The Phenomenon Of Sluicing English Language Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Sluicing is classified as a phenomenon of clausal ellipsis. The term Sluicing was first discovered and named by Ross (1969). It generally expresses a construction with an interrogative clause that has only one wh-phrase for pronunciation, as in (1).

Tina will sell something but I don't remember what [Tina will sell t].

The square brackets contain the material which stands for the construction's unpronounced part, though it should be interpreted. Both elements in the square brackets, the subject Tina and the auxiliary will, are deleted in (1). As we see in (1), modals, occupying T0, and subjects, located in SpecTP, are deleted in Sluicing constructions suggest that we are dealing with TP-ellipsis.

Sluicing can occur in two types of clauses. First, it occurs in embedded clauses, as in (1), second, it also occurs in main clauses, as in (2).

Speaker A: Paul adores somebody.

Speaker B: Who [Paul adores t]?

I will follow the fundamental analysis of Sluicing introduced by Ross (1969) and further elaborated by Lasnik (2001) and Merchant (2001), who regard the elliptical structure as a case of regular wh-movement out of TP, which is followed by the deletion of TP at PF. So, (3) reconstructs the derivations that take place in Sluicing constructions. That means in detail that firstly a wh-phrase is the subject to regular wh-movement to SpecCP, and secondly the deletion of TP takes place at PF.

(3) Step 1: Tina sold something. I wonder [CP what [TP Tina sold t]].

Step 2: Tina sold something. I wonder [CP what [TP Tina sold t]].

However, there are also other analyses of ellipsis, developed by Williams (1977), Lobeck (1995), Chung et al. (1995). According to such analyses, an empty category is found in the place of the deleted TP, then the empty category is moved due to the copy operation of the antecedent TP at LF. Such analyses do not support deletion, since they do not possess any clausal construction in the sluice.

Furthermore, there are strictly semantic analyses as well which are developed by Dalrymple et al. (1991), Jacobson (1992), and Hardt (1993, 1999). Nevertheless, Ross (1969), Merchant (2001) and Stjepanović (2003) represent comprehensive arguments against the non-deletion approaches in their researches. Consequently, I will also follow their way of analysing Sluicing as the PF-deletion.

The phenomenon of Sluicing is quite widespread in many languages. It is also very productive in Russian. In the present paper I will concentrate on Russian for the most part, and relate it to other Slavic languages whenever necessary.

As the English constructions show above, Russian can also allow embedded, as well as main clause Sluicing. The (4a) example demonstrates embedded Sluicing, while the (4b) demonstrates matrix Sluicing respectively.

(4a) Pavel dast komu-to rabotu, no ja ne pomnju komu/*kto [Pavel dast t].

Pavel will give someone DAT job but I not remember who DAT/NOM

'Pavel will be giving someone a job but I don't remember who.'

(4b) Speaker A: Pavel dast komu-to rabotu

Pavel will give someone DAT job

'Pavel will be giving someone a job.'

Speaker B: Komu/*Kto?

who DAT /who NOM


The examples in (4a) and (4b) definitely show that dative case is assigned to every remnant wh-phrases, and so the corresponding antecedent clauses should also carry dative case. Otherwise, the construction is ill-formed. The indirect object of the Russian verb davat'-д°Ð²°Ñ‚ÑŒ, patterning with the English verb give, gets obligatory dative case. If the case of the remnant wh-phrase is changed from dative to nominative, it leads to unacceptability. Therefore, the case evidence maintains that there is in fact the movement of wh-phrases out of TP, where they get dative case assignment. Hence, the examples present true Sluicing cases. However, there is a variant of Sluicing which is called Pseudo-Sluicing. In this case, Pseudo-Sluicing would have a different structure in the sluice, namely a cleft structure, as shown in (5).

(5) Frank invited someone but I don't remember who [it was _ (that Frank invited)].

In Russian, clefted elements obligatory bear nominative case, as in (6) from Russian.

(6) Ivan dal komu-to knigu, no ja ne pomnju kto/*komu eto byl.

