Surprise questions in spoken French

1 Introduction

This paper focuses on two wh-questions used non-canonically in French - c’est quoi, ce N (C’est quoi, ce délire?; literally:`It is what, this delusion?’, What the hell is this mess?) and qu’est-ce que questions (Qu’est-ce que tu te plains?; literally: `What you are complaining?’, Why the hell are you complaining?). Previous studies by Dekhissi (2016, 2018) have shown that these questions are dedicated to expressing surprise. However, Coveney and Dekhissi (2017: 188) make a distinction between these two question types in terms of communicative function. They argue that qu’est-ce que questions are “conflictual rhetorical questions” that emphatically assert that there is no reason for p and express “a reproach, complaint or criticism”. By contrast, in situ questions with a right dislocation (c’est quoi ce N) are reported to be “exclamatory questions” because they do not imply an assertion of opposite polarity and do not carry a conflictual overtone (Dekhissi and Coveney 2018). Their claim is that “exclamatory questions” have the same prosody as exclamatives. We argue that both structures convey a judgment of incongruity and have a similar communicative function. Building on Obenauer (2004, 2005, 2006, we regard these two structures as surprise questions (SQs). This claim is substantiated by the results of a production experiment that show that prosodically (i) both constructions significantly differ from string-identical information-seeking questions (ISQs) (ii) in the surprise reading the two constructions share similar features with respect to lengthening, speech rate and final contour. This suggests that in French, SQs can be defined as a question type alongside ISQs on the basis of a mapping between syntactic-prosodic features and functional meaning.

2 Surprise questions based on previous accounts

Our definition of surprise questions relies on two previous approaches that concur in acknowledging the non-canonical nature of these questions. In a constructional approach to “What’s X doing Y”, Kay and Fillmore (1999) have argued that the function of this construction is to attribute “a judgment of incongruity” to a state of affairs:

In and of itself, the question in (1) is ambiguous between an information-seeking question and a surprise question. However, in the scenario suggested in (1), no information concerning the fly’s activity is expected from the addressee, hence the joke resulting from the informative answer provided by the waiter, as pointed out by Kay and Fillmore. In its intended meaning, the surprise question in (1) is semantically non-canonical as the wh-word is not referential.

In a syntactic approach, Munaro and Obenauer (2002) and Obenauer (2006, 2005, 2004 have identified “surprise-disapproval questions” as a special type of question in Bellunese dialects. They argue that these questions differ syntactically from ISQs, and also from rhetorical questions (RQs) and wh-exclamatives. They express the speaker’s attitude towards the propositional content. This attitude is one of “surprise with a negative orientation (disapproval)” (Obenauer 2005: 58). Obenauer points out that the surprise-disapproval meaning weakens the information-seeking value. In Bellunese, the form of this question indicates that it can only be understood as a SQ, not as an ISQ. Obenauer further notes that cossa (what), as opposed to che (what), can be used non-argumentally meaning why:

In French, non-argumental qu’est-ce que encodes a SQ in a similar way:

However, there is no clear-cut syntactic difference between an information-seeking and a surprise interpretation in argumental qu’est-ce que questions in French. As what in (1), argumental qu’est-ce que is ambiguous between a referential information-seeking reading and a non-referential surprise reading:

We assume that prosodic cues allow differentiating between the two meanings in French and present the results of a production experiment in Section 5. The surprise question arises from a conflict between an expectation and a state of affairs that disconfirms this expectation. It encodes a judgment of incongruity and requests an explanation to resolve an epistemic conflict.

Adopting a sociolinguistic perspective, Dekhissi (2016) and Coveney and Dekhissi (2017) have argued that non-argumental qu’est-ce que questions (3) are “conflictual rhetorical questions”. In these questions, qu’est-ce que is not a direct object. Its meaning is close to why, except that qu’est-ce que “implicates (emphatically) an assertion of the opposite polarity” (Coveney and Dekhissi 2017: 189), which can be paraphrased as “there is no reason or purpose for p”. While pourquoi asks a question about the cause for p, non-argumental qu’est-ce que also expresses the speaker’s negative attitude to the content. This negative attitude may account for a rhetorical flavor, as noted by Coveney and Dekhissi (2017). Yet, the question requests an explanation. This suggests that this type of question is different from RQs:

The RQ in (5) implies an assertion of the opposite polarity (ça ne change rien/It does not make any difference). No conflict arises between the speaker’s prior expectations and the situation. The variable is bound by the wh-interrogative as in other canonical questions. RQs are defined by Rohde (2006) as redundant interrogatives that rely on common ground. As pointed out by Alcázar (2017) and Celle (2018), SQs are a distinct category since the incongruity judgment is not shared by speaker and addressee. SQs convey the speaker’s reaction to a situation that runs counter to expectations. Therefore, we view these questions as biased rather than rhetorical. “Surprise question” is a label we choose for the sake of convenience. The negative bias associated with thwarted expectations accounts for the sense of disapproval and indignation that is often carried by these questions. We argue that the non-referentiality of the wh-interrogative is evidence of the non-canonicity of SQs.

