1-Prosody

Oct 10, 2020 (5 pm - 7 pm, JST)

Sara Myrberg (Lund University)

"Two-peakedness in South Swedish: implications for the Scandinavian tone accent typology"

In this talk, I present production data from two South Swedish dialects, which have interesting implications for the prosodic typology of Swedish and Norwegian.

Swedish and Norwegian are well-known to have a binary lexical distinction between so-called tone accent 1 and tone accent 2. The tonal representation of the accents varies between dialects, and a long line of research has studied the prosodic typology of these dialects (e.g. Gårding 1977; Bruce 1977, 2007; Kristoffersen 2000; Riad 2006, 2018). One of the most fundamental parameters for understanding the Scandinavian tone accent typology has been a distinction between so called one-peak dialects, which have one tonal peak in words with tone accent 2, versus two peak-dialects, which have two tonal peaks in words with tone accent 2 (all dialects have a single peak in words with tone accent 1). South Swedish has been unanimously categorized as a one-peak dialect, and has thereby been assumed to have a profoundly different type of prosodic system than the two-peak dialects of Stockholm Swedish or Oslo Norwegian.

However, previous categorizations have been based on studies of short utterances or spontaneous speech. The present production experiment uses longer sentences, and therefore reveals that South Swedish exhibits two-peak contours in non-final positions. These two-peak contours are characterized by so-called drifting of the second peak, which has also been reported from nuclear accent contours in Oslo Norwegian (e.g. Fretheim 1987; Kristoffersen 2000) and Gothenburg Swedish (e.g. Bruce 2007). Drifting has also been reported from a special type of prenuclear accent in Stockholm Swedish (initiality accent, Myrberg 2010, 2013). However, opposed to other dialects, South Swedish also exhibits truncation of the tonal peak in phrase-final position.

I will argue that the South Swedish data implies that the distinction between one- and two-peaked dialects is not binary. Instead, dialects place themselves on a scale from more to less two-peaked. The extent to which the dialects allow drifting of tonal peaks and truncation determine their place on this scale. Allowing more drifting and truncation makes a dialect more one-peaked, whereas restrictive or no application of these phonological processes makes a dialect more two-peaked.

Lena Borise (Research Institute for Linguistics at the Hungarian Academy of Sciences & Harvard U)

"Verb height, prosodic phrasing, and the flexible ⍳-mapping hypothesis: evidence from Iron Ossetic"

(joint work with David Erschler, Ben Gurion University of the Negev)

This paper provides novel evidence in favor of flexible mapping between an Intonational Phrase (ι) (Nespor & Vogel 1986) and syntactic constituents. In the existing accounts, ι is assumed to map onto a syntactic clause, but a ‘clause’ in the syntax-prosody literature may be defined as a TP (Zerbian 2006), CP (Truckenbrodt 2005, Henderson 2012), or the complement of Force⁰ and C⁰ (Selkirk 2011). Hamlaoui & Szendrői (2015, 2017) propose that ι is flexible and corresponds to the highest projection that hosts verbal material, together with its specifier (HVP, ‘highest verbal projection’), and, therefore, provide a unified, syntax-based account of cross-linguistic variation in ι­-size. A prediction that it makes is that ι­-size is also determined by HVP in a language where the height of the verb depends on utterance type. Iron Ossetic (East Iranian), with several projections available for verb raising, is a uniquely suitable testing ground for this prediction. We adopt the flexible ι-mapping approach and, using experimental prosodic data, show that the HVP indeed determines the size of ι in Iron Ossetic. Specifically, we demonstrate that the prosodic phrasing obtained in clauses that contain narrow foci and negative indefinites directly follows from the flexible ι-mapping hypothesis. We also show how the flexible ι-mapping hypothesis interacts with phonological (i.e., independent from syntax) markedness constrains on prosodic phrasing in wh-questions and neutral SOV clauses. Overall, this paper provides further support to flexible ι-mapping, based on a new language type, while also showing that the surface prosodic facts of Iron Ossetic result from the interaction of factors rooted both in syntax and phonology. It also contributes to documenting the prosodic properties of an understudied language.

