It is a commonly held view that in Italian and in Spanish the fronting of the focus constituent of the sentence to a left-peripheral position (i.e. focus fronting, henceforth FF) is to be associated with a contrastive interpretation. Non-contrastive focus, instead, occurs sentence-finally – or at least postverbally (see, e.g., Rizzi 1997; Zubizarreta 1998; Belletti 2004; López 2009). Since a non-contrastive focus is expected in the answer to a wh-question, only sentences with a postverbal focus are considered to be pragmatically felicitous in the corresponding answer, as shown in (1) and (2):

In recent years, however, this view has been challenged in several studies, which on the basis of experimental data, report a different view: preverbal foci are indeed accepted by native speakers in non-contrastive contexts. Different factors seem to play a role in this controversial issue: crosslinguistic (or dialectal) variation, a different conceptualization of the notion of contrast and of contrastive focus, and the methodology at the basis of the relevant judgements, which sees a contraposition between introspection and experimental data.

In this paper, I propose a solution to this empirical tension based on a more fine-grained notion of contrast. The binary distinction between contrastive and non-contrastive focus, which is traditionally advocated to explain the syntactic distribution of focus, must be abandoned in favour of the identification of different ‘degrees’ or ‘types’ of contrastive focus. The different types and interpretations of focus arise from the way the set of alternatives is pragmatically exploited (Krifka 2007; Cruschina 2012), giving rise to a contextually open set (information focus), unexpectedness with respect to more likely alternatives (mirative focus) or a correction of given alternatives (corrective focus). A special subtype of information focus, namely, exhaustive focus, additionally implies the exhaustive identification or the exclusion by identification with respect to the relevant set of alternatives. On the basis of their degree of contrast against the alternatives (cf. § 2), we could order the types of focus along a scale:

Drawing on a sample of crosslinguistic data, I show that syntactic operations are not sensitive to the mere presence or absence of contrast, but rather to specific types of focus. Languages vary as to the focus types that admit FF, although there seems to be a general tendency to admit FF with the focus types characterized by higher degrees of contrast. We thus need to look at the contrastive meanings that associate with focal constituents in order to gain a better understanding of the grammatical effects that contrast may or may not have in a particular language. As will be shown (cf. § 3), for example, Sardinian and Sicilian allow FF with information focus, but also with the other more contrastive types, that is, with mirative and corrective focus. In Hungarian, FF is typically associated with exhaustive focus, but is also possible with mirative and corrective focus; it is not possible with (non-exhaustive) information focus. In Italian and Spanish, FF is accepted not only with ‘greatest’ and most explicit case of contrast (i.e. corrective focus), but also with contrast against expectations (i.e. mirative focus), while it is not possible with information focus.

The structure of the paper is the following. In the rest of this introductory section, I will discuss the notion of contrast, with special reference to its syntactic effects. In Section 2, I will elaborate on the new perspective on contrast and on the resulting classification of contrastive foci, while Section 3 will be devoted to the association – or lack thereof – between the syntactic operation FF and the different types of focus. In Section 4, I will propose an analysis of the correlation between syntactic movement (FF) and interpretation in terms of conventionalization: following Bianchi, Bocci & Cruschina (2015; 2016), I will argue that FF is in fact triggered not by contrast per se, but by the conventional implicature that is associated with certain types of focus. In Section 5, the relationship between the question-answer context and the type of focus in the answer will be discussed. The use of a focus to answer a wh-question is not per se a sufficient condition for the definition of information focus: other types of foci can occur in answers to questions. The main points of the paper will be summarized in the concluding Section 6, which also contains some considerations for future research.

The empirical basis of this study is mostly constituted by datasets from Italian, Spanish and other Romance varieties such as Sicilian, Sardinian, and French. In relation to exhaustive focus, the discussion will turn to Hungarian.

Contrast is traditionally thought of as a binary notion that admits two possible values: +contrast and -contrast. Attempts exist in the literature to outline a different conception of contrast, especially in relation to focus. Molnár (2002), for instance, arranges the most important criteria in connection with the definition of contrast in a hierarchical order. These criteria have been in fact discussed and judged in different ways in the literature, leading to varying definitions of contrastive and non-contrastive foci. However, they can be considered as minimal requirements for different degrees of contrast along a hierarchy of necessary conditions, as shown in (9) (Molnár 2002: 149):

Focus is ‘highlighted’ with respect to the rest of the sentence, primarily by prosodic means. In other definitions, however, contrast requires the exclusion of certain possibilities, that is, ‘dominant contrast’, or the generation a set of alternatives for the focused constituent. In many approaches, moreover, the condition ‘membership in a set’ is narrowed down to the requirement of a limited set or the explicit mentioning of the alternatives.

