6.1 Introduction
Information structure is most commonly understood in terms of a set of discourse-pragmatic notions relevant for structuring utterances in a context. I will follow Reference KrifkaKrifka (2008) and many others in distinguishing three such notions – focus, givenness, and topic. Each of these notions exists in a privative opposition: focus–background, given–new, and topic–comment, whereby the second member of the opposition is typically defined as the absence of the property ascribed to the first one. It is possible to define these notions in purely semantic and pragmatic terms (that is, without reference to form) and ask how they are formally reflected (if at all). This chapter is concerned with the prosodic reflexes of these information-structure notions in Slavic languages.
Surveys containing information on closely related issues include Reference Jasinskaja, Féry and IshiharaJasinskaja’s (2016) survey on information structure in Slavic languages, Reference Kügler, Calhoun, Gussenhoven and ChenKügler & Calhoun’s (2020) survey on prosodic encoding of information structure across languages, and finally also Titov’s chapter on morphosyntactic reflexes of information structure in Slavic languages (Chapter 22 of this volume). I would like to point out that this chapter is concerned with prosodic reflexes of information structure in declaratives, but not in other types of sentences such as interrogatives or imperatives. The basic findings discussed below should hold independently of the clause type, although non-trivial interactions between information structure and clause-type-related prosody can be expected. Such effects are beyond the scope of this chapter.
In Section 6.2 I introduce the basic notions of information structure – focus, givenness, and topic. In Section 6.3, we will see that the prosodic expression of focus and givenness is obligatory in Slavic languages (for which there is such evidence): a focused constituent must be prosodically prominent and a given one must not be (unless it is also focused). The prosodic expression of topicality is optional and less consistent. The phonetic correlates of stress or its lack are fundamental frequency (f0) (the best-studied acoustic property), but also duration and intensity. Section 6.4 briefly discusses the interaction between word order and prosody in the expression of information structure. Section 6.5 concludes and mentions some open issues.
6.2 Notions of Information Structure
The nearly two-century-old conceptual and terminological history of information structure is rich, but often confusing. This is not a place to attempt any kind of systematic overview of IS concepts and terms (see e.g. Reference Molnár and ReisMolnár 1993, Reference Kruijff-Korbayová and SteedmanKruijff-Korbayová & Steedman 2003, Reference KrifkaKrifka 2008, or Reference ŠimíkŠimík 2021). I follow the development of the past decades, succinctly presented in Reference KrifkaKrifka (2008), and assume three fundamental notions (oppositions): focus (vs. background), given (vs. new), and topic (vs. comment). These notions, or a subset thereof, are also assumed by most recent Slavicist literature (see e.g. Reference JunghannsJunghanns 2002, Reference Junghanns, Zybatow, Kempgen, Kosta, Berger and GutschmidtJunghanns & Zybatow 2009, Reference Jasinskaja, Féry and IshiharaJasinskaja 2016, Jasinskaja & Šimík forthcoming). The notions are defined as semantic and discourse-pragmatic properties of syntactic constituents or more generally linguistic expressions. They are in principle independent of one another, which means that a single expression can be focused, given, and a topic at the same time, although there is a tendency for focused constituents to be new (with the background being given) and for topics to be given (with the comment being new).
6.2.1 Focus (vs. Background)
A constituent is focused if it gives rise to alternative denotations relevant in the current discourse (Reference RoothRooth 1985, Reference Rooth1992, Reference KrifkaKrifka 2008). What is not focused is called the background. A prototypical instance of focus is a constituent which corresponds to the short answer to a wh-question.Footnote 1 In B’s response to A, Moscow is the focus because it gives rise to alternatives – contextually relevant cities that Sally might have visited. The rest of the sentence (Sally visited) is the background because it remains constant across the alternative propositions that might have served as the full answer to A’s question.Footnote 2 The variable nature of the focus and the constant nature of its background is made clear by the alternative propositions in (2), the so-called focus semantic value (Reference RoothRooth 1992), representing the possible answers to A’s question.
(1)
A Which city did Sally visit? B (Sally visited) [Moscow]F.
(2) {Sally visited Moscow, Sally visited Sofia, Sally visited Warsaw, …}
The focus used in answers to wh-questions is called information focus (also answerhood focus). There are other types of focus, however, some of which are illustrated in (3). Focus can be used for contrast (3a), where the alternative(s) are explicitly mentioned in the discourse, correction (3b), where an explicitly mentioned alternative is contradicted, (exhaustive) identification (3c), as in the cleft construction, or in association with focus particles like only (3d). In Reference RoothRooth’s (1992) broadly adopted theory of focus, these focus types do not constitute bona fide focus categories, but simply different uses of focus, which give rise to (or are motivated by) various pragmatic or semantic effects.
