In this paper, I propose that fragmentary utterances as realized in Basaá (Narrow Bantu language spoken in Cameroon) derive from two different sources, namely; a regular source and a copular source. By regular fragments, it is meant those that involve movement of a Negative Polarity Item (NPI) or a CP complement to the clausal left periphery followed by clausal ellipsis (Merchant 2004 and related work). Conversely, copular fragments are the ones that have a biclausal structure made up of a main clause and an embedded one. The main clause is headed by a null verbal copula which selects an external subject argument within the VP domain along the lines of Koopman and Sportiche’s (1991) VP-Internal Subject Hypothesis and an internal object argument, namely; a headless relative clause. The goal of this paper is to show that no matter which syntactic function it fulfills (e.g. subject, object, adjunct etc.); the copular fragment always ends up in the matrix TP where it fulfills the classical EPP requirements i.e. the condition that every clause should have a subject (Chomsky 1982, Chomsky 1995; Grimshaw 1991 etc.). The headless relative clause which is selected as an internal argument of the null copula contains a relative operator which can be overt or covert. It is this relative operator that undergoes A-bar movement to Spec-CP above a functional focus phrase (FocP), the target of ellipsis. As A-bar movement of the relative operator takes place in Spec-CP i.e. above FocP (Rizzi 1997) which is the ellipsis site, this operator also evades ellipsis. This contradicts a standard view that fragment answers target a focus phrase category and that the target of ellipsis is a TP (tense phrase) projection (cf. Brunetti 2003; Merchant 2004; Temmerman 2013; Griffiths and Lipták 2014; Weir 2014 etc.) The paper shows that the null copula analysis and the existence of A-bar movement of a relative operator inside the relative clause are supported by a number of sentence internal properties such as quantifier and adjective floating, islands and connectivity effects. All these arguments lend support to the ellipsis approach.
The paper is organized as follows. Section 2 is concerned with some main properties of the language with focus on phonology, morphology and word order. More precisely, as Basaá is still understudied, the section deals with issues such as word order, tone patterns, pro-drop and briefly discusses some copular and copula-drop structures. Section 3 introduces Merchant’s (2004) ellipsis analysis of fragments and sets the stage for the whole discussion. More specifically, in this section, I make the main proposal of the paper by proposing the two sources of Basaá fragments and their syntactic derivations. Section 4 shows that Basaá fragments have two different sources. This is supported by a number of morphosyntactic evidence. Section 4.1 uses negation as the first argument for the (non) copulahood of certain fragments. In other words, while some fragments cannot be negated, others on the contrary can. As a result, I propose that the presence of negation in such contexts is an indication for their copulahood. In Section 4.2 I discuss copula drop and headless relatives. I propose that focus fronting is a biclausal structure which hides a verbal copula and the nominal head of a relative clause. I resort to the parallel between focus fronting and their paraphrases to support the null copula analysis. The analysis shows that copula drop works in tandem with headless relatives. Section 4.3 argues against a cleft source for Basaá fragments in particular and focus fronting at large. This is supported in light of a wide range of semantic and syntactic tests such as else-modification (see Section 4.3.1), focus fronting of indefinite and universal quantifiers (Section 4.3.2), scope relations as well as quantifier and adjective floating (Section 4.3.3). Section 5 deals with sluicing in Gungbe and Hungarian and attempts to draw a parallel between Basaá and these two languages. Copular fragments in Basaá seem to exhibit interesting similarities with sluicing in these languages, modulo some salient disparities at the level of clause structure. Section 6 provides arguments in support of the ellipsis approach to Basaá fragments with focus on binding connectivity effects (Section 6.1), scope and bound pronouns (Section 6.2). In Section 7, I present arguments in support of the movement analysis. These arguments include preposition stranding (Section 7.1), the distribution of NPIs (Section 7.2), binding ambiguity (Section 7.3), island effects (Section 7.4) and other semantic and morphosyntactic evidence such as the distribution of relative operators and focus markers (Section 7.5.1), and subcategorization (Section 7.5.2). An interim conclusion is provided in Section 7.5.3. Section 8 is concerned with the syntactic derivation of the two types of fragment answers attested in the language. It is suggested that regular fragments are derived by A-bar movement of an NPI or a CP complement into the left periphery followed by clausal ellipsis, while copular fragments rather involve a biclausal structure in which the fragment raises to the matrix TP for the purpose of satisfying the EPP, while a relative operator (overt/covert) moves inside the relative clause into Spec-CP, a position above FocP which is the target of ellipsis. The discussion in Section 8.1 reveals that there exists a null operator in non-emphatic fragment contexts. The existence of this null operator is supported by syntactic and semantic tests such as island and quantifier binding. Section 8.2 is about island sensitivity attested in fragment utterances. It is proposed that island-sensitivity effects attested in Basaá fragments are due to the presence of an offending *marked trace above the ellipsis site as proposed by Merchant (2004). Section 9 is the conclusion.
2 Some main properties of the language
As the language under investigation is still understudied, this section discusses some of its main properties with special focus on its phonology, morphology and syntax in order to familiarize the reader with the data, although all of them will not have a direct impact on the analysis. Basaá belongs to the Narrow Bantu language family and is mainly spoken in the Centre and Littoral Regions of the Republic of Cameroon. Some speakers are also found in the South and South-West Regions of the country. The language has been analyzed as evidenced by a good number of scientific works. Very recent works include (Hyman 2003; Makasso 2008; Bassong 2010, Bassong 2012, Bassong 2014; Hamlaoui and Makasso 2011, Hamlaoui and Makasso 2015; Jenks et al. 2017), to name only a few. From a phonological perspective, there are two main tones namely, a high tone [ ́] and a low one [ ̀], and two contour tones, namely a high-low/falling tone [ ̂], and a low-high/rising one [ ̌]). These are illustrated in turn below.
|‘to escape/run away’|
|‘clean!’ (2PL imperative)|
Syntactically, the canonical word order is SVO (2a), but sometimes, it can change in some contexts such as question formation, focus and topic constructions. These are illustrated in (2b), (2c) and (2d) respectively. In question formation, some wh-phrases (object and referential adjuncts) can be fronted to the so-called Immediately After Verb position (IAV) or to sentence initial position as indicated by the brackets in (2b). 1 Recall that it is not the case that the question word kíí ‘what’ in (2b) is doubly pronounced. (2b) simply shows that wh-movement can take place either immediately after the verb or at clause initial position.
‘The woman gave the doughnuts to the children yesterday.’
‘What did the woman give to the children yesterday?’
‘Doughnuts are what she gave to the children yesterday.’
‘As for the doughnutsi, shej gave themi to themk too.’
Another salient characteristic of Basaá is that it is a pro-drop language. This can be observed in (2c) and (2d), where the subject of the sentence mudaá ‘woman’ is omitted. Note that in this language, pro-drop is possible due to a rich verbal morphology. 2 More specifically, like in many noun class Bantu languages, the subject marker (sm) that underlies subject verb agreement and which agrees in noun class and number with the subject of the sentence helps recover the interpretation of the missing subject. By the same token, recently, it has been observed that focus drop as well as topic drop is attested in the language (cf. Bassong 2014). For instance, in some discourse contexts, more precisely when a discourse referent is known, given, or has been mentioned in the discourse, it can be omitted as shown in (3) and (4).
‘I heard that the woman gave the doughnuts to the children yesterday.’
‘Yes, they (doughnuts) are what she gave them yesterday.’
‘I heard that the woman gave many things to the children. Didn’t she give them the doughnuts?’
‘Yes, as for the doughnutsi, she gave themi to them too.’
The focus-drop construction in (3B) shows the peculiarity of Basaá in the sense that no language in the literature, to the best of my knowledge, has been reported to license focus-drop. The latter seems to be subject to parametric variation, and as pointed out by an anonymous reviewer, Basaá seems to differ from Romance languages where the overtness of focus is a condition on ellipsis. In a very recent analysis of the interaction between focus and ellipsis in Spanish, Ortega-Santos (2016: 166–167) has proposed the EXPRESS-FOCUS constraint within the framework of OT. As shown in (3) and (4), the discourse referent makala ‘doughnuts’ is first of all mentioned in A’s statements so much so that when omitted in B’s, it can be easily recovered due to the presence of an agreeing noun class focus/topic marker.
In negative sentences, negation always follows and never precedes all the other morphemes that make up the verbal complex as shown in (5a)–(5b). 3 Any other positioning of negation gives rise to ungrammaticality. This is illustrated in (5c)–(5d).
