An increasingly popular analysis of object gap sentences in many languages derives them in two steps: (a) V-raising out of VP, and (b) VP-ellipsis of the remnant, stranding the verb (V-stranding VP-ellipsis, VSVPE). For Hebrew, Hindi, Russian, and Portuguese, I show this analysis to be inadequate. First, it undergenerates elliptical objects in various environments, and second, it overgenerates nonexisting adjunct-including readings. For all the problematic data, simple argument ellipsis provides a unified explanation. The absence of VSVPE in languages that do allow V-raising and Aux-stranding VP-ellipsis raises an intriguing problem for theories addressing the interaction of head movement and ellipsis.
Within the growing literature on movement out of ellipsis sites, the interaction of head movement and ellipsis has attracted much recent attention (Van Craenenbroeck and Lipták 2008, Lipták and Saab 2014, Gribanova 2018, Hein 2018, Sailor 2018). A popular analysis of certain object gap (OG) sentences in many languages assumes an instance of this interaction: V-stranding VP-ellipsis (VSVPE). On this analysis, the lexical verb raises out of vP to some functional head (T or Asp), followed by ellipsis of the verbal projection. The ellipsis operation is, importantly, the same one found in English-type VPE. The only difference is that the lexical verb escapes the elided constituent by prior head movement, hence spells out.
[TP Subjk [T′ [Vi-v]j-T [vP tk [v′ tj [VP ti Obj]]]]]
VSVPE has been proposed for quite a few languages.1 By now, the analysis has been “canonized” in all leading surveys of ellipsis (see Van Craenenbroeck and Merchant 2013, Lipták 2015, Van Craenenbroeck 2017, Merchant 2019). Nevertheless, I argue here that VSVPE is not the right analysis for elliptical OG sentences. Rather, the simpler analysis of argument ellipsis (AE) is the right one.
[TP Subjk [T′ [Vi-v]j-T [vP tk [v′ tj [VP ti Obj]]]]]
The argument proceeds in two steps. First, I show that VSVPE is not needed, as AE can generate all the strings and interpretations that VSVPE allegedly does. Second, I show that VSVPE is not available, as it overgenerates nonexisting interpretations. This argument has been recently made for Hebrew (Landau 2018), and I will extend it to the three languages for which VSVPE has been most extensively applied: Hindi, Portuguese, and Russian. Obviously, one would like to subject the other languages mentioned in footnote 1 to similar testing, but this will have to wait for future work.
That VSVPE should be met with skepticism is, in fact, old news from the perspective of East Asian languages. Mandarin Chinese was the first language for which VSVPE was proposed (Huang 1987, 1991). This analysis was criticized in a series of studies (Xu 2003, Aoun and Li 2008, Cheng 2013). Otani and Whitman 1991, which extended VSVPE to Japanese and Korean, has been extensively criticized by Park (1997) and Kim (1999) for Korean, and by Hoji (1998), Oku (1998), Tomioka (1998), Saito (2007), Takahashi (2008), Abe (2009), and Sakamoto (2017) for Japanese. All these authors demonstrate that the AE analysis is much more adequate for these three languages, and this appears to represent the current consensus (for recent overviews, see Saito 2017, Sato 2019).
Among the problems that the VSVPE analysis faces in Chinese, Korean, and Japanese are that (a) the antecedent for the OG can be a subject; (b) subject gaps (in Japanese and Korean) display sloppy readings just like OGs; (c) OGs with sloppy readings can cooccur with a second object in ditransitive VPs; (d) gaps corresponding to PPs are attested; (e) adverbs (VP-adjuncts) are excluded from the ellipsis site; and (f ) verb identity is not imposed between the antecedent and the target clauses. Indeed, most of these properties are found in other languages, as we will see shortly.
While V-stranding VP-ellipsis is arguably unattested, other types of V-stranding ellipsis are real. In particular, Aux-stranding AuxP-ellipsis and V-stranding TP-ellipsis in polar responses are common constructions. In Landau 2020, I discuss the theoretical underpinning of the distinction between possible and impossible head-stranding ellipsis derivations.
The absence of VSVPE derivations is all the more striking in languages possessing the two ingredients that seem individually necessary and jointly sufficient to produce such derivations: V-raising and Aux-stranding VPE. Call these type H languages.
(3) Profile of type H languages
Aux-stranding VPE: ✓
V-stranding VPE: *
Argument ellipsis: ✓
Type H languages lie at the intersection of VPE languages and AE languages; they employ both types of ellipsis. Curiously, though, in these languages VPE is constrained not to generate strings that AE can generate, namely, sentences with an overt lexical verb and an OG. This cannot result from some economy-based competition, for the only conceivable principle that might adjudicate between the two options, MaxElide (Takahashi and Fox 2005, Merchant 2008, Hartman 2011), has the opposite effect from the one desired here: namely, it favors a bigger ellipsis over a smaller one (all being equal), whereas type H languages choose the smaller AE over the bigger VSVPE.2 The latter derivation, therefore, seems to be banned for some independent reason.
As I show in Landau 2018, Hebrew is a type H language. Chinese may well be too, if it has V-raising out of vP, at least over a certain range of constructions (Paul 2000). A number of studies have shown conclusively that Chinese employs AE to generate OG sentences (maybe alongside other null object strategies), but not VSVPE (Xu 2003, Aoun and Li 2008, Cheng 2013). Below, I will argue that Portuguese and Russian are also type H languages. Once again, type H languages are not currently acknowledged in the literature, so establishing their reality is the main focus of this article. The puzzle they raise for syntactic theory (how to rule out the particular interaction of head movement and ellipsis in VSVPE, but not elsewhere) is fully addressed in Landau 2020.
The structure of this article is as follows. In section 2, I present the main empirical test to be used here, the inclusion or exclusion of adjuncts in the ellipsis site, discussing both its force and potential pitfalls. In section 3, I argue against VSVPE in Hebrew, Hindi, Portuguese, and Russian, noting as well that it is not clearly motivated for Ndendeule and Swahili. Particular attention will be given to alleged positive evidence for VSVPE in these languages. I will argue that this evidence has been inconclusive at best or uninformative at worst regarding the proper analysis of OG constructions. Once properly tested, they reveal the hallmarks of AE rather than VSVPE. Section 4 concludes the article.
2 How to Use the Adjunct Test
As often noted in the literature, it is not easy to find empirical properties consistent only with AE and not with VSVPE. The best test so far, due to Park (1997) and Oku (1998), concerns the inclusion/exclusion of adjuncts in the ellipsis site. This test has been successfully applied to East Asian languages (as well as to Turkish; Şener and Takahashi 2010), but has yielded inconclusive results in some other languages—in my view, for reasons of implementation. So before we turn to the crosslinguistic data, let us review the logic of the test.