(3)
a. Sally visited [Moscow]F, even though she normally goes to [Warsaw]F.
contrast b. A Last week, Sally visited Warsaw. B No, she visited [Moscow]F. correction c. It was [Moscow]F that Sally visited last week. identification d. Sally only visited [Moscow]F. associated
Any kind of syntactic constituent can be focused – not just an object or more generally argument, but also a verb, an adjunct, a nominal attribute, and of course also larger constituents like the whole verb phrase, the whole clause or even a complex sentence. Focus thus can be of different ‘sizes’ – from ‘narrow’ focus (illustrated above), via VP/‘intermediate’ focus, to clausal/‘broad’ focus.
We will see in Section 6.3.1 that focus in Slavic languages is reflected by prosodic prominence.
6.2.2 Given (vs. New)
A constituent is given if it has a synonymic or hyponymic antecedent in the immediately preceding discourse (Reference Chafe and LiChafe 1976, Reference SchmerlingSchmerling 1976, Reference LaddLadd 1980, Reference SchwarzschildSchwarzschild 1999, Reference KrifkaKrifka 2008, Reference Wagner, Kučerová and NeelemanWagner 2012, Reference Rochemont, Féry and IshiharaRochemont 2016, Reference Kratzer and SelkirkKratzer & Selkirk 2020). Any non-given part of an utterance is new. Typically, it is the background to a focus that is given, as in (1B).Footnote 3 But not only that. In (4), for instance, the whole sentence I love cabbage/vegetables can be considered focus containing a given constituent, namely the object cabbage/vegetables. The object cabbage is given because it has a synonymic antecedent and the object vegetables is given because it has a hyponymic antecedent – in both cases the object cabbage in A’s immediately preceding utterance.
(4)
A I put some cabbage in the soup. B Sure, I love [cabbage/vegetables]G.
As illustrated by (4), givenness must not be confused with more narrowly defined co-reference (cf. Reference SzwedekSzwedek 2011): the object cabbage in B’s response has a kind reference (Reference ChierchiaChierchia 1998), while the cabbage in A’s lead-in utterance is existential and introduces an individual-level referent. Yet the notions are related and it is often the case that discourse-anaphoric expressions – including pronouns or definite NPs – are also given.
The kind of givenness introduced above is sometimes called discourse givenness because it is constituted by a discourse relation between a constituent and its discourse antecedent. Other kinds of givenness have been postulated, including referential givenness (co-reference) or visual givenness (extralinguistic presence in the utterance situation). It is discourse givenness, however, which has the most clearly detectable formal reflex, namely the lack of prosodic prominence (see Section 6.3.2).
6.2.3 Topic (vs. Comment)
The topic (also: sentence topic, aboutness topic) is the constituent referring to the entity that the sentence is ‘about’. The part of the sentence that conveys what is said about the topic is called the comment. The aboutness relation as the criterion for topichood has its roots in the early nineteenth-century approaches to information structure (Reference WeilWeil 1844, Reference PaulPaul 1880) and was also taken up by Reference MathesiusMathesius (1907, Reference Mathesius1939), although in a more linguistic (and less ‘psychological’) fashion. Terminologically, the opposition topic–comment (along with its current meaning) was introduced by Reference HockettHockett (1958) and Reference ChaoChao (1958). Reference ReinhartReinhart (1981) is generally considered to be the modern seminal work on the aboutness topic. For recent surveys, see Reference Büring, Féry and IshiharaBüring (2016), Reference Tomioka, Gutzmann, Matthewson, Meier, Rullmann and ZimmermannTomioka (2021). For a critical view of aboutness, see Reference JacobsJacobs (2001).
Aboutness topics – in contrast to foci or given expressions – are not easy to reliably elicit. The reason for this is that it is largely up to the speaker whether and in which way they choose to express the entity that the sentence is about.Footnote 4 Most linguists would agree that Viktora is the aboutness topic of (5B1). The issue is much less clear in B2. While the entity (Viktor) is expressed – by means of the clitic ho ‘him’ – opinions differ on whether the expression is prominent enough to be considered the topic of the sentence. This in turn has non-trivial consequences for the formal reflexes of the aboutness topic. If ho in B2 is considered the topic, then its formal expression (whether syntactic – such as the clause-initial position – or prosodic) is clearly optional (counter to the expression of focus and givenness, as we will see). If ho in B2 is not considered the topic, one can entertain the hypothesis that topic has some obligatory formal properties (such as clause-initiality or some pitch accent; see Section 6.3.3).Footnote 5
(5)
A Co mi můžeš říct o Viktorovi? what me.dat can.2sg say.inf about Viktor ‘What can you tell me about Viktor?’ B1 [Viktora]T včera zatkli. Viktor.acc yesterday arrested. ‘Viktor was arrested yesterday.’ B2 Včera hoT? zatkli. yesterday him arrested.pl ‘He was arrested yesterday.’
