1Languages with distinct degrees of word order flexibility, such as European Portuguese (EP), English and French, have one property in common: they admit so-called “locative inversion” (LI), which is a type of subject-verb inversion where the subject occurs postverbally, while a locative X phrase (XP), typically a prepositional phrase (PP), is preposed, as in [1].
- 1 For reasons of space, in this article, we do not provide glosses for sentences (...)
[1a] |
Na floresta vivia uma família de ursos. (EP1) |
[1b] |
Dans la forêt habitait une famille d’ours. (French) |
[1c] |
In the forest lived a family of bears. (English) |
2Over the last decades, research into LI has generally focused on the properties of the verbs and subjects compatible with this word order. A large number of studies have consistently shown that the subject of LIs is required to be (part of) the focus (cf., e.g., Bresnan, 1994, for English and Chichêwa; Culicover & Winkler, 2008, for English; Cornish, 2001 and 2005, for French; Sheehan, 2007, for EP, Spanish and Italian), and that the (sub)classes of verbs permitted in this type of inversion vary across languages. For example, in EP, all classes of intransitive verbs admit LI (examples [1a], [2a] and [3a-b]) (cf. Pereira, 1998). In contrast, in English and French, this word order is only allowed with unaccusative verbs of existence and appearance (examples [1b-c] and [2b-c]) and, in certain cases, with unergative verbs (see [3c-f]) (cf. Levin & Rappaport Hovav, 1995; Cornish, 2001 and 2005). Since researchers have given little attention to the properties of the XPs which occupy the preverbal position of LI structures, the following question remains unanswered: what properties must a clause-initial XP have to license LI? This is the question which the present study aims to answer.
[2a] |
Na cozinha partiu-se um copo. |
[2b] |
*Dans la cuisine s’est brisé un verre. |
[2c] |
*In the kitchen broke a glass. |
[3a] |
No aquário nadavam muitos peixes coloridos. |
[3b] |
Junto à nossa mesa vomitou um jovem rapaz. |
[3c] |
Dans l’aquarium nageaient des poissons colorés. |
[3d] |
*À côté de notre table a vomi un jeune homme. |
[3e] |
In the aquarium swam many colourful fish. |
[3f] |
*Near our table vomited a young man. |
3In the literature, it is possible to identify two competing perspectives on this issue. According to a purely syntactic perspective, followed by Bresnan (1994), Coopmans (1989), Sheehan (2007), among many others, the preverbal XPs that license LI are locative arguments of the verb. According to an alternative perspective, the key to understanding the properties of these XPs lies in the discourse domain, rather than in narrow syntax. Some authors, most notably Birner and colleagues (e.g., Birner, 1994 and 1996; Birner & Ward, 1998; Ward et al., 2002), have advocated that the preverbal XPs which are found in (English) LI correspond to information that is at least as familiar in discourse as that conveyed by the postverbal subject. Other authors, in contrast, have suggested that these XPs are topics (e.g., Stowell, 1981; Rizzi & Shlonsky, 2006). Recently, a new proposal has been put forth by Lahousse (2003 and 2011) to account for the properties of preverbal XPs in French LI. According to her, they are stage topics (sTOPs).
4With a view to contributing to this debate, the present study investigates the properties of the preverbal XPs that allow LI across languages with distinct degrees of word order flexibility. Three languages are considered in the study: (i) EP, a language where subject-verb inversion is very productive and occurs quite freely (cf. Costa, 1998 and 2004); (ii) French, a language which permits subject-verb inversion in a limited number of contexts, such as LI, wh-questions and some adverbial clauses (cf. Riegel et al., 1994; Jones, 1996; Lahousse, 2003 and 2011; Fuchs, 2006; Marandin, 2011, for a complete list); and (iii) English, a language which has a more rigid word order than French, and allows subject-verb inversion almost exclusively in LI and there-constructions (cf. Culicover & Winkler, 2008; Ward et al., 2002). The relation between these languages with respect to word order flexibility can hence be schematically expressed in the following way:
[4] |
FLEXIBLE EP > French > English RIGID |
5Building on Lahousse’s (2003 and 2011) work, in this study, we argue that the XPs which occupy the preverbal position of LI constructions are sTOPs not only in French, but also in English and EP. Moreover, we contend that being a sTOP is both a necessary and sufficient condition for a preposed XP to license LI. However, our study shows that the types of sTOPs found in this structure vary across the three languages under analysis. We claim that this is because of factors pertaining to morphosyntax, rather than discourse.
6The article is organized as follows: section 2 defines the concept of sTOP, which is central to our analysis. In section 3, we present various pieces of empirical evidence in favor of the claim that preverbal XPs are required to be sTOPs to license LI in English, French and EP. Furthermore, we show that none of the other proposals put forth in the literature can account for all the data presented. Section 4 proposes potential morphosyntactic explanations for the cross-linguistic differences found. Finally, the main conclusions of the study are presented in section 5.
- 2 A number of grammatical phenomena have been shown to be sensitive to the stage- (...)
7The notion of sTOP was proposed by Erteschik-Shir (1997, 1999 and 2007), on the basis of earlier work by Kratzer (1989) on spatio-temporal arguments. According to Erteschik-Shir, a sTOP is a type of topic which defines a spatial and/or temporal location. She argues that sTOPs are not admitted by all types of predicates. Assuming with Kratzer (1989) that predicates fall into two classes – those that express temporary properties and events (stage-level predicates), and those that express more permanent properties and characteristics (individual-level predicates) –, the author claims that only the former can take sTOPs2. Nevertheless, this view is challenged by examples such as [5], which prove that prototypical individual-level predicates, such as “to love” and “to hate”, are also compatible with sTOPs.
[5] |
Last year, John read a book by Chomsky. Since then, he loves/hates linguistics. |
|
(Adapted from Lahousse, 2003: 130) |
- 3 In line with previous work by Birner and Ward (Birner & Ward, 1998; Ward et al., 2002), w (...)