Consider first an example like (4), where English words are used for convenience only.
He read the sign loudly, but I didn’t read ___.
He didn’t read the sign loudly, but I read ___.
If the gap in (4a) corresponds to a full VP, it ought to allow (though not force) the inclusion of the adjunct loudly (present in the antecedent VP). This should give rise to the reading ‘I didn’t read the sign loudly’, which in turn allows the interpretation ‘I read the sign but not loudly’. If, on the other hand, the gap corresponds to a bare argument, the reading of the target clause should be ‘I didn’t read the sign’. These two readings are different enough to tease apart the two analyses.
The test should be used with caution, however. Notice that the VSVPE analysis does not have to produce a reading distinct from that of the AE analysis. First, the adjunct need not be included (as already shown in Sag 1976). Second, even if the adjunct is included, there are two ways to make the negative conjunct in (4a) true: by denying the manner or by denying the very event (I didn’t wash the car carefully could be true because I didn’t wash the car at all).
To compound the problem, we must beware of the polar inverse of (4a), namely, (4b). In this sentence, the AE analysis yields ‘but I read the sign’. This interpretation is consistent with my reading the sign loudly; in fact, the contrastive coordinator but facilitates this reading (and similarly for the frame . . . and I read ___ too). Thus, this sentence makes a very poor test for present purposes because the mere pragmatics of the construction, combined with AE, generates the same reading that the syntax of VSVPE is supposed to generate. This point is relevant for the discussion below; examples parallel to (4b) were occasionally cited by VSVPE proponents to “refute” the AE alternative—a fallacious move, as just explained.
In order to sharpen the Park-Oku test, let us test examples with creation verbs. The advantage of these predicates is that under negation, the existence of their object is denied; subsequent reference to this object is then perceived as anomalous. Consider the following format, again using English words for convenience:
(5) He baked a cake with baking powder, but I didn’t bake ___. It came out flat.
Once again, a VSVPE analysis of the gap should recover the sentence I didn’t bake a cake with baking powder, which allows the reading ‘I baked a cake but not with baking powder’. Further reference to the cake in the last clause should be both possible and natural.3 In contrast, on the AE analysis the gapped clause recovers as I didn’t bake a cake. Since my cake was never baked, the continuation It came out flat should be anomalous. This test uses the same logic as the original Park-Oku test, but produces stronger judgments, as we will see.
3 Empirical Results of the Adjunct Test
In sections 3.2–3.5, I apply the adjunct test to Hebrew, Hindi, Portuguese, and Russian, concluding that these languages do not employ VSVPE in their grammars. While I have not been able to test Ndendeule and Swahili, in section 3.1 I show that these languages too do not offer conclusive evidence for VSVPE.
3.1 Ndendeule and Swahili
It has been argued that certain Bantu languages, like Ndendeule and Swahili, employ VSVPE (Ngonyani 1996). However, a closer look reveals that the alternative analysis of AE was not ruled out. For example, Ngonyani shows that OG sentences with a VP antecedent need not contain an object clitic, unlike OG sentences without a VP antecedent. Yet the clitic in the latter is associated with a null object pronoun, according to Ngonyani. Hence, it is still possible that elided objects do not require an object clitic, and merely need a proper antecedent (which very often occurs inside a VP). Other properties, like omission of two objects or sloppy readings, do not favor VSVPE over AE (possibly applying twice). Finally, Ngonyani (1998) cites a grammatical OG sentence in which the antecedent and target clauses contain different verbs; thus, the Verb Identity Requirement (Goldberg 2005:171), normally taken to be a hallmark of VSVPE, appears not to hold in Swahili.4
The VSVPE analysis of OG sentences in Hebrew has been defended and elaborated in a series of works (Doron 1990, 1999, Sherman 1998, Goldberg 2005). The analysis has gained much currency in the field and is regularly cited as a cornerstone in all authoritative survey articles on ellipsis (Van Craenenbroeck and Merchant 2013, Van Craenenbroeck 2017, Merchant 2019). At around the same time that VSVPE was developed for Hebrew, it was also developed for Chinese (Huang 1987, 1991), Japanese and Korean (Otani and Whitman 1991), and Irish (McCloskey 1991).
Significantly, the VSVPE analysis has been retracted and superseded in all these languages. In Chinese, Japanese, and Korean, the current consensus takes OG sentences to involve AE. In Irish, the latest proposal is TP-ellipsis under a Pol(arity) head (McCloskey 2012, 2017). The analysis has survived the longest for Hebrew, Landau 2018 being the first systematic effort to demonstrate the inadequacy of VSVPE for that language.5 In this section, I summarize the main findings of that article.
To begin with, Hebrew OG sentences allow a range of interpretations that are not derivable from a pronominal source ( pro) or a topic-oriented variable. Thus, we find OGs with sloppy, nonspecific, and quantificational readings, as well as obligatory bound readings with reflexives, showing remarkable similarity to the East Asian OG sentences (Takahashi 2014). These data (presented in Landau 2018 but omitted here for space reasons) are not problematic for the VSVPE analysis. Yet I also show there that none of the constraints that are supposed to distinguish between straight null objects and VSVPE (on Doron’s (1990, 1999) and Goldberg’s (2005) accounts) really hold. Such objects are not island-sensitive, not necessarily inanimate, and not necessarily nominal. In fact, every sentence for which VSVPE has been proposed can easily be derived by AE.
But not vice versa. The crucial data that only AE but not VSVPE can generate involve adjuncts.6 Consider (6). The first sentence in response (6B) cannot be used to negate the adverb alone, namely, the source of acquaintance. Rather, the negation scopes over the event itself, with the entailment that B is not acquainted with the relevant woman. Hence, the corrective continuation is infelicitous. In contrast, bare negation, as in response (6B′), can easily be used to convey the intended meaning.
ata makir ota me-ha-tixon?
you know her from-the-high.school
‘Do you know her from high school?’
#lo makir ___. me-ha-cava.
not know from-the-army
‘I don’t know her. From the army.’
‘No, from the army.’ / ‘I don’t. From the army.’
The combination of negation and a creation verb offers another opportunity for testing the VSVPE analysis. The negated OG sentence in (7a) entails that there is no cake baked by Gil. It is therefore infelicitous to refer to this empty set by a pronoun in the following sentence. Crucially, to obtain the reading that the cake was baked but not according to the recipe, Hebrew must resort to stripping (7b), where the entire TP is missing and the remnant is a displaced contrastive focus, not necessarily the subject (see Doron 1999, Depiante 2000, Merchant 2004). Note that the grammaticality of (7b) confirms that the problem in (7a) is not due to the occurrence of the antecedent inside an ellipsis site.
