Verbum – Analecta Neolatina XXVII, 2026/1

ISSN 1588-4309; https://doi.org/10.59533/Verb.2026.27.1.8



Abstract: This paper outlines an approach to determine the prosodic characteristics of a discursive marker (tu vois ‘you see’ and its past tensed variant t’as vu ‘you saw’) commonly used in different non-standard French variations, and the eventual dialectal similarities or differences between authentic speakers from the suburbs of Paris and Marseille. This study consists of a descriptive and a comparative analysis leaning on spontaneous speech samples. These samples provide a rich source of data, enabling us to make two key assertions. Firstly, our examination of the prosodic features of the discourse marker reveals a notable consistency among them. Secondly, our comparative analysis highlights an intriguing finding: despite examining two distinct dialects, we observed no significant difference between the suprasegmental features present in each. This implies a level of similarity or convergence in the overarching suprasegmental aspects of speech between the two dialects under investigation.

Keywords: comparative phonology, prosodic features, French dialects, regional variation, non-standard speech, discourse marker, semantic-enunciative function



1 Introduction

When a speaker expresses themselves in spontaneous speech or written discourse, it is often observable that they employ particular words or constructions that serve interpersonal functions. These may act to convey the speaker’s attitude, capture or direct the interlocutor’s attention, or emphasize the significance of the message. In this regard, Pallaud & Bertrand (2020) argue that certain inserted elements, which they refer to as auto-interruptions, are methodologically significant. Their distinguishing feature lies in their syntactic independence, as they are not structurally linked to either the preceding or following segments of the utterance. In seeking the taxonomic category that various scholars (Andersen 2007, Bolly 2009, 2010, 2012, Détrie 2010, Pallaud & Bertrand 2020, Lee et al. 2020, Rouanne 2022, among others) associate with these functions, one inevitably encounters discourse markers – an extensively studied and long-debated linguistic phenomenon.

For various reasons it is far from straightforward to navigate the realm of discourse markers. Indeed, it may be precisely this very challenge that has spurred a multitude of linguistic investigations, whether conducted from diachronic versus synchronic perspectives or concentrating on morphosyntactic or semantic-pragmatic dimensions. In tracing the roots of this issue, one arrives at the work of Dostie & Pusch (2007), who initially examined the extent to which discourse markers can be regarded as a distinct grammatical class. To verify the legitimacy of this proposed category, the authors enumerate a series of characteristics (pp. 3–4) that effectively serve as a working definition of discourse markers. Although the debate remains ongoing, in this paper we endorse the perspective that offers clarification on the nature of these markers, a stance that aligns with Brinton’s definition: “A pragmatic marker is defined as a phonologically short item that is not syntactically connected to the rest of the clause (i.e., is parenthetical), and has little or no referential meaning but serves pragmatic or procedural purposes” (Brinton 2008: 1). Moreover, through a carefully compiled bibliography, Dostie & Pusch highlight that consensus in this area is emerging only gradually, which is further fragmented by the diversity of approaches and the variation in definitions.

Even if we accept the definition proposed by the author, this is by no means exhaustive. Discourse markers are characterized by significant terminological diversity, which leads to the frequent appearance of varied expressions in linguistic studies. Researchers have encountered designations such as the English (reduced) parenthetical clauses (Schneider 2007), epistemic parentheticals (Dehé & Wichmann 2010), comment clauses (Brinton 2008, Bolly 2009), occasionally discourse particles (Lee et al. 2019) or in some cases verbal fillers (Bolly 2009), as well as the French incise (Cornulier 1978), propositions parenthétiques (Andersen 2007), marqueurs discursifs propositionnels (Andersen 2007), construction à verbe recteur faible (Blanche-Benveniste & Willems 2007), and even ponctuant (Vincent 1993).

To mitigate this inconsistency, Bolly (2009: 107) adopts the term construction parenthétique ‘parenthetic construction’, asserting that it is delineated by distinctive pragmatic and functional criteria. She contends that the multifunctional and multi-categorical nature of these markers necessitates such a designation – a claim she validates by exploring the evolution of parentheticals alongside the process of pragmaticalisation, as illustrated by the discourse marker tu vois ‘you see’. This comprehensive and multifaceted analysis has provided a robust foundation for our own research.

Given the growing body of literature that underscores the importance of the French tu vois construction (or its variants, e.g., t’as vu ‘you saw’) as a discourse marker, our study seeks to further elucidate its underlying mechanisms. Our objective is not to portray the developmental stages of the marker or to scrutinize the pragmaticalisation process; rather, we aim to illuminate the various pragmatic and phonological roles it fulfills. As we shall see, several functions can be ascribed to the tu vois construction, for instance we might say that it is an attention-management tool intricately linked to a range of conversational activities.

