Gapping in European Portuguese Gapping has been classically characterized as an ellipsis. However, recent literature (Johnson 2009) has regarded it as Across-the-Board movement in the English language. This paper aims to analyze these proposals and giving its own for European Portuguese, emphasizing some of the differences between the two languages as the source for these contrasts. 1. Introduction Gapping has never been subject to an in-depth analysis in European Portuguese (EP), although some of its general properties have been described in previous work (Matos 1992, 2003, 2005, 2013). However, detailed analyses of Gapping exist for the English language, both prior to the Principles and Parameters framework, such as in Ross (1971), Sag (1980), and within the Minimalist Program, Johnson (2009) and Vicente (2010). With that in mind, we can compare how Gapping behaves in English and EP so as to determine the degree to which they are either different or alike. This paper will make an initial analysis of Gapping in EP, contrasting it with the more recent proposals made for English. I will especially focus on the connection between Gapping and Across the Board movement and on the compatibility or incompatibility of Gapping with the CP node. The analysis will be developed in the Principles and Parameters framework. I will begin by presenting the main properties of Gapping as described in the literature, briefly analyzing its structure and using it to contrast with the properties proposed for English. Finally, I will problematize the compatibility of Gapping with the CP node. Proceedings of ConSOLE XXII, 2014, 156-166 http://www.sole.leidenuniv.nl
Gapping in European Portuguese 157 2. The main properties of Gapping in European Portuguese Gapping occurs when the verb, or verb sequence, is omitted, as in (1a) and (1b). Optionally some arguments or adjuncts of the main verb may also be omitted, as long as two of them are left overt, as in the examples in (1c) vs. (1d): (1) a. O Paulo vai ao cinema e o Pedro vai ao teatro. The Paulo goes to-the cinema and the Peter goes to-the theatre. Paul is going to the cinema and Peter to the theatre. b. A Joana tem estudado alemão e a Ana tem estudado francês. The Joana has studied German and the Ana has studied French. Joana has been studying German and Ana French. c. A Maria deu chocolates ao Manuel e a Rita deu chocolates ao Rui. The Maria gave chocolates to-the Manuel and the Rita gave chocolates to-the Rui. Mary gave chocolates to Manuel and Rita to Rui. d.*o João tem ido às aulas e o Miguel tem ido às aulas. The John has gone to-the classes and the Miguel has gone to-the classes. *John has been going to classes and Miguel has been going to classes. In EP, Gapping requires the omission of the entire verb sequence when this includes (semi-) auxiliaries and a main verb, as in the examples (2a-c). However, the clauses may only present the omission of a shared auxiliary verb as in (2d). In opposition the omission of a shared main verb is ungrammatical, as shown in (2b) and (2c). These facts will be important later in the paper. (2) a. A Joana tem estudado alemão e a Ana tem estudado francês. The Joana has studied German and the Ana has studied French. Joana has been studying German and Ana French b.*/#a Joana tem estudado alemão e a Ana tem estudado francês. The Joana has studied German and the Ana has studied French. c.*eu tenho ido comprar revistas à loja e ele tem ido comprar livros à feira. I have gone to-buy magazines to-the shop and he has gone to-buy books to-the market. d. O Pedro tem caçado e o João tem pescado. 1 The Pedro has hunted and the John has fished. Peter has been hunting and John fished. Lastly, in Gapping the clauses need to have the same polarity, i.e., they need both to be either positive or negative. 1 A ConSOLE reviewer suggests (2d) is not the best example and a better one would be as in (i) because Gapping requires there to be new information on the second conjunct or Gapping is not favorable: (i) O João tem ido à escola e o Rui tem ido ao cinema. The João has gone to-the school and Rui has gone to-the cinema. While this is true, it is simply more natural to omit the entire verb sequence in this particular case and as such the original example is the clearer one.
158 (3) a. O Pedro é inteligente e o João é simpático. The Peter is intelligent and the João is nice. Peter is intelligent and John nice. b. O Pedro não é inteligente nem o João é simpático. The Pedro no is intelligent nor the João is nice. Peter is not intelligent nor John nice. c. *O Pedro é inteligente e o João não é simpático. *The Pedro is intelligent and the John not is nice. d. *O Pedro não é inteligente e o João é simpático. *The Pedro not is intelligent and the John is nice. These are some of Gapping s main properties in EP. I believe that these properties are adequately accounted for by an ellipsis approach. A proposal for its structure as well as its explanation can be found below in (4): (4) O Paulo vai ao cinema e o Pedro vai ao teatro. The Paulo goes to-the cinema and the Pedro goes to-the theatre. Paul is going to the cinema and Peter to the theatre.