Similar conditions have been carefully analysed by Repp (2010) with reference to the set of alternatives (see also Umbach 2004). According to Repp (2010), the aspects that differentiate (plain) focus from contrastive focus are related to the restrictions on the set of alternatives evoked by the focus itself. These aspects include the size of the alternative set (open vs closed set), and the explicit mention or the implicit presupposition of the alternatives, which is to be connected with their identifiability. A further requirement for contrastive focus seems to be the exclusion of the alternatives, in the sense that the assertion only holds for the element denoted by the contrastive focus and does not hold for its alternatives. Repp (2010: 1336) considers these aspects as conditions that provide different requirements for the opposition between contrastive and non-contrastive focus: an open set characterizes non-contrastive focus (10), but a closed set is necessary for contrastive focus; similarly, explicit mention is a typical component of contrastive focus (11), while an implicit presupposition of the alternatives defines non-contrastive focus (10) (see also Krifka 2007):

Following this approach, but giving up on the binary distinction between contrastive vs non-contrastive focus, I would like to propose that different types of focus must be identified and that their different interpretations arise from the way the set of alternatives is pragmatically exploited (Krifka 2007; Cruschina 2012):

In this typology, I deliberately avoid the terms contrast and contrastive, assuming that all the different types of focus are characterized by a degree of contrast: the more ‘active’ or ‘given’ the alternatives are in the context, the greater the contrast (see also Titov’s 2019 notion of D-linking). Contrast is rather weak in the case of information focus, where the focus constituent contrasts with the alternatives of the focal set by virtue of being a member of the same set (13). The identification of the focus constituent implies the exclusion of the alternatives: this exclusion is typically an inferred pragmatic meaning, as with information focus (cf. (4a) above; see Büring 2016: 64), but it could also be a stronger and constant value associated with focus, as with exhaustive focus (see example (14); cf. also § 3.3):1

According to É. Kiss (1998), the preverbal focus in (14B) identifies the exhaustive subset of a potential set for which the predicate phrase actually holds, indicating that for all of the alternatives triggered by focus, the proposition is not true.

The set of focal alternatives can be ordered so as to coincide with the speaker’s expectations: mirative focus is defined as contrast against the expectations, whereby the proposition asserted is presented as being more unlikely or unexpected with respect to the alternative propositions (15):

The sentence with mirative focus in (15c) expresses a violation of the expectations, either because of the complete unexpectedness of the event – in the context of (15a) – or because the content of the asserted proposition was less likely than other focal alternatives – in the context of (15b) (i.e. we saw a tiger yesterday at the zoo is less likely than we saw a zebra yesterday at the zoo, we saw a lion yesterday at the zoo, etc.).

The set of alternatives can be said to be open with information and exhaustive focus: the set is of course pragmatically delimited by the encyclopaedic or shared knowledge of the conversation participants, but what is crucial is that no alternatives are explicitly active or given in the context.2 The alternatives to mirative focus may be given in the context (15b), but need not be (15a) (see Bianchi, Bocci & Cruschina 2015; 2016). It is only with corrective focus that one or more alternatives have to be active in the discourse having been explicitly mentioned or deictically identified, as in (16), where the corrective proposition ‘we saw a tiger’ corrects the previous statement ‘you saw a lion’:

This classification makes reference not only to the implicit vs explicit presence of the alternatives in the context, but also to discourse relations. Indeed, Repp (2016) proposes that different degrees of contrast emerge not only from the way in which alternatives are construed, but also from the discourse relations between the contrasted elements. Contrast always implies a relation of similarity or dissimilarity between discourse segments, but additional aspects of the contrastive relation may add to the degree of contrast. The discourse relation involved in an interrogative discourse consisting of a question and a congruent answer is intuitively non-contrastive – or the least contrastive (as with information focus) – although an implicature of exclusion may assign a higher contrastive value to the relation (as with exhaustive focus). The violation of expectations (as with mirative focus) and the incompatibility between two states of affairs (as with corrective focus) feature higher degrees of contrast. Since corrections involve explicit alternatives, corrective focus is even more contrastive than mirative focus, with which the alternatives need not be explicit but can just be situated in the interlocutors’ belief systems and common ground. Indeed, mirative assertions are not corrections.