A more tangible subtype of topic is the so-called contrastive topic (Reference BüringBüring 2003, Reference Büring, Féry and Ishihara2016). A contrastive topic denotes what the sentence is about and at the same time implies the relevance of other topics, which may but need not be explicitly mentioned in the discourse. A contrastive topic is similar to focus in that it involves reference to alternative denotations. It is different in that it is always coupled – within the same sentence – by an additional focus. A’s question in (6) introduces the Novák siblings. I assume that it is A and B’s common ground that the siblings are Viktor and his sister Eva. In B1 Viktor is introduced as the contrastive topic – implying the contrast with the other sibling, namely Eva. The comment (subscripted by C) functions as the focus (F) of the sentence, as indicated by the F-subscript.Footnote 6 Another sentence can be added – this time about the contrasting entity Eva and with a different comment/focus, which in turn contrasts with the focus of the first sentence. What (6B2) shows is that contrastive topics (unlike ordinary aboutness topics, arguably) entail some formal properties: they are typically clause-initial and they must be able to carry a contrastive stress; the clitic ho does not satisfy either of these properties and hence is not a felicitous contrastive topic.
(6)
A Co mi můžeš říct o sourozencích Novákových? what me.dat can.2sg say.inf about siblings Novák ‘What can you tell me about the Novák siblings?’ B1 [Viktora]CT [včera zatkli]C/F (ale [Eva]CT je stále na svobodě]C/F). Viktor.acc yesterday arrested.pl but Eva is still on freedom ‘Viktor was arrested yesterday, but Eva is still free.’ B2 Včera ho*CT zatkli (…) yesterday him arrested.pl ‘He was arrested yesterday (…).’
6.3 Prosodic Reflexes of Information-Structural Notions
Thanks to their flexible word order, there has been a great deal of research into how information structure is reflected by word order in Slavic languages. Yet Slavic languages not only have a flexible word order, they also have a flexible prosody at the phrasal and sentence level. Slavic languages thus cannot easily be categorized in terms of Reference Vallduví, No and LibuchaVallduví’s (1990) and Reference Vallduví and EngdahlVallduví & Engdahl’s (1996) plasticity parameter (see Reference Šimík and WierzbaŠimík & Wierzba 2017 and Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. 2019 for a recent explicit argument along these lines): they are ‘plastic’ both in their word order (like some Romance languages) and prosody (like Germanic languages).
In what follows, I concentrate on the prosodic reflexes of information structure in Slavic languages. See Titov (2023) for the word order aspect. Section 6.4 briefly discusses the relation between the two.
The three IS notions have been argued to be associated with well-defined prosodic reflexes in Slavic languages: focus obligatorily attracts prosodic prominence and is associated with a falling tone; (contrastive) topic optionally attracts prosodic prominence and is associated with a rising tone; and finally, givenness is associated with the lack of prosodic prominence. Similar correspondences are attested in many other European languages (mostly Germanic, but also Greek or Finno-Ugric) and also beyond Europe; for discussion and references see Reference Kügler, Calhoun, Gussenhoven and ChenKügler & Calhoun (2020). Yet they are by no means universal; for recent relevant work (on Kinyarwanda and Rwandan English) see Reference Hamlaoui, Engelmann, Szendrői, Lee, Patin and RiedelHamlaoui et al. (2021).
6.3.1 Prosodic Reflexes of Focus
Slavic languages systematically exhibit the so-called stress–focus correspondence, which can be formulated as in (7). Sentence stress is understood as perceptually the most prominent stress in a sentence/clause.
(7) The focused constituent is realized with sentence stress.
An initial illustration from Czech is provided in (8). In B’s answer, the subject ‘the new pupil from the next class’ is focused and, at the same time, it is the constituent where sentence stress is realized (indicated by small caps on the focus exponent – the syllable carrying the stress) – no matter whether it is placed sentence-finally, as in B1, or sentence-initially, as in B2. The position of the stress within the focused phrase is determined by the nuclear stress rule, assigning greatest prominence to the rightmost stressed word within the phrase, which happens to be třídy ‘class’ in this case. Notice that the nuclear stress rule itself is independent of information structure. It applies by default, but can be overridden by information-structural considerations. Sentence stress in B1 is realized in accordance with the nuclear stress rule applied to the whole clause: the greatest prominence goes to the rightmost phrasal stress within the clause, which in turn happens to satisfy the stress–focus correspondence. If the nuclear stress rule were to apply in B2, the result would violate the stress–focus correspondence, which is why the stress is “shifted” to the focused phrase.