As we have noted, the literature does not merely offer an uncountable array of terms, often at odds with one another. Andersen (2007: 13–14) presents six arguments to support the claim that constructions such as tu vois, that is, propositions employing a verb devoid of its rective capacity, are grammaticalized as discourse markers. In the course of grammaticalization, these markers secure fixed positions within the clause, with each position corresponding to distinct functions. Andersen also writes that this marker mainly signals an invitation for the other communicative partner to actively participate in the speaker’s reasoning, thus reinforcing the understanding. In this manner, it functions not just as a signal of the intended message of the speaker but as also an appeal to the assent of the other, thereby fostering an interactive communicative dynamic. This dual function is further underscored by paraphrastic forms such as tu vois ce que je veux dire ‘you see what I mean’, which reiterate the call for mutual comprehension.

Regarding the semantics of the tu vois construction, the literature is already considerably advanced. In this context, Bolly’s work is once again indispensable (2012); she argues that the semantic shift of tu vois primarily emerged from Modern French, a period marked by a notable increase in abstract cognitive usages and pragmatic functions of parentheticals. This is consistent with what Détrie (2010) stated, who simultaneously draws our attention to the marker’s role and functions. In its initial position, tu vois operates as a reactivation of the interlocutor’s attention – a form of actantial inversion in which the speaker invites the listener to share his perspective by presenting it as mutually endorsed. In medial positions, tu vois appears to reintroduce discursive antecedence and facilitate the cognitive alignment among the conversational partners. Finally, in the utterance-final position, the marker seems to question the communicability of the speaker’s message.

Rouanne (2022) further contributes to this framework by positing that the tu vois discourse marker functions at a metadiscursive level, therefore conveying the speaker’s attitude or commentary on his own discourse. She distinguishes two types of tu vois: in one, the marker links utterances that are interpreted as arguments supporting a conclusion, whereas in the other, it serves as an evaluative remark on a preceding discourse.

In this study, we aim to examine the tu vois construction through the lens of the diasystem, focusing on its prosodic characteristics. Additionally, we endeavor to compare two regional variations of French. To achieve this successfully, it is essential to closely analyze the works of Anzai (2009), Avanzi (2014), Lee et al. (2020), and Pálvölgyi (2020, 2022).

Anzai (2009), among others, examines the intonational behavior of the tu vois marker to identify its discursive and enunciative characteristics, as he extends the theoretical framework developed by Morel & Danon-Boileau (1998). Their findings indicate that an increase in intensity signals a claim to the right to speak and maintain control of the discourse, whereas a decrease or a lower intensity level suggests that the speaker is stepping back or allowing another participant to take the floor. Similarly, Lee et al. (2020) examined the pragmatic functions of three discourse markers, subsequently analyzing the distribution of pauses around discourse markers according to their pragmatic function.

At this point, we wish to emphasize the methodological rigor of Pálvölgyi (2022), whose approach serves as a foundation for the methodology outlined later in this study. Furthermore, in a separate article (2020), Pálvölgyi conducts a comparative analysis of Spanish dialects to highlight distinctions between them. She categorizes European Spanish into northern and southern varieties before examining the duration of filled pauses and prolongations. Although she did not succeed in identifying significant differences, we argue that his approach to regional classification can be adapted to other languages and linguistic contexts.

Avanzi (2014) stands as a valuable reference in this regard, as he conducted an extensive investigation into regional prosodic variations in French. His research is grounded in the fundamental premise that “the fact that French is not as dialectally fragmented as other languages spoken in Europe does not mean it is uniform across the different areas where it is spoken” (2014: 309). Theoretically, Avanzi classifies France into two main regions: NoF (North of France) and SoF (South of France), asserting that speakers from the North of France are characterized by the least pronounced regional accent, with their pronunciation aligning most closely with “Reference French” (Avanzi 2014: 310).

We build upon his observation that despite the renewed focus on prosodic regional variation in French, our understanding of how prosody contributes to differentiating these distinct accents remains limited. Moreover, he critiques prior studies for their predominant focus on Paris, which he considers a limitation in the broader examination of French prosody. In response, Avanzi expanded his scope to three Francophone countries (Belgium, France, and Switzerland) dividing them into 14 distinct regions. His analysis centered on four prosodic features: AP metrical length, the proportion of initial accents in disyllabic lexical words, articulation rate, and pitch span.