Ivan gave someone book but I not remember who NOM/DAT it was

'Ivan gave someone a book but I don't remember who it was.'

The remnant wh-phrases in (4a) and (4b) are completely different to the wh-phrase in (6). Thus, the examples in (4a) and (4b) are true cases of Sluicing.

Furthermore, Besides Sluicing with a singular wh-remnant which can be found in English, Russian also allows Sluicing with multiple wh-remnants, as in (7). Takanashi (1994) describes this phenomenon, as an operation of TP-deletion with multiple wh-remnants. Multiple Sluicing shares some properties with Sluicing. For example, like Sluicing, multiple Sluicing can occur in embedded clause, as the (7a) example shows, and also in main clause, as the (7b) example does below.

(7a) Každyj iskal kogo-to v zale, no ja znaju kto kogo.

everyone searched someone in hall but I not know who whom

'Everyone searched someone in the hall but I don't know who whom.'

(7b) Speaker A: Každyj iskal kogo-to v zale.

everyone searched someone in hall

'Everyone searched someone in the hall.'

Speaker B: Kto kogo?

who whom

'Who whom?'

Since Slavic languages, and Russian in particular, are multiple wh-fronting languages, it is reasonably to suggest that such languages also allow multiple Sluicing. Strictly speaking, all bare wh-phrases move to the front in multiple questions in Russian, as in (8a) and (8b).

(8a) Kto1 komu2 [t1 zvonit t2]?

who whom calls

'Who calls who?'

(8b) *Kto1 [t1 zvonit komu]?

who calls whom

2. Licensing TP-Deletion

A central question in ellipsis analysis is the definition of the categories that can license the deletion of their complements. Since the Ross's study on Sluicing (1969), researchers regard that the interrogative +wh complementizer licenses the deletion of the complement TP. The evidence for it comes from Sluicing in Germanic that includes only interrogative clauses which wh-phrase is in SpecCP. Moreover, Lobeck (1995) explores other different constructions in English. She observes, however, that TP-deletion is impossible in the following constructions as finite declarative clauses, or in lexically governed TP-s, as well as in relative clauses. Therefore, Merchant (2001) supposes, that in Sluicing the TP is the complement of an interrogative wh-complementizer. In other words, what licenses the TP deletion of the complement is the complementizer that possesses the +Q feature and the +wh feature. The following structure of the sluice - C0 bearing [+Q] and [wh] features - is shown in (9), where the wh-phrase is in SpecCP and the interrogative C0 licenses the deletion of its complement at PF.

(9) Tina sold something. I wonder [CP what C0 [TP Tina sold t]].

Since the Slavic languages do not represent a similar pattern of wh-movement than the one in Germanic, it is quite questionable how the analysis presented above can be taken to the Slavic languages. According to Bošković (2002), wh-fronting in Slavic languages, like Russian and Serbo-Croatian, includes another phenomenon called focalization. However, Bulgarian, for example, has the +focus feature which is found in the position of the interrogative C0, in addition the strong +wh feature is located there as well. Thus, Bulgarian resembles English in having the same target position of wh-movement, which is then SpecCP. Therefore, the licenser of Sluicing is clearly C0.

However, Sluicing in Russian and Polish and certain contexts in Serbo-Croatian needs some more profound explanation. The first issue to explain is why the remnants of Sluicing manage to remain after the deletion.

2.1. Multiple wh-fronting and Contrastive focus

The analysis of wh-fronting with the help of focus-movement is basically the same for Russian and Polish. Stepanov (1998) explores the phenomenon of wh-movement in Russian and asserts in his approach that it is not the +wh feature of C0 that triggers the wh-movement in Russian. Hence, the wh-phrases do not reach SpecCP in overt syntax. The argument of Stepanov (1998) goes back to the lack of superiority effects in Russian. Stepanov (1998) looks for the parallels with the Minimal Link Condition of Chomsky (1995). He suggests that superiority determines a strong feature that drives movement and reveals the Economy approach for superiority. Under the Economy approach to superiority, C0 possesses a strong +wh feature that attracts the closest item with a +wh feature to SpecCP in order to check features, as formulated in Chomsky's (1995) Minimal Link Condition. Thus, this approach gives explanation for the existence of superiority effects in English, illustrated with the help of the paradigm in (10).