8In order to draw a clear line between what is and is not a sTOP, one must briefly examine a question which has generated a great deal of debate among researchers: what is a topic? In the extensive literature on information structure, there are various definitions for this concept, but no general consensus about which one is the most precise (see, e.g., Gundel & Fretheim, 2004, and Barbosa, 2005, for an overview of the literature). Frequently, topic is taken to be the part of the sentence which conveys discourse-old information (i.e., information explicitly introduced in or inferentially linked to prior discourse3), because it is assumed that the notions of “topic” and “focus” are grounded, to some extent, in the discourse notions of “old” and “new” information. However, there is not always a one-to-one correspondence between these notions. Consider the following example:
[6] |
I can’t find broccoli anywhere. [TOP Crack] they sell at every corner, but broccoli it’s like they don’t grow it anymore. |
|
(Vallduví, 1990: 25) |
- 4 It should be noted that topics are typically discourse-old due to independent reasons rel (...)
9Here the topicalized noun phrase (NP) “crack”, though a topic, does not represent old information. Rather, it represents discourse-new information, since it is neither introduced in nor inferentially linked to the prior discourse. This example demonstrates that, as some authors (e.g., Reinhart, 1981; Vallduví, 1990, among others) argue, encoding old information is not a necessary condition for a constituent to be topic4. Consequently, it is not adequate to define topic and focus in relation to the notions of old and new information.
10Following Chomsky (1971 and 1976), Jackendoff (1972) and Zubizarreta (1998), among others, in the present study, we assume that the notions of focus and topic are best defined in relation to the discourse notion of presupposition, which we take to be “the set of propositions lexicogrammatically evoked in a sentence which the speaker assumes the hearer already knows or is ready to take for granted at the time the sentence is uttered” (Lambrecht, 1994: 52; emphasis added). In other words, a presupposed constituent conveys information that is taken for granted, but not necessarily discourse-old. On the basis of this notion, the distinction between focus and topic is generally made in the following way: the focus is defined as the non-presupposed part of the sentence, whereas the topic is taken to be a constituent which is part of the presupposition associated with the sentence.
11In the light of this view of the topic-focus distinction, sTOPs can be defined in more precise terms as constituents that are presupposed (i.e., they convey information that the addressee either knows, by virtue of its discourse-old status, or is ready to take for granted) and define a spatio-temporal location. From this definition it follows that a temporal PP such as “in 1945” has a different discourse status in the question-answer pairs [7a] and [7b].
[7a] |
A: When did the Second World War end? B: The Second World War ended in 1945. Presupposition: “The Second World War ended in date x”; Assertion: “that date x = in 1945” |
[7b] |
A: What happened in 1945? B: In 1945 the Second World War ended. Presupposition: “event x happened in 1945”; Assertion: “that event x = the Second World War ended” |
- 5 Underlying this claim is the assumption that, in a question-answer pair, the co (...)
12In [7a], the temporal PP “in 1945” is the focus, since it corresponds to the non-presupposed part of the sentence, i.e. the assertion5. In contrast, in [7b], the same PP is a sTOP, because it is part of the presupposition associated with the sentence and specifies the temporal location of the event.
13According to Erteschik-Shir (1997, 1999 and 2007), sTOPs can be not only overt, as in [7b], but also covert, as in the example below:
[8] |
A: What’s going on? B: It’s raining. [sTOP = here and now] |
14Here the assertion “it’s raining” is made in relation to a spatio-temporal location which is taken for granted by the addressee, even though it is not explicitly mentioned in the discourse: the here and now of the speaker. For this reason, Erteschik-Shir postulates that there is a covert sTOP in sentences like [8].
15That this type of topic is necessarily associated with [8] is suggested by three pieces of evidence. First, when uttered in an out-of-the-blue context like the one in [8], the sentence “it’s raining” is synonymous with [9], but, crucially, not with a sentence whose location is indeterminate, like [10].
[9] |
It’s raining here now. |
[10] |
It’s raining somewhere now. |
16Second, this sentence can only be understood as not referring to the here-and-now of the speaker when an overt locative is present, as in [11], or when the sentence is embedded in a discourse context which induces a non-deictic interpretation, as in [12]. Note that, in the latter case, there is an implicit location which is recoverable from the previous linguistic context.
[11] |
(A speaker in Lisbon says:) It’s raining in Edinburgh. |
[12] |
They have just arrived in Edinburgh and it’s raining. [sTOP = in Edinburgh, now] |
17Third, the truth value of [8] can be challenged by a sentence like [13]. Assuming that one can only deny something that is effectively part of the interpretation of a sentence, this fact provides robust evidence in favor of the claim that sentence [8] does have a covert sTOP, which is interpreted as the here-and-now of the speaker.
[13] |
No, it’s not raining here right now (but it rained this morning / it’s raining in my village). |
18On the basis of these pieces of evidence, it can be concluded that covert sTOPs do exist. In line with previous work by Lahousse (2003, 2007 and 2011), we consider that covert sTOPs can be subdivided into two types, which we label “deictic sTOPs” and “anaphoric sTOPs”. The former are interpreted as referring to the spatio-temporal location of the speaker, as in the case of [8]. The latter, on the other hand, are interpreted in relation to the spatio-temporal parameters of the preceding linguistic context. This is what happens in [14]:
[14] |
We arrived in Edinburgh this morning. It was raining. [sTOP = in Edinburgh, at the moment of the speaker’s arrival] |
19In brief, for a (c)overt constituent to be a sTOP, it has to meet two requirements: (i) be part of the presupposition associated with the sentence, and (ii) specify a location. As shown above, there are three types of sTOPs: (i) overt, (ii) covert anaphoric, and (iii) covert deictic sTOPs.