3 Non-canonical c’est quoi, ce N ?! - a surprise question

In addition to non-argumental qu’est-ce que questions, Dekhissi (2018) has identified another pattern dedicated to expressing surprise: c’est quoi/qui, ce N (it is what/who, this N), i.e. an in situ question followed by a right dislocation.

Dekhissi (2018) labels this pattern “exclamatory question” because it is supposedly closer to exclamatives in terms of meaning. She claims this structure can be associated with a positive emotion:

As stressed by Dekhissi (2018), both non-argumental qu’est-ce que questions and c’est quoi ce N questions are expressive questions. Based on prosodic cues, we suggest in Section 5 that these questions form a paradigm of SQs that is functionally different from ISQs.

As such, C’est quoi, ce N is an in situ question about the definition of the subject complement.^[1] The nominal referent is designated by the demonstrative ce in the right dislocation and coindexed with the clitic pronoun c’. Two ingredients are necessary for the detachment construction to take on a surprise meaning: a right dislocation and a demonstrative. In (8), the right-dislocated phrase contains a noun that is unambiguously interpreted figuratively in that syntactic position:

The demonstrative designates a state of affairs the speaker has direct experience of. This state of affairs is judged to be incongruous because it violates the speaker’s expectations. In the right dislocation, the noun délire (delusion) cannot be interpreted in its literary sense. The following answer would not be felicitous:

The construction in (8) conventionally expresses the speaker’s surprise and her negative evaluation of the state of affairs. As argued by Stefani and Horlacher (2017), the right dislocation has become a routinized structure to express evaluation grounded in immediate sensory evidence. The right dislocation blocks the information-seeking reading of the in situ question. In other words, quoi cannot be interpreted referentially. The question will request an explanation if the nominal lexeme can be taken to designate an odd situation (8) or if the noun is postmodified by a PP as in (6), (7) and (9):

In (9), the addressee’s activity is judged to be incongruous. It follows from the incongruity judgment that the evaluation is a negative one.^[2] Even in contexts that might look positive at first sight, such as a surprisingly fortunate situation (7), this construction is pragmatically different from an exclamation.^[3] In (7), the speaker expresses bafflement at her friend’s good fortune, which induces a search for an explanation. Similarly in (9), surprise arises as a result of a belief discrepancy between the perceived event (i.e. someone kicking molehills in a high-mountain Swiss pasture) and the speaker’s background knowledge. The question indicates both this inconsistency and the speaker’s request for an explanation. Once an explanation has been provided, the unanticipated perceived event can become part of the speaker’s expectation sets and induce belief revision in an adaptive process (Meyer et al. 1997).^[4]

4 Prosody in non-canonical questions

Although no studies to our knowledge have specifically examined the prosody of surprise questions, several papers have been published on rhetorical questions in a variety of languages – English (Dehé and Braun 2019), German (Braun et al. 2019; Wochner et al. 2015), Icelandic (Dehé et al. 2018), and French (Delais-Roussarie and Beyssade 2019). They all followed the same protocol, and consistently found prosodic differences between ISQs and RQs. The wh-word is pronounced with a softer voice quality (Dehé and Braun 2019; Wochner et al. 2015) and with a lower pitch (Delais-Roussarie and Beyssade 2019; Wochner et al. 2015) in RQs than in ISQs. ISQs are uttered with a higher F0 key and a greater pitch range than RQs (Delais-Roussarie and Beyssade 2019). In French, ISQs are realized with a rising contour most of the time, while more different final contours are observed for RQs (Delais-Roussarie and Beyssade 2019). Duration seems to be a particularly reliable indicator as well: RQs have longer utterance durations (Dehé and Braun 2019; Wochner et al. 2015) and a slower speech rate (Delais-Roussarie and Beyssade 2019), and several constituents are longer: the first word and nuclear syllable (Dehé et al. 2018), and the questioned part (Delais-Roussarie and Beyssade 2019). Although no study has yet compared string-identical ISQs and SQs, Lai (2009) found that right and really are realized with a larger pitch range when they express surprise in isolation, although, in the case of really, surprise cannot be dissociated from the questioning semantics. A study directly comparing the realization of ISQs and SQs is therefore of particular interest.