(10) a. Who sold what?

b. ??What did who sell t?

c. Who did Frank make t to do what?

d. *What did Frank make who to do t?

According to the sentences in both (10b) and (10d), which signalize that C0 does not in fact attracts the closest item, namely what, to C0. The closer item is who, thus the wh-movement in (10b) and (10d) is not economical.

Stepanov (1996) asserts that Russian wh-questions do not exhibit superiority effects in typically any contexts, no matter if it is in main clause, as in the (11a, b) example, or in embedded questions, as in the (11c, d) example below.

(11) a. Kto1 kogo2 [t1 nenavidit t2]?

who whom hates

b. Kogo2 kto1 [t1 nenavidit t2]?

c. Ja ne znaju [kto kogo nenavidit].

I not know who whom hates

'I don't know who hates who.'

d. Ja ne znaju [kogo kto nenavidit].

In English, only one wh-phrase can be fronted, as demonstrated in the paradigm in (10). Unlike English, Russian multiple questions do not have superiority effects in practically any context, as shown in (11).

The facts from those examples cannot, however, correspond to the Economy approach to superiority. Since the economy issues of MLC make sense only when there is actually a Comp with a strong +wh feature present in the structure. Therefore, Stepanov (1996) supposes, that Russian does not, actually, have a strong +wh feature. It rather has a weak +wh feature, which does not drive overt wh-movement and hence, does not generate superiority effects.

In Russian, wh-phrases obligatory front and Stepanov (1996) is the first to observe that such fronting attributes to contrastive focalization. The analysis is based on the idea of interaction between wh-fronting and fronting of contrastively focused R-expressions in Slavic. Contrastively focused R-expressions are fronted as well as wh-phrases in Slavic, as demonstrated by the Russian paradigm in (12).

(12) a. IVANA ja vstretila t.

IvanACC i met1.FEM.SG

‚I met Ivan'

b. ??Ja vstretila IVANA.


Thus, Stepanov (1998) sums up, that wh-phrases in Russian are fronted to a focus position below CP. As mentioned before, superiority effects are absent in Russian. Bošković (1998) assumes that each wh-phrase itself carries a strong +focus feature and therefore there is no competition between the wh-phrases with respect to the closeness to C0. Following Bošković's (1998) way of analyzing, Stepanov(1998) explains that such focalization is insensitive to superiority.

2.2. Focus-Licensed Sluicing

As far as Sluicing is concerned, it is supposed that the C0 is the structural licenser of TP-deletion. Anyway there is still one issue to explain, namely it is not clear how the remnant wh-phrases in Russian Sluicing survive the deletion if they are not in SpecCP. Grebenyova (2006) proposes, that not only an interrogative C0 can license TP-deletion, but any functional category bearing a + focus feature can license the deletion of its complement as well, producing the structure as in (13) below.

(13) Ivan kupil čto-to, no ja ne znaju [ čto X0 [TP Ivan kupil t]]?


Ivan bought something but I not know what Ivan bought

'Ivan bought something but I don't know what.'

So, the wh-phrases in Russian survive TP-deletion due to this functional category.

It follows from this proposal, that Sluicing should be also possible with contrastively focused remnants that are not wh-elements. And indeed the evidences from Russian confirm the proposal that the contrastively focused R-expressions can in fact be remnants of Sluicing.

(14) Speaker A: Ty skazal, čto on budet uvažat' Mašu.

you said that he will respect Maša ACC

'Did you say that he will respect Maša?'

Speaker B: Net. Ja skazal, čto IVANA [on budet uvažat' t].

no I said that Ivan ACC he will respect

'No. I said that (he will respect) IVAN.'