20Traditionally, it is assumed that the XPs admitted in the preverbal position of LI are locative arguments of the verb (cf. Bresnan, 1994; Coopmans, 1989; Sheehan, 2007; among others). Nonetheless, this account is disconfirmed by sentences like [15], where the preverbal constituents are adjuncts of the verb and yet allow LI.
[15a] |
Precisely at nine o’clock came the Princess in the carriage with four horses. |
|
(A. Lang, The Blue Mountains, London, Sovereign Classic, 2014, p. 11) |
[15b] |
Plus tard sont arrivés des gens de divers groupes hema “dans le graben du lac et sur la crête”. |
|
(E. Thiry, Une introduction à l’ethnohistoire des Hema du Nord (Congo du Nord-Est), Tervuren, Musée royal de l’Afrique centrale, 2004, p. 62) |
- 6 “Lit.” stands for “literal translation”.
[15c] |
Ao fundo do quintal ladravam os cães da Dona Maria. Lit.6 ‘at the back of the backyard barked the dogs of Mrs Maria’ |
21In the face of this evidence, we are left with the following question: what precise properties must a clause-initial XP have to be able to license LI? In the present section, we argue that the answer to this question can be found in the discourse domain. More specifically, building on Lahousse’s (2003, 2007 and 2011) work on French inversion, we consider that a clause-initial XP must be a sTOP to license LI, and further, that this requirement applies not only to French, but also to other languages, such as English and EP.
22Our hypothesis makes two testable predictions. On the one hand, it predicts that, in EP, English and French, the XPs in preverbal position must be presupposed, but not necessarily discourse-old, for LI to be felicitous. On the other hand, it predicts that not all the presupposed topic XPs allow LI; only sTOPs do. Both predictions will be examined in detail below.
23In English, French and EP, LI is possible with both discourse-old and discourse-new preverbal XPs. While the former can co-occur with new and old postverbal subjects, the latter are only compatible with subjects which introduce new information in discourse – cf. examples [16] to [19]. Crucially, as we will show in this section, the (un)acceptability of each of these combinations of constituents can be successfully explained by our proposal that a preverbal XP must be presupposed to license LI.
[16] |
Discourse-old preverbal constituent + discourse-new postverbal constituent |
[16a] |
A Sofia tem um piano no centro da sala. Em cima do piano estão várias fotografias antigas. Lit. ‘the Sofia has a piano in the middle of the living room. On the piano are various photographs old’ |
[16b] |
Marie a une table en marbre dans la salle de séjour. Sur la table se trouve un vase de cristal. |
[16c] |
They have a great big tank in the kitchen, and in the tank are sitting a whole bunch of pots. |
|
(Ward et al., 2002: 1386) |
- 7 Note that, in English, French and EP, when the information presented by the pre- and po (...)
[17] |
Discourse-old preverbal constituent + discourse-old postverbal constituent7 |
[17a] |
Obama foi cumprimentado pelo diretor do museu, Jean-Luc Martinez, que lhe ofereceu um livro sobre o Louvre. A visita do Presidente dos EUA durou cerca de duas horas. Ao seu lado esteve sempre o diretor do museu, que lhe mostrou as obras primas em exposição. Lit. ‘Obama was greeted by the director of the museum, Jean-Luc Martinez, who gave him a book about the Louvre. The visit of the President of the USA lasted about two hours. By his side was always the director of the museum, who showed him the masterpieces on display.’ |
[17c] |
Tich made tea in a blackened billy and McPherson filled a telescopic cup he took from a pocket. Seated on a form, he helped himself to sugar and then proceeded to cut chips from a tobacco plug, the cold and empty pipe dangling from his lips against the full grey moustache. Seated opposite him was Tich, waiting for gossip, wandering, hoping. |
|
(Birner & Ward, 1998: 169) |
[18] |
Discourse-new preverbal constituent + discourse-new postverbal constituent |
[18a] |
Almocei na Marshall Field’s ontem e nem vais acreditar quem lá estava. Atrás de um aglomerado de microfones estava a Hillary Clinton. |
[18b] |
J’ai pris le déjeuner à Marshall Field’s hier, et tu ne peux pas imaginer qui était là. Derrière une forêt de microphones se trouvait Hillary Clinton. |
[18c] |
I had lunch at Marshall Field’s yesterday, and you wouldn’t believe who was there. Behind a cluster of microphones was Hillary Clinton (holding yet another press conference). |
|
(Ward et al., 2002: 1387) |
[19] |
Discourse-new preverbal constituent + discourse-old postverbal constituent |
[19a] |
A Sofia tem muitas fotografias antigas na sala. ??Em cima do/de um piano estão algumas fotografias. Lit. ‘the Sofia has many photographs old in the living room. On the/a piano are some photographs’ |
[19b] |
Marie a deux vases de cristal dans la salle de séjour. ??Sur la/une table se trouve un vase de cristal. |
[19c] |
They have a whole bunch of pots in the kitchen, and ??in a great big tank are sitting all of the pots. |
|
(Ward et al., 2002: 1386) |
- 8 Discourse-oldness is a gradient notion. As noted by Birner (1998: 309), “more recently (...)
- 9 For reasons of space, these tests are just presented in English. However, the (...)
24Let us first consider the LI structures in [16] and [17]. In these sets of examples, the preverbal XPs convey information that is more familiar than that of the postverbal subjects. In the case of [16], the preverbal PPs are the most familiar constituents, since they have a discourse-old status while the subjects do not. In the case of [17], the XPs that appear preverbally are also the most familiar constituents, because the information they convey has been more recently mentioned in discourse than the information presented postverbally8. Given the facts just described, it can be concluded that, in the LIs [16] and [17], the information in postverbal position cannot be readily taken for granted, whereas the information conveyed by the preverbal constituents can. Hence, these constituents have a presupposed status. This is why they can license LI. Significantly, this claim is supported by the following question-answer tests9, which prove that the acceptability of LIs like [16] and [17] is greatly reduced when they are embedded in discourse contexts where the preposed XPs are not part of the presupposition associated with the sentence.