Oct 24, 2020 (5 pm - 7 pm, JST)

Frank Kügler (Goethe University Frankfurt)

"On different degrees of post-focal compression"

Prosodic focus is usually realised on the focused constituent highlighting or demarcating this constituent from other constituents. Cross-linguistically however, it remains on open issue to which extent post-focal, usually given elements and the prosodic variation found in that domain relates to the expression of focus marking. This talk explores the conditions of the variation found in post-focal prosody. Post-focal compression (PFC) is widespread among languages of the world, yet a non-universal cue to encode focus. Typologically diverse languages like Finnish, German, Greek, Hindi, and Mandarin show post-focal pitch events in a compressed pitch register. However, the degree of compression differs: partial compression is observed for instance in Finnish, Hindi and Mandarin, almost complete compression in German and Greek. In this talk, I argue for a theory of two distinct post-focal compression types. The claim is that the different degree in register compression is a consequence of the phonology of a language governing the need to express the functional load of tones or accents located in the post-focal domain. If the phonology of a given language demands a high need to maintain phonological contrasts post-focally the post-focal register compression is partial. If on the other hand the phonology demands a low need to express post-focal pitch events register compression is complete.

Nancy Kula (University of Essex)

"Pre-focus expansion and prosody-phonology mismatches"

This talk will discuss some robust intonational patterns in Bemba (Central Bantu) that correlate to different kinds of declarative sentences and questions. Specific intonational features such as, final lowering, pitch range expansion and pitch register raising will be discussed to illustrate robust intonational patterns in tone languages. The aim of the talk will be to make a comparison of controlled data with more spontaneous data but controlled for genre, here specifically from sermons, to see how comparable these data are given what we already know about Bemba intonation as discussed in Kula & Hamann (2017).

Nov 7, 2020 (5 pm - 7 pm, JST)

Shin Ishihara (Lund University)

"On the (lack of) correspondence between syntactic clauses and intonational phrases"

In this talk, I will discuss the syntax–prosody mapping at the clause-level, or the potential absence thereof. The main goal of this talk is to bring up some of the empirical and theoretical questions related to the MatchClause constraint in Match Theory (Selkirk 2009, 2011), and to reexamine the validity and the necessity of the clause-level mapping principle proposed in the theory. Empirical discussion includes, among other things, a critical review of Selkirk’s (2009) discussion of the wh-prosody in Fukuoka Japanese. It is shown, based on additional data from Fukuoka and Tokyo Japanese, that Selkirk’s claim that Fukuoka Japanese provides empirical support for MatchClause is untenable. The theoretical questions discussed in this paper also challenge the validity and the necessity of the clause-level mapping. An alternative model is suggested in which the intonational phrase (ι) is related to the pragmatics–prosody mapping rather than the syntax–prosody mapping.

Cédric Patin (Université de Lille)

"Domains as tools for analyzing the tone-intonation interface"

In the introduction of a paper dedicated to the prosody of Shekgalagari, a Bantu language of Botswana and Namibia, Hyman & Monaka (2011) list three main possibilities of interaction between tones and intonation: Accommodation (“peaceful coexistence”) – the lexical and intonational tones have little interaction with each other; Submission (“surrender”) – the intonational tones overrule the lexical tones; Avoidance (“blockade”) – the lexical tones override the intonational tones. Accounting for such a diversity of situations, both in terms of representation and analysis, is far from simple. In a seminal paper, Beltzung et al. (2010) propose that ToBi-like intonational tones evolve on a separate tier from lexical tones, allowing the former and the latter to be superimposed or to follow each other, an approach that was later adopted in Downing & Rialland (2016). As interesting as it may be, it is not obvious how such an approach can account for the different configurations evoked by Hyman & Monaka (2011). Above all, it struggles to explain certain interaction phenomena: in Shingazidja (a Bantu language from the Comoros), for example, the peak of F₀ associated with polar questions, which normally emerges on the penultimate syllable, is shifted to the antepenultimate when the final syllable is high-toned. In this talk, I will explain how the use of domains as developed in Optimal Domains Theory (Kisseberth 1994; Cole & Kisseberth 1994; Cassimjee & Kisseberth 1998) offers an interesting alternative for examining some of the interface phenomena between the tones and intonation of Niger-Congo languages.

Nov 21, 2020 (5 pm - 7 pm, JST)

Nicole Dehé (Universität Konstanz)

"The prosodic marking of rhetorical questions: A cross-linguistic view"

In this talk, I will present results from an ongoing research project on (the prosody of) rhetorical questions (RQs) as compared to neutral, information-seeking questions (ISQs).