‘The child did not buy the doughnuts for his/herfriend yesterday.’
‘The girl was not making/causing the woman (to) cook nice food.’
Negation plays a central role in the morphosyntax of the language under discussion. In the context of this paper, it is a good indicator of copulahood in copula-drop constructions. This is illustrated in the following sentences.
‘Who is your friend?
‘My friend is Tonye’
‘Who is not your friend?’
‘My friend is not Tonye’
Sentences (6) and (7) are instances of positive and negative question-answer pairs respectively. A striking issue in these sentences is the optionality of the verbal copula 4 jé (present tense, positive form) along with the subject marker lí (class 8). This is also the case with its negative verbal counterpart tâ (negative present form) along with the subject marker lí-. Note that in usual speech, speakers tend to omit the copula. In the absence of negation and the verbal complex (subject marker and copula), it seems hard to decide whether the above constructions are full sentences or not. The presence of negation and the optional copulae in these constructions indicate that they are full sentences. As the discussion goes on, I capitalize on the presence of negation to support the view that some fragment answers involve a copula structure.
3 Basaá fragments and ellipsis
Merchant (2004: 673) proposes that fragmentary utterances such as (8b), although subject to ellipsis, contain fully-fledged syntactic structures as their non-elliptical counterparts in (8c).
Who did she see?
She saw John
According to Merchant, the fragmentary DP John in (8b) undergoes syntactic movement into the clausal left periphery, followed by ellipsis of the clause containing its trace (silent copy). This is illustrated in (9) where angled brackets indicate ellipsis.
(Merchant 2004: 675)
In Merchant’s analysis, FP is a functional projection which may be identified with Rizzi’s (1997) Focus Phrase. The functional head F hosts the E-feature which has phonological, syntactic and semantic properties (cf. Merchant 2004 for further details on these properties).
The goal of this paper is to analyze Basaá fragments along the lines of Merchant’s (2004) ellipsis approach and related work, and to show that they have two different sources, namely; regular fragments like in (10) and (11), which undergo focus movement followed by ellipsis as illustrated in (12), and copular fragments like in (13) and (14), whose derivation is given in (15).
‘Who did the woman meet?’
‘She didn’t meet anybody’
‘What do you believe?’
‘That the woman will arrive tomorrow’
‘I believe that the woman will arrive tomorrow.’
I propose that regular fragments like (10b) and (11b) are similar to English-like fragments as analyzed by Merchant (2004). Clearly put, I propose that regular fragments are moved to the clausal left periphery followed by clausal ellipsis (angled brackets indicate ellipsis). The fragmentary utterance in (10b) is given the syntactic derivation in (12).
Conversely, I propose that fragment answers such as (13b) and (14b) are copular structures as depicted in (15) where angled brackets indicate ellipsis while strikethrough indicates movement.
‘What has the woman bought?’
‘Bananas are what she has bought’
‘She has bought the bananas’
‘What is it that the woman has bought?’
‘Bananas are what she will buy’
‘Bananas are what she will buy’
‘She has bought the bananas’
As can be seen in (15), the only difference between (15a), which is the derivation of (13b) and (15b), the derivation of (14b) is that the relative operator in the former is covert while it is overt in the latter. In fact, there is no fundamental syntactic disparity between the two. Whether the relative operator is overt or covert, the target of ellipsis in both cases is a focus phrase. However, it should be noted that while a non-emphatic question such as (13a) (repeated below as 13aʹ for convenience) can be answered negatively (13bʹ), 5 its emphatic counterpart in (14a) (repeated as 14aʹ) is incompatible with a negative answer as shown in (14bʹ).
‘What has the woman bought?’
‘Nothing’ (Intended: She hasn’t bought anything’)
‘What is it that the woman has bought?’
‘Nothing’ (Intended: She hasn’t bought anything’)
In concrete terms, in a context like (13ʹa) the speaker does not assert that the woman bought something. Rather, s/he only defeasibly presupposes it. On the contrary, (14ʹa) denotes a presupposition of existence, that is, it is known to discourse participants that the woman has bought something which is contextually salient, although that thing remains unknown to the speaker. As a result, while a felicitous answer to the non-emphatic interrogative can be negative (13ʹb), an emphatic wh-question such as (14ʹa) is incompatible with a negative answer (cf. 14ʹb).
In this paper, I propose that in copular fragments, the remnant originates from the subject position within the matrix VP, and along the lines of the VP-Internal Subject Hypothesis (cf. Koopman and Sportiche 1991 among others). This fragment raises to Spec-TP to satisfy the EPP requirements. Adopting Kayne’s (1994) head raising approach to relative clauses, I propose that what undergoes A-bar movement inside the headless relative clause is a silent nominal head along with a relative operator (overt or covert). This operator stands in a predication and agreement relation with the fragment DP by virtue of sharing the same noun class, person and number features. It is this semantic identity which allows reconstruction effects within the embedded clause. The idea that the subject DP in copular fragments originates in the matrix VP is supported by a number of diagnostic tests such as scope relations as well as adjective and quantifier floating. In Section 4, I provide arguments in support of the copulahood of certain fragmentary utterances and their non-elliptical counterparts.
4 On the (non) copulahood of fragments
In Section 3, I have briefly introduced the two instances of fragment answers discussed in this paper. The aim of this section is to provide evidence that these two fragment types have different sources as they do not share the same morphosyntactic and semantic properties.
4.1 Interaction with negation
I use negation as the first argument in support of the copular nature of some fragments and their corresponding non-elliptical counterparts. While some fragments can be negated as in (16) and (17), others cannot, as shown in (18) and (19). As will be observed throughout the discussion, each time a focused constituent is fronted to clause initial position, whether in elliptical or non-elliptical contexts, it is morphologically marked for focus.
The teacher was giving the presents to somebody, but the speaker didn’t know who it was (Here it is known to A, B and C that someone is receiving the presents. Participant C contradicts B)
Lit: ‘Who was it that the teacher was giving the presents to?’
Intended: ‘The girl is whom she/he was giving the presents to.’
Lit: ‘The girl is not whom he was giving the presents to.’
It is known that the teacher was going to give the presents to someone but A’s speaker has no complete knowledge of the beneficiary (i.e. the beneficiary is unknown).
‘Who was the teacher giving the presents to?’
Intended: ‘The girl is whom he was giving the present to.’
Lit: ‘The girl is not whom she/he was giving the presents to.’
The above examples in (16)–(17) are instances of emphatic and non-emphatic information-seeking questions. In a context where participant C contradicts B’s statement, one can realize that B’s statements can be negated as shown in C. This can be done in the form of a fragment answer or in the form of a fully-fledged clause (recall that angled brackets indicate ellipsis).
As mentioned earlier, some fragments cannot be negated. Sentences (18B)–(18C) convey the same meaning and are felicitous answers to (18A). While (18B) is a regular fragment, I suggest its counterpart in (18C) is a biclausal copular focus construction (angled brackets indicate ellipsis).
‘What was the teacher giving to the girl?
Nothing (Intended: She was not giving her anything (cf. C below).
‘Nothing’ (Intended: She was not giving her anything).
The following examples in (19) show that although a clausal complement can be a regular fragment, it cannot be negated. The illicitness of (19D) indicates that a clausal complement can be neither negated nor focalized.
‘What was the woman telling to the children?’
‘That they will go to school tomorrow.’
‘She was telling them that they will go to school tomorrow.’
In this section, I have shown that when a constituent is focus fronted to clause initial position it is morphologically marked and can be negated. I have used the presence of negation as a first assumption for the copulahood in these constructions. Conversely, CP complements can be fronted without being negated and/or focalized. The following section deals with copula drop. It will be shown that the fronted foci (fragmentary and non-elliptical) actually contain a silent verbal copula and a silent relative clause head.
4.2 Copula and relative head-drop
In this section, I provide evidence that focus fronting in Basaá is actually a bi-clausal construction which hides an underlying verbal copula and a relative clause head. That fronted foci hide a verbal copula and the head of a relative clause is supported in the following examples.
Lit: ‘Who was it that the teacher was giving the presents to?’
‘Who is the man to whom the teacher was giving the presents?’
Intended: ‘The student is whom she/he was giving the present to.’
Intended: ‘The student is the person whom she/he was giving the present to.’
‘No, the student is not whom s/he was giving the presents to’
Lit: ‘The STUDENT is not the person to whom he was giving the presents’.