[20a] |
A: What is sitting in the tank? B: In the tank are sitting [FOC a whole bunch of pots]. |
[20b] |
A: Where is a whole bunch of pots? B: ??[FOC In the tank] are sitting a whole bunch of pots. |
[20c] |
A: Why are you upset? B: (because) ??[FOC in the tank are sitting a whole bunch of pots]. |
[20d] |
A: Who was seated opposite him? B: Seated opposite him was [FOC Tich]. |
[20e] |
A: Where was Tich? B: ??[FOC Seated opposite him] was Tich. |
[20f] |
A: Why was he upset? B: (because) ??[FOC Seated opposite him was Tich]. |
25Contrary to what we observe in [16] and [17], in the LI structures in [18], there are no differences between pre- and postverbal constituents in terms of discourse familiarity: they both convey discourse-new information. However, the preceding context, namely the sequence “you won’t believe who was there”, makes it clear that what is being asserted is the presence of Hillary Clinton, and not where she was. Therefore, in this context, the information presented by the preposed constituent is not under assertion. On the contrary, it appears to be taken for granted. Put differently, it seems to be presupposed.
- 10 For reasons of space, these tests are just presented in English. However, t (...)
26Two pieces of evidence lend support to this claim. First, the LI in [18] can be replaced by [21] without losing its core meaning. Second, and more importantly, this structure is infelicitous in reply to an out-of-the-blue question, which forces the entire sentence to be interpreted as focus, i.e. as non-presupposed (example [22])10.
[21] |
[…] you wouldn’t believe who was there: Hillary Clinton. |
[22a] |
A: Why did everybody look so surprised? B: ??(Because) [FOC behind a cluster of microphones was Hillary Clinton]. |
[22b] |
A: Why did everybody look so surprised? B: (Because) [FOC Hillary Clinton was behind a cluster of microphones]. |
27Crucially, the felicity contrast between [18], where the preverbal PP is simultaneously [+ presupposed] and [+ discourse-new] (cf. above), and [22a], where the PP is [- presupposed] and [+ discourse-new], indicates that, as predicted, the factor which determines whether a clause-initial XP can license LI is its [± presupposed] status, rather than its [± discourse-new] status.
28The proposal that a fronted constituent must be presupposed to allow LI accounts not only for the felicity of the structures in [16], [17] and [18], but also for the infelicity of sequences like [19], where a discourse-new preverbal constituent is followed by a discourse-old postverbal constituent. In this case, LI is infelicitous because the preverbal PPs are not presupposed while the postverbal subjects are.
29Even though the evidence presented so far strongly suggests that only preverbal XPs with a presupposed status can license LI, there are (apparent) exceptions to this rule in literary texts. As the following examples taken from the Portuguese, French and English versions of Ali Baba and the forty thieves illustrate, literary texts often begin with a LI structure. In this context, none of the sentence constituents seems to be presupposed.
30Interestingly, this type of (apparently) non-presupposed LI has a strongly literary flavor. As shown in [24], outside of a literary context it is much less acceptable to use an inversion structure whose pre- and postverbal constituents have a non-presupposed status.
[24a] |
A: João, ouviste a reportagem insólita que deu nas notícias? B: ??Num bairro de lata vive um ex-presidente brasileiro. / Um ex-presidente brasileiro vive num bairro de lata. |
[24b] |
A: Jean, est-ce que tu as écouté le reportage insolite dans le journal? B: ??Dans un bidonville vit un ex-président brésilien. / Un ex-président brésilien vit dans un bidonville. |
[24c] |
A: John, did you hear the weird report on the evening news? B: ??In a slum lives a former Brazilian president. / A former Brazilian president lives in a slum. |
31These facts raise the following question: why are non-presupposed fronted XPs accepted in LI structures (almost) exclusively in literary texts? Following Birner and Ward (1998), we argue that this occurs because the hearer/reader expects stories to have spatio-temporal settings. As proposed by Birner and Ward (1998: 176), “in the context of the beginning of a story, the notion of a setting may be assumed to be situationally evoked (Prince, 1981)”, i.e. salient in the situational context of the discourse. Significantly, in non-literary contexts, the preposed constituents which convey situationally evoked information also admit LI. For example, one can point to a table in a room and utter sentences like [25] out of the blue.
[25a] |
João, em cima daquela mesa está uma caneta azul; podias trazer-ma, por favor? |
[25b] |
Jean, sur cette table se trouve un stylo bleu; est-ce que tu pourrais me l’apporter, s’il te plaît? |
[25c] |
John, on that table is a blue pen; could you bring it to me, please? |
32In line with Prince (1981), we assume that situationally evoked information constitutes given, rather than new information. We further assume that situationally evoked XPs, such as those in [23] and [25], have a presupposed status, corresponding to propositions which “the speaker assumes the hearer already knows or is ready to take for granted at the time the sentence is uttered” (Lambrecht, 1994: 52). Based on these assumptions, we propose that, contrary to what might appear at first sight, the structures in [23] and [25] conform to the generalisation that preverbal XPs must have a presupposed status to be accepted in LI.
33Taken together, the empirical data examined in this section provide robust evidence in favor of the prediction that a clause-initial XP must be presupposed to license LI. Simultaneously, these data bring into question the hypotheses that appeal to the notion of discourse familiarity to account for the properties of preposed XPs in LI. In particular, the felicity contrast between [18] and [22a], repeated under [26] for ease of exposition, constitutes a key piece of evidence against a widely adopted generalisation, proposed by Birner and colleagues (e.g., Birner, 1994 and 1996; Birner & Ward, 1998; Ward et al., 2002), according to which LI requires the preverbal constituent to represent information that is at least as familiar within the discourse as that represented by the postverbal constituent. This is because the contrast in [26] demonstrates that, unlike what Birner and colleagues’ predict, the combination of a discourse-new preverbal constituent with a discourse-new postverbal constituent is not always felicitous. The facts illustrated by [26] are only explainable if one assumes – as we do – that the (in)felicity of a preverbal XP in LI is determined by its [± presupposed] status, rather than by its discourse familiarity.