(8)
A Kdo bude hrát volejbal? who will play.inf volleyball ‘Who will play volleyball?’ B1 Volejbal bude hrát [ten nový žák z vedlejší třídy]F. volleyball will play.inf demnew pupil from next class B2 [Ten nový žák z vedlejší třídy]F bude hrát volejbal. dem new pupil from next class will play.inf volleyball ‘The new pupil from the other class will play volleyball.’
The obligatory stress–focus correspondence has been observed in most Slavic languages (e.g. Czech: Reference DanešDaneš 1957, Reference Daneš1959, Reference Daneš1960, Reference Sgall, Hajičová and PanevováSgall et al. 1986, Reference Groeben, Šimík, Kügler, Oseki, Esipova and HarvesGroeben et al. 2017, Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. 2019; Polish: Reference DogilDogil 1980, Reference Dogil, Williams and van der HulstDogil & Williams 1999, Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. 2019; Russian: Reference BryzgunovaBryzgunova 1977, Reference Bryzgunova and Švedova1980, Reference Alter, Junghanns and ZybatowAlter 1997a, Reference Alter, Lindseth and Franks1997b; Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn 2000; Bulgarian: Reference Arnaudova, Franks, King and YadroffArnaudova 2001, Reference Arnaudova2003, Reference Andreeva, Oliver, Franks, Gladney and Tasseva-KurktchievaAndreeva & Oliver 2005, Reference Andreeva, Barry and KoremanAndreeva et al. 2016; Slovenian: Reference Stopar, Komar and StoparStopar 2017; Serbo-Croatian: Reference GodjevacGodjevac 2000, Reference Godjevac2006; Ukrainian: Reference Féry, Paslawska and FanselowFéry et al. 2007). For a more theoretically oriented discussion, see Reference Chomsky and HalleChomsky & Halle (1968), Reference JackendoffJackendoff (1972), Reference GussenhovenGussenhoven (1983), Reference TruckenbrodtTruckenbrodt (1995), and Reference ReinhartReinhart (2006), among many others.
The primary and best-studied acoustic correlate of what I refer to as ‘stress’ here (a perceptual/phonological category) is fundamental frequency f0 (pitch or more precisely pitch movement on the stressed syllable), but other acoustic parameters also play a role, including duration (stressed syllables last longer) or amplitude/intensity (stressed syllables are louder); see the above-cited literature. In what follows, we concentrate on fundamental frequency/pitch.
The pitch movement usually ascribed to the focus exponent (the stressed syllable) in many Slavic languages is high tone followed by a fall (and sometimes preceded by a rise from a low tone), typically indicated by the sequence HL (or LHL) notation in the ToBI system (abbreviation of ‘tones and break indices’).Footnote 7 The alignment of the (rise-)falling tone with the stressed syllable can differ depending, among other factors, on the particular language, focus position within the clause, pragmatic focus type, and syntactic focus size. As an initial illustration, consider Figure 6.1, representing the fundamental frequency development while uttering (9), a Russian statement with narrow non-contrastive focus (response to a wh-question). Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn (2000) note that the focus exponent – the first syllable of Jaltu – is aligned with the low tone preceded by a fall from a high tone, whence the HL* ToBI notation.
(9)
Russian (Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn 2000: 417) Miroslava uexala [v Jaltu]F. Miroslava left to Jalta ‘Miroslava left for Jalta.’
Figure 6.1 Fundamental frequency/f0 (vertical axis; Hz) development in time (horizontal axis; msec) of utterance (9) used non-contrastively
Note. The portion of the contour with square data points corresponds to the focus exponent.
Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn (2000) further note that the HL* contour is characteristic for focus exponents of rightmost foci of any size, not just for narrow focus as in (9), but also for VP and sentence focus (focus on uexala v Jaltu and Miroslava uexala v Jaltu, respectively), although the overall contour differs somewhat.Footnote 8 If the focus is narrow as in (9) the pre-focal (backgrounded) part of the utterance retains a high pitch and the pre-focal stress (on the verb) is not that pronounced. That is in line with the verb being backgrounded/given (see Section 6.3.2). The high pitch on the verb is in turn preceded by a pitch movement related to the subject, which may (but need not be) interpreted as a topic accent (see Section 6.3.3).