Yosi afa et ha-uga lefi ha-matkon. hi hayta me’ula. Gil lo
Yosi baked ACC the-cake according the-recipe it was fabulous Gil not afa ___. #hi hayta mag’ila.
baked it was gross
‘Yosi baked the cake according to the recipe. It was fabulous. Gil didn’t bake the cake. #It was gross.’
GIL, LO ___. hi hayta mag’ila.
Gil not it was gross
‘Gil didn’t. It was gross.’
Even if VPE (and by extension, VSVPE) supports adjunct-excluding readings, it does not force them. The fact that Hebrew OG sentences do force them, then, suggests that they involve not ellipsis of VP but ellipsis of the internal argument alone.
Recall that in addition to having AE and lacking VSVPE, type H languages, as characterized in (3), should harbor the syntactic machinery required for VSVPE: V-raising and Aux-stranding VPE. Indeed, Hebrew has V-to-T raising (at least optionally) (8a) (Doron 1983, 1990, Shlonsky 1987, Borer 1995) and exhibits canonical VPE in periphrastic constructions (8b).
Gil [V šaxax] [Adv le-itim krovot] tV et ha-maftexot ba-oto.
Gil forgot to-times frequent ACC the-keys in.the-car
‘Gil often forgot the keys in the car.’
Gil haya maskim la’azor lanu?
Gil was agree.PRTC to.help to.us
‘Would Gil have agreed to help us?’
Batuax hu haya ___.
surely he was
‘Surely he would have.’
Therefore, the absence of VSVPE in Hebrew cannot be blamed on any obvious language-internal factor. It represents a genuine puzzle.
Simpson, Choudhury, and Menon (2013) have argued that alongside AE, Malayalam, Bangla, and Hindi display genuine VSVPE; similarly, Manetta (2018, to appear) has argued for VSVPE in Hindi. Following Kumar (2006) and Bhatt and Dayal (2007), Manetta assumes that the verb in Hindi raises to a vP-external head position, specifically Asp0, escaping vP-ellipsis. Because V-to-Asp movement is obligatory in Hindi, Aux-stranding vP-ellipsis never materializes, as the aspectual auxiliary never surfaces without a lexical verb. Strictly speaking, then, Hindi is not a type H language. Nevertheless, the VSVPE analysis for this language is quite popular, so it merits close examination.
The purported evidence for VSVPE presented by Simpson, Choudhury, and Menon (2013) consists in sentences that allegedly pass the Park-Oku test for adjunct inclusion in the ellipsis site. The Hindi example in (9) is analyzed with V-raising followed by ellipsis of the VP, which contains both an object and an adjunct.
Amit-ne dheere-dheere ek vritt banaya. Gita-ne bhi ___ banaya.
Amit-ERG slowly one circle make.PRES.M.SG Gita-ERG also make.PRES.M.SG
‘Amit drew a circle slowly. Gita also drew (a circle slowly).’
The second conjunct is indeed compatible with the adjunct (Gita may have drawn the circle slowly), but given the discussion of (4b), this does not demonstrate that the adjunct was copied together with the object. Adjuncts are optional to begin with. This is evident in (10), in which the presence of an overt object in the second sentence rules out a VSVPE derivation. Nonetheless, the adjunct reading is no less accessible than it is in (9).7
Amit-ne dhiire-dhiire ek vritt banaayaa. Gita-ne chaukor banaayaa.
Amit-ERG slowly one circle make.PRES.M.SG Gita-ERG square draw.PRES.M.SG
‘Amit drew a circle slowly. Gita drew a square (slowly).’
More revealingly, just as in Japanese, Chinese, Turkish, and Hebrew, and unlike in English, negation in the second conjunct necessarily negates the event and cannot solely target the hypothetical, elided adjunct.
Amit-ne dhiire-dhiire ek vritt banaayaa, lekin Gita-ne nahiiN ___
Amit-ERG slowly one circle draw.PRES.M.SG but Gita-ERGNEG
‘Amit drew a circle slowly, but Gita didn’t draw a circle.’
The second conjunct in (11) can only mean that Gita did not draw any circle; it cannot mean that she drew a circle but not slowly. Finally, the combination of negation and a creation verb in the second conjunct implies that no object came into being, making subsequent reference to the object infelicitous (parallel to Hebrew (7a)).
John-ne apnaa cake recipe-ke anusaar banaayaa. vo bahut tasty thaa. Bill-ne
John-ERG self cake recipe-GEN according made it very tasty was Bill-ERG
nahiiN ___ banaayaa. #vo bahut bekaar thaa.
NEG made it very bad was
‘John baked his cake according to the recipe. It was very tasty. Bill didn’t make a cake.
#It was very bad.’
Careful application of the adjunct test, then, indicates that Simpson, Choudhury, and Menon’s (2013) conclusion from (9) was premature. There is no evidence that Hindi employs VSVPE in its grammar and there is, in fact, strong evidence that it does not. While Malayalam and Bangla should be subjected to parallel scrutiny, I will assume, in the absence of counterevidence, that they are no different from Hindi in this respect (Simpson, Choudhury, and Menon’s examples from these two languages are analogous to (9)). Of course, evidence for the existence of AE in these languages that the authors cite is unaffected and remains conclusive—in fact, convergent with what we learn from (10)–(12).
The absence of the adjunct-including reading is explicitly addressed in Manetta to appear and especially in Manetta 2018, and recognized as a challenge to the VSVPE analysis. Manetta (2018) then discusses special circumstances in which this reading becomes available. Crucially, she analyzes these circumstances as involving contrastive polarity ellipsis, where a remnant TP (from which a contrastive XP has been extracted) is deleted under a Pol head, which hosts the stranded verb. Such derivations are indeed possible (and sanctioned on the analysis proposed in Landau 2020) and may be realized in Portuguese, too, for some speakers (see section 3.4). However, Manetta’s account for the absence of adjunct-including VSVPE derivations in Hindi contains a lacuna. She presents the following example:
Sita-ne kah-aa ki Ram Chomsky-ka naya lekh dhyaanse paRh-egaa.
Sita-ERG say-PST.M.SG that Ram Chomsky-GEN new article carefully read-FUT.M.SG
‘Sita said that Ram will read the new article by Chomsky carefully.’
Raj nahiiN ___ paRh-egaa.
Raj NEG read-FUT.M.SG
‘Raj will not read the new article by Chomsky (?*carefully).’