Building on the success of Avanzi’s research, we have been inspired to similarly investigate the prosodic diversity of French. The aim of our study is to assess whether his assertion still holds true on an alternative corpus a decade later. We hypothesize that the function of the selected discourse marker primarily determines its prosodic characteristics, thereby rendering dialectal or regional differences less pronounced. Accordingly, we will analyze and compare measurable features, specifically, pitch, intensity, and duration of the tu vois construction across a bifurcated corpus.

2 Corpus

Our corpus consists of 17 videos (in total 1 hour, 9 minutes, 11 seconds) involving 8 rappers, all of whom were in their 20’s, 30’s (the oldest rapper was born in 1981 and the youngest in 2000). The corpus was sampled in consonance with the 2 most salient, geography-based French variations, thus 4 speakers coming from different departments of the suburbs of Paris which highly correlates with the Parisian communication, and 4 other rappers with Marseille as birthplace which reflects language use in the South of France. Our selection of rappers was purposeful, because we anticipate that they may employ a vocabulary rich in words, expressions, and markers that are emblematic of non-standard linguistic variations. By choosing these specific individuals, we aim to explore and analyze the unique linguistic features present in their lyrical content. These artists are likely to utilize colloquialisms, slang terms, and regional dialectical expressions. The table below (Table 1) shows the details of the corpus (for further information see Appendix).

Table 1: The parameters of the corpus centered around the speakers (the age, the place of birth, the name and date of the interview, and the number of the analyzed utterances

The name of the rapper (birth name) Date of birth (the age of the rapper at the time of the interview) Place of birth Analyzed videos Date of release of the video Number of “tu vois”
Hornet la Frappe (Mounir Ben Chettouh) 1992 Seine-Saint-Denis Konbini – Lexique de la Street (Hornet la Frappe) Konbini – Passion (Hornet la Frappe) 2020 4
Koba LaD (Marcel Junior Loutarila) 2000 Essonne Konbini – Lexique de la Street (Koba LaD) 2019 1
Timal (Ruben Louis) 1997 Seine-Saint-Denis Lexique de la Street (Timal) Kultur – Posé avec Timal 2020 3
Zikxo (?) 1995 Seine-Saint-Denis Lexique de la Street (Zikxo) 2019 6
L’Algérino (Samir Djoghlal) 1981 Marseille Lexique de la Street (L’Algérino) RAPRNB – Son 1er telephone, SCH, son Emoji, Zidane, Anakin Skywalker 2021 2
Naps (Nabil Arlabelek) 1986 Marseille Lexique de la Street (Naps) Konbini – Stylé (Naps) Konbini – 8 choses à savoir (Naps) 2019 3
Soso Maness (Sofien Hakim Manessour) 1987 Marseille Lexique de la Street (Soso Maness) GQ France – Soso Maness juge le rap français 2020 4
TK (?)  ? Marseille Lexique de la Street (TK) Booska-P – Wesh (TK) 2020 8

We must highlight some methodological aspects regarding both the conduct of the interviews and the characteristics of the interviewees. First, with the exception of one case (Booska-P – Wesh (TK)) all interviews were registered in a noiseless environment, furthermore, the video recordings were edited in a way that the interviewer’s questions were removed (although these questions are provided in written form), so the entire video consists of the rappers’ statements. In order to avoid the silent pauses between the questions, they linked the parts of the interview with music which did not deflect the measurements of the targeted prosodic features. We must note that, in certain cases, the tu vois discourse marker appears more frequently during the interviews than was documented in our table above. However, we assumed that including the unselected instances could introduce bias. Therefore, we only kept those examples that do not distort our findings. As for the interviewees, certain rappers in the corpus use stage names to maintain anonymity, as a result, neither their real names nor their dates of birth are known. These cases are indicated by a question mark in Table 1.

The interviews in question contain 68 sentences with the discourse marker tu vois (or its past tense form: t’as vu), and it is noteworthy that this marker, among others, is typical of non-standard variations, since they predominantly appear in spoken discourse and primarily serve to reinforce the informal character of speech (Bolly 2012). Although not all the question tags were formed in the same way and there are differences among the linguistic properties of the marker across the 68 occurrences, some of them were clause-initial and others were in clause-internal or clause-final, they exhibit a consistent pattern by displaying the reductions and compressions typical of nonstandard language, as observed at the word level (for instance, the word tu ‘you’ is reduced to [t] when it precedes a vowel) (Gadet 1991: 68). So, the pronunciation also contributes to this unconventional aspect of the discourse, because the discourse marker can be regarded as a non-standard linguistic element, it gives the sentence the same stylistic value.