In (14), the remnant is Mašu and, in (15), there are three remnants: a wh-phrase and two R-expressions. This considerably confirms the proposal that wh-fronting and contrastive-focus-fronting function alike in Slavic, and in Russian in particular.

(15) Speaker A: Ty ne pomniš', kogda Ivan vstretil Mašu?

you not remember when Ivan NOM met Maša ACC

'You don't remember when Ivan met Maša?'

Speaker B: Net. Ja ne pomnju GDE SERGEY LENU.

no I not remember where Sergey NOM Lena ACC

'No. I don't remember WHERE SERGEY (met) LENA.'

It is also necessary to mention, that there are several possibilities to analyze the properties of the constructions in (14) and (15).

The first possibility is to consider a Pseudogapping analysis for (14) and (15). However, since Pseudogapping is not generally possible in Russian, the examples of (14) and (15) cannot be the cases of Pseudogapping. The unavailability of Pseudogapping in Russian is shown in (16). Moreover, Sag (1976), Lashnik (1995) analyze Pseudogapping as VP-ellipsis. What is elided in (14) is the auxiliary budet 'will', indicating that a larger constituent then VP is elided.

(16) * Maša budet čitat' knigu, a Ivan budet gazetu [čitat' t].

Maša NOM will read book ACC and Ivan NOM will newspaper ACC

'Maša will read a book and Ivan will a newspaper.'

The other possibility to regard is that the examples of (14) and (15) above are the instances of Gapping. It is quite unlikely to analyze these constructions as Gapping. Since in English as well as in Russian, Gapping is restricted to local coordinations and (a) and or (ili), while the conjunction but cannot occur in Gapping structures, as illustrated in (17a-c).

(17) a. Maša budet čitat' knigu, a Ivan budet čitat' gazetu.

Maša NOM will read book ACC and Ivan NOM will read newspaper ACC

'Maša will be reading a book and Ivan a newspaper.'

b. Ili Maša budet čitat' knigu, ili Ivan budet čitat' gazetu.

either Maša NOM will read book ACC or Ivan NOM will read newspaper ACC

'Either Maša will be reading a book or Ivan a newspaper.'

c. * Maša budet čitat' knigu, no Ivan budet čitat' gazetu.

Maša NOM will read book ACC but Ivan NOM will read newspaper ACC

'Maša will be reading a book but Ivan a newspaper.'

Therefore, Gapping analysis cannot also account for the constructions (14) and (15), since these easily contain but­ and are completely grammatical, as shown in (18) below.

(18) Ty skazal, čto on budet uvažat' Mašu, no ja dumaju, čto IVANA.

you said that he will respect Maša ACC but I think that IvanACC

'You said that he will respect Maša, but I think that he will respect Ivan. '

In both English and Russian, Gapping cannot occur in embedded clauses, as demonstrated by the difference between (19a) and (19b).

(19) a. Maša budet čitat' knigu, a Ivan budet čitat' gazetu.

Maša NOM will read book ACC and Ivan NOM will read newspaper ACC

'Maša will be reading a book and Ivan a newspaper.'

b. *Maša budet čitat' knigu, a Lena dumala, čto Ivan gazetu.

Maša will read book and Lena NOM thought that Ivan newspaper

'Maša will be reading a book and Lena thought that Ivan a newspaper.'

Furthermore, the antecedent of Gapping cannot occur in an embedded clause, as the difference between (20a) and (20b).

(20) a. Ili Maša budet čitat' knigu, ili Ivan budet čitat' gazetu.

either Maša NOM will read book ACC or Ivan NOM will read newspaper ACC

'Either Maša will be reading a book or Ivan a newspaper.'

b. Ili Lena dumala, čto Maša budet čitat' knigu, ili Ivan budet čitat' gazetu.

either Lena thought that Maša will read book or Ivan will read newspaper

'Either Lena thought that Maša will be reading a book, or Ivan a newspaper.'

Consequently, none of the main requirements of Gapping are met in (14) and (15).