(13b) cannot involve any ellipsis, Manetta argues. The missing object is pro, and therefore there is no silent copy of the antecedent manner adverb. The reasoning is as follows. Normal polarity ellipsis deletes the subject along with the TP, so it cannot underlie (13b), where the subject survives. Because of V-to-Asp-to-T-to-Pol raising, however, the parallelism domain for evaluating ellipsis is Pol′ (the minimal projection containing the binder of the verbal trace). MaxElide then dictates that TP rather than vP be deleted. Consequently, TP-ellipsis is unavailable (generating a string without the subject), and vP-ellipsis is unavailable (due to MaxElide), leaving only the pro analysis, which correctly does not support the adjunct-including reading.
This account is problematic in several respects, internally and also externally to the MaxElide system. Note first that the notion of blocking invoked here is rather curious. Unlike the original cases of MaxElide, where a grammatical bigger ellipsis blocks an ungrammatical smaller one, here the bigger ellipsis is in fact not available, for it generates a different string (namely, “Neg-V ___”). How, then, can it block the smaller ellipsis? Second, (13a–b) offer just the right opportunity for bypassing MaxElide, namely, by including a contrastive subject in the elliptical clause. The presence of such contrastive elements is known to bypass MaxElide, as in the following pair:
I know what books Bill likes and I know what movies (*he does).
I know what books Bill likes and I know what movies MIKE does.
Therefore, even the MaxElide account predicts vP-ellipsis to be possible in (13b), leaving Manetta’s account with the puzzle of why the adjunct-including reading is absent.
Furthermore, some elliptical parse of (13b) must be available. Evidence for this comes from the familiar sloppy-reading test. (15) is modeled on Manetta’s (13), only now the elided object supports a sloppy reading (15b), in contrast to an overt pronoun (15c). Importantly, the adjunct including reading is still unavailable throughout.
Rami apnaa lekh dhyaanse paRh-egaa.
Ram self.M.SG article carefully read-FUT.M.SG
‘Rami will read his article carefully.’
Rajj nahiiN ___ paRh-egaa.
Raj NEG read-FUT.M.SG
‘Rajj will not read hisi/j article (*carefully).’
Rajj use nahiiN paRh-egaa.
Raj it.DATNEG read-FUT.M.SG
‘Rajj will not read it (= hisi/*j article) (*carefully).’
Given the sloppy reading, (15b) must be elliptical. However, it cannot be elliptical on Manetta’s approach, which analyzes it with pro, predicting only a strict reading, contrary to fact. Even worse, the MaxElide explanation favors vP-ellipsis here. TP-ellipsis is not a “competitor” because of the contrastive subject that should not be elided. Therefore, vP-ellipsis solely competes with AE, both sparing the subject and yielding equivalent interpretations. The bigger vP-ellipsis should win over the smaller AE, yet the absence of the adjunct-including reading, once again, proves that vP-ellipsis is not an option.8
Hence, Portuguese is expected to display VSVPE, and indeed, the literature is unanimous in analyzing OG sentences like (17) as involving VSVPE (see Martins 1994, 2016, Kato 2003, Cyrino and Matos 2005, Santos 2009, Rouveret 2012). Different analyses posit different landing sites for V-raising in VSVPE (Asp, T, or Σ), but they all share the core idea that the stranded V originates within the VP projection undergoing ellipsis.10 A single recent paper (Cyrino and Lopes 2016) acknowledges that Brazilian Portuguese permits AE alongside VSVPE.
Ela está lendo livros ás crianças, mas ele não está lendo __.
she is reading books to.the children but he not is reading
‘She is reading books to the children, but he is not.’
I will argue that while (16b) is indeed a standard case of VPE, (17) involves not VSVPE but AE. Just as in the East Asian languages, Hebrew, and Hindi, once the proper tests are applied, it will be seen that (possibly multiple) AE provides a superior account of the data. The strength of this alternative has not previously been appreciated, because most studies have counterposed the VSVPE analysis not to AE but to pro or Ā-variable analyses. Indeed, Portuguese OGs display a wider distribution than what the latter analyses would predict (e.g., nonpronominal reference, island-insensitivity), but these properties are no more problematic to the AE analysis than to the VSVPE one.
First, it has been suggested that Portuguese has no process of PP-ellipsis, so that ditransitive VPs missing both objects, like (17), can only arise from VPE (Matos 1992, Rouveret 2012). But in fact this is false, and PPs can undergo independent ellipsis in Portuguese just as they can in Japanese and Hebrew (see Santos 2009:29, 39). Santos claims that PP-drop is island-insensitive but DP-drop is island-sensitive because only the latter is derived by Ā-movement (similarly to the claims made by Doron (1990, 1999) and Goldberg (2005) for Hebrew). However, alongside Santos’s examples (18a–b) (where obligatory PPs undergo ellipsis), we find (19a–b), which indicate that even direct object gaps may occur inside islands.11
A Teresa entregou as chaves ao porteiro e eu conheço a senhora que the Teresa gave the keys to.the porter and I know the lady that entregou o carro ___.
gave the car
‘Teresa gave the keys to the porter and I know the lady who gave him the car.’
Pus os meus filhos nesta escola e conheço uma senhora que pôs put.1SG the my children in.this school and know.1SG a lady that put.3SG os nestos ___.
‘I put my children in this school and I know a lady who put her grandchildren there.’
A Teresa deu o seu lugar a um amigo e eu vi uma senhora que deu ___ the Teresa gave the her space to a friend and I saw a woman that gave a um desconhecido.
to a stranger
‘Teresa gave her space to a friend and I saw a woman who gave her space to a stranger.’
Pus os meus móveis antigos na cave mas conheço um vizinho put.1SG the my furniture old in.the basement but know.1SG a neighbor
que pôs ___ na sala.
that put.3SG in.the living.room
‘I put my old furniture in the basement but I know a neighbor who put his old furniture in the living room.’
That multiple AE, including PP-ellipsis, must be available in Portuguese can also be established on the basis of object-oriented secondary predicates (OOSPs). As Rouveret (2012) points out, VPE cannot strand an OOSP, which merges too low to escape deletion.
(20) *Lucy submitted the manuscript unfinished and Jan did badly typed.
Rouveret (2012:928, (69c)) then notes that in Portuguese, a ditransitive VP with a missing direct object and an overt indirect object accepts an OOSP, concluding that this option arises from a null object. However, grammaticality is unharmed if the PP goes missing too.
O João devolveu o livro à Maria em bom estado e o Pedro devolveu the João returned the book to.the Maria in good condition and the Pedro returned
___ ___ mais estragado.
‘João returned the book to Maria in good condition and Pedro returned it to her more damaged.’
(21) can only be derived by multiple AE of the direct and indirect objects; VSVPE is not possible given its inability to strand an OOSP. Hence, AE in Portuguese, as in the other languages discussed above, does not distinguish DPs from PPs (the null hypothesis, it would seem).