Phonologically speaking, the marker tu vois is realized in the majority of cases as a monosyllable. The pronoun tu is about to be neutralized in this phrase by the deletion of the /y/ and the spirantization of the /t/. The stop becomes an affricate pronounced as [t͡ʃ], which precedes the diphthong [wa]. As a result, we see the following phonemic transcription [t͡ʃwa]. It is not surprising then that having looked for the prosodic features of this marker such as the pitch, the intensity, and the duration, generally we measured one value by criterion (Figure 1), and we found merely a few occurrences with more than one value (Figure 2).

Fig. 1: Praat image of the simple, monosyllabic discourse marker tu vois uttered by Naps (4th analyzed sentence of the speaker)

Fig. 2: Praat image of the complex, disyllabic discourse marker t’as vu uttered by Naps (5th analyzed sentence of the speaker)

Regarding the position of the discourse marker within the clause, our corpus confirms the observations made by Andersen (2007) and Bolly (2009): tu vois is typically extra-propositional, that is, positioned at the periphery of the clause. More precisely, although its placement in our sample is somewhat flexible, the marker is most often found postpositionally, at the end of the clause. Out of 68 occurrences, it appears within the clause only seven times and in sentence-initial position just four times. Consequently, its interactional function is emphasized, as it allows the speaker to ascertain that the interlocutor is following the discourse.

3 Method

We chose the scope of this study in the light of the contemporary and rapidly changing features of the French language about which it is commonly known that it is mainly influenced not only by the geographical data, but also by the sociocultural and socioeconomic background of French speakers, this is, the sociolect presents a large impact on the speech and language (Gadet 1991, 2017). According to mainstream perspectives, the dialects and sociolects of French encompass a wide range of non-standard linguistic variations that can be studied through the analysis of a corpus consisting of both recently recorded and spontaneous speech. In terms of phonology, informal, thus spontaneous speech is especially instructive, because its primary features and eventual distinctions among various forms of spoken language reveal the genuine, non-negligeable side of nowadays communication. To pursue this inquiry, we selected interviews conducted with popular French rappers who were not provided with the questions beforehand and did not read any text in advance.

We divided the rappers into two groups based on their geographical location, one combining speakers from Paris and another created with speakers from Marseille. During the interviews, the rappers uttered 68 sentences containing the discourse marker tu vois or t’as vu. Using the computer software called Praat, we collected the three prosodic features of each discourse marker, namely pitch, stress, and duration. We annotated the statements and summarized the data for each feature in an Excel spreadsheet. In the first step, to determine the relations between the prosodic features, we used the Excel program. As a second step, we performed an independent two-sample t-test on the means of both groups using IBM SPSS Statistics, version 26. The research results for this study were obtained from these two analyses.

3.1 The three prosodic features

All analysis, annotations and measurements concerning the prosodic attributes of the discourse marker were done using the acoustic software, Praat. The tonal values are composed of data about pitch, intensity, and duration. The extracted values were then transformed and replaced with standardized values in percentages. This phase of standardization was crucial to generate an independent set of data from the inherent bias and irrelevant values. During the analysis, we investigated the changes in curve of the monosyllabic construction in the graph drawn in Praat. We paid special attention to the past tensed form of the marker t’as vu /tɑ vy/, which, being disyllabic unit, takes up 2 distinct prosodic values.

To analyze prosodic features, our process involved three steps. As a first step, we identified the syllable(s) of the discourse marker. Typically, we looked for the intensity and the pitch peak in the middle of the vowel of the monosyllabic construction. These are the values used for analysis. As for the duration, the absolute duration of the syllable is calculated from the lowest point at which the intensity value of the discourse marker started to rise until it fell back to the lowest point. The second step was to represent the prosodic features. To focus solely on relevant prosodic features, irrelevant information must be rejected, hence the data was reduced to one value of each attribute. The three attributes were given in the following units: Hz for intonation, dB for intensity, and seconds for duration. In the third step of the analysis, we undertook the process of standardizing the values obtained from the prosodic data. This standardization was essential to ensure a uniform and consistent basis for comparison across the various data curves. Instead of using absolute values which might vary significantly in magnitude from one observation to another, we opted for relative values. These relative values were calculated in a manner that expressed the prosodic characteristics as a percentage of change relative to the preceding syllable.

3.1.1 Pitch

Regarding the inflections of the marker, its pitch variations (e.g., changes in loudness or softness), the whole utterance, including the marker under investigation, was measured within a pitch range of 50 to 300 (in Hertz) using an autocorrelational analysis method optimized for intonational research instead of voice research. Sometimes, the tonal segment waved at a high degree on the tag, that is, we considered the marker as a two-part segment if the rising or falling standardized value was equal to or was over 10% (Pálvölgyi 2022). In cases where the inflection was prominent, we measured the onset of the syllable to determine whether the rise or fall began at that point, at the end of the preceding syllable, or whether there was a micro-inflection that influenced the intonation of the entire marker (Figure 3). More than one example shows that the inflection of the marker did not start at the end of the previous syllable, these cases are marked and treated apart (Figure 4).