Therefore, Sluicing analysis seems to be the only plausible explanation for these constructions.

2.3. Overt material in COMP

Lobeck (1995), Chung et al. (1995) and Merchant (2001) among others notice that nothing besides the overt material in SpecCP can survive Sluicing. In other words, their proposal is that no overt material in C0 itself survives Sluicing. Merchant (2001) analyzes the data from several languages, such as English, Danish, Dutch, Frisian, German, Norwegian, Slovene. For example, English situations in (21a) - (21d) show that the auxiliary cannot remain unelided under Sluicing, even though T-to-C movement is obligatory in the main clauses.

(21) a. What will John buy?

b. *What John will buy?

c. John will buy something but I don't know what.

d. *John will buy something but I don't know what will.

The proposal is valid for the elements that move to C0 and for the elements that are base generated in C0 as well. Lasnik (1999a) and Merchant (2001) research the moved elements and base their researches on Economy and feature-movement. Its concept states that if the element in T0 does not move to C0 overtly, then this material in T will cause a PF crash. The reasons for PF crash are either because the strong feature of T0 will remain unchecked, or because this material will be unpronounceable at PF on the feature-movement account. At the same time, if ellipsis deletes the structure with the inadequacy at PF, there are no more complications with the derivation.

Obviously, in (21) will does not move to C0, and creates a problem which is later removed by Sluicing.

As for the base-generated elements, they are analyzed as clitics which must cliticize to the right, and therefore cannot remain unsupported in C0 under Sluicing.

Furthermore, Russian base-generated particle li, which is an interrogative yes/no question complementizer, can be a remnant of Sluicing, in case there is a focused element in SpecCP, as in (22).

(22) Ivan vstretil kogo-to, no ja ne znaju LENU li.

Ivan met someone ACC but I not know Lena ACC li C

'Ivan met someone but I don't know where he met LENA.'

In Russian, C0 can carry +focus feature. Thus, there seem to be two focus positions in Russian above TP: the first in CP and the other below CP. In (23), the focused element follows the declarative complementizer čto, representing the position below CP.

(23) Maria ne znala, čto IVANA ona dolžna vstrečat'.

Maria not knew that Ivan ACC she must meet

'Maria didn't know that it was Ivan who she was supposed to meet.'

Grebenyova (2006) suggests that Russian li is a clitic that cliticizes to the left, and therefore can remain in C0 under Sluicing. Since Russian li is base-generated in C0, the Economy considerations do not apply to li. Hence, Sluicing provides a diagnostic for the properties of certain clitics.

The argument, that Russian li cliticizes to the left, comes from a structure where li joints to a number of focused constituents and it always goes after these constituents, as illustrated in (24).

(24) a. Ivan li, Maša li napišet, mne vsjo ravno.

Ivan li Maša li will-write, to-me all equal

'Whether Ivan or Maša will write, doesn't matter to me.'

b. *Ivan li, li Maša napišet, mne vsjo ravno.

c. *Li Ivan, li Maša napišet, mne vsjo ravno.

The construction of (25) proposes that if li is supported by another morpheme from the left, then it can precede the focused constituents. Here, to- does not really give any extra meaning to the sentence. Therefore, it is quite proper to suppose, that it functions in a dummy 'do-support'-like manner.

(25) To-li Ivan, to-li Maša napišet, mne vsjo ravno.

'Whether Ivan or Maša will write, doesn't matter to me.'

Thus, it is possible to conclude that contrastive focus can in fact license Sluicing in Russian and some other Slavic languages. Hence, the facts above determine that contrastive focus is capable of licensing Sluicing in Russian. Moreover, Merchant (2004) uses the conception that focus can license deletion of its complement also in his analysis of fragment answers in English. Craenebroeck and Lipták (2005) also come to this conclusion in their analysis of Ellipsis in relative clauses in Hungarian. Thus, focus has an ellipsis-licensing capability in a number of languages.