Let us now consider another salient feature of the VSVPE analysis, the Verb Identity Requirement. This condition was taken to hold in Portuguese in Kato 2003 and Cyrino and Matos 2005, but Santos (2009:56–58) and Cyrino and Lopes (2016) convincingly refute it with examples that either lack both objects or permit a sloppy reading for the OG (hence, must involve VSVPE on the standard account).
O João venedeu livros à Teresa ontem e a Ana ofereceu ___ ___.
the João sold books to.the Teresa yesterday and the Ana offered
‘João sold books to Teresa yesterday and Ana offered them.’
Ontem o João pôs o dinheiro na gaveta, mas Pedro guardou ___ na
yesterday the João put the money in.the drawer but Pedro kept in.the cofre.
‘Yesterday, Joãoi put hisi money in the drawer, but Pedroj kept hisj money in the safe.’
It is sometimes claimed that verb identity still holds in this type of construction but that “focalized or weakly contrasted material is exempted from this requirement” (Rouveret 2012:933; see also Lipták 2012, Gribanova 2013b). However, the fact that the unshared material in the target clause must be at least contrastive if not focused is a hallmark of ellipsis in general, so it is not clear what is gained by duplicating this condition just for deviations from verb identity. Indeed, if VSVPE does not exist and these are all AE cases, then the peculiar condition of verb identity can simply be dispensed with, and the pattern of verb (non)identity can be made to follow from the general felicity conditions on the residue of ellipsis (Rooth 1992, Merchant 2001).
So far, the data suggest that nothing rules out an AE analysis of Portuguese OG sentences, either with ditransitive VPs or inside islands. I now turn to data that pose serious difficulties for the accepted VSVPE analysis.
Consider adjunct inclusion. In the following example, it is claimed that the adjunct in the antecedent clause is also construed in the target clause (Santos 2009:28):
A Raquel não limpou o carro cuidadosamente. Mas a Ana limpou ___.
the Raquel not cleaned the car carefully but the Ana cleaned
Preferred: ‘Raquel didn’t clean the car carefully, but Ana did.’
Dispreferred: ‘Raquel didn’t clean the car carefully, but Ana cleaned it.’
Santos (2009) claims that the antecedent adjunct is obligatorily copied to the ellipsis site, but Rouveret (2012) claims that this is only the preferred, default reading.12 Indeed, given the optionality of adjuncts, even VPE is not forced to apply to them. In this sense, Portuguese is like English, where VPE only optionally copies adjuncts (Sag 1976). However, as pointed out above for Hindi (9) (see also the comments on (4b)), OG sentences constructed with too or also are not informative with regard to the syntactic issue of adjunct inclusion, because they very easily facilitate and even favor that reading solely for pragmatic reasons. To probe the syntactic question, one needs a different setup. Two useful setups that were demonstrated above involve either a negated target clause or attempted reference to the missing object of a negated creation event.
Applying these adjunct tests to Portuguese, we obtain a pattern similar to that found in Japanese, Chinese, Korean, Turkish, Hebrew, and Hindi, although it is less robust.
(24) A Raquel limpou o carro dela cuidadosamente, mas a Ana não limpou ___. the Raquel cleaned the car her carefully but the Ana not cleaned
‘Raquel cleaned her car carefully, but Ana didn’t clean it/hers.’
‘Raquel cleaned her car carefully, but Ana cleaned it/hers not carefully.’
Out of seven speakers consulted, two rejected the OG sentence altogether. Of the remaining five, three only accepted reading (a), that is, the one generated by AE. Two speakers, however, reported an ambiguity, accepting both readings. One could attribute reading (b) to VSVPE, but this is not necessary; it could also arise from “adjunct ellipsis” applying in tandem with AE (similarly to DP- and PP-ellipsis cooccurring in ditransitive PPs) and placing the silent adjunct under the scope of the overt negation. This should be a marginally available option, explaining why only a minority of speakers employ it (below I consider an alternative explanation for the adjunct-including judgments).
Consider next the following question-answer pair:
Q: Entregaste o artigo à Maria na biblioteca?
gave.2SG the paper to.the Maria at.the library
‘Did you give the paper to Maria at the library?’
Santos claims that the locative adjunct must be understood as part of the confirmation (‘Yes, I gave the paper to Maria at the library’). However, this again may be a reflection of a default reading (i.e., unless specified otherwise, a positive answer confirms all parts of the question). If we just switch the polarity of the answer, a different result emerges.
Q: Entregaste o artigo à Maria na biblioteca?
gave.2SG the paper to.the Maria at.the library
‘Did you give the paper to Maria at the library?’
A: Não entreguei ___ ___. #Entreguei-lho no bar.
not gave.1SG gave.1SG-CLDat+CLAcc at.the cafeteria
‘I didn’t give it to her. #I gave it to her at the cafeteria.’
Out of seven speakers consulted, six rejected the continuation of the negative reply, suggesting that the missing material in it consists of the two arguments without the adjunct. Only one speaker allowed the answer to be understood as ‘I gave it to her not at the library’.13 The majority of speakers, then, employ multiple AE, while a minority further allow adjunct ellipsis (see below, though, for an alternative explanation).
As a final confirmation that VSVPE is not part of Portuguese grammar, consider attempted reference to the object of a creation verb under negation.
O João escreve poemas por desespero. O Pedro não escreve ___. #Os dele
the João writes poems out.of despair the Pedro not writes those of.him
‘João writes poems out of despair. Pedro doesn’t write (poems). His are cheerful.’
Out of five speakers consulted, four rejected the continuation of the negative reply, implying that they only allowed it to be understood as ‘Pedro doesn’t write poems’—that is, the outcome of AE. The single speaker who accepted the interpretation ‘Pedro writes poems not out of despair’ (which allows further reference to the poems) appeared to apply adjunct ellipsis in addition to AE.
To repeat the argument: If VSVPE had been a productive option in Portuguese grammar as Aux-stranding VPE is (see (16b)), one would have expected a much higher rate of acceptance of adjunct-including interpretations in OG sentences, similar to the status of (4a) in English. The fact that the pervasive pattern excludes the adjunct thus strongly supports the view that such sentences result from AE and not VSVPE. On this analysis, Portuguese is another type H language.