[26a] |
Discourse-new preverbal constituent + discourse-new postverbal constituent: I had lunch at Marshall Field’s yesterday, and you wouldn’t believe who was there. Behind a cluster of microphones was Hillary Clinton (holding yet another press conference). (= [18c]) |
[26b] |
Discourse-new preverbal constituent + discourse-new postverbal constituent: A: Why did everybody look so surprised? B: ??(Because) behind a cluster of microphones was Hillary Clinton (holding yet another press conference). (= [22a]) |
34The hypothesis that the XPs in the preverbal position of LIs are required to be sTOPs predicts that having a presupposed topic status is a necessary, but insufficient condition for a preverbal XP to license subject-verb inversion. An additional condition must be satisfied: the XP must define a location. This prediction is confirmed by various pieces of empirical evidence, which will be presented and discussed in the remainder of this section.
- 11 Note that, in participial phrases with a locative PP, omitting the PP reduces the (...)
- 12 English admits bare inversion in the context of stage directions, as illustrated (...)
35The first piece of evidence comes from the well-attested fact that, in English, French and EP, LI typically occurs with preverbal XPs which define a spatio-temporal location. These include locative PPs and adverbial phrases (AdvPs) (examples [27] and [28]), temporal PPs, AdvPs and NPs (examples [29] and [31]), and participial phrases with a locative PP11 (example [32]) (cf. Birner, 1994 and 1996; Birner & Ward, 1998; Cornish, 2001 and 2005; Lahousse, 2003 and 2011; Pereira, 1998, among others). In addition to overt spatio-temporal XPs, preverbal covert locatives are also compatible with LI, but only in EP and French (cf. Lahousse, 2003, 2007 and 2011; Sheehan, 2007 and 2010)12.
[27a] |
Em frente deles estavam mais de cem jornalistas dos quatro cantos do mundo. |
[27b] |
Devant eux se trouvaient plus de cent journalistes venus des quatre coins du monde. |
[27c] |
In front of them were over a hundred journalists from the four corners of the world. |
[29a] |
Às 10:02 horas partiu do pódio o concorrente número 1, Miguel Nunes/Victor Calado, em Mitsubishi Lancer X. Lit. ‘at 10:02 o’clock departed from the podium the contestant number 1, Miguel Nunes/Victor Calado, in Mitsubishi Lancer X’ |
|
(Internet: the page is no longer online) |
[29c] |
Precisely at nine o’clock came the Princess in the carriage with four horses. |
|
(A. Lang, The Blue Mountains, London, Sovereign Classic, 2014, p. 11) |
[31c] |
Four years later occurred another attempted invasion that, like its predecessors, was a fiasco. |
|
(W.M. Armstrong, E.L. Godkin: A Biography, Albany, State University of New York Press, 1978, p. 109) |
[32c] |
Coiled on the floor lay a one-hundred-and-fifty-foot length of braided nylon climbing rope three-eighths of an inch thick. |
|
(Birner & Ward, 1998: 121) |
- 13 According to this principle, the canonical subject position (i.e., the specifier of the (...)
36In EP, inversion structures without any overt preverbal XP (henceforth, bare inversion) only occur in two contexts (cf. Costa, 1998 and 2004; Sheehan, 2007 and 2010): (i) when the subject receives narrow focus, independently of the type of verb (example [33]), and (ii) in sentence-focus contexts, with a group of intransitive verbs which cut across the traditional unergative/unaccusative distinction (e.g., the unaccusative verb chegar [“to arrive”] and the unergative verb telefonar [“to telephone”]) (see [34]). According to Sheehan (2007 and 2010), who follows earlier work by Pinto (1997) on Italian bare inversion, these verbs, labeled “inversion verbs”, optionally select an extra null loco-temporal argument (LOC), which typically has a deictic interpretation, and is a possible candidate for satisfying the extended projection principle (EPP)13. For this reason, unlike all other verbs, the so-called “inversion verbs” can exploit two alternative strategies for satisfying the EPP: (i) the subject can be moved to Spec, IP or (ii) LOC can be raised to Spec, IP. The latter strategy allows the subject to remain low, in a postverbal position. Thus, as proposed by Pinto (1997) and Sheehan (2007 and 2010), the availability of “bare” inversion in sentence-focus contexts is dependent on the availability of LOC as a possible EPP-satisfier.
[33a] |
A: Quem chegou? B: Chegou o João / ??O João chegou. Lit. ‘A: who arrived? B: arrived the John / the John arrived’ |
[33b] |
A: Quem falou? B: Falou o João / ??O João falou. Lit. ‘A: who spoke? B: spoke the John / the John spoke’ |
[33c] |
A: Quem comeu o bolo? B: Comeu(-o) o João / ??O João comeu(-o). Lit. ‘A: who ate the cake? B: ate (it) the John / the John ate it’ |
[34] |
A: O que aconteceu? Lit. ‘what happened?’ |
[34a] |
B: Chegou o João. / O João chegou. Lit. ‘arrived the John / the John arrived’ |
[34b] |
B: ??Descongelou o frigorífico / O frigorífico descongelou. Lit. ‘defrosted the fridge / the fridge defrosted’ |
[34c] |
B: ??Falou o João / O João falou. Lit. ‘spoke the John / the John spoke’ |
[34d] |
B: Telefonou o João / O João telefonou. Lit. ‘telephoned the John / the John telephoned’ |
- 14 For further evidence, see Pinto (1997) and Sheehan (2007, 2010).