2.4. Unifying the Theory of Licensing TP-Deletion

On the one hand, we have Merchant's (2001) conclusions concerning English, that the feature that licenses Sluicing in English is the +wh feature. On the other hand, we have the conclusions of the previous section, that the feature that licenses Sluicing in Slavic is the +focus feature. Consequently, we have two options, either +wh and +focus features are both capable of licensing TP-deletion, or the +focus feature is the licenser of TP-deletion in general. The last option seems to be the stronger one and, for this reason, is more complicated to assert.

Particularly a language like English has difficulties with the +focus feature as the licenser of TP-deletion, because contrastively focused phrases in English always remain in situ. However, Grebenyova (2006) proposes that Sluicing is licensed by the +focus feature with an overtly realized specifier of the head carrying this feature. This proposal is illustrated in (26).








In English, the +focus feature is weak, while in Russian it is strong. However, the feature strength should not be so important for licensing Sluicing, if the mechanism of Sluicing licensing is attempted to be unified for both languages. So, the CP layer looks the following way in English, as in (27).









This signifies that wh-movement in English represent by itself the operation that produces the needed configuration for licensing TP-deletion. While the +wh feature itself is unimportant for licensing TP-deletion. This hypothesis is actually favourable, since the conditions that do not allow Sluicing in English are likely to contain elements that cannot be focused. Such elements are the relative pronouns in relative clauses and complementizers like that and if.

3. Multiple Sluicing and the Semantics of Multiple Interrogatives

Russian seems to allow multiple Sluicing, but, however, the appropriate contexts for multiple Sluicing depend on the interpretation of multiple interrogatives in this language. Russian does not have single-pair readings in multiple interrogatives, as the contrast between (28) and (29) shows.

(28) Každyj priglasil kogo-to na tanec, no ja ne pomnju kto kogo.

everyone invited someone to dance but I not remember who whom

'Everyone invited someone to a dance but I don't remember who whom.'

(29) ??Kto-to priglasil kogo-to na tanec, no ja ne pomnju kto kogo.

someone invited someone to dance but I not remember who whom

'Someone invited someone to a dance but I don't remember who whom.'

Usually, multiple interrogatives possess a Pair-List or a Single-Pair reading. Although the Single-Pair reading is more restricted crosslinguistically, as Bošković (2003) and Grebenyova (2004) among others describe.

The scenarios in (30) and (31) represent the readings with respect to the English question in (32). The question is inadmissible on the Single-Pair scenario in (31), because English as Russian lacks Single-Pair readings.

(30) Scenario 1 (PL): John is at a formal dinner where there are diplomats and journalists. Each journalist was invited by a different diplomat. John wants to find out all the details, so he asks the host: (32)

(31) Scenario 2 (SP): john knows that a very important diplomat invited a very important journalist to a private dinner. John wants to find out all the details, so he asks the caterer: (32)

(32) Who invited who to the dinner?


Patterning with English, Russian also lacks the SP reading in multiple interrogatives, as illustrated in (33).

(33) Kto kogo priglasil na tanec? PL/*SP

Who whom invited to dance

'Who invited who to a dance?'

Since Russian lacks the SP reading, it seems logical to suppose that the antecedent clause distributes a single-pair reading on the interrogative clause in the sluice and therefore, the Russian multiple Sluicing in (29) is degraded. Obviously, a multiple wh-question in Russian cannot have exactly this type of reading.

There is one more type of reading, which is very close to the SP reading. This type is called Order reading, as in (34).

(34) John and Bill were fighting. Who hit who first?

In Russian, multiple Sluicing is possible with Order reading, as long as the context provided by the antecedent is relevant.

(35) Maša i Ivan pošli na večer. Kto-to iz nih priglasil drugogo na tanec, no ja ne znaju kto kogo.

Maša and Ivan went to party. One of them invited the other to dance but I not know who whom.

'Maša and Ivan went to a party. One of them invited the other to a dance but I don't know who invited who.'