As pointed out by Repp (2016: 275), “if contrast comes in different degrees we may expect that these degrees correlate with the application of additional or different grammatical means.” This is indeed what happens when we investigate the syntactic effects of focus and, in particular, the association between FF and focus types.3 Given languages appear to selectively associate the syntactic operation FF with distinct types of focus, but not necessarily with all of them. Let us consider the behaviour of FF in our sample of languages:4

As a tendency, and as suggested by the title of this paper,5 the greater the degree of contrast, the greater the potential is for focus to give rise to syntactic effects and, thus, to undergo FF. In Italian and Spanish, FF is possible not only with the ‘greatest’ and most explicit case of contrast, that is, with corrective focus, but also with mirative focus expressing contrast against expectations, while FF is not accepted in the absence of contrast (i.e. with information focus), independently of its exhaustivity.6 In Hungarian, FF is typically associated with exhaustive focus, but is also possible with mirative and corrective foci. In Sicilian and in Sardinian, the occurrence of FF with information focus goes hand in hand with its acceptability with all other types of focus, with the exception of exhaustive focus.

These facts will be reviewed in more details in the next sections, but before turning to the empirical basis of the present proposal, a clarification is in order. Speaking of a hierarchy of contrast or of different degrees may lead to a scalar conceptualization of contrast, as a kind of continuum. Although, conceptually, contrast may be indeed conceived of as a gradable notion, and even if the exact boundaries may sometimes be difficult to define in a clear-cut way, the focus types identified above have nonetheless to be envisaged as discrete categories. We need cut-off points and hence discrete categories to capture the discourse conditions that license FF.

Building on the definition of contrast elaborated in Section 2 and on the resulting classification in (12), in this section the associations between the syntactic operation FF and the different types of focus are presented. Let us start with corrective focus.

In the previous sections, we saw that FF depends on the type of focus and that particular languages vary with respect to the focus types that admit FF. The next question to be addressed is the nature of the meanings that distinguish the different focus types. What we need is an analysis that is able to account both for the different interpretations of the focus types and for the crosslinguistic variation. Following Bianchi, Bocci & Cruschina (2015; 2016), I assume that the different interpretations associated with focus are conventional implicatures (in the sense of Potts 2005; 2007), namely, grammaticalized associations between a linguistic form and a special interpretive import, which need the set of alternative propositions evoked by the focus structure in order to be interpreted correctly. It is thus the specific conventional implicature that is ultimately responsible for the syntactic operation of FF.

The corrective import of corrective focus has been analysed as a conventional implicature in Bianchi & Bocci (2012), and has been redefined as in (48) in Bianchi, Bocci & Cruschina (2015: 12):

Indeed, corrective focus implies that the proposition expressed by corrective claim (49B) is incompatible with one salient alternative proposition, in that it expresses an incompatible description of one and the same event:

The incompatibility between the corrective claim and the antecedent gives rise to the correction effect, typical of corrective focus, whereby speaker A’s assertion is rejected by speaker B. Along the same lines, the mirative import associated with mirative focus is defined as follows (Bianchi, Bocci & Cruschina 2015: 13):16

In (51a), for example, the other focal alternatives belonging to the set ‘they gave me x’ are considered as more likely than the asserted proposition with the focus value denoting a diamond necklace. Being a speaker’s commitment, the mirative conventional implicature cannot be cancelled or denied by the speaker herself. It would thus be odd for the same speaker to continue the mirative utterance with a sentence like (51b):

Unexpectedness is then analysed in terms of the comparative likelihood of alternative propositions (following Grosz 2012) with respect to a contextually relevant modal base (the context set) shared by the speakers and a stereotypical ordering source defining the normal course of events (Kratzer 2012 [1981]).17 On this account, mirative focus evokes a set of alternatives, which provides exactly the alternative propositions that are necessary for a correct interpretation of the mirative implicature.

