This special issue is dedicated to the syntax-prosody interface in non-canonical questions and originated in the international workshop Non-Canonical Questions at the Syntax-Prosody Interface, organised at the Université Paris Cité and held online in November 2020. Recent research has demonstrated that the phonology-syntax relation cannot solely account for prosodic structure, prosody being closely intertwined with discourse organisation, information structure and focus structure (Gussenhoven Reference Gussenhoven1983, Féry Reference Féry, Féry and Sternefeld2001). Questions are a case in point, as they crucially call on the addressee before a proposition may be added to the common ground.
The meaning of a question is generally considered to be the set of propositions that can answer it (Hamblin Reference Hamblin1973). By contrast, the denotation of a declarative sentence consists in a unique alternative. Building on Roelofsen & Farkas (Reference Roelofsen and Farkas2015), the denotation of an interrogative sentence may be viewed as non-informative, whereas the denotation of a declarative sentence is non-inquisitive. Interrogative sentences result in an inquisitive context, while declarative sentences are proposals to update the common ground (Farkas & Bruce Reference Farkas and Bruce2010, Stalnaker Reference Stalnaker2002). The distinction between canonical and non-canonical questions goes on the assumption that canonical questions are unmarked (Farkas Reference Farkas2020), while non-canonical questions are marked because they involve a mismatch between clause type and speech act and, therefore, distort the clause type-speech act relationship. Pragmatically, they are not used to request information. They flout the principle that the speaker is ignorant as opposed to the addressee. Syntactically, non-canonical questions may not have an unmarked interrogative form, but this is not a necessary condition. In English, for instance, they may rely on the declarative structure followed by a tag. Prosodically, we assume that non-canonicity is marked. It is the purpose of this special issue to investigate how prosody and syntax interact in marking non-canonical questions.
Recent developments in inquisitive semantics (Ciardelli, Groenendijk & Roelofsen Reference Ciardelli, Groenendijk and Roelofsen2013) have proposed a unified treatment of sentences by capturing the inquisitive and informative contents of both interrogative and declarative sentences. Farkas & Roelofsen (Reference Farkas and Roelofsen2017) have further laid out the desiderata to treat rising declaratives and tag interrogatives in English in terms of the semantics and the conventional discourse effects contributed by their respective forms. On their account, rising and falling intonation signals differences in the speaker’s credence in a highlighted alternative. The conventional discourse effects of non-canonical questions are connected to their semantics, syntax and intonation. However, this account raises theoretical issues. First, it assumes that rising intonation is a uniform parameter that encodes questioning. Second, this model ‘predicts that biased questions can only be expressed by polar interrogatives’ (Farkas Reference Farkas2020: 18). In that framework, biased questions are associated with a unique highlighted state. The contributions that make up this volume show that the picture is much more nuanced as regards rising intonation. They also demonstrate that surprise questions are a type of content question that involves bias, which is marked prosodically. Prosodic features may help to disambiguate between biased questions and rhetorical questions. Furthermore, a number of theoretical insights can be gained by taking an experimental approach to the syntax-prosody interface.
Cross-linguistically, the difference between questions and assertions is often marked prosodically. Phrasing patterns are related to the information structure and the illocutionary force of a sentence. Tonal patterns allow distinguishing between assertions and questions when the syntax is ambiguous. In French, for instance, declarative sentences are generally reported to be either assertions or questions, depending on their intonation. In the case of questioning declaratives, a rising contour is responsible for conveying the question’s meaning (Delattre Reference Delattre1966, Di Cristo Reference Di Cristo, Hirst and Di Cristo1998). However, recent studies on intonation suggest that questions are compatible with several tonal patterns. Delais-Roussarie et al. (Reference Delais-Roussarie, Post, Avanzi, Buthke, Di Cristo, Feldhausen, Jun, Martin, Meisenburg, Rialland, Sichel-Bazin, Yoo, Frota and Prieto2015) argue that nuclear configurations are not determined by clause types. Rather, there is evidence that tonal patterns convey information about both the speaker’s attitude and the speaker-addressee relationship by specifying whether the speaker commits to the truth of the content or whether this responsibility is attributed to the addressee (see also Bartels (Reference Bartels1999) and Gunlogson (Reference Gunlogson2003) for similar claims on English). The nature of a rise will thus vary depending on how the speaker commits to a proposition and what they expect of the addressee.