Consequently, the rather direct conclusion can be reached at this stage, namely, if the interpretations are available to wh-interrogatives in a given language, then they are the only interpretations of wh-interrogatives available under Sluicing in that language. Therefore, the sluices could be analyzed as full interrogative clauses.

Moreover, it is also possible to predict out of this conclusion that multiple Sluicing should not be available with adjunct wh-questions since the Order reading is impossible with adjuncts. The prediction is confirmed, as demonstrated in (36).

(36) *Kto-to sprjatal gde-to zdes' klad, no ja ne znaju kto gde.

someone hid somewhere here treasure bu I not know who where

'Someone hid the treasure somewhere here bu I don't know who hid it where.'

4. Superiority under Sluicing

In this section, another property of Sluicing in Russian will be explored, namely the Superiority effects. First Sandra Stjepanović (2003) observes for Serbo-Croatian that Sluicing enforces superiority effects in contexts where parallel non-elliptical structures do not exhibit any superiority effects.

Superiority effects in Serbo-Croatian are generally present in embedded but not in main clauses.

(37) a. Kto šta1 o njemu govori t1?

Who what about him says

'Who says what about him?'

b. Šta1 ko o njemu govori t­1?

(38) a. Pavle je pitao ko šta1 o njemu govori t1.

Pavle aux asked who what about him says

'Pavle asked who says what about him.'

b. ??Pavle je pitao šta1 ko o njemu govori t1.

Nevertheless, Stjepenović (2003) goes into detail and asserts that superiority effects emerge in Serbo-Croatian in main clauses under Sluicing, as in (39).

(39) Speaker A: Neko voli nekog.

somebody loves somebody

'Somebody loves somebody.'

Speaker B1: Ko koga?

who who

Speaker B2: *Koga ko?

So, Serbo-Croatian multiple Sluicing exhibits superiority effects in main clauses with null C0.

However, the same effects emerge also in embedded clause in Serbo-Croatian, but they are of no importance, because the generalization relates to the issues in parallel non-elliptical structures.

Coming back to Russian, according to Stepanov (1998), it is a language that does not exhibit superiority effects either in main or embedded clauses in non-elliptical structures. Russian patterns with Serbo-Croatian and so superiority effects also emerge in Russian under Sluicing in main and in embedded clauses as well, as shown in (40) and (41).

(40) a. Speaker A: Každyj priglasil kogo-to na tanec.

everyone invited someone to dance

'Everyone invited someone to a dance.'

b. Speaker B: Kto kogo?

Who whom

c. Speaker B: *Kogo kto?

(41) a. Každyj priglasil kogo-to na tanec, no ja ne pomnju kto kogo.

everyone invited someone to dance but I not remember who who

'Everyone invited someone to a dance but I don't remember who who.'

b. * Každyj priglasil kogo-to na tanec, no ja ne pomnju kogo kto.

Although it is known that Sluicing sometimes repairs the derivation. And indeed, according to Merchant (2001), who expects no island effects under Sluicing. However, in the cases above, Sluicing seems to destroy the derivation.

If superiority effects are, for example, taken as essentially minimality effects and minimality is encoded into the definition of Attract (Chomsky 1995), hence, such violations cannot technically exist in any derivation and therefore cannot be repaired by deletion. Obviously, this means that superiority effects in non-elliptical structures in a language like Bulgarian are not expected to disappear under Sluicing. Since Bulgarian is a language with strong superiority effects.

Merchant (2001) explores the data demonstrating that this is indeed the case in Bulgarian. Then, he indicates that this produces an additional evidence for the deletion approach to Ellipsis, since superiority is a diagnostic of movement and movement could have taken place out of the ellipsis site only if a full clause is present in the structure from the beginning and is deleted at PF. Nevertheless, the data shows that Sluicing invokes superiority effects in languages that actually lack superiority effects without Ellipsis, as in Serbo-Croatian and Russian.