Figure 6.2 demonstrates the contour of a contrastive/corrective use of (10). There are two differences to the non-contrastive use: the target of the pitch accent is a high tone, which is preceded by a low tone – LH*. The high tone is then followed by a fall to a low tone, which Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn (2000) interpret as a part of the complex pitch accent (LH*L).Footnote 9 Two other interpretations of the complex tone are imaginable. First, it could be that the pitch accent is terminated by the high tone (LH*) and is followed by gradual fall towards a low phrase accent (L−; cf. Reference GodjevacGodjevac’s 2000 treatment of Serbo-Croatian) or perhaps to a low boundary tone (L%) associated with the intonation of a statement. Second – and more consistently with the received view – the tone could be primarily falling – H*L – and could be preceded by a left boundary tone %L.
(10)
Russian (Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn 2000: 419) V Jaltu uexala [Miroslava]F. to Jalta left Miroslava ‘Miroslava [not Ljuda] left for Jalta.’
Figure 6.2 Fundamental frequency/f0 (vertical axis; Hz) development in time (horizontal axis; msec) of utterance (10) used contrastively
Note. The portion of the contour with square data points corresponds to the focus exponent.
The contrastive pitch accent observed in Figure 6.2 is realized independently of the position of the contrastive constituent – not just clause-finally as in (9), but also clause-initially and clause-medially – and not just for argument focus, but also verum/verb focus.Footnote 10 The pitch range (frequency difference between the low tone and the high target) is greater for initial and medial foci. In all cases, non-focal constituents have a relatively flat intonation (with the exception of the potential high tone of a topic, as visible in Figure 6.2; see Section 6.3.3), which corresponds to the lack of stress of given constituents (Section 6.3.2).Footnote 11
A comparable pitch contour with a high target has been reported for many other Slavic languages, too. For Slovenian, Reference Stopar, Komar and StoparStopar (2017) reports a high tone preceded and followed by a low tone on the focus exponent in non-contrastive uses, independently of focus size. Like Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn (2000), he observes a greater pitch range in contrastive focus exponents. The high target is preceded by sharp pitch fall on the pre-exponent syllable and followed by a similarly sharp fall located on the exponent (Stopar does not use ToBI, but the accent could probably be labeled H*L).
Reference Andreeva, Oliver, Franks, Gladney and Tasseva-KurktchievaAndreeva & Oliver (2005) analyzed focus accents in Bulgarian and Polish. Unlike Reference Zybatow, Mehlhorn, King and SekerinaZybatow & Mehlhorn (2000) or Reference Stopar, Komar and StoparStopar (2017), whose data were averaged over a large number of speakers, Andreeva & Oliver only had two speakers per language, making it difficult to draw solid generalizations. And indeed, they observe significant individual differences. If we abstract away from details, we can conclude that Bulgarian focus exponents generally carry high tone (H*) in broad focus contexts and a rising tone (LH*) in narrow and contrastive foci. Polish broad focus accent (and for one speaker also narrow/contrastive) is realized by a falling tone according to the authors (either HL* or H*L), while a rising accent (LH*) is reserved for narrow or contrastive foci.
Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. (2019) provide a detailed acoustic analysis of focus marking in Polish and Czech. They concentrate on stress shift within adjective–noun sequences, where the target structures involve focus on the adjective. They demonstrate that focus – independently of its type (answerhood, correction, contrast, parallelism) – is consistently expressed by a tonal rise in both languages (presumably targeting at H*), followed by a sharp fall on the post-exponent syllables. Overall, no significant differences between the different focus types are observed. This may well follow from the fact that the focus exponent was not located in the standard rightmost position within the phrase investigated and hence the stress was treated as ‘contrastive’ across the board.Footnote 12
A qualitatively different situation obtains in Serbo-Croatian. Serbo-Croatian is a language with lexical pitch accents, which means that pitch accent – in particular two tones (rising and falling, both of which can be short and long; see e.g. Reference Browne and McCawleyBrowne & McCawley 1965/Reference Browne, McCawley and Fudge1973; Reference Inkelas and ZecInkelas & Zec 1988) – can be used to distinguish different word meanings. For this reason, Reference GodjevacGodjevac (2000, Reference Godjevac2006) argues that pitch accent is not used for the expression of focus exponents. Instead, focus exponence is reflected by a low phrase accent (indicated by L−). While the stressed word realizes the lexical pitch accent, L− affects the subsequent (post-focal) material. The phonetic effect is a pitch range compression (an overall lowering of the pitch) relative to which any further lexical pitch accents are realized. Despite this difference, the overall pitch pattern appears to be close to that observed for other Slavic languages: the focus exponent is realized in a relatively high tone register (corresponding to the H tone typically implicated in focus exponence in other Slavic languages), followed by a tonal compression, which is attributed to the L− phrase accent in Serbo-Croatian by Godjevac, and possibly to the destressing of given material in other Slavic languages.