Fig. 3: A stagnant micro-inflection at the beginning of the discourse marker (1st analyzed sentence of Hornet La Frappe)

Fig. 4: A rising micro-inflection on the onset of the marker (1st analyzed sentence of Koba LaD)

3.1.2 Intensity

Marking the intensity of the construction tu vois or t’as vu seemed to be done with little effort because it follows a simple path. The one syllabic marker worked with only one value measured between 40 and 100 dB, however in the cases of t’as vu, two units have been registered in this range. Out of the 68 tagged utterances, 5 operate with 2 peaks (Figure 2). During the analysis, we determined the ratio of the syllable(s) compared to the value of the syllable before it.

3.1.3 Duration

The question of the duration of the marker also merits a more detailed description or a whole new study as well, since the plausible syllable lengthening can expose prosodic roles of the construction. Although the construction in question occurred usually at the end of the utterance and we might think that this fact defines the manner of its pronunciation, we want to clarify that the duration presenting values manifest merely the length of the syllable(s) seen on the spectrogram, the following respiratory noises and pause have been excluded from the analysis. It is important to note that the past tense discourse marker composed of two phonological elements /tɑ vy/ possesses 2 values, one value each. As an example, in Figure 5 below, we observe a disyllabic, pas tense variant of our marker, which is in a clause-internal position.

Fig. 5: Double durational value of t’as vu with a highlighted area focused on the whole length (5th analyzed sentence of Naps)

3.2 Correlation between the prosodic features of the markers

After data collection, we handled the data describing the three attributes separately, and further we divided them based on the origin of the speakers (Paris or Marseille), which resulted in six columns of data. Since the two groups consisted of distinct speakers whose expressions can be considered as spontaneous language data, we conducted an independent two-sample t-test. The test helped us to compare the value means of the three prosodic features between the two groups of nonstandard speakers who represented different regions of France, and to determine whether there is statistically proved difference between them.

4 Results

We now turn our attention to the results of the study, where we analyze the phonological and prosodic features that characterize the French discourse marker. Following this, we will discuss the similarities and differences observed between the northern and southern dialects of French. To effectively visualize the explanatory analysis of the 68 occurrences of the discourse marker, we employed several graphical methods. Initially, we illustrated the prosodic features using both column and line diagrams to depict a straightforward visualization of the data. Additionally, a box-plot diagram was constructed to offer a more comprehensive view, displaying the mean and median values, as well as the dispersion of the data set with respect to regional variations. Finally, we will present a detailed account of the outcomes derived from three independent two-sample t-tests.

4.1 Co-occurrence of the three prosodic features

Using the foregoing method, we found that in 83.82% of the utterances, an intonational change was perceivable, i.e., 57 discourse markers out of 68 showed an inflection of a rise higher than 10%. Within this subset, the inflection was complex 34 times. 31 tu vois marker had two inflectional parts, 2 comprised three, and 1 marker featured four (Figure 6–8).

Fig. 6: A complex, past-tense discourse marker with three inflectional parts (2nd analyzed sentence of Koba LaD)

Fig. 7: The tu vois marker as a construction with 3 inflectional values (the blue inflection on the tu marks a background noise; 1st analyzed sentence of Soso Maness)

Fig. 8: 4 inflectional values on the discourse marker (2nd analyzed sentence of Soso Maness)

The detailed analysis allowed us to point out that 45 discourse markers (66.17%) exhibited a rising final inflection – 15 with simple inflections and 30 with complex ones. Regarding the complex inflections, in most of the cases (29 out of 34, 85.29%) the segment of the contour which precedes the final inflection was falling. We need to emphasize that in many examples micro-inflections were detectable, but these subtle variations did not significantly impact the general amplitude of the intonation.

The intensity values were divided into two categories; hence we processed the present tensed and the past tensed markers separately. For the past tense variant of t’as vu, two intensity peaks were observed, and in these cases the second peak was compared to the first. Among the interview data, we identified four markers with dual intensity peaks; notably, in only one instance did the first peak exceed the second, indicating that in the remaining three cases a rise in intensity occurred. This trend, typical of the dual-peak markers, is surprisingly at odds with our overall findings, where the marker’s intensity increased in only 36.76% of cases (25 out of 68), while no increase was observed in 43 cases.