To begin with, in an emphatic question context like (20), each of the sentences in (20A)–(20C) is a reduced form of (20Aʹ), (20Bʹ) and (20Cʹ). Here, the italicized copula je/ta ‘be’, the subject marker (sm) and the nominal head mut ‘person’ of the relative show up. As can be observed, whenever the verbal copula and the relative head are overt, the presence of the focus marker is ruled out. This is shown in (20Bʹ) and (20Cʹ). In other words, there is incompatibility between the focus marker, the verbal copula and the relative head. The source of this PF incompatibility is still unexplained so far, and shall be explored in future research. The above examples show that the reduced forms hide a copula and a relative head that are overtly realized in the full forms (in prime).
The idea that focus fronting hides a verbal copula and the head of a relative clause is also attested in the context of non-emphatic questions as illustrated in the following examples. In (21), A, B and C are discourse participants. B answers to A’s question and C contradicts B by negating the focus of his statement. Sentences (21Bʹ) and (21Cʹ) are paraphrases of (21B) and (21C) respectively. Like in (20), the presence of the focus marker is ruled out in the full forms (i.e. the one with an overt copula and the relative head (See 21B’ and 21 Cʹ).
‘Who (lit. ‘the who’)did the woman invite?’
Intended: ‘The children are whom she invited’
Intended: ‘The children are the persons that she invited’
Intended: No, ‘The children are not whom she invited’
Intended: ‘No, the children are not the persons whom she invited’
A closer look at the C and Cʹs examples enables to see that what resists ellipsis in C is the fragment DP and negation, while ellipsis targets the remaining part containing the focus marker and all the elements following it. In the paraphrases in (21Bʹ) and (21Cʹ), it seems that everything should normally delete except the focused constituent ɓɔ̀ɔ̀ŋgɛ́ ‘children’, the subject marker, the copula, negation and the nominal head of the relative clause.
4.3 Against a cleft analysis
This section discusses arguments against a cleft analysis of Basaá fragments in particular and focus fronting in general.
Else-modification has been used in the literature as one of the main arguments against a cleft analysis of English sluicing (cf. Merchant 2001: 122; van Craenenbroeck 2010: 1716). However, it has been noted crosslinguistically that else-modification is subject to parametric variation. Rodrigues et al. (2009: 183-184) have shown that while else-modification is compatible with clefts in Brazilian Portuguese (BP), such compatibility is impossible in Spanish. According to them, this (in) compatibility is linked to the fact that clefts in BP are associated with a non-exhaustive reading while in Spanish they are associated with an exhaustive reading. I build on this test to argue against a cleft-based analysis of Basaá fragments and sluicing. 6 The following examples show that else-modification is possible in sluicing and in fragment utterances. Note that in (22), part of the sentence can be elided as shown by the angled brackets.
The teacher has just given the presents to some students. But there are still others to whom he is giving the presents as well.
Lit: ‘Who else is the teacher giving the presents to?’
‘Another student’ (Lit: *Another student is whom he is giving the presents to’
The grammaticality of (22) provides semantic evidence in support of non-exhaustivity in the else-modification test as expected. As a result, I use this to argue against a cleft-based analysis. In other words, as clefts are generally said to be associated with exhaustivity (cf. Rodrigues et al. 2009 for contradictory arguments), the preceding example clearly indicates that the list of the people who receive the presents is not exhaustive (i.e. it is not closed), rather, it remains open (i.e. it indicates that in addition to ‘some students’, others were given the presents as well’). The set of students who received the presents is not closed. At this juncture, a simple copula analysis seems to be more plausible as it overrides a cleft analysis.
4.3.2 Negative indefinites and universal quantifiers
The fact that negative indefinites and universal quantifiers can be focused and negated is another argument in favour of a copular analysis. This is illustrated in the following examples whereby angled brackets indicate ellipsis.
‘Who will come to the meeting tomorrow?’
‘Who will come to the meeting tomorrow?’
‘No, everybody will not come’
Assuming that negative and universal quantifiers are inherently exhaustive, in the sense that they cover a closed set of entities such that nothing out of that set can be excluded, it is obvious that the focus constructions containing the indefinite tɔ-mut ‘nobody’ and the universal quantifier ɓot ɓɔ-ɓásónâ ‘everybody’ in (24c)–(24d) are not clefts, but simple bi-clausal structures with a hidden verbal copula. In a well-defined context, neither the negative quantifier tɔ-mut ‘nobody’ nor its universal counterpart ɓot ɓɔ́-ɓásónâ ‘everybody’ excludes anything from a closed set of entities. On the contrary, each of them conveys an exhaustive interpretation and needs not be cleaved. É. Kiss (1998: 253) analyzes almost similar facts in English by showing that universal quantifiers, also-phrases and even-phrases are incompatible with clefts. The following section deals with two semantico-syntactic arguments against the cleft analysis.
4.3.3 Scope relations, quantifier and adjective floating
Before the analysis proceeds, I would like to briefly recall the syntactic derivation proposed in (15) for the derivation of copular structures. This would enable us to understand the discussion on scope relations as well as quantifier and adjective floating. It is proposed that the null verbal copula acts like a light verb that originates within the matrix VP and raises into T. The focus constituent (here DP) is initially merged with the quantifier/adjective in Spec-VP such that it can raise to Spec-TP, leaving the quantifier within VP as shown in (25). The relative clause (DP) is merged as the complement of the null copula.
In the following examples, scope relations hold both for elliptical (26 B) and non-elliptical (26 C) constructions (angled brackets indicate ellipsis).
‘It seems to me that many students speak Basaá’
‘No, many students do not speak Basaá’[many >not / not >many]
There is a list of many students such that they do not speak Basaá (it is the case that none of them speaks Basaá).
There is a list of students such that many do not Basaá (It is case that some speak Basaá).
Intended: ‘No, many students do not speak Basaá’ [many >not/not >many]
(iii) There is a list of many students such that they do not speak Basaá (it is the case that none of them speaks Basaá).
(iv) There is a list of students such that many do not Basaá (It is case that some speak Basaá).
The preceding examples show that scope relations between negation ɓéé ‘not’ and the quantified phrase ŋgandag ɓáúdú ‘many students’ are possible either at S-Structure or at a certain stage of the derivation; namely, prior to A-movement of the quantifier phrase ŋgandag ɓáúdú ‘many students’ to Spec-TP as briefly depicted. In other words, each of the readings (not > many and many > not) is obtained as follows: First of all, considering the VP-Internal Subject Hypothesis (Koopman and Sportiche 1991 a.o.), the quantified subject originates within VP i.e. under the scope of negation, hence the ‘not > many reading. Secondly, when the quantified subject ŋgandag ɓáúdú ‘many students’ is raised to Spec-TP, it occupies a position where it c-commands negation at S-structure, hence the many > not interpretation.
Last but not least, another argument in support of a biclausal structure for some fragmentary and non-elliptical utterances discussed in this paper has to do with quantifier and adjective floating. The fact that quantifier and/or adjective floating is/are allowed only in the matrix and never in the embedded clause is an indication that the fronted element does not originate elsewhere (i.e. inside the embedded relative), but within the matrix VP (headed by a null verbal copula) before raising to Spec-TP where it complies with the EPP requirements. The following are instances of adjective and quantifier floating.
‘The woman met the lonely students.’
Lit: ‘The lonely students are not whom she met (i.e. they were accompanied by their parents).’
Lit: ‘The lonely students are not whom she met (i.e. they were accompanied by their parents).’
In (27), the adjective ɓɔ́támâ ‘alone’ and the head noun ɓáúdú ‘students’ co-occur in sentence final position (27A) whereas in focus fronting contexts, the whole noun phrase occurs in sentence initial position (27B). Adjective floating is disallowed in sentence final position (27B) while it is permitted in-between negation and the focus marker (27C). The same results are obtained with quantifier floating as illustrated below.
Intended: ‘The teacher read all the books.’
Intended: ‘He did not read all the books’ [not>all/ all>not]
There is a list of books such that he read some of them (not >all)
There is a list of book such that he read none of them (all >not)
Lit: ‘No, the books are not all that he read’ not >all / *all>not
Just like adjective floating, quantifier floating is possible in-between negation and the overt relative operator βí in (28C) while quantifier floating in sentence final position is disallowed in (28A). Interestingly, quantifier floating is permitted not only in full sentences, but also in elliptical ones. As can be seen, only (28B) as opposed to (28C) is ambiguous. The ambiguous reading in (28B) is possible if and only if one assumes that scope relations between negation and the QP ‘all the books’ take place at S-structure and prior to movement. Conversely, in (28C) the quantifier gwɔ́βísónâ ‘all’ cannot scope over a c-commanding negation. This explains why only the ‘not >all’ reading is possible. The syntactic derivation for (28C) also holds for (27C).