37One of the key pieces of evidence that supports this proposal is the fact that inversion triggers a subtle change in meaning in sentence-focus contexts. In fact, in verb-subject orders like [34a] and [34d], the location of the event/action is associated with that of the speaker (i.e., a sentence like chegou o João is interpreted as “John arrived here”). In contrast, in subject-verb orders, such a deictic interpretation is not available. Here, a location neutral interpretation is strongly preferred (i.e., o João chegou is interpreted as “John arrived somewhere”) (see in particular Sheehan, 2007 and 2010). In the light of this evidence14, we assume with Pinto (1997) and Sheehan (2007 and 2010) that, in sentence-focus contexts, bare inversion is licensed by a preverbal null locative, thus constituting an instance of covert LI.
38We propose that LOC acts as a covert sTOP at the level of discourse, since it defines a location and its content is readily taken for granted by the addressee. In other words, it is presupposed. We further propose that, even though LOC has a deictic interpretation by default, it can also receive an anaphoric reading, as in [35]. In short, depending on the context where it is embedded, LOC can function either as a covert deictic sTOP or as a covert anaphoric sTOP.
[35] |
A: Eles andavam a brincar no parque e resolveram entrar na casa abandonada. Lit. ‘they were playing in the park and decided to enter the abandoned house’ B: E o que aconteceu? Lit. ‘and what happened?’ A: Apareceu um fantasma. [sTOP = ‘afterwards, in the abandoned house’] Lit. ‘appeared a ghost’ |
39Like EP, French has a type of bare inversion, traditionally called “absolute inversion” (e.g., Le Bidois, 1952; Gournay, 2006), where the location of the event/action described in the sentence is implicit (Lahousse, 2003, 2007 and 2011). Interestingly, this type of bare inversion is only admitted with a group of verbs which cuts across the unaccusative/unergative distinction and roughly corresponds to the group of “inversion verbs” which allow covert LI in EP. Thus, this appears to be another manifestation of covert LI.
[36a] |
Tout le monde s’assoit… Arrive le chef d’orchestre, Eliahu Inbal, un Israélien… |
|
(Lahousse, 2007: 14) |
[36b] |
Tout le monde s’assoit… ??Se brisent les chaises. |
[37a] |
Tout le monde s’assoit… Sonnent les cloches. |
[37b] |
Tout le monde s’assoit… ??crient les enfants. |
- 15 Possibly, this type of LI only occurs in narrative sequences because the he (...)
- 16 As Haegeman and Guéron (1999: 63) point out, “the ‘core grammar’ of a language […] (...)
40Crucially, the type of covert LI admitted in French differs from that of EP in two ways. First, unlike EP, French appears to admit this inversion only in (oral or written) narrative sequences15, which suggests that it is not part of its core grammar16. Second, French allows covert LI exclusively in discourse contexts where the spatio-temporal location of the postverbal subject is recoverable from the preceding context. Put differently, this type of inversion is just admitted with covert anaphoric sTOPs (example [38]). As Lahousse (2003, 2007 and 2011) points out, these topics always get a temporal reading, and tend to be interpreted as meaning “then, afterwards (in the same place)”.
[38a] |
Elle sonne. Arrive une infirmière: “Ah! Mais madame, ce n’est pas l’heure”. |
|
(Lahousse, 2007: 13) |
[38b] |
(Context: opening sentence of a text) *Arrive une infirmière: […]. |
[38c] |
A: Qu’est-ce qui s’est passé? B: *Est arrivée une infirmière. |
41Taken together, the cross-linguistic data presented above concerning the type of preverbal XPs found in LI suggest that this type of word order is licensed by (c)overt constituents which are presupposed and define a spatio-temporal location, just as sTOPs characteristically do.
42Significantly, the spatio-temporal XPs which do not specify a location are unable to trigger inversion in these languages. This is the case of the non-specific indefinite adverb “somewhere” (example [39]) and of adverbs of duration, like “temporarily” (example [40]). Only in EP do we find (apparent) exceptions to this rule. As shown in [40a], EP permits inversion with fronted adverbs like “temporarily”. Contrary to what might appear at first sight, this verb-subject order is not licensed by the fronted adverb. Rather, it is an instance of bare inversion and licensed by other mechanisms, which are unavailable in English and French. This is why the structure in [40] is admitted in EP – but crucially not in the other languages – without any overt preverbal XP (example [41]). In the light of these facts, it can be concluded that the word order illustrated by examples [39] and [40] constitutes a second piece of empirical evidence in favor of the proposal that a preverbal XP must define a location to license LI.
[39a] |
??Algures apareceu o João. |
[39b] |
??Quelque part est apparu Jean. |
[39c] |
??Somewhere appeared John. |
[40a] |
Temporariamente ocorreram algumas halucinações. |
[40b] |
*Temporairement se sont produites quelques hallucinations. |
[40c] |
*Temporarily occurred hallucinations. |
[41a] |
Ocorreram algumas halucinações. |
[41b] |
*Se sont produites quelques hallucinations. |
[41c] |
*Occurred some hallucinations. |
- 17 Note that the term “inversion” is used throughout this article to refer to inversion wi (...)
- 18 As an anonymous reviewer rightly pointed out, French admits inversion with the manner a (...)
43The last piece of evidence which supports this proposal comes from the behavior of English and French with non spatio-temporal XPs. In these languages, subject-verb inversion without a contrastive meaning17 is only allowed in declarative matrix clauses in two constructions: (i) LI and (ii) expletive-associate inversion (e.g., “there occurred something strange”; “il s’est passé quelque chose de bizarre”). Just as our hypothesis predicts, in these languages, the XPs which do not denote a spatio-temporal location are generally incompatible with inversion, even when their content is presupposed. This is exemplified in [42], with an instrumental PP, and in [43], with a manner AdvP18.