Gapping in European Portuguese 159 In (4), the double-striked constituents have been moved from their base position while the single-striked constituents have been gapped. I assume Kayne s (1994) coordination configuration for these structures. According to the representation in (4), the clauses are connected at the TP level through a Conj head that projects ConjP and takes these TPs as its Specifier and Complement. Each clause is independently derived with each verb moving to T and leaving a copy at its base position. The ellipsis then omits the verb and other redundant constituents from the second clause. This representation will be what I will base my arguments on for the problem addressed below. 2. Gapping: a structure of ellipsis or ATB movement? Considering English, Larson (1988) and Johnson (2006, 2009) have argued that Gapping works as Across the Board movement and not as an elliptic construction. As Colaço (2006) stresses, ATB movement requires two coordinate terms, each one presenting an empty category that is bound by a phonetically overt constituent in a position that allows it to license the empty categories through c-command. This configuration does not occur in the structures which I assume to capture Gapping in EP, like (4). In his article, Larson suggests the following representation (which only features the relevant part of the derivation for this paper s purposes): (5) John sent a letter to Mary and a book to Sue.
160 Larson was the first to suggest an analysis for Gapping in which the constituents are coordinated at the VP level and ATB movement of the verbs raises them to the highest V in the VP Shell (Larson s Shell). Johnson (2006, 2009) presents a similar although not identical structure. 2 He develops this treatment in detail under the Minimalist Program (Chomsky, 1995) and relates Gapping to Pseudogapping. The example (6), taken from Johnson s article, illustrates his analysis for Pseudogapping: (6) Some will eat beans and others eat rice. Adopting the Minimalist Program, Johnson assumes that the verb s subject is firstly merged in vp and defends vp coordination with the verb s occurrences raising across the board to a functional projection he calls XP. To defend the coordination at this low level, Johnson mentions the existence of Pseudogapping, which is distinguished from regular Gapping by leaving the auxiliary verb(s) overt, such as will in example (6). However, this analysis presents some problems, especially regarding European Portuguese. First of all, the structure in (6) violates the Coordinate Structure Constraint by extracting the subject some of the first clause from the coordinate structure while leaving its correspondent others in the second clause in its basic position. Second, there are several empirical arguments in favor of the fact that Gapping in EP requires the coordination to be, at least, at the TP level. The arguments are as follows: i) As previously said, Pseudogapping does not exist in EP. The (semi-)auxiliary verbs omission is mandatory when they form part of the sentence, as seen in example (2), repeated below as (7): 2 A ConSOLE reviewer points out that the two structures given are fundamentally different because Larson s example shares the same subject between the conjuncts, while in Johnson s example the subjects are disjoint. As such, they should not be treated equally but as separate conditions. However, these were the examples given by the respective authors and thus, it is not appropriate to analyse them differently here.
Gapping in European Portuguese 161 (7) *Eu tenho ido comprar revistas à loja e ele tem ido comprar livros à feira. I have gone to-buy magazines to-the shop and he has gone to-buy books to-the market. Furthermore, it is important to recall that in EP we have the opposite of Pseudogapping, which is the possibility to omit the auxiliary while leaving the main verb overt, cf. (2d). Additionally EP is a language of generalized verb movement, which is to say that both auxiliary and main verbs may raise to T. Thus, if a verb sequence is composed of one (or several) auxiliary verb(s) as well as the main verb, the (first) verb moves to T. Given all of this, there is no plausible justification for a vp or another kind of low-level coordination in EP as presented in English. ii) In EP, it is only possible to value the uninterpretable Φ-features of T when the verb moves to T head. In addition, although in EP the subject may occur in a post-verbal position and remain in situ, in Spec vp, (cf. Costa 2004) it is often claimed that even in these cases there is an expletive null pro in Spec TP to check T s strong EPP feature. If ATB movement applies and the coordinate terms have different subjects, it becomes impossible to check these features in the second term of the coordinate sentence. iii) The compatibility between Gapping and the topicalization of constituents. Matos (2013) presents an example like (8), which shows that Gapping in EP is compatible with topicalization: (8) À Maria eles ofereceram flores e ao Paulo eles ofereceram um livro. To-the Maria they offered flowers and to-the Paulo they-offered a book. To Mary they offered flowers and to Paul a book. We can confirm from the example in (8) that coordination necessarily includes both topicalized constituents because the alternative would mean that not only their movement would be asymmetrical but also that the Coordinate Structure Constraint would be violated. I propose structure (9) for example (8):