Among the cases analyzed, we noticed that 44 simple markers and 3 complex markers presented syllabic lengthening which is in total 69.11% of the instances. The durational aspect of the complex discourse marker mirrored the patterns found in the intensity analysis: in each instance, the final segment “vu” of the marker was consistently pronounced with extended duration. Therefore, this prolongation of the syllable is a prominent feature in our data.

After the separate analysis of the prosodic features, it is now time to compare the received values. We studied 68 occurrences from 3 points of view, and this next phase shows great relevance especially for 2 aspects. For pattern recognition, we constructed a comparative table with 3 columns representing pitch, intensity, and duration. In this section, we will glance through the positive values in one, two or three aspect(s) and the exceptions. Only 4 discourse markers do not obey the rule of accommodation, none of the mentioned features have been changed in them. It represents 5.88% of the total. To a rough approximation, one-third of the examples (35.94%) were sensitive in one dimension which means that the value of only one feature passed the needed threshold. The most tuned feature was the duration with 47.82% of the cases. 45.31% of the cases involved a set of two positive values, this is, two criteria fulfilled the needed conditions. 22 markers out of 29 (75.86%) showed similitudes between the pitch and the duration, which we consider highly persuasive. Finally, 18.75% (12/64) of the cases triggered prominence in all aspects. Figure 9 provides the relative proportion of the co-occurrences.

Fig. 9: The proportion of the co-occurrences of the prosodic features

4.2 Comparison of the dataset of the discourse marker

Even upon first hearing, the contrast in speech between representatives of the two dialects is striking. However, this does not necessarily mean that the analyzed features will differ from one another. To figure out any pattern specific to the two regions, our analysis extended beyond the characteristics of the discourse marker itself to include the quality of its surrounding linguistic environment. Having established the sets of each value, we systematically compared the three prosodic properties of the marker as produced by speakers from different geographic areas, contrasting Marseille with Paris. In addition, we drew a summary diagram including all the data by criterion serving as evidence of correlation between the features. For clarity and interpretability, we opted for a box-plot visualization. The diagram shown below helps to understand the intonation patterns associated with the marker as produced by speakers from the suburbs of Paris (Figure 10).

Fig. 10: Comparison of the intonational values of the two groups of speakers

The upper quartile traces out the line at 23.42%, while the lower quartile is found at −21.69%, the overall median value is 7.17%. The maximum and the minimum values are 70.83% and −66.45%.

In contrast to our findings in the northern part of France, the intonation of the speakers of Marseille on the marker is slightly different. As we can see on the diagram, the median of the upper half climbs to 21.11% and the median of the lower half of the data falls to −14.74%. The median is seen at 5.13%, the maximum at 57.76% and the minimum at −52.05%. About the intonation, we state that our analysis indicates that for speakers from the Paris agglomeration, the median of the upper half of the dataset is higher, while the lower quartile shows a decrease.

The intensity of the discourse marker reveals a less significant difference between the two dialectal groups. As shown in Figure 11, the fluctuations of the intensity remain relatively minor in both cases. As fascinating as it seems, despite these small variations, the intensity drops below 0.

Fig. 11: Comparison of the intensity values of the two groups of speakers

The closing part of the results is about the durational values of the discourse marker. In contrast to what we have experienced so far, the duration between the two dialects is not nearly as consistent. In the diagram, the box of the speakers from Marseille spreads out much more, the quartiles, and the maximum and minimum values are more extreme, they are higher and lower than the ones seen for speakers from Parisian region. Nevertheless, what the two dialects have in common is the fact that the median values are highly above 0 (Figure 12).

Fig. 12: Comparison of the durational values shown by regions

4.3 The difference between the means

Ultimately, we present the result of the final stage of our research, in which independent two-sample t-tests were conducted to compare the values of the three previously described prosodic features of the marker between the two populations. This rigorous statistical analysis was essential in determining whether significant differences existed between Francilien (speakers from the Île-de-France region) and Marseillais (speakers from the Marseille region) dialects. By focusing on these two distinct varieties, the study sought to explore the nuances of French regional linguistic variation.

The three prosodic features examined were pitch, which reflects the intonation and frequency; intensity, which measures the loudness or energy; and duration, which accounts for syllable or word length. To ensure a precise and thorough examination, separate t-tests were made for each feature, and the results were systematically summarized. Through this comprehensive analysis, we aimed to provide a clear and detailed comparison of the prosodic characteristics of discourse marker in different French dialects, contributing to a deeper understanding of commonly assumed regional linguistic variations.

The results showed no significant difference in the intonational aspect of the discourse marker between the speakers of the region of Île-de-France (M = 6.6, SD = 44.53) and those of Marseille (M = 8.79, SD = 38.72); t(113) = −.282, p = .779.

