds, J. (1976). A Transformational Approach to English Syntax. Academic Press, New York. Feldhausen, I. (2008). The prosody-syntax interface in Catalan. PhD thesis, Universität Potsdam. Foley, W. and van Valin, R. (1985). Language Typology and syntactic description, volume 1, chapter Information packaging in the clause, pages 282–364. CUP, Cambridge. Frascarelli, M. (2000). The syntax-phonology interface in focus and topic constructions in Italian, volume 50 of Studies in Natural Language and Linguistic Theory. Kluwer Academic Publishers, Dordrecht. Frascarelli, M. and Hinterhölzl, R. (2007). On information structure, meaning and form, chapter Types of topics in German and Italian, pages 87–116. John Benjamins, Amsterdam. Frey, W. (2004). Notes on the syntax and the pragmatics of German left dislocation. In Lohnstein, H. and Trissler, S., editors, The syntax and semantics of the left 22 Bantu 5 Paris Talk given the 06/13/2013 periphery, pages 203–233. Mouton de Gruyter, Berlin. Frey, W. (2005). Pragmatic properties of certain German and English left peripheral constructions. Linguistics, 89–129. Gadet, F. (1989). Le français ordinaire. Armand Colin, Paris. Grimshaw, J. (1997). Projections, Heads and Optimality. Linguistic Inquiry, 28 :373–422. Hamlaoui, F. (2009). La focalisation à l’interface de la syntaxe et de la phonologie : le cas du français dans une perspective typologique. PhD thesis, Université Paris III Sorbonne Nouvelle. Hamlaoui, F. and Makasso, E.-M. (Under Revision). On focus-marking and the unvailability of inversion structures in the bantu language basaa (A43). Hyman, L. (2003). The Bantu languages, chapter Bàsàa (A43), pages 257–282. Routledge. Ishihara, S. (2007). Major phrase, focus intonation, and multiple spell-out. The Linguistic Review, 24 :137–167. Ito, J. and Mester, A. (2011). Recursive prosodic phrasing in Japanese. In Borowsky, T., Kawahara, S., Shinya, T., and Sugahara, M., editors, Prosody matters : Essays in honor of Elisabeth Selkirk. Equinox Publishers. Kawasha, B. (2007). Passivization in Lunda. Journal of African Languages and Linguistics, 28 :37–56. Keenan, E. and Dryer, M. (2007). Clause structure, language typology and syntactic description, volume 1, chapter Passive in the World’s languages., pages 325–361. CUP, Cambridge. Kimenyi, A. (1980). A relational grammar of Kinyarwanda. University of California Press. Kimenyi, A. (1988). Passives in Kinyarwanda. In Shibatani, M., editor, Passive and Voice, volume 16 of Typological Studies in Languages. John Benjamins, Amsterdam/Philadelphia. Kratzer, A. and Selkirk, E. (2007). Phase theory and prosodic spellout : The case of verbs. The Linguistic Review, 24 :93–135. Kula, N. and Marten, L. (2010). Argument structure and agency in Bemba passives. In Legère, K. and Thornell, C., editors, Bantu languages : Analyses, description and theory, pages 115–130. Rüdiger Köppe, Cologne. Morimoto, Y. (1999). An optimality account of argument reversal. In Proceedings of the LFG99 Conference. CSLI Publications. Morimoto, Y. (2006). Agreement properties and word order in comparative bantu. ZASPiL, 43 :161–187. Ndayiragije, J. (1999). Checking economy. Linguistic Inquiry, 30 :399–444. Nespor, M. and Vogel, I. (1982). Prosodic domains of external sandhi rules. pages 225–256. Nespor, M. and Vogel, I. (1986). Prosodic Phonology. Foris, Dordrecht. Noonan, M. (1977). On subjects and topics. pages 372–385. Noonan, M. (1992). A grammar of Lango. Mouton de Gruyter. Noonan, M. and Bavin Woock, E. (1978). The passive analog in lango. pages 23 Hamlaoui & Makasso Prosody/Syntax Bàsàa LD 128–139. Prince, A. and Smolensky, P. (2004). Optimality Theory : Constraints Interaction in Generative Grammar. Blackwell, Oxford. Rizzi, L. (1997). The fine structure of the left periphery. In Haegeman, L., editor, Elements of Grammar, pages 281–337. Kluwer, Dordrecht. Selkirk, E. (1978). On prosodic structure and its relation to syntactic structure. In Fretheim, T., editor, Nordic Prosody, volume 2, pages 111–140. TAPIR, Trondheim. Selkirk, E. (1984). Phonology and Syntax. MIT press, Cambridge, MA. Selkirk, E. (2005). Comments on Intonation Phrasing in English. In Frota, S., Vigario, M., and Freitas, M. J., editors, Prosodies. Monton de Gruyter, Berlin. Selkirk, E. (2011). The syntax-phonology interface. In Goldsmith, J., Riggle, J., and Yu, A., editors, The handbook of phonological theory. Blackwell Publishing, Oxford, 2nd edition edition. Szendrői, K. (2001). Focus and the syntax-phonology interface. PhD thesis, University College of London. Truckenbrodt, H. (1999). On the relation between syntactic phrases and phonological phrases. Linguistic Inquiry, 30(2) :219–282. Truckenbrodt, H. (2005). A short report on intonation phrase boundaries in german. Linguistische Berichte, 203 :273–296. Truckenbrodt, H. (2007). The syntax–phonology interface. In de Lacy, P., editor, The Cambridge handbook of phonology, pages 435–456. Cambridge University Press. Ura, H. (1999). Checking Theory and Grammatical Functions in Universal Grammar. Oxford University Press. Woolford, E. (1991). Two subject positions in lango. In Proceedings of the 17th Annual Meeting of the Berkeley Linguistics Society, pages 231–243. Zeller, J. (2009). On clitic left dislocation in Zulu. Frankfurt African Studies Bulletin 18 (2006), Focus and Topic in African Languages :131–156. Zerbian, S. (2007a). Phonological phrasing in Northern Sotho (Bantu. The Linguistic Review, 24 :233–262. Zerbian, S. (2007b). Subject/object-asymmetry in Northern Sotho. In Schwabe, K. and Winkler, S., editors, Information Structure and the Architecture of Grammar : A Typological Perspective, pages 323–346. John Benjamins, Amsterdam. 24
© Copyright 2026 Paperzz