The only disparity lies in the fact that there is no overt relative operator in (27C), while the operator is overtly realized in (28C). I propose that both sentences have the same syntactic derivation, the difference being at the level of overtness/covertness of the relative operator. Besides, I assume that both (27C) and (28C) contain a silent relative nominal head. This silent relative head raises along with the relative operator (overt or covert) from AgrSP/TP to Spec-CP in line with Kayne’s (1994) Head Raising Approach to relative clauses (angled brackets indicate ellipsis, while strikethrough indicates movement).
As illustrated in (29), the target of ellipsis in copular fragments is a functional Focus Phrase (FocP) (Rizzi 1997 onwards), the head of which is occupied by an agreeing noun class focus marker. What resists ellipsis is the fragmentary DP along with negation and the relative operator. The Basaá data are reminiscent of the Hungarian and Gungbe sluicing inside relative clauses (cf. van Craenenbroeck and Lipták 2006; Lipták and Aboh 2013 respectively). I briefly talk about their analyses and show the similarities that exist between Basaá and these languages. The data, when compared with Basaá, offer interesting results which bear on clause structure and the theory of ellipsis crosslinguistically.
5 Sluicing within relatives in Gungbe and Hungarian
As reported in Lipták and Aboh (2013), sluicing in Gungbe and Hungarian is not a biclausal construction like copular fragments in Basaá. Nevertheless, there seem to be some morphosyntactic similarities between these typologically unrelated languages. In Gungbe (Kwa), not only the relative operator, but also the focus marker survive ellipsis in sluicing (Aboh 2010; Lipták and Aboh 2013). 7 This is illustrated in (30).
Lit. ‘Kofi will call someone, but I don’t know the person who.’
Lit. ‘Kofi will call someone, but I don’t know the person who Kofi will call.’
(Lipták and Aboh 2013: 106, Ex. (9) and (11)).
First of all, note that the sequence Rel-Foc that is attested in Basaá copular fragments (see, 28C and 29) is the one attested in relative sluicing in Gungbe. In addition to the fact that Gungbe sluicing is monoclausal while Basaá copular fragments are bi-clausal, these two African languages also differ at the level of the target of ellipsis. In Basaá, ellipsis targets a functional FocP, the complement of C, the head of CP that hosts the [E] feature. On the contrary, in Gungbe, 8 the target of ellipsis is a TP the complement of a functional Foc head that hosts the focus marker.
Similarly, and as pointed out by an anonymous reviewer, Basaá and Hungarian also seem to be similar to some extent. In their analysis of relative clause deletion (RD) in Hungarian, van Craenenbroeck and Lipták (2006) show that in this specific construction, everything deletes except the relative pronoun and one more constituent. This is illustrated below.
‘The girl who Kornél invited was the one who ZOLTÁN did’
‘The girl who Kornél invited was the one who ZOLTÁN invited’
(van Craenenbroeck and Lipták (2006: 248–249))
Interestingly, relative clause deletion in Hungarian exhibits the same clause structure as Gungbe, modulo a slight difference with respect to the landing site of the remnants. Like in Gungbe, the target of ellipsis is a TP projection, the complement of Foc, the head of the Focus Phrase (cf. van Craenenbroeck and Lipták 2006:262). In Hungarian, the focused constituent lands in Spec-FocP while the relative pronoun is hosted in Spec-CP. The resulting structure is the one in which only the relative pronoun and the focalized material survive ellipsis. 9 Gungbe and Hungarian seem to be similar at the level of the trigger and target of deletion, namely; Foc0 and TP respectively. However, both languages differ at the level of the landing site and nature of the remnants (cf. fn. 8 and 9). Basaá and Hungarian are similar to the extent that the relative operator resists ellipsis in both languages, although the relative operator may be silent in Basaá (see 15a). This discussion has revealed the crosslinguistic variation of ellipsis from typologically unrelated languages. So far, we assume that Basaá fragments are derived from fully-fledged sentential structures that are subject to PF-deletion. The following section deals with the arguments that support the ellipsis approach to fragments in Basaá.
6 Evidence for ellipsis
In this section, I capitalize on well-known arguments in the literature that show that fragment answers contain an unpronounced structure. These arguments, also known as sentence internal dependencies, have been used crosslinguistically (cf. Morgan 1973; Merchant 2001, Merchant 2004; Nishigauchi 2006; Ince 2012 etc.). They show that connectivity effects hold between fragment answers and their correlates in non-elliptical clauses.
6.1 Binding connectivity effects
Fragmentary reflexives and reciprocal pronouns exhibit the same connectivity effects regulated by Binding Theory (henceforth, BT) as their in-situ correlates in full sentential utterances.
‘Who does Tonye admire?’
‘Tonyei admires himselfi.’
Principle A of Binding Theory requires that an anaphor (a reflexive and /or a reciprocal) should be properly bound by a c-commanding antecedent. This applies to the fragmentary non-fragmentary utterances in (32a)–(32b) above and in the following example in (33) where the matrix DP subject Tɔnjɛ́i (+male) as well as its embedded counterpart ŋgɔ ɓásɔ́ŋj (+female) successfully binds the reflexive njɛ́-mɛdɛ́ ‘himself/herself’ in the embedded clause. As the grammaticality of (33a) shows, the same binding and ambiguity effects hold for both the fragmentary reflexive njɛ́-mɛdɛ́ ‘himself/herself’ and its counterpart in the fully-fledged clause in (33b).
‘Who does Tonyei think that ŋgɔ ɓásɔ́ŋj invited?’
‘Tonyei (male) thinks that ŋgɔ ɓásɔ́ŋj (female) invited himselfi/herselfj?’
Note that in Basàá the reflexive njɛ́-mɛdɛ́i/j can corefer either with Tonye, the subject of the main clause or with the embedded subject ŋgɔ ɓásɔ́ŋj. Besides, in Basaá there is no male/female distinction in the distribution of pronouns. The distinction is simply sensitive to noun classes. Given that all kinship names and proper names bear human characteristics in the language, they belong to the same noun class; namely class 1. It is certainly the reason why in the absence of a specific context, it is not easy to notice whether the anaphor njɛ́-mɛdɛ́i ‘him/herself’ is bound by the matrix subject Tonye or by its embedded counterpart ŋgɔ ɓásɔ́ŋj.
Similarly, reciprocals can be bound by a c-commanding DP both in elliptical and non-elliptical contexts as shown in (34). This state of affairs is tantamount to saying that at a certain stage, precisely prior to movement and PF-deletion, the anaphors are properly bound by their corresponding antecedents.
‘Who do the students love?’
‘The childreni love [one another]i.’
The same connectivity effects are obtained in emphatic questions in the following examples.
‘Who is it that Tonye is admiring?’
‘Tonyei admires himselfi.’
‘Who does Tonyei think that ŋgɔ ɓásɔ́ŋj invited?’
‘Tonyei (male) thinks that ŋgɔ ɓásɔ́ŋj (female) invited himself/herselfi/j?’
‘Who is that the students like?’
‘The childreni like each otheri.’
In the same vein, Principle B effects are attested in the distribution of fragmentary pronouns and their counterparts in non-elliptical sentences. This holds for regular and emphatic question-answer pairs.
‘Whom does the womani want to serve the food to?’
‘*The womani wants to serve heri the food’
‘Who is it that the woman wants to serve the food to?’
‘*The womani wants to serve heri the food’
Sentences (38) and (39) show that Principle B of BT is violated because the pronoun njɛ́ ‘heri’ is not free. It is bound by the antecedent mudaá ‘woman’ in the fragmentary answers in (a) and in their non-elliptical counterparts in (b).
Similarly, Principle C which bans coreference between a name/an epithet and a c-commanding name or pronoun is illustrated in (40). The name Bikun in (40) corefers with the epithet í maaŋgɛ́ wándá nú ‘that young man’, hence the illicitness of the sentences.
‘What does Bikuni think?’
*‘That that young mani is handsome.’
*‘Bikuni thinks that [that young man]i is handsome.’
The ungrammaticality of (40b), (40c) stems from coreference between the proper name Bíkûn in the subject position and the epithet í maaŋgɛ́ wándá nú ‘that young man’. Both cannot refer to the same person, hence the illicitness of these sentences.