[42a] |
A: Qui est arrivé par le train? B: *Par le train est arrivé Jean. |
[42b] |
A: Who arrived by train? B: *By train arrived John. |
[43a] |
A: Qui est venu volontairement? B: *Volontairement est venu Jean. |
[43b] |
A: Who came voluntarily? B: *Voluntarily came John. |
44There are, nevertheless, some exceptions to this rule. In fact, various corpus-based studies have observed that English and French permit verb-subject orders with preverbal XPs which are not semantically spatio-temporal, namely adjective phrases (AdjPs) (example [44]), NPs with a predicative function (example [45]) and participial phrases (example [46]) (Birner, 1994 and 1996; Birner & Ward, 1998, for English; Lahousse, 2003 and 2011, for French).
[44a] |
Republican senators on the conference committee expressed outrage when these gimmicks were proposed. Angriest of all was Judd Gregg. |
|
(R.G. Kaiser, Act of Congress: How America’s Essential Institution Works, and How It Doesn’t, New York, Alfred A. Knopf, 2013, p. 361) |
[45] |
She is a nice woman, isn’t she? Also a nice woman is our next guest… |
|
(Ward et al., 2002: 1385) |
- 19 Underlying this proposal is the assumption that the location of an entity E can be defi (...)
45The inversion structures presented above have one characteristic in common: their sentence-initial XPs generally repeat an adjective (e.g., “angry”), noun (e.g., “woman”) or participle (e.g., “arrested”) mentioned in or inferable from the preceding linguistic context, thus situating the postverbal subject in relation to or within a group of entities in the prior discourse. For example, in [44a], the AdjP “angriest of all” situates “Judd Gregg” in the group of Republican senators who were outraged. Similarly, in [45], the NP “also a nice woman” locates the postverbal subject “our next guest” in the class to which the woman mentioned in the prior sentence belongs − the class of “nice women”. Finally, in [46a], the preverbal verb phrase (VP) “arrested” situates “Condy Warren Ashe Jr”., “Cory D. Brown”, “Jeffery Blake Hasty” and “Joshua David Mixter” in the group of four people arrested in Twiggs County. These examples, therefore, suggest that the non spatio-temporal XPs compatible with inversion are presupposed and define a notional, abstract location in context. As a result, they function as sTOPs19. The difference between them and the typical sTOPs lies in the fact that they provide a notional, rather than a spatio-temporal, location.
46Crucially, as this analysis predicts, when a non spatio-temporal XP is unable to fulfil a notional “locative” function, subject-verb inversion is blocked. Consider the examples below:
[47a] |
I went to the police station yesterday, and you wouldn’t believe who was there. ??Arrested were Condy Ashe Jr. and Cory Brown. |
[47b] |
I went to the police station yesterday, and you wouldn’t believe who was there. Behind a cluster of microphones were Condy Ashe Jr. and Cory Brown. |
47In [47a], the preverbal VP “arrested” is not linked to any entities mentioned in or inferable from the preceding discourse and, as a result, it cannot locate the postverbal subject in relation to any entity. Due to this fact, the VP cannot function as a sTOP, which, in turn, leads to the infelicity of inversion. Significantly, when the sentence-initial position is filled by a sTOP like the locative PP “behind a cluster of microphones”, inversion becomes felicitous, as shown in [47b]. Note that, in both [47a] and [47b], the preverbal XPs represent discourse-new information and are not under assertion. What differentiates them is that the PP in [47b] has a locative function, while the VP in [47a] does not. Therefore, the comparison between [47a] and [47b], on the one hand, and between these and [46a], on the other hand, makes it clear that the infelicity of inversion in [47a] is caused neither by the fact that the VP is not intrinsically locative nor by its discourse-new status. Rather, the infelicity of this inversion structure stems from the fact that the VP does not have a locative function in the discourse context in which it is embedded and is consequently unable to act as a sTOP. These data thus support the proposal that English and French only allow verb-subject orders with (referential) preverbal XPs which define a location and are interpreted as sTOPs.
48To sum up, from the empirical evidence presented throughout section 3, three conclusions can be drawn. First, preverbal XPs must be presupposed to license LI (prediction 1 confirmed). Second, not all the preverbal XPs with a presupposed topic status allow this inversion – only the ones that define a location, i.e. sTOPs, do (prediction 2 confirmed). Finally, it appears that being a sTOP is not only necessary, but also a sufficient condition for a preverbal XP to license LI. At least, in English, French and EP, any sTOP can trigger this type of subject-verb inversion, regardless of two factors typically deemed decisive in the literature: (i) the adjunct or argument status of the preposed constituent, and (ii) the discourse-oldness or -newness of the information it conveys.
49Even though all the languages under analysis require the preverbal position of LI to be filled by a sTOP, the subtypes of sTOPs allowed in this position vary cross-linguistically. As shown in Table 1, English is the strictest of the three languages considered in this study, as it only permits LI with overt sTOPs. EP, on the contrary, admits this inversion with all types of sTOPs. French is situated between these opposite poles. As illustrated below, this language allows LI not only with overt sTOPs, but also with covert anaphoric sTOPs. In the present section, we will argue that these cross-linguistic differences stem from morphosyntactic factors.
Table 1. Configurations of LI permitted in English, French and EP
LI structure |
English |
French |
EP |
overt sTOP verb subject |
✓ Mary owns a marble coffee table. [On the table is a crystal vase.] |
✓ Marie possède une table basse en marbre. [Sur la table se trouve un vase de cristal.] |
✓ A Maria tem uma mesa de café em mármore. [Em cima da mesa está uma jarra de cristal.] |
covert sTOP[+ anaphoric] verb subject |
✗ *The door slowly opens and [appears a man in his sixties]. |
✓ Une porte s’ouvre doucement et [apparaît un homme d’une soixantaine d’années]. |
✓ A porta abre-se lentamente e [aparece um homem com uns sessenta anos]. |
covert sTOP[+ deictic] verb subject |
✗ A: What happened? B: *[Died Prince]. |
✗ A: Qu’est-ce qui s’est passé? B: *[Est mort Prince]. |
✓ A: O que aconteceu? B: [Morreu o Prince]. |
50The differences between EP, on the one hand, and French and English, on the other, with respect to covert LI are related to a macro-parametric difference: EP fixes the null subject parameter at a positive value, while English and French have a negative setting for this parameter – cf. [48] and [49].