Most analyses of the syntactic and semantic properties of FF in Hungarian agree that the immediately preverbal position dedicated to fronted foci is associated with an exhaustive interpretation, which is absent when the focus constituent occurs postverbally (see, a.o., Szabolcsi 1981; Kenesei 1986; 2006; Horvath 1986; 2000; É. Kiss 1998; 2002). In answers to questions, Hungarian FF is thus exhaustive, in the sense that it expresses both the identification of the focus value from a set of alternatives and the simultaneous exclusion of other members of the same salient set of alternatives which are identifiable or highly predictable from the context and for which the predicate can potentially hold (cf. § 3.4). Despite this general consensus about the exhaustivity of FF, different hypotheses have been formulated on the source of the exhaustive interpretation (see Pintér 2017 for an overview). On the one hand, exhaustivity has been analysed as the interpretive import of a semantic device, either an exhaustive operator or a semantic presupposition, defending the idea that the immediately pre-verbal focus in Hungarian is semantically exhaustive and, thus, that exhaustivity is part of the truth conditions of the corresponding sentences (Szabolcsi 1981; 1994; Kenesei 1986; Kálmán & van Leusen 1993; É. Kiss 1998; 2006; 2010; Horvath 2007; Bende-Farkas 2009). On the other hand, other approaches reject the semantic nature of the exhaustivity associated with FF and claim that such an interpretive effect simply results from a pragmatic effect or inference (Wedgwood 2003; 2005; 2007; Onea 2007; 2009; Brody & Szendrői 2011; Onea & Beaver 2011; Gerőcs, Babarczy & Surányi 2014).

In line with the latter approach, I defend the hypothesis that the exhaustive interpretation associated with Hungarian FF is an implicature, which is directly responsible for the syntactic displacement of the focus constituent. Importantly, however, I analyse it as a conventional implicature. Unlike conversational implicatures, conventional implicatures are part of the meaning of a word or, as in our case, of a construction, but are not part of the at-issue truth conditions. Even if the term implicature is traditionally associated with the realm of pragmatics, it must be noted that, according to Potts (2005; 2007), conventional implicatures are fully semantic, not pragmatic meanings, even though they pertain to a separate non-at-issue dimension that is independent from the at-issue content (see also Balogh 2013 and Gerőcs, Babarczy & Surányi 2014). Various tests have been proposed in the literature to demonstrate the semantic nature of the exhaustive interpretation associated with FF (see Szabolcsi 1981; É. Kiss 1998; 2002). Crucially, these tests only show that exhaustivity is an essential and constant component of the focus meaning and are not really able to tell at-issue and non-at-issue meanings apart. Insofar as conventional implicatures are not cancellable and, once the conventionalization has taken place, they are a constant part of the meaning, the results of these tests are fully compatible with the hypothesis defended here that exhaustivity is a conventional implicature.

Despite being a prominent one – perhaps the most prominent –, the exhaustive reading is not the only possible interpretation that can be associated with FF in Hungarian. In particular, as shown in Section 3.4, exhaustivity is not a necessary condition for FF in contexts other than answers to wh-questions.18 In this sense, a semantic approach to the interpretation of the fronted focus proves too strong and would make the incorrect prediction that fronted focus is always exhaustive, irrespective of the context.

In sum, all the special meanings that characterize FF are conventional implicatures, which add up to the at-issue content of the sentence. These implicatures need a focal set of alternatives in order to be interpreted correctly and are thus intimately related to focus. The association between one such implicature and the syntactic operation FF is then a matter of conventionalization and may differ from language to language. FF can be associated with a corrective or with a mirative implicature in Spanish and in Italian, but only with a mirative implicature in French (see (27) above). In Hungarian, both the mirative and the corrective implicature may trigger FF in contexts where there need not be any implicit question under discussion or presupposition that could justify an exhaustive identification. By contrast, in answers to wh-questions, exhaustivity is generally associated with focus as the result of a language-specific process of conventionalization (see also Brody & Szendrői 2011; Onea & Beaver 2011; Gerőcs, Babarczy & Surányi 2014).