6.2 Scope and bound pronouns
First of all, an indefinite direct object can take scope over a quantified subject. This is possible in the fragment utterance in (41c) and its non-elliptical counterpart in (41b).
‘Who did every student meet?’
‘Every student met a teacher.’
‘Every student met a teacher.’
Each of the readings below holds for both (41b) and (41c).
There is a list of teachers such that every student x met a teacher y from that list.
There is one teacher y such that every student x met that teacher.
Similar ambiguity effects are attested in the distribution of bound pronouns. In (42), the possessive dʒéi/j ‘his/her’ contained in the DP líwándá dʒéi/j ‘his/her friend’ in the object position can bind either the noun maaŋgɛ́i ‘child’ in the quantified subject híkií maaŋgɛ́i ‘every child’ or any other salient subject in the discourse, hence the ambiguity as shown by co-indexation.
‘Who did every childi see?’
Intended: ‘Every childi saw hisi/j/heri/j friend’
Intended: ‘Hisi/j/heri/j friend’
There is a list of friends s.t. every child x saw his/her friend y
There is a one individual X’s friend s.t. every child saw that friend
Notwithstanding the reading in (42ii) which indicates that the direct object líwándá dʒéi/j ‘his/her friend’ does not bind the noun maaŋgɛ́i ‘child’ inside the quantified subject, the reading in (42i) clearly shows that such binding effects are still possible. Parallel grammatical dependencies exhibited in fragments and their sentential counterparts in (41) and (42) are expected under an ellipsis approach which assimilates fragment utterances with the well-known syntactic properties. The binding binding effects obtained in (42) are interesting as they do not only show evidence for ellipsis, but also for syntactic movement operations.
7 Evidence for the movement analysis
This section addresses arguments which support the idea that fragments undergo syntactic movement prior to ellipsis. These arguments are taken from a wide range of sentence internal dependencies such preposition stranding, complementizer deletion, subcategorization properties, binding ambiguity, island effects, the distribution of negative polarity items etc.
7.1 Preposition stranding
Although Basaá is a non-preposition stranding language in wh-question contexts as shown in (43a) and (44a), preposition stranding is successfully operative in fragment answers as illustrated in (43b) and (44b).
‘Who did you meet Maah with?’
‘I met him with Pondí’
‘Please, tell us with whom you are living now’
‘With my friend’
‘I am living with my friend now.’
The optionality of pied-piping in the above examples indicates that preposition stranding under fragments is possible. As a result, the possibility of having preposition stranding is an argument in favor of a movement plus deletion analysis. In more concrete terms, the fact that the preposition ni ‘with’ can be left out in (44b) indicates that its DP complement liwándá dʒɛ̂m ‘my friend’ can move alone without any illicitness. When this happens, it is assumed that the complement of this preposition, namely the DP liwándá dʒɛ̂m undergoes A-bar movement, followed by PF deletion of the clause which contains the stranded preposition ni ‘with’.
These facts seem to run counter to Merchant’s (2001: 92) cross-linguistic generalization on ellipsis, and sluicing in particular, but with similar results in fragment answers:
Form-identity generalization II: P-stranding
“A language L will allow preposition stranding under sluicing iff L also allows preposition stranding under wh-movement.” (Merchant 2001: 92).
In fact, Basaá is not an isolated case because similar facts have been reported in the literature cross-linguistically. For example, it has been revealed that Brazilian Portuguese (BP) and Spanish are two non-preposition stranding languages under regular wh-movement. However, they allow preposition stranding in sluicing (cf. Rodrigues et al. 2009 and references cited therein for further details). Of interest is the fact that these authors demonstrate that only in those cases where sluicing has a cleft source is preposition stranding possible. 10 Although the present paper argues against a cleft analysis of copular fragments, a unifying factor between Basaá (Bantu) and Brazilian Portuguese & Spanish (Romance) is that these typologically unrelated languages have a biclausal structure in ellipsis (fragments and sluicing). The following subsection deals with the distribution of NPIs as additional evidence in support of the ellipsis analysis.
7.2 The distribution of NPIs
Negative Polarity Items (NPIs) in Basaá are made up of the item tɔ ‘no’ used in yes/no questions followed by a nominal element. 11 The following examples show that NPIs should occur in the presence of negation ((46a) and (47a)), and are not licensed in positive contexts ((46b) and (47b)).
‘The boy didn’t buy anything’
‘I haven’t invited anybody’
They can be used as fragment answers as in (48b) and (49b) and can be fronted in fully-fledged sentences as in (48c) and (49c).
‘What has the woman bought?’
Intended: ‘She hasn’t bought anything.’
‘Who does the woman love?’
Intended: ‘She doesn’t love anybody.’
I argue that these NPIs are licensed by the negative marker ɓé ‘not’ and postulate that the grammaticality of the (b) and (c)’s sentences is explained under the assumption that NPIs are c-commanded by negation either at S-Structure or LF. In this case, c-command relations between negation and NPIs take place at a certain stage of the derivation i.e. prior to the movement of the NPI to clause initial position. In the context of (46a) and (47a), it is obvious that c-command between negation and NPIs takes place on the surface while in the context of (48b)–(48c) and (49b)–(49c) it takes place at LF. I consider the licitness of these sentences as a core evidence that Basaá NPIs undergo syntactic movement followed by ellipsis of part of the clause containing a licensing negation. Whatever the case, I consider that the moved NPI leaves a copy inside the clause such that that copy is activated and licensed by a c-commanding negation.
7.3 Binding ambiguity
Ince (2012) uses binding ambiguity as an argument in favour of a movement analysis of fragments in Turkish (cf. Ince 2012 for more details on Turkish). I capitalize on his ideas and show that this also holds for Basaá fragments.
The following sentence in (50) is ambiguous because the third person possessive jě ‘his’ in the embedded clause can be bound either by the matrix DP subject Ntogue, the embedded DP subject Konde or any other contextually salient discourse referent. Both the matrix and embedded subjects c-command the possessive jě ‘his’.
‘Ntoguei says that Kondej lost hisi/j book’
The same binding ambiguity holds in the context of fragment utterances and fully-fledged sentences with focus fronting as illustrated in (51b) where ellipsis is indicated by angled brackets.
‘What does Ntogue say that Konde lost?’
Lit: ‘Hisi book is what Ntoguei says that Kondei lost.’
The grammaticality of (51b) is understood if we assume that prior to movement, the possessive jě ‘his’ contained in the fronted object DP kaat jěi/j/k ‘his book’ is bound by the c-commanding subjects. However, if such an analysis is true, it nevertheless raises a striking issue about the syntactic derivations proposed in (15), whereby the fronted constituent in copular structures is said to be based-generated in the main clause headed by a null verbal copula. I postpone this discussion to Section 8.
7.4 Island effects
Island sensitivity is used as another argument for the ellipsis approach (cf. Merchant 2004: 688–689, Merchant 2008; but see also Ince 2012 for a contradictory approach). Since Ross (1967), a number of restrictions on movement have been proposed. These restrictions serve as diagnostic tests for movement. If movement applies out of a domain which is not transparent to extraction, the resulting structure becomes illicit. It is true that testing for island sensitivity with fragment answers is not simple as noted by Merchant (2004: 687). This is due to the fact that the questions that are used to test for island sensitivity are themselves found in islands, that is, they constitute island violations. Merchant resorts to two strategies to test for islands in the context of fragment answers. The first strategy proposed consists in asking multiple wh-questions where one wh-phrase occurs inside an island as in the context in (52).
In Basaá, a full sentential answer is perfect when the targeted constituent occurs in an in-situ position inside an island (52b), but a corresponding fragment answer as well as its non-elliptical counterpart is ruled out in the same context as shown in (52c)–(52d) respectively (strikethrough indicates deletion of the moved constituent).
Question-answer pairs in a multiple question context
‘Which teacher wants to meet the man who speaks which language?’
Intended: ‘Ntogue wants to meet [the person who speaks ENGLISH].’
‘*English is what Ntogue wants to meet the person who speaks.’
The second strategy consists in asking a polar (yes/no) question with stress on a specific constituent within an island as in (53). In a context where the target constituent is in-situ inside an island in a yes-no question, a fully-fledged sentential answer is successful as in (53b) while a fragment is ruled out (53c). In the same vein, a fully-fledged sentential answer with a fronted constituent is ruled out (53d).