In summary, focus in all Slavic languages is reflected by prosodic prominence, as captured by the stress–focus correspondence in (7). The phonetic realization may differ slightly depending on the language (and probably even speaker; cf. Reference Andreeva, Oliver, Franks, Gladney and Tasseva-KurktchievaAndreeva & Oliver 2005), focus type (non-contrastive vs. contrastive), and possibly focus size (narrow vs. broad). The common denominator appears to be a high pitch accent preceded and/or followed by a low tone. Individual accents may differ in the exact alignment with the focus exponent (stressed syllable) – in most cases, the syllable is aligned with the high tone, but alignment with a low tone is also attested. Contrastive focus exponents are characterized by an extended pitch range in the complex accent (a greater rise/fall to/from the high tone). A different type of analysis has been put forth by Reference GodjevacGodjevac (2000, Reference Godjevac2006) for Serbo-Croatian, whose pitch accents are primarily used not for information-structural but lexical purposes.
6.3.2 Prosodic Reflexes of Givenness
Givenness has mostly been studied in tandem with focus. The reason for this is that the background to focus counts as given under any standard definition of givenness, including the one provided in Section 6.2.2. It therefore does not come as a surprise that there is a correspondence between being backgrounded and not being stressed. Yet the generalization is stronger: it turns out that given constituents in general (not just backgrounded ones) cannot be realized with sentence stress; (11).
(11) A given constituent is not realized with sentence stress.
An initial illustration from Czech is provided in (12). B’s answer to A’s question involves broad (sentence) focus. Yet the focused clause contains a given constituent, namely the object Martu. While in all-new settings, the order of a direct object and a directional PP (here: do nemocnice ‘to hospital’) is free (Reference Šimík and WierzbaŠimík & Wierzba 2017: experiment ‘all new’), the given object Martu in B’s answer must not be located in the rightmost position where sentence stress is normally realized. That is, not only must a background to focus lack sentence stress, the same holds of a given constituent located within a focused constituent.Footnote 13
(12)
Czech (Reference Šimík and WierzbaŠimík & Wierzba 2017: 688; judgments in accordance with exp. results) A ‘Do you have an idea why Marta made a phone call?’ B1 #Protože prý [teta poveze do because allegedly aunt. nom take to nemocnice MartuG]F. hospital Marta. acc B2 Protože prý [teta poveze MartuG because allegedly aunt. nom take Marta. acc do nemocnice]F. to hospital ‘Because allegedly the aunt will take Marta to the hospital.’
The observation that given expressions avoid sentence stress (or even stress more generally) in Slavic languages is by no means new, but has been discussed much less frequently than the stress–focus correspondence (Czech: Reference PetříkPetřík 1938, Reference DanešDaneš 1957, Reference Šimík and WierzbaŠimík & Wierzba 2015; Polish: Reference SzwedekSzwedek 1974, Reference Szwedek2011; Czech/Slovak/Polish: Reference Šimík and WierzbaŠimík & Wierzba 2017; Ukrainian: Reference Antonyuk-Yudina, Mykhaylyk, Kan, Moore-Cantwell and StaubsAntonyuk-Yudina & Mykhaylyk 2013). For a theoretically oriented discussion, see Reference Féry and Samek-LodoviciFéry & Samek-Lodovici (2006), Reference TruckenbrodtTruckenbrodt (2012), and Reference Wagner, Kučerová and NeelemanWagner (2012), among others.
When it comes to the phonetic realization of givenness, it is generally assumed that ‘lacking stress’ does not necessarily correspond to lacking any kind of pitch accent.Footnote 14 In fact, given constituents can be realized with word and phrase stresses if these occur in pre-nuclear positions, that is, prior to sentence stress. In (12B2), for instance, Martu can be realized with word/phrasal accent despite its givenness.Footnote 15 This is because it is the rightmost stress that is perceived as the most prominent one. See for example Reference Antonyuk-Yudina, Mykhaylyk, Kan, Moore-Cantwell and StaubsAntonyuk-Yudina & Mykhaylyk (2013) for Ukrainian production data which are consistent with this stance. Whether post-focal word/phrase stress on given constituents is attested is an open issue subject to further inquiry; to date, there is little systematic empirical investigation into this question. One relevant recent finding comes from Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. (2019), a study mentioned already in Section 6.3.1. Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. (2019) investigated the prosodic realization of focus in Czech and Polish adjective–noun sequences with focus on the adjective. A side effect of the focusing of the adjective was the backgrounding/givenness of the post-focal noun.