A legitimate question is why the Portuguese data are not as clear-cut as the data reported above from other languages. Trivially, this may be due to insufficient data; perhaps a more careful investigation will reveal that OG constructions in other languages are just as noisy. A marginal strategy of adjunct ellipsis is also a possibility, as noted above. There is, however, a potential theoretical explanation. In Landau 2020, I discuss a construction (in Irish, Finnish, and Hungarian) in which V-stranding ellipsis does occur. Crucially, though, the stranded V raises to a polarity head (Σ/Pol) at the left periphery and the elided category is bigger than VP, probably a TP. This type of ellipsis is expected to be adjunct-inclusive just like simple “Yes!” or “No!” responses are (the polarity particle realizing the Σ /Pol head, as is standardly assumed).
In fact, just this sort of analysis has long been advocated for Portuguese OG constructions (Martins 1994, 2016, Costa, Martins, and Pratas 2012). It may well be the case, then, that Portuguese speakers generate OG sentences either via AE or via V-stranding TP-ellipsis (under some polarity focus).14 This would account for the nonuniform distribution of judgments seen above. In any event, the fact that some (maybe most) speakers exclude adjuncts in properly constructed OG sentences like (24)–(27) indicates that AE is the only elliptical strategy available to them.
As a final consequence, the proposal that Portuguese employs V-stranding AE alongside Aux-stranding VPE allows a simple account of certain “parallelism” puzzles that have been debated in the literature (see Matos 1992, Cyrino and Matos 2005, Rouveret 2012). The following contrast holds in European Portuguese:
O João não tinha [vP dado [DP o presente] [PP à mãe]], mas a prima
the João not had given the gift to.the mother but the aunt
tinha ___ / dado ___ ___.
‘João had not given the present to his mother, but his aunt had.’
O João deve ter lido esse livro esta tarde e o Luis deve também
the João must have read this book this afternoon and the Luis must too
ter *(lido) ___ ___.
‘João must have read this book this afternoon and Luis must have also.’
The generalization is that ellipsis may strand a perfect participle that follows an inflected auxiliary but must strand it after an uninflected auxiliary. Since all of these elliptical sentences represent VPE for the above-mentioned authors, this contrast is a puzzle. Matos (1992) states that ter ‘have’ is a proper governor (for ellipsis purposes) only when in T. Cyrino and Matos (2005) assume that VSVPE in (28a) depends on incorporation of the participle into Aux, but there is evidence against it from (lack of) adjacency effects. Rouveret (2012) advances a phase-based approach, where only phase heads license ellipsis (of their complement). By assumption, Part0 (to which the verbal root raises to form a participle) is a phase head, but Perf 0 (the base position of ter) is not. An inflected ter licenses ellipsis not from its surface position (in T) but from a v position above PerfP, which is another phase head.
The present approach, I believe, offers a simpler solution. To understand the contrast in (28), we only need to assume, with Cyrino and Matos (2005), that the relevant licensor of VPE in European Portuguese is T. Thus, (28a) without the participle represents a simple case of VPE. (28b) without the participle is not possible because the elided constituent is a sister of Aux, a nonlicensor. Crucially, the versions of (28a) and (28b) with the participle do not involve VSVPE, which is impossible on the current proposal; rather, they involve multiple AE, as in (21), (22a), (25), and (26). No further assumptions about the variable scope of ellipsis or incorporation need to be invoked.
The status of verb movement in Russian has been under dispute for more than two decades (Bailyn 1995, King 1995, Slioussar 2011, Gribanova 2013b). It is generally agreed that V does not raise to T in SVO clauses (although it might under “inversion,” in XP-V-S clauses; see Bailyn 2004). However, Gribanova and Bailyn argue that V does raise to some low aspectual projection. Indirect evidence comes from across-the-board movement of V to a position below the Aux-filled T (29a). In addition, Russian displays Aux-stranding vP-ellipsis (29b) (Kazenin 2006), so we may take it to be another type H language.
Petja budet priglašat’ Mašu v muzej segodnja, a Dinu v
Petya.NOM will invite.INF Masha.ACC in museum today and Dina.ACC in
‘Petya will invite Masha to the museum today, and Dina to a movie tomorrow.’
Ja budu pomogat’ Pete, a Kolja ne budet ___.
I will help.INF Petya but Kolya not will
‘I will help Petya, but Kolya won’t.’
Gribanova (2013b) claims that OG sentences in Russian can arise either by some island-sensitive Ā-dependency or by VSVPE. Similarly to Doron’s (1990, 1999) and Goldberg’s (2005) approach to Hebrew OG sentences, she then uses island environments to isolate the VSVPE construction. As in Rouveret’s (2012) analysis of Portuguese, Gribanova suggests that VSVPE in Russian strands the verb in Asp. Consequently, elements of the verbal complex that originate within the vP are predicted to fall under the Verb Identity Requirement whereas those originating in Asp are predicted to allow a mismatch between the antecedent verb and the stranded verb in the target clause.
Včera ja ne poznakomila Mašu s Sašej, i poka ne
yesterday I NEG introduced.1SG.F Masha.ACC with Sasha.INSTR and until NEG poznakomlju ___ ___, ne uedu.
‘I didn’t introduce Masha to Sasha yesterday, and I won’t leave until I do.’
Gribanova notes that PPs can be independently dropped; hence, the choice of a ditransitive verb in OG sentences in itself cannot rule out a multiple AE analysis. In fact, Gribanova explicitly sets aside the AE alternative (see her footnote 3). Importantly, a key identifier of VSVPE in her account—the obligatory presence of a linguistic antecedent—holds equally of AE, both being instances of ellipsis. Gribanova only remarks (2013b:93n3) that “the strong requirement for verb matching” should serve as evidence against AE. However, this requirement is easily defeasible, even by Gribanova’s own account (see below), and is perfectly consistent with the AE analysis: nonelided constituents in the target clause should be in some contrast with the corresponding constituents in the antecedent clause.
These empirical considerations indicate that VSVPE has no advantage over AE in accounting for the properties of Russian OG sentences.15 In fact, it faces some problems that the AE analysis does not, as pointed out in Erteschik-Shir, Ibnbari, and Taube 2013 and Bailyn 2017. I will mention some of these objections and add others to them.16
First, Erteschik-Shir, Ibnbari, and Taube point out that there is no clear evidence for any island-sensitive OG strategy in Russian. Allegedly ungrammatical examples involve insufficient prominence, in the deictic context, of the “continued” topic that is the required antecedent for the OG. Once this prominence is ensured, OGs inside islands are acceptable even without any linguistic antecedent.
More relevantly to our concerns, Bailyn shows that there is no formal identity requirement holding between the antecedent and the target verbs. While Gribanova goes to considerable lengths to demonstrate that lexical prefixes (being VP-internal) have to match and that superlexical prefixes and aspectual suffixes do not, Bailyn cites counterexamples of every type. What seems to be involved is a general constraint (familiar from other AE languages) on the range of permissible contrastive semantic relations between the two verbs (e.g., antonymy, intensification). Lexical prefixes may or may not express contrasts of that sort, which leads to variable results in verb mismatch situations. Crucially, different verb stems are possible too.17
Kto-to skazal, čto vse nenavidjat Ivana, tak čto menja udivil fakt, čto
someone said that everyone hates Ivan so that me surprised fact that
Nadja ljubit ___.