Yes-no question context
‘Can a Basaá teacher teach [the students who speak German]?’
‘No, he can teach the students who speak Búlu’
[Bantu language spoken in Cameroon].’
‘*Búlu is what he can teach the students who speak’
These island-sensitivity effects can be explained iff one admits that the fragmentary answers ŋgísi ‘English’ in (52c) and búlu ‘Búlu’ in (53c) have been extracted from an island by A-bar movement, the phonetic realization of which goes missing at the PF component.
7.5 Other morphosyntactic and semantic diagnostic tests for movement
This section is concerned with three other arguments that militate for a movement analysis of Basaá fragments. The arguments include the distribution of relative operators, focus markers as well as subcategorization.
7.5.1 The distribution of relative operators and focus markers
In Basaá, neither the relative operators nor the focus markers can show up when the focused constituent appears in-situ (54A1). The relative operator and the focus marker show up in every context where the focalized constituent is fronted to clause initial position as in (54A2). As illustrated in (54A2), both the relative operator and the focus marker are mandatory when constituent fronting takes place; otherwise, one ends up with an illicit construction. As it has been shown all along, only the relative operator resists ellipsis in the context of emphatic questions such as (54). The focus marker is banned from fragmentary utterances as shown in (54A3), (see also Example 15, among others).
‘Whom is the teacher congratulating?’
Lit: ‘Whom is it that the teacher is congratulating?’
‘He/she is congratulating the student.’
‘The student is whom he is congratulating.’
The distribution of relative operators and focus markers lends support to the movement plus deletion analysis of fragmentary utterances.
In Basaá, the declarative lexical complementizer lɛ́ ‘that’ and its interrogative counterpart tɔɔ́ ‘if/whether’ always select for clausal complements (see 55B and 56B) and never for DPs as shown in (57a)–(57c).
Context: A and B are out of the classroom
‘What’s happening inside there?’
Lit: ‘I think that the students are whom the teacher is talking to.’
Intended: ‘I think that the teacher is talking to the students.’
‘Has the teacher arrived?’
‘I don’t know if he has arrived.’
The sentences in (57) in which a lexical complementizer (57a)–(57b) or a series of two complementizers (57c) is followed by a DP complement are unacceptable in out-of-the-blue contexts. In other words, the sentences in (57) can only be correct in specific discourse contexts, precisely when the complement clause is overtly realized, salient or implied in the discourse. In the last two cases, the missing constituent is subject to ellipsis.
The following sentences indicate that in a specific discourse context, precisely in a question-answer pair context, (57a)–(57c) can be improved both in fragments and fully fledged sentential answers.
‘Who is it that the teacher is talking to?’
Lit: ‘I think that the students are whom s/he is talking to.’
Lit: ‘I don’t know if the students are whom s/he is talking to.’
Lit: ‘*I wonder myself that if the students are whom s/he is talking to.’
It is clear from (57) and (58) that the fragment answer ɓaúdú ‘students’ is not subcategorized/selected by the lexical complementizers. It is rather the case that a fully-fledged clause has undergone PF-deletion after syntactic movement of the remnant DP has taken place. These subcategorization facts do not only constitute evidence for ellipsis, but also for movement. The movement analysis holds along the lines if we assume that relativization and focus fronting in Basaá involve A-bar movement. In this case, the presence of both the relative operator ɓâ and the focus marker ɓɔ́-n (within the elliptical part) in (58) is clearly suggestive of syntactic movement. In other words, the proposal made throughout is that there is A-bar movement inside the relative clause. If this is true along the lines, it cannot be the case that the relative operator ɓâ and the focus marker ɓɔ́-n in (58) occur in an in-situ context. On the contrary, their presence is a signal of syntactic movement as these markers never occur in in-situ contexts.
7.5.3 Interim conclusion
Hitherto, the discussion has revealed that Basaá fragments include a syntactic structure that is subject to ellipsis at PF. Arguments such as connectivity and island effects as well as other morphosyntactic and semantic tests have been used to support this view. The following section comes back to the syntactic derivation of Basaá fragments with some clarifications on the motivations for having a null operator in those structures that lack an overt relative operator.
8 The syntactic derivation of fragments
In Section 2, I have proposed three derivations for Basaá fragments (cf.12 and 15). More precisely, on the one hand, I suggested that regular fragment answers i.e. the ones that realize as CP complement clauses and/or NPIs undergo run-of-the-mil A-bar movement to the left periphery of the clause (Spec-CP) followed by PF-deletion of the clause (TP/AgrSP). This analysis is in line with Merchant’s (2004) proposal (see also Merchant 2003). In this vein, fragmentary utterances such as (59b) and (60b) are syntactically derived as briefly depicted in (61a) and (61b) respectively.
‘What is the woman afraid of?’
‘That she will be late for the meeting’
|1.sm-prs-be sick||3.fear||that||1.sm-fut2-be late|
‘That she will be late for the meeting.’
‘What did the woman buy?’
‘She didn’t buy anything’
As can be observed in (59b)–(59c), complementizer deletion is not possible in both the elliptical and non-elliptical structures (see also Merchant 2004: 689–692 for similar cases in English). Complementizer deletion in these sentences will result in ungrammaticality. Following Merchant’s (2001, 2003, 2004, etc.) [E]-feature analysis, the null C in (61) does not only attract the fragment to Spec-CP, but it also triggers non-pronunciation of its TP/AgrS- complement.
On the other hand, copular fragment answers to emphatic questions (62) and their non-emphatic counterparts (63) are said to involve a biclausal structure that requires two major independent movement operations, namely; A-movement of the fragment DP into the matrix TP and A-bar movement of an overt or covert relative operator to the embedded Spec-CP via Spec-FocP. These two structures are represented in (64a)–(64b) respectively.
‘Who is it that you are talking to?’
‘The woman is whom I am talking to.’
‘What did you buy?’
‘The books are what I bought’
I propose that copular fragments are derived in two main stages: the fragment DP originates as the internal subject of the null verbal copula inside the matrix clause before raising to the matrix TP. I suggest this movement takes place to satisfy the EPP requirements, namely; the condition that the main T should have a subject in its specifier position (Chomsky 1982, Chomsky 1995; Grimshaw 1991). This movement is necessary because focus fronting constructions in Basaá lack expletive subjects that would otherwise fulfill the EPP requirements. More precisely, since focus fronting in this language does not follow the English pattern (it-be-XP) and/or French (ce-être-XP) cleft structures (cf. Akmajian 1970; É. Kiss 1998; Lambrecht 2001; Belletti 2005 among others.), the focalized constituent needs to raise into the subject position to comply with the EPP condition.
The second step in the derivation takes place into a headless relative clause in the complement position of the null verbal copula. It involves successive cyclic A-bar movement of a silent NP head along with a relative operator (overt/covert) into Spec-CP via FocP. I suggest that A-bar movement into Spec-FocP takes place because the relative operator on its way to Spec-CP needs to check the focus features against those of the focus marker in Foc. More precisely, in the context of (62b), the derivation of which is given in (64a), the focus marker njɛ́-n and the relative operator nû agree in noun class (class 1) and number features (singular). These features, along with the focus features hosted in Foc need to be checked by a matching relative operator. One way of accounting for (64b) is to assume that although the relative operator and the nominal head of the relative clause are silent, they nevertheless bear the same features as the focus marker gwɔ́-n in Foc, so much so that although the silent relative NP head and operator have a null spell-out, feature checking is still possible as in (64a). Adopting Kayne’s (1994) head raising approach to relative clauses, 12 the functional D head of DP selects a CP complement such that the target constituent that raises into Spec-CP is a silent NP head along with a relative operator (overt/covert). Recall that in this approach, the null copula in the matrix clause selects the fragment DP as its subject and a headless relative clause as its complement. It is D, the head of this DP in the object position that selects a CP complement. If this analysis is plausible, one can raise the question about the existence and implications of a null operator in non-emphatic fragment contexts as in (63c) and (64b) above.
8.1 The syntax and semantics of the null operator
The preceding analysis has shown that in cases such as (64b), there is a null operator that corresponds to an overt relative operator in emphatic structures (see (63c) and (64a)). In this section, I provide structural and interpretative arguments in favour of the existence of a null relative operator.
Note that whether focus fronting takes place in emphatic or non-emphatic contexts, one always ends up with the same semantic and syntactic effects. This is supported by reconstruction and island effects.
‘Which pictures does every woman like?’