Figures 6.3 and 6.4 demonstrate the contours discovered by Reference Hamlaoui, Żygis, Engelmann and WagnerHamlaoui et al. (2019). In both languages, the focus exponent in the focus conditions (parallelism, wh, contrast, correction) is realized with the main stress – presumably an (L)H* pitch accent. This accent is followed by a sharp pitch fall after the accented syllable (mo in Czech, wa in Polish), which is motivated by the givenness/backgrounding of the noun. Compare the focus conditions with the new and coordinated condition, where the noun is new and where the pitch fall is much less pronounced. What is also of interest is the tonal realization of the given noun: in Czech, the noun (its first syllable) clearly carries a pitch accent; in Polish, it does not. Extrapolating from Hamlaoui et al.’s finding, we could hypothesize that the situation in Czech vs. Polish may well represent two ways of dealing with the requirement for given constituents not to be stressed. While Polish given expressions in a post-focal area are completely destressed, in Czech such expressions retain stress (at least word-level stress), realized as a pitch accent, but the overall pitch register in the post-focal area is significantly compressed.
Figure 6.3 Time-normalized f0 contours for item 24 (modré balonky ‘blue balloons’) in Czech, averaged across all speakers
Figure 6.4 Time-normalized f0 contours for item 16 (murowane domy ‘brick houses’) in Polish, averaged across all speakers
6.3.3 Prosodic Reflexes of Topic
Unlike focus and givenness, aboutness topic seems to have no obligatory prosodic reflex in Slavic languages, although the issue is heavily dependent on the exact definition of the notion of topic (see Section 6.2.3). What does have a very consistent prosodic expression is the notion of contrastive topic, which, similarly to focus, is obligatorily stressed. This can be captured by the generalization in (13). Notice that (13), unlike the previous statements concerning focus and givenness, does not make reference to sentence stress, but just stress. This is because contrastive topics bear stress, but not sentence stress (which remains with the focus).
The prosodic reflexes of topic have received relatively little systematic attention in Slavic linguistics. Reference Mehlhorn and ZybatowMehlhorn & Zybatow (2000) and Reference JunghannsAlter & Junghanns (2002), who systematically investigate topic prosody in Russian, claim that topics are reflected by a rising accent – either LH* or L*H. This accent is put into opposition with the Russian focus accent, which (as discussed in Section 6.3.1) is primarily falling (i.e. HL*) and which typically if not necessarily follows the clause-initial topic accent.
Reference JunghannsAlter & Junghanns (2002) report the same kind of topic accent for both ‘internal’ and ‘external’ topics. Internal topics are plain aboutness topics illustrated in Section 6.2.3. External topics are syntactically external to the core clause and are resumed by a pronominal. The only difference observed is the tendency for an intonational pause to occur after external topics; the pitch accent remains unaffected.
Moreover, Reference Mehlhorn and ZybatowMehlhorn & Zybatow (2000) report the same kind of accent for aboutness topics and contrastive topics. The only difference is that contrastive topics are optionally realized with a greater pitch excursion, that is, a greater rise. It remains an open issue to what extent the difference is distinctive (phonological).
Despite the above-mentioned findings, there is an important concern about the putative ‘topic accent’. As Reference Jasinskaja, Féry and IshiharaJasinskaja (2016) correctly notes – in line with the above-cited works – the rising accent associated with topics is optionally used for any pre-nuclear phrasal stress, independently of the information status of the stressed expression. The question is, then, whether it is substantiated to call the rising accent a dedicated ‘topic accent’. As I showed in Section 6.2.3, the very status of aboutness topic is questionable, shedding further doubt on the linguistic reality of this information-structural notion (see Reference Büring, Féry and IshiharaBüring 2016 for a similar general concern).
The situation looks more hopeful for contrastive topics, in line with the conjectured generalization (13). Contrastive topics are obligatorily prominent – they cannot be realized by clitics, for instance. This prominence very often corresponds to prosodic prominence or, more precisely, to the rising accent discussed above.
6.4 Interactions between Prosody and Word Order
Both prosody and word order are traditionally assumed to play important roles in the expression of information structure. Slavic flexible word order has attracted a lot of attention over the many years of research, and linguists have frequently claimed the primacy of word order alternations in expressing information-structural categories (recent arguments to this effect can be found in Reference SlioussarSlioussar 2007 for Russian or Reference KučerováKučerová 2007 for Czech). Yet there is also a competing view according to which prosody is more important and the word order flexibility is exploited for prosodic optimization.Footnote 16 In this short section, I use two kinds of examples to illustrate this approach.