A syntactic implementation of this account is offered in Bianchi, Bocci & Cruschina (2015) within the cartographic approach. In line with a T-model of the grammar, Bianchi, Bocci & Cruschina (2015) propose that the corrective and the mirative implicature are triggered by active features in the syntactic structure, so that they provide specific instructions to the interfaces. More specifically, they propose that these implicatures are conventionally associated with the activation of a left-peripheral functional projection where the focus-associated implicatures (FAI) are encoded as features. Since these implicatures depend on the availability, in their scope, of a set of focal alternative propositions, this projection has to be higher than FocP. It is ultimately the implicature feature that activates the focus projection in its scope and that thus acts as the syntactic trigger of focus movement. This implementation is illustrated in (52) where the exhaustive implicature has been added as a possible feature of the FAI projection:

This unified analysis of the focus types maintains a uniform notion of focus and at the same times accounts for the different interpretations. Referring to Horvath’s (2007) analysis of the exhaustive interpretation in Hungarian, an anonymous reviewer points out that exhaustive focus may involve a different mechanism than mirative and corrective focus. Indeed, Horvath proposes to analyse the exhaustive interpretation as the result of a null Exhaustive Identification Operator that triggers movement but that does not necessarily associate with focus (see also Horvath 2010). My proposal shares some important assumptions with Horvath’s analysis, in particular, the claim that focus movement must be separated from focus, but it also differs from it in significant respects: a crucial component of my analysis is that the trigger of the movement operation requires a direct association with the focus of the sentence, an association that is less central in Horvath’s analysis. If one accepts Horvath’s account for exhaustive focus, moreover, corrective and mirative focus have to be treated as different phenomena; this would imply the need to postulate different mechanisms to derive focus structures with very similar syntactic and prosodic properties.

The present account relies on the assumption that the different types of focus require a direct association with focus in order to be interpreted correctly, an assumption that is also pervasively present – either implicitly or explicitly – in the specialist literature on exhaustivity (see, e.g., É. Kiss 1998; Kenesei 2006; Krifka 2007). As illustrated in (52), the mechanism ensuring the association with all different types of focus is the same: there is only one focus, the different interpretations come from a syntactically encoded conventional implicature. Another advantage of this unifying account is that it grants a semantic status to the different types of focus (as uncancellable conventional implicatures that are part of the meaning), but it attributes their origins to a conventionalization process rather than to a (language-specific) semantic operation or to a specialized semantic operator. This analysis is thus more flexible and provides a more straightforward explanation to the observed variation.

Information focus introduces new information and it mainly occurs in answers to wh-questions. In a context consisting of a question and a congruent answer, the discourse relation at play with information focus is non-contrastive (Repp 2016) or the least contrastive, insofar as it contrasts with the other alternatives of the focal set. What indeed distinguishes information focus from other focus types is that these alternatives are not necessarily included in the speakers’ common ground, either as previous knowledge or as expectations; nor are they explicitly present in the context. In this sense, information focus constitutes the neutral type of focus which does not involve any implicatures about the alternatives.

The common association between the question-answer context and information focus, however, has led to a frequent misunderstanding. This context is not a sufficient condition for information focus, in that not all foci in answers to questions are necessarily information foci, especially when fronted. The fact that a constituent that is syntactically and/or prosodically marked as focus is used in the answer to a wh-question does not convert per se this focus into an information focus (Escandell Vidal & Leonetti 2019: 208). Other types of focus can occur in the very same context. This misconception of information focus is most probably at the basis of the empirical challenges and controversies about FF in answers to wh-questions discussed in Section 3.3, misleading several scholars both on conceptual and on methodological grounds.

The methodological problems have already been highlighted by Uth (2014) and Escandell Vidal & Leonetti (2019) (see also Uth & García García 2018). In some of the experimental and/or quantitative studies that have challenged the traditional view on the conditions licensing FF in Spanish, written stimuli were employed. As pointed out by Escandell Vidal & Leonetti (2019), in the absence of prosodic information it was not possible to ascertain whether speakers were choosing or accepting as grammatical, depending on the task, a sentence with a marked initial focus subject or rather a sentence with an unmarked SVO order. Even when acoustic information was available (e.g. in production tasks), confirming the prosodic marking of the fronted focal constituent, the interactions between the experimenter and the participants that led to the production of a preverbal focal subject (or to the choice of such an option among different possibilities) were barely natural and therefore inadequate for a genuine question with a neutral information focus in the answer.19 In the tasks adopted in Gabriel (2007; 2010), the participants were shown short picture stories introducing the scenario and the referents that would be involved in the questions. The same pictures were later shown once again together with the questions asking about the scenario and the referents that had already been introduced before (see Uth 2014). Moreover, the participants were explicitly asked to repeat in their answer all the constituents appearing in the question (cf. (53) which repeats the example (35) above). In other words, the background information and the questions were probably too detailed for an interpretation of the question as a neutral question:

Another problem casts doubt on the naturalness of the questions asked following this methodology. An essential property of canonical information-seeking questions is speaker’s ignorance, that is, the pragmatic assumption that the speakers asking a canonical question are not able to commit themselves to any of the possible alternative answers (see, e.g., Farkas 2020). Even if this condition is admittedly difficult to be met in an experimental environment, the fact that the pictures were already illustrating the answer to the question presumably made the interpretation of the question as a canonical question even harder, irrespective of the simulation exercise.

These methodological problems might have thus favoured different interpretations of the focus, rather than eliciting the desired information focus. In particular, mirative focus can also occur in an answer to a question. Consider the Spanish example in (54) and the Italian examples in (55) and (56):

In the answer (56B), for example, the constituent that corresponds to the wh-phrase in the question (i.e. the proper name Temistocle corresponding to the wh-phrase Che nome ‘What name’) is focussed, thus providing a congruent answer to the wh-question in (56A). At the same time, the assertion with mirative FF in (56B) implies that one or more focus alternative propositions would have been more likely in that given context (i.e. other dog names were more expected to be chosen), yielding a sense of unexpectedness and surprise.

Rather than a less likely alternative, the focus constituent in the answer may denote a more likely alternative with respect to the set of focal alternative propositions – or indeed the most likely. In this case, a sense of obviousness is the source of the surprise effect. A natural reading of the answer in (57B) – although presumably not the only one – involves an interpretation of obviousness: it is Juan who buys the newspaper every morning and the hearer expects this to be part of the shared knowledge of the conversation participants. Being surprised at the question itself, the hearer thus replies with (57B) to (re)state the most likely alternative (Juan bought the newspaper, who else could have bought it?):

The import of obviousness is discussed in Uth (2014) and Escandell Vidal & Leonetti (2019) in reference to the marked preverbal foci in the experimental studies on Spanish FF. Indeed, in a context in which the answer is self-evident (cf. (53)), and as a consequence of the adopted methodology and of the nature of the stimuli, “the informants very probably end up with the most likely pragmatic enrichment, i.e. the encoding of the obviousness and/or the certainty of the assertions they make when answering the corresponding Wh-questions. […] The participants are questioned about what they see on the pictures, and even if they manage to properly reenact the ‘game’, the answers are still more than obvious” (Uth 2014: 95).

Uth (2014) analyses this sense of obviousness as the result of evidential and epistemic marking (see also Fließbach 2016; Reich 2018). I would instead like to propose that obviousness results from a subtype of the mirative implicature, whose specific interpretation derives from a reverse order of the expectations. Both unexpectedness and obviousness result from a violation of the expectations. Unexpectedness involves the assertion of a less likely alternatives in contrast to more likely alternatives; obviousness, on the contrary, involves the attempt to assert and restore the most likely alternative, which according to the speaker should be already part of the common knowledge, beliefs, and expectations shared by the interlocutors.

To sum up, what at first sight appears to be an instance of information focus, for the simple reason of occurring in an answer to a wh-question, at closer scrutiny turns out to be in most cases either an unmarked constituent occurring in the basic order or an instance of mirative focus. Dialectal variation is not to be excluded as a factor of variation, but I believe that the simplistic distinction between contrastive and non-contrastive focus, coupled with the methodological problems reviewed in this section have played a major role in the empirical controversies around the distribution of focus in Spanish.