162 (9) With these three arguments, I conclude that in EP, in Gapping the coordination occurs at least at the TP level. 4. Gapping and the CP Node Matos (2005) claims that in Gapping, the clauses must be coordinated at the TP level and the whole structure must remain inside the CP phase. However, if we accept the articulated analysis initially presented in Rizzi (1997), Matos proposal must be revised. As seen in the structure in (9), the coordination includes TopP which, despite being included in CP(=ForceP), is part of the left periphery of the sentence above TP (cf. Duarte 1996; Rizzi 1997, 2004). 3 It should be noted that Clitic Left Dislocation doesn t seem to differ from topicalization as far as being compatible with Gapping is concerned, as we can see in (10): (10) Ao Pedro, a Maria dá(-lhe) livros e ao Paulo, a Maria dá-(lhe) chocolates. To-the Pedro, the Maria gives (him) books and to-the Paulo, the Maria gives (him) chocolates. To Peter, Mary gives books and to Paul, chocolates. As such, if we accept Rizzi s proposals, it is possible to update Matos (2005) assumption by claiming that Gapping is not necessarily limited to TP coordination. However, it still stands that it does not go beyond CP(=ForceP) in subordinate clauses, both in islands and in some non-island contexts. The ungrammatical examples of Gapping in (11) and (12) present island domains, Adjunct Island in (11), Sentential Subject Island in (12a) and Complex DP Island in (12b): 3 It should be noted that while Duarte (1996) doesn t adopt the TopP representation per se for topicalization, she adopts it for Clitic Left Dislocation and also assumes that TopP occurs above TP.
Gapping in European Portuguese 163 (11) *A Maria come maçãs quando a Joana come pêras. The Maria eats apples when the Joana eats pears. (12) a. *A Maria come maçãs e que a Ana come pêras é óbvio. *Maria eats apples and that Ana eats pears is obvious. b. *A Ana gosta de livros e nós temos um amigo que gosta de chocolates. *Ana likes books and we have a friend who likes chocolates. An initial explanation for this fact is that Gapping cannot go beyond the CP phase. However, consider the ambiguity in example (13a) and its possible interpretations in (b) and (c): 4 (13) a. O João disse que convidava a Joana e o Pedro a Maria. The John said that invited the Joana and the Pedro the Maria John said that he would invite Joan and Peter, Mary. b. O João i disse que [pro i convidava a Joana e o Pedro convidava a Maria]. John i said that [he i would invite Joan and Peter would invite Mary]. c. O João i disse que [pro i convidava a Joana] e o Pedro j disse que [pro j convidava a Maria]. John i said that [he i would invite Joan] and Peter j said that [he j would invite Mary]. If we accept the reading in (13c), there is clearly a CP, the subordinate completive clause that he would invite Joan/Mary, in each coordinate clause and the complex sentence is still completely grammatical in EP. 4 Regardless of the given examples, Matos (2013) admits that Gapping can co-occur with subordinate completive infinitive clauses such as: Ana claims to go to the cinema and Paul claims to go to the theatre.
164 In that case, what causes the ungrammaticality in a non-island embedded domain like the one illustrated in (14a)? Notice that the ungrammaticality of (14a) contrasts with the acceptability of the example (14b), which has the structure represented below. (14) a. *O Pedro pediu que a Maria comesse a pêra e que a Joana comesse a maçã. *Peter asked that the Mary ate the pear and that the Joana ate the apple. b. O Pedro pediu que a Maria comesse a pêra e a Joana comesse a maçã. The Pedro asked that the Maria ate the pear and the Joana ate the apple. Peter asked that Mary ate the pear and Joan the apple. With this contrast and the previous examples in mind, I propose that Gapping is incompatible with CP(=ForceP) when this node is directly involved in the articulation of the two clauses. If the CP node is within the elliptic clause, such as in the (non-island) subordinate completive clause in (13), then there is no incompatibility whatsoever. Thus, the grammaticality contrast between (14a) and (14b) is due to the coordination level. On one hand, for the structure in (14a), the Conj head would coordinate the maximal projections of the C (Force), the CPs, which would cause the sentence to be ungrammatical because then one would be trying to gap constituents across distinct CP(=ForceP) phases. On the other hand, in (14b), as seen in the representation above, the maximal projection of Conj is the complement of a single C(=Force) node.