Lit: ‘Every womani likes pictures of herselfi.’
Lit: ‘The pictures of herselfi/j are what every woman likes.’
Lit: ‘pictures of herselfi.’
‘Which are those pictures that every woman likes?’
Lit:‘Every womani likes pictures of herselfi.’
|híkií||mudaái||a- ŋ́-gwês||Focus fronting|
Lit: ‘The pictures of herselfi/j are what every woman likes.’
Lit: ‘The pictures of herselfi.’
The above examples show that the quantified subject híkii mudaá ‘every woman’ can bind a possessive and a reflexive found inside the focused constituent [bititî gwééi/j njɛ́-mɛdɛ́i/j] ‘pictures of herself’ not only in non-elliptical contexts but also in the elliptical ones. The licitness of these constructions is an indication that the reflexive and possessive contained in the fronted constituent [bititî gwééi/j njɛ́-mɛdɛ́i/j] ‘pictures of herself’ reconstruct. The reconstruction effects in (65c) and (65d) are a good signal that there is a null operator inside the relative clause that has the same semantic properties as the fronted constituent in the matrix subject position. In the same vein, the idea that there is a null operator even in non-emphatic contexts is also supported by the following examples.
Lit: ‘*The pictures of herselfi/j are what I saw every woman who likes.’
Lit: *‘The pictures of herselfi/j are what I saw every woman who likes.’
These structural dependencies and interpretative effects support the view that even in cases like (67a) where the relative operator has a null spellout, this operator nevertheless exists in the syntax and is subject to A-bar movement inside the relative clause. There is feature identity and a predication relation between the relative operator and a co-indexed fragmentary constituent in the matrix subject position.
Implementing Merchant’s (2001, 2004, 2008) ellipsis approach and considering (64) to be the correct derivation for copular fragments in Basaá, I propose that Merchant’s abstract [E] feature which is said to be responsible for PF-deletion is the same in Basaá. However, while the target of ellipsis is claimed to be a functional TP projection in English (Merchant 2004), in this Bantu language, two different cases arise. The target of ellipsis in regular fragments in Basaá is a TP (or AgrSP) category like in English, while the target of PF-deletion in copular fragments is a functional FocP.
8.2 Island sensitivity and Merchant’s (2004) *trace approach
In (52) and (53) (see Section 6.4.), it was shown that Basaá fragments are island-sensitive. In this section, I will build on Merchant (2004) and assume that these island effects are due to the presence of a *marked trace that resists ellipsis. In the case of Basaá, I suggest that this *marked trace evades ellipsis because it is located above the C head that hosts the [E] feature which triggers non-pronunciation of FocP. Following Merchant’s analysis, I assume that there is an extra CP layer above the [E]feature so much so that A-bar movement of the relative operator on its way to the highest CP layer leaves an offending *marked trace in the lower Spec-CP (called FP) such that this trace resists ellipsis. In this vein, the ungrammatical structure such as (68c) will be briefly derived as in (69) (here, the fragment answer ŋgisi ‘English’ is extracted out of a complex DP).
Question-answer pairs in a multiple question context
‘Which teacher wants to meet the man who speaks which language?’
Intended: ‘Ntogue wants to meet [the person who speaks English].’
With (69) in mind, the proposal is easily accommodated by postulating an extra FP layer between CP and FocP. The null operator raises in a cyclical fashion to Spec-CP via Spec-FocP and Spec-FP. Given that the *marked trace is located above FocP the target of ellipsis, an island violation arises in (68c) as depicted in (69).
Although the use of *trace in Merchant’s approach seems to violate Chomsky’s (1995) Inclusiveness Condition, his proposal is nevertheless appealing as it easily accounts for island violation in fragments. 13
In this paper, I have discussed two instances of fragment answers in Basaá; a Bantu language spoken in Cameroon. The discussion has shown that there exist two sources of fragment answers in this language; namely a regular source and a copular one. It was proposed that regular fragments are derived by A-bar movement of the fragment in the left periphery of the clause followed by clausal deletion like in Merchant (2004). As for copular fragments, I proposed that they involve a structure in which a null verbal copula selects the fragment as a subject and a complex DP (relative clause) as an object. The null verbal copula analysis has been supported by a number of arguments such as adjective and quantifier floating, scope effects etc. That fragmentary utterances are derived by movement has been supported by sentence internal dependencies such as connectivity effects, preposition stranding, island effects and subcategorization requirements. In addition, it was shown that copular fragments do not originate inside the embedded clause, but are rather generated in the subject position of the main clause as supported by quantifier and adjective floating among others. It was revealed that what undergoes A-bar movement inside the relative clause is a relative operator which has the same featural make-up as the fragment in the matrix clause. Consequently, reconstruction effects come out as a result of feature identity between the fragment and the relative operator. Overall, the analysis of Basaá fragments provides additional evidence that fragmentary utterances are derived from fully-fledged sentential structures that are subjected to ellipsis, supporting therefore the ellipsis approach. This paper has also contributed to current advances in the crosslinguistic variation of the syntax of ellipsis at large and fragment answers in particular.
First person singular
Agreement Complementizer Phrase
Agreement Subject Phrase
Extended Projection Principle
Polarity/yes-no question particle
Tense, Aspect and Mood
Null subject in tensed clauses
Past Tense one
Past Tense two
I am grateful to two anonymous reviewers of Linguistics for their valuable and insightful comments and suggestions. Special thanks are due to Ann Kelly for helping throughout the process.
Aboh, Enoch O. 2007. Leftward focus versus rightward focus: The Kwa-Bantu conspiracy. SOAS Working Papers in Linguistics 15. 81–104. Google Scholar
Aboh, Enoch Oladé. 2010. Information structure begins with the numeration. Iberia 2(1). 12–42. Google Scholar
Akmajian, Adrian. 1970. Aspects of the grammar of focus in English. Cambridge, MA: Massachusetts Institute of Technology dissertation. Google Scholar
Bassong, Paul Roger. 2010. The structure of the left periphery in Basaá. Yaounde: University of Yaounde I MA thesis. Google Scholar
Bassong, Paul Roger. 2012. Left peripheral focus and topic and complementizer agreement in Basaá. University of Potsdam. Ms. Google Scholar
Bassong, Paul Roger. 2014. Information structure and the Basaá left peripheral syntax. Yaounde: University of Yaounde I dissertation. Google Scholar
Belletti, Adriana. 2005. Answering with a cleft: The role of the null subject parameter and the VP periphery. In Laura Brugé, Giuliana Giusti, Nicola Munaro, Walter Schweikert & Giuseppina Turano (eds.), Proceedings of the thirtieth ‘incontro di grammatica generativa’, 63–82. Venice: Cafoscarina. Google Scholar
Brunetti, Lisa. 2003. Information focus movement in Italian and contextual constraints on ellipsis. In Gina Garding & Mimu Tsujimura (eds.), WCCFL 22: Proceedings of the 22nd West Coast Conference on Formal Linguistics, 95–108. Somerville, MA: Cascadilla Press. Google Scholar
Chomsky, Noam. 1982. Some concepts and consequences of the theory of government and binding. Cambridge, MA: MIT Press.Google Scholar
Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press. Google Scholar
É. Kiss, Katalin. 2009. Deriving the properties of structural focus. In Arndt Riester & Edgar Onea (eds.), Focus at the syntax-semantics interface: Working Papers of the SFB 732(3), 19–33. Stuttgart: University of Stuttgart.Google Scholar
Gilligan, Gary. 1987. A cross-linguistic approach to the pro-drop parameter. Los Angeles, CA: University of Southern California dissertation. Google Scholar
Grimshaw, Jane. 1991. Extended projection. Brandeis University. Ms.Google Scholar
Hamlaoui, Fatima & Emmanuel-Moselley Makasso. 2011. Bàsàa wh-questions and prosodic structuring. In Laura Downing (eds.), Questions in Bantu languages: Prosodies and positions (ZASPiL 55), 47–63. Berlin: Zentrum Allgemeine Sprachwissenschaft. Google Scholar
Hamlaoui, Fatima & Emmanuel-Moselley Makasso. 2015. Focus marking and the unavailability of inversion structures in the Bantu language Bàsàá (A43). Lingua 154. 35–64 CrossrefWeb of ScienceGoogle Scholar
Hyman, Larry. 2003. Bàsàa (A43). In Derek Nurse & Gerard Philippson (eds.), The Bantu languages, 257–282. London: Routledge. Google Scholar
Hyman, Larry M. (ed.). 1979. Aghem grammatical structure. (Southern California Occasional Papers in Linguistics (SCOPIL) 7), 137–197. Los Angeles: University of Southern California. Google Scholar
Hyman, Larry M. & Maria Polinsky. 2010. Focus in Aghem. In Malte Zimmermann & Caroline Fery (eds.), Theoretical, typological and experimental perspectives, 206–233. Oxford: Oxford University Press. Google Scholar
Jenks, Peter, Emmanuel-Moselley Makasso & Larry Hyman. 2017. Relative clauses in Basaá (A43). In Gratien Gualbert Atindogbé & Rebecca Grollemund (eds.), Relative clauses in some Cameroonian languages: Structure, function and semantism, 17–46. Berlin & Boston: De Gruyter Mouton. Google Scholar
Kayne, Richard. 1994. The Antisymmetry of syntax. Cambridge, MA: MIT Press.Google Scholar
Lambrecht, K. 2001. A framework for the analysis of cleft constructions. Linguistics 39(3). 463–516. Google Scholar
Makasso, Emmanuel-Moselley. 2008. Intonation et mélismes dans le discours oral spontané en Bàsàá. Marseille: University of Provence (Aix-Marseille) dissertation. Google Scholar
Merchant, Jason. 1998. Pseudosluicing: Elliptical clefts in Japanese and English. In Artemis Alexiadou, Nanna Fuhrhop, Paul Law & Ursula Kleinhenz (eds.), ZAS Working Papers in Linguistics 10, 88–112. Berlin: Zentrum fur Allgemeine Sprachwissenschaft. Google Scholar
Merchant, Jason. 2001. The syntax of silence: Sluicing, islands and the theory of ellipsis. New York: Oxford University Press. Google Scholar
Merchant, Jason. 2003. Section excised from submitted version of fragments and ellipsis. University of Chicago. http://home.uchicago.edu/merchant/pubs/stripping.pdf.