I will first present the results of a series of production experiments testing three head-prominence intonation languages (German, English, Icelandic) and one head-prominence tone language (Standard Chinese). Results for all languages show that prosody disambiguates between ISQs and RQs. Speakers of all languages use F0 (intonation languages: pitch accents, boundary tones; tone language: local and global F0 height), as well as duration and voice quality to signal rhetorical meaning (e.g., Braun et al 2019 for German; Dehé & Braun 2020b for English; Dehé & Braun 2020a for Icelandic; Zahner et al to 2020 for Standard Chinese; and work in progress). The four languages differ in the way the prosodic parameters are used (e.g., pitch accent types signal rhetorical meaning, but different ones for each language).

Second, I will present results from two further studies testing German: one corpus study studying the prosodic differences between ISQs and RQs in spontaneous language (Braun et al. 2020), and one experiment focusing on the prosodic, syntactic and lexical choices that speakers make to express rhetorical meaning (in progress).

References

Braun, Bettina, Nicole Dehé, Jana Neitsch, Daniela Wochner & Katharina Zahner. The prosody of rhetorical and information-seeking questions in German. Language and Speech 62(4): 779-807.

Braun, Bettina, Marieke Einfeldt, Gloria Esposito & Nicole Dehé. 2020. The prosodic realization of rhetorical and information-seeking questions in German spontaneous speech. Proceedings of Speech Prosody, 10th International Conference (2020), 24-28 May 2020, Tokyo, Japan. 342-346. DOI: 10.21437/SpeechProsody.2020-70.

Dehé, Nicole & Bettina Braun. 2020a. The intonation of information seeking and rhetorical questions in Icelandic. Journal of Germanic Linguistics 32(1): 1-42.

Dehé, Nicole & Bettina Braun. 2020. The prosody of rhetorical questions in English. English Language and Linguistics 24(4): 607-635.

Zahner, Katharina, Manluolan Xu, Yiya Chen, Nicole Dehé, & Bettina Braun. 2020. The prosodic marking of rhetorical questions in Standard Chinese. Proceedings of Speech Prosody, 10th International Conference (2020), 24-28 May 2020, Tokyo, Japan. 389-393. DOI: 10.21437/SpeechProsody.2020-80.

Adam J. Chong (Queen Mary University of London)

"Singapore English intonation: An Autosegmental Metrical perspective"

Singapore English is a nativised English variety that displays a number of systematic linguistic features that distinguish it from Western varieties (e.g. British English). In this talk, I present work in progress on Singapore English intonation, an under-examined aspect of the linguistic system. I first present a preliminary phonological model of Singapore English intonation couched in the Autosegmental Metrical framework. I then present results from two production studies that provide qualitative and quantitative support for aspects of the initial model. In particular, I will focus on evidence for the existence of a prosodic level above the word (the Accentual Phrase), as well as a preliminary examination of the relation between phrasal position, stress and tonal alignment. I end by discussing the implications of this work for our understanding of prominence and rhythm in Singapore English prosody. I also discuss the important contribution New Englishes have for prosodic typology.

Dec 5, 2020 (10 AM - 12 PM, JST)

Emily Elfner (York University)

"Evaluating Evidence for Covert Prosodic Structure"

Research on the syntax-prosody interface concerns the relationship between syntactic constituent structure and prosodic phrasing. A crucial aspect and challenge of this research has been the detection of prosodic boundaries, through the study of phonetic and phonological cues to phrase edges and domains. However, in much work on syntax-prosody interface, it is implicitly assumed that we may have only partial evidence for prosodic structure in a given language. For example, a language may overtly mark right edges of prosodic phrases, but leave the left edges unmarked, as assumed in the End/Edge-based approach to syntax-prosody mapping (e.g. Selkirk 1984, 1995) or in Match Theory (Selkirk 2011), yet we may presume that these boundaries exist to do assumptions about mapping between syntactic and prosodic constituents or the well-formedness of prosodic structure. This underspecification of certain phrase boundaries is at odds with what is assumed in much of the literature on psycholinguistics or “intonation-first” approaches to prosodic structure, in which an overt phonetic/phonological cue is both a sufficient and necessary diagnostic of a prosodic boundary. In this talk, I review some of the arguments for the positing of “covert” phrase edges based on syntactic structure, and discuss a case-study from Irish, which shows an apparent mismatch between tonal prosodic cues for phrase edges and the relative strength of prosodic boundaries as determined by gradient juncture phenomena.