‘Someone said that everyone hates Ivan, so the fact that Nadya loves him surprised me.’
Russian thus joins all the other V-stranding languages in which identity is but one possible relation between the antecedent and target verbs. One can go further and show that OG constructions in Russian do not even respect the weaker condition of “valence identity” (i.e., parallel argument structures) that is an absolute prerequisite for VSVPE (see Landau 2018 for parallel facts in Hebrew). An OG in a ditransitive VP can take the object of a monotransitive VP as antecedent, and vice versa; note that the OGs in (32) occur inside islands, so under Gribanova’s approach, they would have to be analyzed as cases of VSVPE.
Ivan sfotografiroval čašku pered tem kak postavit’ ___ na polku.
Ivan.NOM photographed.3SG cup.ACC before as put.INF on shelf.ACC
‘Ivan photographed the cup before putting it on the shelf.’
Ivan postavil čašku na polku posle togo kak sfotografiroval ___.
Ivan.NOM put.3SG cup.ACC on shelf.ACC after as photographed.3SG
‘Ivan put the cup on the shelf after photographing it.’
Next, consider adjunct inclusion. Contrary to Gribanova’s (2013b:102) claim that all adjuncts are included in the ellipsis site, a closer inspection reveals that Park’s (1997) and Oku’s (1998) discovery for Korean and Japanese carries over to Russian too: adjuncts are not included. According to Bailyn (2017), the target clause in (33) cannot be understood as ‘I recorded Vasiliev but not often’.
Ty snimal Vasil’eva často, a ja ne snimal ___.
you recorded Vasiliev often but I NEG recorded
‘You recorded Vasiliev often but I didn’t record him.’
The test with creation verbs applied above yields consonant results.
Ivan pišet stixi ot otčajanija. Ja uveren čto Sergej ne pišet ___.
Ivan writes.3SG poems from despair I sure that Sergei not writes.3SG
#Oni vsegda radostnyje.
they always cheerful
‘Ivan writes poems out of despair. I am sure that Sergei does not write (poems). #They are always cheerful.’
Because the OG in the second sentence of (34) results from AE and not VSVPE, no copy of the antecedent adverb is present under the scope of negation. The sentence thus means that Sergei does not write poems and cannot mean that he writes poems but not out of despair. The continuation, referring to nonexisting poems, is infelicitous. Importantly, felicity is obtained when ellipsis genuinely targets a larger constituent—for example, the complement of Neg, which obviously includes the adverb ( . . . čto Sergei net ___ ‘that Sergei not’), mirroring the Hebrew contrast in (7a–b).
As noted, Gribanova (2013b) uses antecedentless OG sentences to show that Russian null objects are Ā-variables constrained by islands. The following example is intended to show their sensitivity to the indicative island constraint in Russian:
[Something is lying on the floor.]
#Ja byl uveren, čto kto-to uže podnjal ___.
I was sure.SG.M that someone.NOM already under.held.SG.M
Intended: ‘I was sure that someone already picked it up.’
This example, however, cannot be taken as conclusive evidence that indicative complements with an OG must be derived by VSVPE. The reason is that AE requires a linguistic antecedent no less than VSVPE does. The way to distinguish the two is to introduce a DP antecedent (for AE) without introducing a VP antecedent (strictly speaking, a lexical VP, as (36A) may involve a null copula).
Smotri, košeljok ješčo zdes’!
look wallet.NOM still here
‘Look, the wallet is still here!’
Ja byl uveren, čto kto-to uže podnjal ___.
I was sure.SG.M that someone.NOM already under.held.SG.M
‘I was sure that someone already picked it up.’
Thanks to the DP antecedent košeljok ‘wallet’, B’s response is much more acceptable than utterance (35), although they are identical in form. But it is not derivable on Gribanova’s assumptions. A null object is not licensed inside indicative complements and VSVPE cannot apply without a proper VP antecedent. The AE analysis, in turn, provides a principled explanation for the contrast between (35) and (36): only in the latter is the OG preceded by an explicit DP antecedent.18
In conclusion, once considered in its entirety, the range of OG sentences in Russian can be fully accounted for by AE and possibly some null object strategy, with no recourse to VSVPE. Moreover, VSVPE overgenerates nonexisting readings, hence must be blocked. Given that Russian has both V-raising and Aux-stranding vP-ellipsis, the absence of VSVPE goes beyond any language-specific constraints and calls for some principled explanation.
VSVPE is not an adequate analysis for OG sentences in a number of languages for which it has been proposed (Hebrew, Hindi, Russian, and Portuguese). The dethroning of VSVPE in these languages joins and reinforces the abandonment of this analysis for East Asian languages. Establishing this result requires careful application of the adjunct inclusion/exclusion test, which had been neglected or misinterpreted in previous works. While it is still possible that VSVPE is adequate for languages not examined here, skepticism is fully warranted: if the analysis does not withstand scrutiny where it has been defended most vigorously, perhaps it is inherently misconceived.
The absence of VSVPE even in languages that contain both V-raising out of VP and VPE (so-called type H languages) generates a puzzle for Universal Grammar: what principles are responsible for barring this derivation? The puzzle is especially acute, for other types of extraction out of elided VPs are possible, and V-movement out of different ellipsis sites is also possible. In Landau 2020, I address this puzzle and develop a solution in the context of a derivational view of ellipsis, coupled with a PF-visibility condition that applies to the head of any constituent undergoing PF deletion. A general methodological lesson from this investigation is that claims for head-stranding ellipsis should be scrutinized more critically and always be evaluated against the simpler alternative of ellipsis without head stranding.
Finally, I have skirted throughout the question of what the AE parameter is. This topic is hotly debated in the literature (see Sato 2019 for a recent discussion). Proposals include the “antiagreement parameter” (ellipsis of an argument is allowed if and only if no functional head agrees with that argument; Saito 2007, Takahashi 2008, 2013, 2014, Şener and Takahashi 2010), the presence/absence of scrambling (Oku 1998), the absence of a DP layer in the nominal projection (Cheng 2013), the agglutinative nature of the nominal φ-morphology (Simpson, Choudhury, and Menon 2013, Otaki 2014), or the property of radical pro-drop (Sakamoto 2017). Clearly, the finding that languages like Hebrew, Russian, and Portuguese fall together with AE languages throws a new wrench into these discussions. Currently, no existing proposal can do justice to the typological diversity of AE languages.