[48] |
Referential null and overt subjects |
[48a] |
Cozinho muito bem. / Eu cozinho muito bem. |
[48b] |
*Cuisine très bien. / Je cuisine très bien. |
[48c] |
*Cook very well / I cook very well. |
[49] |
Expletive null and overt subjects |
[49a] |
Está a chover. / *Ele está a chover. |
[49b] |
*Pleut. / Il pleut. |
[49c] |
*Is raining. / It is raining. |
- 20 Languages which allow referential and expletive null subjects.
- 21 Languages which allow expletive, but not referential null subjects.
- 22 Recall that LOC functions as a (deictic or anaphoric) sTOP in EP covert LI.
- 23 Note that it is well attested in the literature that certain constraints of non-null (...)
51In fact, the behavior of consistent null subject languages (NSLs)20 (e.g., EP, Spanish and Italian), partial NSLs21 (e.g., Brazilian Portuguese and Cape Verdean Creole) and non-NSLs (e.g., English and French) with respect to inversion suggests that the order covert sTOP verb subject is only admitted in the grammar of languages which license expletive null subjects (see, e.g., Costa & Figueiredo Silva, 2006; Costa & Pratas, 2004; Nicolis, 2008; Pratas, 2002; Sheehan, 2007 and 2010). In these languages, the EPP does not require phonetic material in the canonical subject position, Spec, IP. As a result, the (deictic/anaphoric) null argument LOC22 can move to this position to satisfy the EPP, which frees the subject from the need to move from its base-generated low position. In contrast, in the languages which do not admit expletive null subjects, like English and French, the EPP must be satisfied by phonetically realized material. For this reason, the movement of LOC to the canonical subject position is not a possibility in these languages. Consequently, the order covert sTOP verb subject is generally unacceptable in English and French. Only in the latter language do we find an exception to this general rule. As described in section 3.2, French admits the order covert sTOP[+ anaphoric] verb subject in a special context: narratives. We argue that this order is acceptable in a narrative style because the requirements made by the EPP in the core grammar of French (i.e., the grammar that governs unmarked registers and styles) are relaxed in the peripheral grammar which governs this special, marked style23. Put simply, in a narrative style, aesthetic concerns seem to overrule the core grammar of French.
52We are now left with one last question: why is it that French allows inversion with covert anaphoric sTOPs in special styles, but English does not? We tentatively propose that this difference is caused by a morphologic factor: the fact that French has a richer tense morphology than English (example [50]).
[50a] |
Past tense of “to sing”: I/you/he/she/it/we/you/they sang |
[50b] |
Past tense of “chanter”: J’ai chanté / tu as chanté / il a chanté / nous avons chanté / vous avez chanté / ils ont chanté |
53Underlying this proposal are two assumptions: (i) that French covert LI is triggered by the fronting of a weak form of LOC, whose interpretation has to be anaphorically established; and (ii) that, as suggested by Lahousse (2003, 2007 and 2011), it is the tense morphology of the verb that indicates the temporal link between the clause and the preceding context, thus allowing the (loco-temporal) interpretation of LOC to be established in relation to prior discourse. In the light of these assumptions, it can be hypothesized that French admits the structure covert sTOP[+ anaphoric] verb subject because its tense morphology enables the sTOP to establish the necessary temporal link with the prior discourse. The poor tense morphology of English, on the contrary, does not allow such a link to be established. As a result, covert LI is completely absent from this language.
54In brief, the cross-linguistic differences found in LI seem to stem from purely morphosyntactic factors. At the level of discourse/pragmatics, there are no asymmetries across English, French and EP. In all of these languages, LI is subject to exactly the same discourse constraints: (i) the preverbal XP must be a sTOP and (ii) the postverbal subject must be (part of) the focus.
55In conclusion, the notions of argumenthood, topicality and discourse familiarity, which have been systematically used in the literature to describe the properties of fronted XPs in LIs, cannot account for the behavior of English, French and EP with respect to these constituents; only the generalization according to which preverbal XPs must be sTOPs can. The empirical data presented and discussed in this study indicate that being a sTOP is both a necessary and sufficient condition for a preverbal XP to license LI. In addition, these data show that this type of subject-verb inversion is a more widespread phenomenon than generally assumed. In addition to the structures typically classified as LI, the following verb-subject constructions have been argued to be licensed by a (covert) sTOP and hence constitute instances of LI: (i) English and French inversion structures with non-spatio-temporal XPs, such as participial and adjectival phrases, which provide an abstract, notional location in context; (ii) cases of subject-verb inversion traditionally labeled “absolute inversion” in French; and (iii) the bare inversion structures which occur in sentence-focus contexts in EP. Thus, the data from English, French and EP suggest that there is a strong correlation between sTOPs and subject-verb inversion cross-linguistically.
56This research was funded by Fundação para a Ciência e a Tecnologia (grant no. PD/BD52263/2013). I would like to thank all the native speakers of English, European Portuguese and French who kindly accepted to share their judgements on locative inversion with me. Thanks in particular to Célia Caravela, Elyse Jamieson, Jordan Person, Melanie Carrie, Nicole Dube and Romance Porrot for their help with judgments and their insightful comments. I am also grateful to Ana Madeira, two anonymous reviewers from Discours and the audiences of the “VII Encuentro de Gramática Generativa” and the “XXXI Encontro Nacional da Associação Portuguesa de Linguística” for their helpful comments and suggestions on previous versions of this work. Needless to say, all remaining errors are mine.