Natalie Weber (Yale University)

"On the misalignment of prosodic edges and syllables"

There is robust evidence from many languages that prosodic constituents may mismatch from syntactic constituents in order to satisfy prosodic wellformedness constraints (e.g. Nespor & Vogel 1986; Selkirk 1986). One such constraint requires the alignment of prosodic edges and syllables, and prosodic edges may misalign to the left or right in order to align with a syllable onset (Downing 1998). In this paper I argue that Blackfoot (Algonquian; Frantz 2017) is a language of the opposite type: syntactic and prosodic constituents are isomorphic at the expense of edge alignment. Specifically, the left edge of the Prosodic Word (PWd) often falls in the middle of a syllable, without aligning to a syllable onset. The left edge of the PWd is defined phonologically by an edge restriction against *[-cont] segments. This is an active constraint which drives multiple patterns of √rootalternations, including a process of epenthesis at the left edge of the PWd. Independent evidence for syllable structure, as well as a regular process of vowel coalescence, indicates that a syllable can span the left edge of the PWd. The left edge of the PWd cannot align to the edge of a syllable without violating either the syntax-prosody correspondence constraints or the constraint against *[-cont] segments at the left edge. For the analysis to converge, the syntax-prosody correspondence constraints must be violated by the inclusion of prefixal material or exclusion of PWd material, but crucially not by epenthesis at the left edge of the stem. I provide a redefined version of MatchWord(Elfner 2012; Guekguezian 2012; Selkirk 2011) which is not violated by epenthesis.

Dec 19, 2020 (10 AM - 12 PM, JST)

Fatima Hamlaoui (University of Toronto)

"Influence of lexical tones on calling melodies: A comparison between Metropolitan and Bàsàa-Cameroonian French"

(joint work with Marzena Zygis, Jonas Engelmann and Sergio Quiroz)

In Metropolitan French, the simple vocative or chanting contour (e.g. Marina! A table! “Marina! Dinner!”) consists of a F0 peak on the penultimate syllable and a sustained, mid-plateau on the final syllable of the utterance (Fagyal 1997, Di Cristo 1999, Ladd 2008). Phonologically, this contour has been represented as a LHM sequence (Dell 1984), H*H-L% (Jun & Fougeron 2000) or more recently as H+!H*!H% (Delais-Roussarie et al. 2015). In the present talk, we present the results of a discourse completion task (Arvaniti et al. 2016) whose first aim is to test whether context (here routine vs.urgent) affects contour choice in French, as recently shown for Catalan by Borràs-Comes et al. (2015). Our study involves two groups: 14 speakers (4 male) of Metropolitan French (FR) and 12 bilingual speakers (8 male) of Cameroonian French (CM) with Bàsàa (Bantu A43) as their L1. As in the latter variety proper names are lexically specified for tone, we are interested in establishing whether speakers make use of the chanting contour too. Just as in Bàsàa, CM names in isolation form present a L/H-HL contour where HL either aligns with the last or the last two syllables, e.g. [àlîs], [màgdàlénà]. As Bàsàa is a tone language in which post-lexical information has little impact on F0 (Makasso et al. 2016), a strong effect of our speakers’ L1 on their L2 predicts that lexical tones should be preserved in both routine and urgent contexts. Results show that the choice of contour is strongly context dependent in FR, where speakers use three main intonational calling contours: the vocative chant, a rising (interrogative-like) contour and a falling parabolic contour. CM speakers, in contrast, show little effect of context on choice of contour and predominantly preserve lexical tones. A final rise is the second most frequent pattern and very little presence of the typical chanting contour is found.

Lauren Clemens (University at Albany)

"Prosodic Phrasing in Rutooro, a Bantu Language of Uganda".

Rutooro is a Bantu language of Uganda that lacks lexical tone. Instead, prominence in Rutooro is marked with a High tone (H) on the penultimate syllable of the phonological phrase (φ-phrase). In this talk, which represents joint work with colleague Lee Bickmore, I argue that the distribution of H tones in adnominal phrases serves as a diagnostic for whether it is generated in a DP-internal or external position. Reduced object RCs with overt subjects are a special case: the relativized head bears an unexpected H tone, while the subject is all-Low despite the fact that it is a self-contained XP. Also in the realm of reduced RCs, when a relativized head is separated from the RC by an additional modifier, e.g. an adjective, that modifier is realized as all-Low even though it is phrasal. We attribute non-isomorphic phrasing to the prevention of i) ambiguity and ii) prosodic indeterminacy—when prosodic structure could be the output of more than one syntactic configuration.