In this paper, I have shown that the theoretical and empirical problems related to the licensing of FF can be better explained if the binary distinction between contrastive and non-contrastive focus is abandoned, and different subtypes of focus are identified (information, exhaustive, mirative and corrective focus). In the sample of languages examined in this paper, focus types associated with greater degrees of contrast are more likely to be marked syntactically and, thus, to admit FF. This is the case not only with corrective focus, which involves explicit alternatives as asserted antecedents and which thus features the highest degree of contrast, but also with mirative focus, which expresses a violation of expectations against alternatives that need not be explicit. This finding is in line with recent analyses of information focus. Outside the Romance languages, and on the basis of evidence from the syntax-phonology interface, Kratzer & Selkirk (2020) argue that in Standard British and American English information focus remains unmarked. The apparent prosodic effects associated with material that is merely new (i.e. information focus) are the result of default prosody.20

Following Bianchi, Bocci & Cruschina (2015; 2016), I have further argued that the different imports associated with FF and the different focus types are to be analysed as conventional implicatures. Indeed, the marked types of focus are associated with conventional implicatures which need a focus structure in their domain to be interpreted correctly. Syntactically, it is one of these implicatures that triggers FF, explaining the relationship between syntactic movement and focus. The crosslinguistic variation can be related to the ‘conventionalization’ of a given subtype of focus with a specific context, as is the case of exhaustive focus in answers to questions in Hungarian.

As observed by an anonymous reviewer, the fact that in some languages FF is compatible with more than one reading could instead favour the hypothesis that such readings are all inferential elaborations of a single meaning which depend on specific contexts. In this sense, these readings would qualify as conversational rather than conventional implicatures. However, even if the interpretations associated with the different focus types are linked to the same syntactic operation, under this account, it is essential that the different implicatures are characterized by different grammatical properties, so as to prove that they are not simply pragmatic variants of the same implicature. Indeed, for Italian, Bianchi, Bocci & Cruschina (2015; 2016) show, on the basis of experimental evidence, that the corrective and the mirative implicature are marked by distinct prosodic contours. A context may well support either implicature, but whether the implicature is corrective or mirative does not (only) depend on the context or on pragmatic conditions; the two implicatures are rather grammaticalized, in the sense that they are associated with distinct formal properties. Native speakers of the other languages considered in this paper (e.g. Spanish, Hungarian) seem to confirm that the different implicatures are characterized by distinct intonational properties, but these intuitions need to be confirmed in a more systematic way and with the aid of experimental procedures that take focus types as a factor. I leave this issue open to future research.

In answers to questions, FF does not necessarily involve information focus, but may involve other focus types such as mirative focus. Indeed, in crosslinguistic research, answers to constituent questions are still “the most widespread and most widely used test for focus” (Van der Wal 2016: 265). As we have seen, however, the new part of an answer to a wh-question might not merely express new information, but it may also involve the violation of expectations or a sense of obviousness. In the case of an imperfect methodology, it could be that the question does not succeed in eliciting the desired information structure, but is rather (incongruently) answered with an unmarked sentence exhibiting the default syntax and prosody.

Since a simple question-answer test does not distinguish between different interpretations of the focus constituent, the available data and findings from the experimental and quantitative studies are somewhat inconclusive and cannot be taken as a real challenge to the traditional view on the distribution of FF. These considerations, together with a series of methodological shortcomings, provide an explanation for the descriptive controversies about the syntactic marking of information focus in languages like Spanish.

In this respect, I would like to conclude this paper with a reflection on the elicitation of information focus in order to combine as many empirical methods as possible and, more specifically, quantitative testing with introspection. But how can we investigate information focus in an experimental or quantitative study? It is indeed not easy to exclude a mirative interpretation of FF in an answer. I only have a suggestion for future studies in this direction. To elicit information focus is necessary to identify contexts that do not easily lend themselves to a mirative interpretation of the focus. This is admittedly not an easy task, but we can think of scenarios in which the interlocutors do not know each other and still exchange questions to seek new information. This could be the case with quiz questions or special types of interrogations such as police interrogations. In these contexts, the interlocutors usually do not share knowledge or beliefs that may generate expectations with respect to the content of the answer, as in the following examples from Italian where FF proves pragmatically infelicitous:21

However, the researcher has to be careful also in these contexts and should avoid situations that could license non-neutral interpretations of the focus. We could think of quiz questions in an informal context, for example at home among friends, which could sound obvious:

In the case of (60), FF proves more acceptable, but would trigger a mirative implicature of obviousness. It goes without saying that, as typical of mirative implicature, whether a proposition is unexpected or obvious depends on the contextually relevant modal base shared by the speakers (cf. § 4). A context in which the interlocutors have no expectations about previous knowledge of Leonardo’s famous painting (e.g. a quiz question for children) would not support such an implicature and would therefore doom the sentence (60A1) to be judged as infelicitous.