Similarly, intensity features did not significantly differ between speakers from the suburbs of Paris (M = −1.7, SD = 5.08) and those from Marseille (M = −2.41, SD = 4.66); t(70) = .621, p = .537.

Finally, the third t-test did not indicate a significant difference in the durational values between the two populations, with means (Paris, M = 30.69, SD = 63.98) for Paris and for Marseille (Marseille, M = 22.47, SD = 57.63), with a t-value of .556, 66 degrees of freedom and a p-value of .58.

5 Conclusion

Our goal in doing this thorough investigation was to advance the understanding of regional prosodic variations in French. This study tried to contribute to the fields of sociolinguistics and dialectology by examining the widely assumed distinctions between the Francilien and Marseillais dialects. It provided a detailed description of prosodic patterns and laid the groundwork for future research on the dynamics of spoken French.

The conclusions derived from the results can be categorized into two primary sections. The first reveals the prosodic significance of the analyzed discourse markers tu vois, along with its variant in the past tense t’as vu. Beyond a purely phonological perspective, we also sought to clarify the sociolinguistic and pragmatic roles of this construction based on an extensive body of literature. In this chapter, we will endeavor to explain the characteristics of each examined feature and emphasize why there is no relationship between them. After that, we will then shift focus to the core of our research: regional variations of the tu vois marker.

We think that the nature of the corpus selection influenced the findings. In spontaneous speech, pragmatic markers often carry significant weight and are of interest from a phonological perspective. Furthermore, the fact that the chosen speakers typically use non-standard language varieties shaped the research. Their speech reflects identity-related linguistic features and remains true to their everyday communication styles. The use of non-standard language by these speakers can be seen as a special characteristic of the identity construction (not excluding the influence of the informal atmosphere of the interview), as they are faithful to their everyday styles of speech. This results in grammatical structures that are not typical of this genre, with their function(s) being conveyed through phonetic characteristics.

Regarding every prosodic characteristic, it can be said that it was mainly the duration and pitch that changed on the marker(s), if one of them has changed, the other followed the alteration in half of the cases. Despite the high number of co-occurrences, the situations where all three characteristics changed at the same time were few, which means that the change of one of the characteristics did not directly trigger shift in another. We argue that the rise and the fall in the prosody serve multiple functions and that changes in duration fulfill a different communicative role than those in pitch.

The t-tests revealed no significant difference between the two dialectal groups. This finding aligns with our initial hypothesis that the function of the selected discourse marker remains consistent across both regions, leading to no expected prosodic divergence. We attribute this lack of difference to two main factors. First, the interviewee’s personal style decides the prosodic values, i.e., individual variations in speech habits, such as intonation, rhythm, and stress, play a significant role in shaping the prosody of the discourse marker. Second, the marker’s inherent communicative function highly determines its prosody. As stated above, discourse markers can have various communicative functions, such as seeking agreement, eliciting a response, or maintaining the flow of conversation. As we saw, for instance, an agreement-seeking may exhibit rising intonation, whereas one facilitating conversational continuity may have a more neutral or falling pattern.

Moreover, we believe that our marker serves to prevent interruptions and marks the necessary pauses that speakers require to maintain the flow of conversation, in addition to facilitating feedback acquisition and raising awareness. Given that our study primarily focused on the durational aspects of the tu vois discourse marker, further research is needed. Future studies should compare the duration of the discourse marker with other components of the sentence in a broader context to accurately determine its precise role in structuring speech. This approach would enable us to study how discourse markers interact with their contextual elements and yield substantial evidence of their function in communication and impact on conversation structure. A more detailed examination of how they interact would also provide a deeper insight into prosodic and functional characteristics of discourse markers and hence into an even more precise perspective of how they are used in different dialects and situations.

References

Andersen, H. (2007): Marqueurs discursifs propositionnels. Langue française 154: 13–28. https://doi.org/10.3917/lf.154.0013

Anzai, Y. (2009): Deux fonctionnements du marqueur français « tu vois » dans les dialogues spontanés : relation entre les faits intonatifs et la structure morphosyntaxique. Paper presented at the Interfaces Discourse & Prosody: IDP 2009, Paris, September 9–11, 2009.

Avanzi, M. (2014): A Corpus-Based Approach to French Regional Prosodic Variation. Paper presented at the Swiss Workshop In Prosody: SWIP, Geneva, September 10–11, 2012.