Merchant, Jason. 2004. Fragments and ellipsis. Linguistics and Philosophy 27. 661–738. Google Scholar
Merchant, Jason. 2008. Variable island repair under ellipsis. In Johnson Kyle (eds.), Topics in ellipsis, 132–153. Cambridge: Cambridge University Press.Google Scholar
Morgan, Jerry. 1973. Sentence fragments and the notion “sentence”. In Braj Kachru, Robert Lees, Yakov Malkiel, Angelina Pietrangeli & Sol Saporta (eds.), Issues in linguistics, 719–751. Urbana: University of Illinois Press. Google Scholar
Müller, Gereon. 2011. Pro-drop and morphological richness. http://home.uni-leipzig.de/muellerg/prodrop1.pdf.
Nishigauchi, Taisuke. 2006. Short answers as focus. Theoretical and applied linguistics at Kobe Shoi 9, 1–22. Kobe: Kobe Shoin Institute for Linguistic Sciences, Kobe Shoin Women’s University. Google Scholar
Ortega-Santos, Iván. 2016. Focus-related operations at the right Edge in Spanish: Subjects and ellipsis (Issues in Hispanic and Lusophone Linguistics 7). Amsterdam & Philadelphia: John Benjamins. Google Scholar
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (eds.), Elements of grammar, 281–337. Dordrecht: Kluwer. Google Scholar
Rodrigues, Cilene, Andrew Nevins & Luis Vicente. 2009. Cleaving the interactions between sluicing and preposition stranding. In Daniéle Torck & W. Leo Wetzels (eds.), Romance languages and linguistic theory: Selected papers from ‘going Romance’, Amsterdam 7–9 December 2006, 175–198. Amsterdam & Philadelphia: John Benjamins. Google Scholar
Ross, John Roberts. 1967. Constraints on variables in syntax. Cambridge, MA: MIT dissertation. Google Scholar
Temmerman, Tanya. 2013. The syntax of Dutch embedded fragment answers: on the PF-theory of islands and the wh/sluicing correlation. Natural Language and Linguistic Theory 31. 235–285. Web of ScienceCrossrefGoogle Scholar
Watters, John. 1979. Focus in Aghem: A study of its formal correlates and typology. In Larry M. Hyman (ed.), Aghem grammatical structure: With special reference to noun classes, tense-aspect and focus marking, 137–197. Los Angeles: University of Southern California. Google Scholar
Weir, Andrew. 2014. Fragments and clausal ellipsis. Amherst, MA: University of Massachusetts, Amherst dissertation. Google Scholar
Starting from Watters (1979), Hyman (1979), Aboh (2007), Hyman and Polinsky (2010) etc. it has been proposed that this position is a focus position in many Bantu languages. Bassong (2014) shows that only some wh-phrases are allowed to front to this position in the language and that there exists a lower FocP in the VP left periphery of Basaá. As pointed out by an anonymous reviewer, this lower FocP in the VP periphery is a good indication of phasehood of the verb phrase in the language. This means that when the argument kíí ‘what’ in (2b) occurs in sentence initial position, it moves cyclically via a lower phase position, namely the VP periphery (FocP) as briefly depicted in (i) where angled brackets indicate movement.(i)This is an appealing argument that goes beyond the scope of this paper and shall be discussed in future research.
[CPkííi[C[AgrSPmudaá[AgrSa[TP [Tbí [AsP[ Asptíj [FocP<kííi>[Foc[<kííi>]]]]]] 9.what 1.woman 1.smpst2- give VP[V<tíj>
There have been crosslinguistic evidence that pro-drop does not necessarily depend on the richness of agreement. Pro-drop licensing is subject to parametric variation (Gilligan 1987; Muller 2011). There are pro-drop languages with rich morphlogical verbal properties just as there are languages that license pro-drop without any morphological richness. On the other hand, there exist languages like French which exhibit rich verbal morphology with no pro-licensing. Gilligan (1987) crosslinguistic survey shows that of twenty-six languages with no subject agreement, only seventeen of them allow null subjects (see also Muller 2011 for more details on pro-drop).
Note that sentential negation is marked by the particle ɓéé that can be realized as ɓé depending on its structural position in the clause. When it is followed by a lexical item, it usually realized as ɓé while when it occurs at the end of a phonological phrase, it realized as ɓéé. However, its realization does not affect the morphosyntax.Consider the following examples:(i)
maaŋgɛ́ a-bí-lɔ ɓé jáání 1.child 1.sm-pst2-come neg 1.yesterday ‘The child did not come yesterday’(ii) maaŋgɛ́ a-bí-lɔ ɓéé 1.child 1.sm-pst2-come neg ‘The child did not come’
In Basaá, the English equivalent for ‘BE’ is ɓá. When inflected in the simple present tense, the latter realizes as je/jé (am/are/is) in the positive form and as tá/tâ/ta in the negative form. The fact that the copula can bear different tones boils down to its syntactic environment.
A detailed study of sluicing in Basaá is beyond the scope of this paper (see Bassong 2014 for details).
Sluicing in Basaá exhibits almost the same morphosyntactic properties as fragments, but the similarity between Basaá copular fragments and sluicing in Gungbe and Hungarian is much more obvious. A detailed study of sluicing in Basaá is beyond the scope of this paper (see Bassong 2014 for a detailed analysis).
Following van Craenenbroeck and Lipták (2006) implementation of the [E]-feature, Lipták and Aboh (2013: 112) propose that relative sluicing in Gungbe is represented as follows where angled brackets indicate ellipsis (cf. Example (30) above).
According to these authors, sluicing in Brazilian Portuguese and Spanish has two different sources:
Consider this example where Pol stands for polarity question marker.(iii)Q.
ɓaá mudaái a-n-sɔ́mb bikaat ɛ Pol 1.woman 1.sm-pst1-buy 8.books Pol
‘Has the woman bought the books?’A.
tɔ, proi a-n-sɔ́mb ɓé gwɔ̂ no 1.sm-pst1-buy neg 8.them
‘No, she hasn’t bought them’
In the spirit of Kayne’s (1994) head raising analysis, a relative clause such as (i) is derived as (ii).In this analysis, the definite artcle the is externally merged under D. D selects CP while what raises into Spec-CP is the relative head pictures along with the relative pronoun which (cf. Kayne 1994 chap 8 for more details). In this analysis, D does not selects for the NP/DP category containing the relative head pictures. The complement of D is CP but DP/DP only appears as the complement of D as a result of syntactic movement.
The pictures which I saw
[DP [D the [CP [DP picturesi [which [e]i]][C [IP…[e]i…]]]