Stjepanović (2003) assumes that the feature that licenses TP-deletion must be on C0, and therefore she sums up that C0 must be merged in overt syntax in Sluicing constructions in Serbo-Croatian. Thus, superiority effects in Serbo-Croatian matrix clauses are driven by the strong +wh feature of C0.

Russian, in comparison to Serbo-Croatian, possesses the weak +wh feature and this fact makes it difficult to extend the analysis for Serbo-Croatian to Russian. Since the +wh feature is weak in Russian, merging C0 in overt syntax cannot lead to superiority effects. However, Grebenyova (2006) researches another account and proposes that the superiority effects detected under Sluicing follow from an independent property of elliptical constructions, namely, quantifier parallelism.

Grebenyova (2006) accepts the conception of parallelism, which was first explored by Fiengo and May (1994), and is further developed by Fox and Lasnik (2003). Parallelism requires that variables in the elided and antecedent clauses be bound from parallel positions. Grebenyova (2006) also supposes that the variable, which is introduced by an indefinite in the antecedent clause, is bound by existential closure, as Kratzer (1997) claims.

At first, Grebenyova (2006) takes a case of Russian multiple Sluicing, as in (42) and analyzes the LF of the antecedent in it.

(42) a. Speaker A: Každyj priglasil kogo-to na tanec.

everyone invited someone to dance

'Everyone invited someone to a dance.'

b. Speaker B: Kto kogo [priglasil na tanec]?

who whom invited to dance

c. Speaker B: *Kogo kto [priglasil na tanec]?

The LF representation of the antecedent in Russian multiple Sluicing in (45a) is given in (43).

(43) x y [x priglasil y na tanec]

invited to dance

The reading represented in (43) is the only available meaning available in (42a), since surface quantifier scope is preserved in Russian. The example in (44) shows this even more clearly and the unacceptable variant in (45), which is an English based case in Fox (2000:70), does it as well.

(44) Kakoj-to paren' poceloval každuju devušku. x y / *y x

some guy NOM kissed every girl ACC

'Some guy kissed every girl.'

(45) #Odin/kakoj-to časovoj stoit naprotiv každogo zdanija.

one/some guard is-standing in-front-of every building

'One/some guard is standing in front of every building.'

Returning to the multiple Sluicing in (42), their following LF representations of the acceptable sluice in (42b) and the unacceptable one in (42c) are given in (46b) and (46c) respectively. The repeated version of (43) is given in (46a). Since superiority effects follow from quantifier parallelism, all three LF representations in (46a-c) are supposed to meet the parallelism requirement. In other words, the variables in these sluices and in the LF of the antecedent should be bound from parallel positions.

(46) a. x y [x priglasil y na tanec] ← LF (antecedent)

invited to dance

b. kto x kogo y [x priglasil y na tanec] ← LF (wh1 > wh2)

who whom invited to dance

c. kogo y kto x [x priglasil y na tanec] ← LF (wh2 > wh1)

whom who invited to dance

Obviously, there is the parallelism in variable binding between (46a) and (46b), but there is no parallelism between the (46a) and (46c). This means that the quantifier, binding the object variable, is inside the scope of the quantifier, binding the subject variable in the antecedent clause. While in the sluice in (46c), it is outside the scope of the parallel quantifier.

For the further tests, the object quantifier is scrambled over the subject in the antecedent clause, as in (47a). Consequently, we get an acceptable sluice with the wh2 > wh1order in (47b), as the parallelism account postulates, since now the object quantifier is outside the scope of the subject quantifier in both the antecedent and the sluice.

(47) a. Speaker A: Každogo1 kto-to priglasil t1 na tanec.

everyone ACC someone NOM invited to dance

'Someone invited everyone to a dance.' (with x y )

b. Speaker B: Kogo kto?

whom who

c. Speaker B: *Kto kogo?

who whom

Since in (47c), the subject >object order of the wh-phrases cannot be accepted, this in fact reinforces the parallelism account stated above.

Furthermore, as the above presented tests show, the actual source of the apparent superiority effects under Sluicing in Russian is indeed parallelism and not minimality.