In Romance languages, the declarative clause type may be used to form both canonical questions and non-canonical questions. In French, Delais-Roussarie (Reference Delais-Roussarie, Fischer and Gabriel2016) points out that the questioning declarative ‘Tu as faim?’ (you are hungry? ‘Are you hungry?’) may either be used to seek information or to convey the speaker’s incredulity. In the non-canonical use, a higher pitch encodes the speaker’s incredulity in French (Michelas, Portes & Champagne-Lavau Reference Michelas, Portes and Champagne-Lavau2013) as well as in Catalan (Crespo-Sendra, Vanrell & Prieto Reference Crespo-Sendra, Vanrell, Prieto and Hasegawa-Johnson2010). A wider pitch range may be assumed to be indicative of the speaker’s attitude in non-canonical questions.
Turning to surprise questions, their non-canonicity has been amply described in syntactic terms. Semantically, these questions convey surprise and disapproval, which weakens their information-seeking function. In Bellunese, they were defined as a question type in its own right by Obenauer (Reference Obenauer, Lohnstein and Trissler2004, Reference Obenauer2005, Reference Obenauer, Doetjes and González2006) on the basis of the following syntactic distinctive properties: they require a bare wh-interrogative in an initial position; the pronoun cossa replaces che (both meaning ‘what’), and they allow in-situ che to double with sentence-initial cossa. Cossa may be used argumentally or non-argumentally. In the latter case, cossa expresses both the speaker’s surprise and their attempt to search for a cause. Munaro & Obenauer (Reference Munaro, Obenauer, Leonetti, Soriano and Vidal2002) and Obenauer (Reference Obenauer, Lohnstein and Trissler2004, Reference Obenauer2005, Reference Obenauer, Doetjes and González2006) claim that surprise-disapproval questions must be distinguished at the speech-act level from both information-seeking questions and rhetorical questions on the one hand and exclamations on the other hand. They formulate this claim by means of the so-called ‘cartographic’ framework, which represents discourse-related meaning components as left-peripheral syntactic projections (Rizzi Reference Rizzi and Haegeman1997, Reference Rizzi and Reboul2014). Munaro & Obenauer (Reference Munaro and Obenauer1999) propose that the surprise-disapproval effect is encoded in a functional projection other than the interrogative and exclamation force. Drawing upon these findings, Celle & Pélissier (Reference Celle and Pélissier2022) have investigated the syntax-prosody interface in surprise questions in French. They argue that the negative bias associated with thwarted expectations accounts for the sense of surprise and disapproval and demonstrate that surprise questions differ from string-identical information-seeking questions with respect to lengthening, speech rate and final contour. Lengthening seems to be a prosodic correlate of surprise questions that is shared with rhetorical questions (Delais-Roussarie & Beyssade Reference Delais-Roussarie and Beyssade2019) in French and in German (Wochner et al. Reference Wochner, Schlegel, Dehé and Braun2015, Braun et al. Reference Braun, Dehé, Neitsch, Wochner and Zahner2019). By adopting the view that rhetorical questions are an extreme case of biased questions (Beyssade & Marandin Reference Beyssade and Marandin2010), in the sense that they contain their own resolution, one might predict that there will be some overlap between the prosodic realizations of surprise questions and rhetorical questions. At the same time, the conflict that arises from the disconfirmation of prior expectations may be expected to be marked only in the prosody of surprise questions.
In sum, there appears to be no direct mapping between prosody and syntax. Nonetheless, recent findings at the syntax-phonology-prosody interface suggest that offering a compositional account of intonational meaning is an attainable goal, provided an abstract phonological representation level is assumed (Portes & Beyssade Reference Portes and Beyssade2015). Tonal configurations may be conceived of as primitives of intonation structure and should be connected to the semantics of dialogue. This special issue aims to determine to what extent the syntactic structure of questions correlates with prosody and how prosody mirrors the speaker’s level of commitment, their anticipation of the addressee’s (non-)commitment in view of contextual evidence and their call on the addressee.