Gapping in European Portuguese 165 Furthermore, on the basis of the contrasts in (15)-(16a) vs. (16b), I also claim that the coordinate constituents in Gapping must be symmetrical even when they involve subordinate clauses. Consider the following example: (15) *A Ana lê romances e penso que a Maria leia poemas (Matos 2003:902) *The Ana reads romances and think that the Maria reads poems. In this example, it would seem clear that the ungrammaticality is due to the existence of CP(=ForceP) between the two coordinate terms. The following paradigm will illustrate this idea: (16) a*o João foi ao cinema e o Pedro disse que foi ao teatro. The John said went to-the cinema and the Peter said that went to the theatre. b. O João disse que foi ao cinema e o Pedro disse que foi ao teatro. The John said that went to-the cinema and the Peter said that went to-the theatre. John said he went to the cinema and Peter said he went to the theatre. This pair of examples seems to show that Gapping may include subordinate completive clauses as long as they occur inside both of the coordinate clauses, respecting parallelism. Additionally, we can coordinate sentence functional projections selected by a single embedded CP phase as in (14b). 5. Conclusion The study presented in this paper is a first analysis of Gapping in European Portuguese. I have contrasted the proposals made for English with EP and have presented empirical evidence against Gapping as Across the Board movement in EP. I have also attempted to give an initial explanation for the apparent incompatibility of the CP node with gapping in EP and have presented some proposals that can explain the examples given in previous literature. Acknowledgements This work is based on my Master s thesis. I would like to thank my supervisor, Professor Gabriela Matos, for the patience with which she guides me and for all the support she gives me whenever I need it. I would also like to thank my colleagues Filipe Marques and Nádia Canceiro for supporting me as friends as well as providing me with useful criticism whenever necessary. I would also like to thank the organization as well as participants of ConSOLE XXII for giving me objective insight about the research at hand as well as the reviewer who gave me further objective criticism. All errors are my own. Faculdade de Letras da Universidade de Lisboa Goncalo.bl.silva@gmail.com
166 References Colaço, M. (2006). Omissão de Material Idêntico em Estruturas Coordenadas: Elipse vs ATB. Actas do XX Encontro Nacional da Associação Portuguesa de Linguística. Associação Portuguesa de Linguística. Coimbra, pp. 261-271. Duarte, I. (1996). A Topicalização em Português Europeu: uma análise comparativa. Actas do Congresso Internacional sobre o Português, vol1, APL, Colibri, Lisboa, pp. 327-358. Johnson, K. (2006). Gapping isn t (VP) Ellipsis. [Ms]. University of Massachussets at Amherst. Johnson, K. (2009). Gapping isn t (VP) Ellipsis. Linguistic Inquiry 40.2 (Spring 2009). 289-328. Kayne, R. S. (1994). The Antisimmetry of Syntax. Linguistic Inquiry Monograph Twenty-Five. MIT Press. Larson, R. K. (1988). On the Double Object Construction. Linguistic Inquiry 19. 335-391. Matos, G. (1992). Construções de Elipse do Predicado em Português SV Nulo e Despojamento. [PhD thesis]. Faculdade de Letras da Universidade de Lisboa. Matos, G. (1995). Estruturas Binárias e Monocêntricas em Sintaxe: algumas observações sobre a coordenação de projecções máximas. Actas do X Encontro Nacional da Associação Portuguesa de Linguística. Edições Colibri, APL, Lisboa, pp. 301-315. Matos, G. (2003). Construções elípticas. Mateus, M.H.M., A. Brito, I. Duarte, I. Faria, S. Frota, G. Matos, F. Oliveira, A. Villalva & M. Vigário (eds). Gramática da Língua Portuguesa, (5ª edição) Editorial Caminho, Lisboa, 99 869-914. Matos, G. (2005). Parataxe: Coordenação e Justaposição evidência a partir da elipse. Actas do XX Encontro Nacional da Associação Portuguesa de Linguística. Associação Portuguesa de Linguística, Lisboa, pp 687-699. Matos, G. (2013). Elipse. Raposo, E., M. F. Nascimento, A. Mota, L. Segura & A. Mendes (eds.), Gramática do Português, vol. II. Fundação Calouste Gulbenkian, Lisboa, pp. 2349-2407. Rizzi, L. (1997). The Fine Structure of the left Periphery. Haegeman L. (ed.), Elements of Grammar. Kluwer Academic Publishers, Netherlands, pp. 281-337. Rizzi, L. (2004). Locality and Left Periphery. Belletti A. (ed.) The Structures and Beyond- the Cartography of Syntactic Structures. Volume 3. Oxford University Press, Oxford, pp. 223-251. Ross, J. R. (1971). Gapping and the order of constituents. Bierwish, N. & K. Heidolph (eds.) Recent developments in Linguistics. The Hague: Mouton. Sag, I. (1980). Deletion and Logical Form. [PhD thesis]. Garland Publishers, New York. Vicente, L. (2010). A note on the movement analysis of Gapping. Linguistic Inquiry 41.3 (Summer 2010), pp. 509-517.