Because AE languages—at least the documented cases—do not seem to distinguish between DP, PP, and CP arguments (all are elidable), the licensing mechanism may be quite different from what it is for VP- and NP-ellipsis. Null complement anaphora (NCA) may turn out to fall under AE in certain languages, producing another typological split between “surface” and “deep” NCA (Cheng 2013); and within the former group, there may be another split between NCA permitting all extraction and NCA permitting only covert extraction (on Japanese as a language of the latter kind, see Sakamoto 2017). All these matters are well worth investigating if we are to gain a deeper understanding of the workings of ellipsis in natural language.
1 Among them: Hebrew (Doron 1990, 1999, Goldberg 2005), Basque (Laka 1994), Portuguese (Martins 1994, Cyrino and Matos 2005, Santos 2009, Rouveret 2012), Ndendeule and Swahili (Ngonyani 1996, 1998), Serbo-Croatian (Stjepanović 1997), Finnish (Holmberg 2001, 2016), Russian, Polish, and Czech (McShane 2000, Gribanova 2013a,b, Ruda 2014), Egyptian Arabic (Tucker 2011), Welsh (Rouveret 2012), Malayalam, Bangla, and Hindi (Simpson, Choudhury, and Menon 2013, Manetta 2018, to appear), Hungarian (Lipták 2013, 2019), Greek (Merchant 2018), and Persian (Sato and Karimi 2016, Rasekhi 2018).
2 Assuming that all is indeed equal. This is far from obvious, given that VPE and AE may well serve distinct discourse functions and so would never be members of the same reference set for economy comparisons. In this case, MaxElide or economy in general would be irrelevant.
4 One challenging property discussed by Ngonyani (1996) is that applied objects (benefactive, locative, and instrumental) cannot go missing when the theme is overt, but the theme can go missing when the applied object is overt. This follows from a VSVPE analysis on the assumption that the Applicative projection dominates the basic VP. On the AE analysis, one would need to develop a licensing account of AE (needed anyway) that would distinguish basic from applied objects.
6 In Landau 2018, I present a second argument to show that VSVPE overgenerates OG sentences with raising verbs. I do not reproduce it here.
7 The data in (10)–(12) and (15) are due to Rajesh Bhatt (pers. comm.). Simpson, Choudhury, and Menon (2013) note examples similar to (10) with other types of adjuncts. When the objects in the two conjuncts are identical, an antecedent adjunct cannot be construed by itself in the second conjunct. The authors take this as evidence of the need for VSVPE independently of AE in Asian languages, but in fact a simpler focus-sensitive analysis captures all the facts under the AE approach (see Oku 2016).
9 Unless otherwise indicated, the data obtain both in European and in Brazilian Portuguese.
10Martins (1994) argues that V raises to Σ, which projects above AgrSP and TP. Ellipsis, however, targets VP and not Σ’s sister. I return below to this possibility.
11 The judgments reported in this section were collected from five to seven native speakers. Examples (19a–b) were found acceptable by all those consulted, except one, who rejected object gaps everywhere.
The Ā-movement analysis of null objects in Portuguese is originally due to Raposo (1986). It remains to be seen how to reconcile his data with these data. Note that counterexamples to Raposo’s claim (involving island-insensitive nominal OGs) are also documented in Cyrino and Lopes 2016.
12Rouveret’s (2012) broader claim is that ellipsis always targets the complement of a phase head (e.g., the VP complement of v). To accommodate the optionality of adjunct inclusion, he proposes that adjuncts may attach to VP (captured by ellipsis) or to vP (spared by ellipsis). This account is challenged by examples in which the same VP antecedent licenses one adjunct-including VPE and another adjunct-excluding VPE (as in Sag 1976). Portuguese allows this too.
O Pedro foi a Paris antes do Natal, a Ana também foi ___, mas a Maria só foi
the Pedro went to Paris before of.the Christmas the Ana also went but the Maria only went
___ depois do Natal.
after of.the Christmas
‘Pedro went to Paris before Christmas, Ana did too, but Maria did only after Christmas.’
The two ellipsis sites place conflicting constituency requirements on the antecedent vP.
13 At the same time, speakers readily accept that reading with a bare negative answer (“Não!”).
14 When more than just the verb is stranded (e.g., the subject), it is necessary to assume an evacuating movement—usually motivated by contrastive topicality—prior to TP-ellipsis.
15Gribanova (2013a) argues that VSVPE can be further distinguished from AE by its greater discourse selectivity: the remnant of VSVPE must be a contrastive topic but the remnant of AE admits other functions (contrastive, backgrounded, or presentational focus). However, her purported AE examples (of the form V-[DP e]-PP) do not rule out the option of a pro object, hence cannot reliably distinguish VSVPE from AE.
17 While Gribanova (2013b) downplays deviations from verb identity (suggesting they occur only under special circumstances, for certain speakers), in Gribanova 2017, 2018 she fully acknowledges that a contrastively focused verb in a verbal answer to a polar question need not match its antecedent (apparently, for all speakers). These recent works also retract the earlier claim that lexical prefixes must match in the antecedent and the target verbs. However, Gribanova continues to link the focus condition on verbal mismatch to extraction out of an ellipsis site (hence, indirectly supporting the VSVPE analysis); for vP-ellipsis sentences, this has the result of requiring verb extraposition in both the antecedent and the target clause, yielding SOV order in the former. This expectation is contradicted by Bailyn’s (2017) data, which evidence verbal mismatches in standard SVO configurations. More generally, the assumption that only extracted material is subject to the focus condition is incorrect. Nonelided material in the target clause must also stand in some contrastive relation to its counterpart in the antecedent clause, including polarity particles and adverbs, which are clearly generated outside the elided VP (i)–(ii); consequently, the interaction between verbal mismatch and focus does not favor the VSVPE over the AE analysis.
Mary cleaned her room and John did *(not).
Mary cleaned her room quickly and John did slowly/*quickly.
18Gribanova (2013a) presents a novel argument in favor of VSVPE in Russian, which draws on the interpretation of OGs in the context of coordinated ditransitive VPs. I tackle this argument separately in Landau 2019. Gribanova (2017) extends the V-stranding analysis to TP-ellipsis under polarity focus, however, still maintaining the VSVPE analysis for nonpolar contexts.
For helping me construct my test examples, I thank Anabela Proença Gonçalves and Pilar Barbosa (on Portuguese), Auva Shahnovich (on Russian), and Rajesh Bhatt (on Hindi). Part of the research reported here was presented in the workshop “The Timing of Ellipsis” at SLE 51, Tallin University, in August 2018. I thank the audience for their helpful comments.