Blanche-Benveniste, C. & Willems, D. (2007): Un nouveau regard sur les verbes « faibles ». Bulletin de la Société de linguistique de Paris 102: 217–254. http://dx.doi.org/10.2143/BSL.102.1.2028205

Bolly, C. (2009): Constructionalisation et structure informationnelle: Quand la grammaticalisation ne suffit pas pour expliquer “tu vois”. Linx 61: 103–130. https://doi.org/10.4000/linx.1342

Bolly, C. (2010): Pragmaticalisation du marqueur discursif “tu vois”. De la perception à l’évidence et de l’évidence au discours. In: F. Neveu & J. Durand & T. Klingler & S. Prévost & V. Muni-Toké (eds.) Proceedings of the Congrès Mondial de Linguistique Française : CMLF 2010. Paris: Institut de Linguistique Française. 673–693. http://dx.doi.org/10.1051/cmlf/2010243

Bolly, C. (2012): Du verbe de perception visuelle au marqueur parenthétique « tu vois »: Grammaticalisation et changement linguistique. Journal of French Language Studies 22: 143–164. https://doi.org/10.1017/S0959269511000044

Brinton, L-J. (2008): The Comment Clause in English: Syntactic Origins and Pragmatic Development, Cambridge: CUP.
https://doi.org/10.1017/CBO9780511551789

Cornulier, B. de (1978): L’incise, la classe des verbes parenthétiques et le signe mimique. Cahier de linguistique 8: 53–95. https://doi.org/10.7202/800060ar

Dehé, N. & Wichmann, A. (2010): The multifunctionality of epistemic parentheticals in discourse: prosodic cues to the semantic-pragmatic boundary. Functions of Language 17: 1–28. http://dx.doi.org/10.1075/fol.17.1.01deh

Détrie, C. (2010): De voir à tu vois / vous voyez : Fonction sémantico-énonciative et postures énonciatives construites par ces particules interpersonnelles. In: F. Neveu & J. Durand & T. Klingler & S. Prévost & V. Muni-Toké (eds.) Proceedings of the Congrès Mondial de Linguistique Française : CMLF 2010. Paris: Institut de Linguistique Française. 755–766. https://doi.org/10.1051/cmlf/2010035

Dostie, G. & Pusch, C. D. (2007): Présentation. Les marqueurs discursifs. Sens et variation. Langue française 2: 3–12. http://dx.doi.org/10.3917/lf.154.0003

Gadet, F. (1991): Simple, le français populaire ? Linx 25: 63–78. https://doi.org/10.3406/linx.1991.1227

Gadet, F. (2017): Variatio delectat : variation et dialinguistique. Langage & Société 160–161: 75–91. http://dx.doi.org/10.3917/ls.160.0075

Lee, L., Bartkova, K., Jouvet, D., Dargnat, M. & Keromnes, Y. (2019): Can prosody meet pragmatics? Case of discourse markers in French. Paper presented at the 19th International Congress of Phonetic Sciences: ICPhS2019, Melbourne, August 5–9, 2019.

Lee, L., Jouvet, D., Bartkova, K., Keromnes, Y., & Dargnat, M. (2020): Étude comparative de corrélats prosodiques de marqueurs discursifs français et anglais selon leur fonction pragmatique. Paper presented at the 6e conference conjointe Journées d’Études sur la Parole: JEP 33e édition, Nancy, Juin 8–19, 2020.

Morel, M-A. & Danon-Boileau, L. (1998): Grammaire de l’intonation, l’exemple du français oral. Paris: Ophrys.

Pallaud, B., & Bertrand, R. (2020): Auto-interruptions et disfluences à l’oral. C’était euh tu vois complètement loufoque comme si ouais euh comme situation. In: F. Hirsh, Dirdiskova, I. & Dodane, C (eds.) Manuel de pausologie: Recueil de recherches sur la parole et le discours. Paris: L’Harmattan. 21–47.

Pálvölgyi, K. (2020): The duration of filled pauses and prolongations in northern and southern dialects of Spanish. Revista de Estudos Linguísticos da Univerdade do Porto 15: 71–93. http://real.mtak.hu/148873

Pálvölgyi, K. (2022): Prosodic perceptive thresholds in Spanish: intensity and duration. Revista da Abralin 21: 193–208. http://dx.doi.org/10.25189/rabralin.v21i2.2077

Rouanne, L. (2022): Tu vois/vous voyez en synchronie : contraintes distributionnelles et valeurs sémantico-pragmatiques. Langages 227: 119–133. http://dx.doi.org/10.3917/lang.227.0119

Schneider, S. (2007): Reduced parenthetical clauses as mitigators: A corpus study in Spoken French, Italian and Spanish. Amsterdam/Philadelphia: John Benjamins. https://doi.org/10.1075/scl.27

Vincent, D. (1993): Les ponctuants de la langue et autres mots du discours. Québec: Nuit Blanche.