Contributions to this collection
Eva Liina Asu, Heete Sahkai and Pärtel Lippus focus on a specific type of non-canonical questions in Estonian, that is, surprise questions, and show that these questions have a formally entirely canonical interrogative syntax. The only difference between information-seeking questions and these questions is prosodic. By comparing elicited string-identical surprise questions and information-seeking questions, Eva Liina Asu, Heete Sahkai and Pärtel Lippus show that they differ in terms of duration characteristics, pitch accents and pitch range. In particular, pitch accents are distributed differently, with an accentuation of deictic elements in surprise questions only. This implies that there is a difference in information structure between surprise questions and information-seeking questions and that surprise questions are a distinct speech act.
This specific illocutionary force of surprise questions is further supported by studies examining other languages, where surprise questions can also take non-canonical syntactic features. Carolina González and Lara Reglero compared intonational features of Spanish canonical and non-canonical wh- in situ questions, expressing an information request and repetition or surprise, respectively. Although there were some prosodic similarities between the two question types, they also found higher contour variability and lower tonal range in canonical than non-canonical questions, in particular, surprise questions, as well as a specific upstepped final boundary tone. Interestingly, surprise questions involved more specific prosodic features than repetition questions.
Two of our collected papers looking at non-Indo-European languages show that non-canonicity can be marked by mixed intonational features and more declarative syntactic properties and propose a more detailed syntactic analysis of those properties.
Duk-Ho An investigates stranded embedded clauses in Korean, a structure which can be interpreted as an assertion or a question requesting confirmation, with undertones of surprise and disbelief. The precise illocutionary meaning of the sentence is determined by its prosody. Duk-Ho An argues that clause roots include functional categories encoding speech act properties, which are realized prosodically. The syntactic derivation of stranded embedded clauses causes the characteristic rising/falling intonation to be displaced.
Basque has a specific discourse particle, ba, to turn a canonical polar question into a knowledge confirmation question, as demonstrated by Aitor Lizardi Ituarte. This structure has declarative syntactic properties and intonational characteristics of both questions and declaratives. Aitor Lizardi Ituarte claims that the mixed intonational properties of these non-canonical questions are related to the varying degree of commitment of the speaker towards the proposition and the meaning of the discourse particle. This paper also argues for a distinction between clause type and speech act, reinforcing the claim that non-canonical surprise questions belong to their own category.
Finally, one of the collected papers examines rhetorical questions, enabling us to make direct comparisons between these non-canonical questions and more surprise-oriented ones. Nicole Dehé, Daniela Wochner and Marieke Einfeldt found characteristic intonational properties for German rhetorical questions, in particular, an (L+H*) pitch accent and show that rhetorical properties can be expressed prosodically only with a canonical interrogative structure or in conjunction with lexical means, such as discourse particles. Eva Liina Asu, Heete Sahkai and Pärtel Lippus found similar differences between information-seeking questions and rhetorical questions in Estonian, and the latter are only partially similar to the characteristics of surprise questions, justifying the qualification of surprise questions and rhetorical questions as separate speech acts from canonical information-seeking questions, despite similar syntactic features. The emotional expressivity of surprise questions is reflected in their wider pitch range, while rhetorical questions have a narrower pitch range than information-seeking questions.
Together, these collected papers show that non-canonical questions are marked by specific prosodic features across languages, which reflect the speaker’s commitment to the content of the utterance. They confirm that surprise questions are a specific category of non-canonical questions, differing from rhetorical questions and repetition questions, although some common characteristics distinguish these non-canonical questions from canonical information-seeking questions. These papers also show that studying prosodic characteristics of non-canonical questions has implications for our understanding of the clause type – speech act relationship and for theoretical syntax, demonstrating the promise of this line of inquiry.