5 Experiment

We conducted an experiment to test whether the surprise and the information-seeking readings of string-identical questions with C’est quoi and Qu’est-ce que would differ in terms of prosodic features. We also wanted to determine whether the surprise reading of the two constructions exhibits similar prosodic cues. In the surprise condition, we expected surprise to be realized through expanded pitch range (Lai 2009) in both constructions, and possibly through lengthening, in a way similar to RQs.

5.2 Results

5.2.1 Whole sentences

Duration (see Table 1; Figure 1a; Appendix 2.1.1): There was a significant Condition:Question word interaction (χ ² (1, 14) = 11.65, p = 0.0006, p _adj  = 0.002): with C’est quoi, the utterance lasted longer for SQs than ISQs (β = 0.15^[8], SE = 0.03, t (30) = 5.81, p/p _adj < 0.0001).

Table 1:

Mean duration and speech rate for the whole sentences with each of the question words (standard deviation in parentheses).

	Duration (ms)		Speech rate (syllables/second)
	C’est quoi	Qu’est-ce que	C’est quoi	Qu’est-ce que
Information-seeking	928 (205)	1,177 (475)	5.19 (1.11)	5.54 (1.01)
Surprise	1,084 (234)	1,211 (426)	4.48 (1.12)	5.35 (1.05
Difference	156***	34	0.71***	0.19

1. ***p < 0.001, **p < 0.01, *p < 0.05, p < 0.1.

Figure 1:

Duration and pitch range differences between the two question conditions for the whole sentences and the target syllables of each question word.

Speech rate (see Table 1; Appendix 2.1.2): There was a significant Condition:Question word interaction (χ ² (1, 14) = 11.63, p < 0.001, p _adj = 0.002): The effect of Condition was only significant with C’est quoi (β = −0.15, SE = 0.03, t (30) = −5.80, p/p _adj < 0.0001): The speech rate was faster for ISQs (5.19 syllables/s) than SQs (4.48 syllables/s).

Pitch (see Figure 1b, Appendices 2.1.3 and 2.1.4): There was no effect of Condition on the mean pitch or the pitch range.

Rising final contour (see Appendix 2.1.4): There was a significant main effect of Condition (χ ² (2, 13) = 14.85, p < 0.001, p _adj = 0.002) on the proportion of rising final contours: They were more common for ISQs (39.17%) than SQs (13.47%).

5.2.2 Target syllables

Duration (see Table 2; Figure 1c and d, Appendix 2.2.1): There was a significant interaction between Condition and Syllable with C’est quoi (χ ² (1, 14) = 9.52, p = 0.002, p _adj = 0.005). SQs were characterized by a longer duration of the first (β = 0.14, SE = 0.035, t (28) = 3.95, p = 0.0005, p _adj = 0.002) and second syllable (β = 0.24, SE = 0.04, t (28) = 6.68, p/p _adj < 0.0001) of the question word. The syllable Qu’est-ce was also longer in SQs than ISQs (χ ² (1, 9) = 7.72, p = 0.005, p _adj = 0.017).

Table 2:

Mean duration in ms for each target syllable (SD in parentheses).

	C’est	Quoi	Qu’est-ce
Information-seeking	174 (55)	180 (58)	191 (54)
Surprise	202 (67)	229 (92)	225 (79)
Difference	28***	49***	34***

1. ***p < 0.001, **p < 0.01, *p < 0.05, p < 0.1.

Mean Pitch (see Appendix 2.2.2): There was no effect of Condition for sentences with C’est quoi (p > 0.1) or with Qu’est-ce que (p > 0.1).

6 Conclusion and perspectives

We have established that SQs differ from ISQs with respect to lengthening, speech rate and final contour. Our next step will be to conduct a perception experiment to check whether these prosodic features are necessary and sufficient for a listener to make the distinction between ISQs and SQs. We will also compare our spoken data from ESLO and OFROM with our production data.

More detailed prosodic annotations would be of interest in our data for a more fine-grained analysis of prosodic contour. It would be relevant to compare the prosodic realization of SQs with that of RQs and exclamatives to determine whether they share prosodic features. Lengthening in SQs is reminiscent of rhetorical questions. As noted by Delais-Roussarie and Beyssade (2019a), lengthening is a distinctive feature of RQs in French. In German, recent studies (Braun et al. 2019; Wochner et al. 2015) have also shown that RQs have longer duration than ISQs. We might speculate that lengthening characterizes both SQs and RQs, which differ with respect to pitch range. However, in order to ascertain whether lengthening is a correlate of both expressivity and rhetoricity, RQs would have to be included in the experimentation.

Corpora

ESLO: Enquêtes Sociolinguistiques à Orléans

(Abouda and Skrovec 2018)

OFROM: Corpus Oral de Français de Suisse Romande

(Avanzi et al. 2012–2020)