Slavic languages exhibit a strong preference for default rightmost stress placement. This is normally captured by the so-called nuclear stress rule.Footnote 17 This rule is not absolute and can be violated if there is sufficient – typically information-structural – motivation. One relevant example was provided in (8), where the B2 answer involved clause-initial subject focus and therefore clause-initial stress nuclear stress, in accordance with the stress–focus correspondence. As illustrated by (8B1), however, a word order alternation (the OVS order) can satisfy the stress–focus correspondence without violating the nuclear stress rule. Indeed, this latter possibility has been proposed, for example by Reference Arnaudova, Franks, King and YadroffArnaudova (2001, Reference Arnaudova2003) for Bulgarian subject-final orders, illustrated in (14). The idea is that non-canonical order is used in order for rightmost stress to be applied at the clausal level.
(14)
Bulgarian (Reference ArnaudovaArnaudova 2003: 115–116) Včera pročete knigata Marija. yesterday read.past.3sg book.def Marija ‘Marija read the book yesterday.’
A similar logic can be applied to givenness-related word order alternations (see Reference Šimík and WierzbaŠimík & Wierzba 2015, Reference Šimík and Wierzba2017, but also Reference SzwedekSzwedek 1974). In Section 6.3.2, we have seen that given expressions avoid sentence stress. If a given expression appears by default in the clause-final position, there are in principle two ways of not stressing the word: either the stress is shifted to a non-final word/phrase or the given expression is realized in a non-final position, giving rise to a non-canonical word order. Reference Šimík and WierzbaŠimík & Wierzba (2017) provided experimental evidence that Czech, Slovak, and Polish allow for both strategies, albeit with different preferences: Czech and Slovak prefer non-canonical word order (satisfying the nuclear stress rule) and Polish prefers to shift the stress and stick to the canonical order of constituents (SVO). The respective preferred variants are illustrated for Czech and Polish in (15). It is also good to realize that both the variants are in accordance with the requirement for given expressions not to be realized with sentence stress. Any variant of (15) where the given object was realized with sentence stress was rated as significantly less acceptable.Footnote 18
(15)
From experimental materials of Reference Šimík and WierzbaŠimík & Wierzba (2017) A ‘I heard that we have to show our IDs at the municipality.’ B (i) Czech Myslím, že Frank [občanku]G ztratil. think.1sg that Frank ID lost (ii) Polish Myślę, że Frank zgubił [dowód]G. think.1sg that Frank lost ID ‘I think that Frank lost his ID.’
It is good to note that not all syntactic configurations allow for constituent reordering. In cases like this, illustrated by the focus on the prenominal attribute in Russian (16), stress shift might be the only way of satisfying the correspondences between information structure and prosody.Footnote 19
(16)
Russian Ja kupil [novye]F [knigi]G. I bought new books ‘I bought new [not old] books.’
In summary, we have seen that word order alternations do not necessarily imply a word-order-based expression of information structure. They could also be motivated by complying with independent prosodic requirements, in particular the default (information structure-independent) nuclear stress rule, the stress–focus correspondence, and the ban on stressing given expressions. Yet prosodically motivated reordering is typically not necessary (and sometimes not even possible) in Slavic languages. Given that Slavic languages are flexible in both word order and prosody, as noted already in the introduction to Section 6.3, prosodically motivated reordering is merely a matter of preference and can be subject to individual and cross-Slavic variation.
6.5 Conclusion and Open Issues
There is solid evidence that information structure is reflected prosodically in Slavic languages. Focus is realized with sentence stress, givenness by the lack of such stress, and topic (optionally) by pre-nuclear stress. The primary acoustic correlate of stress is fundamental frequency, and in particular various kinds of pitch accents; focus is often related to a falling accent and topic/pre-nuclear stress to a rising accent. However, other phonetic parameters, including duration and intensity also play a role.
Information structure is of course also expressed by word order in Slavic languages. At least to some extent, however, the word-order-based expression is arguably derivative of prosody in the sense that word order alternations are motivated by optimal prosodic realization. An example of this is the tendency for clause-final focus placement, which follows from the conjunction of stress–focus correspondence and the information-structure-independent nuclear stress rule.
Despite many valuable findings and generalizations, systematic and methodologically robust investigations into the interface between information structure and prosody are still missing for most Slavic languages. While the general tendencies for focus to be stressed and for given expressions to be unstressed are strong and likely to be replicated across individual Slavic languages, many particular issues remain unresolved, including the distribution of different accent types (across focus types or sizes), the precise nature of the tonal events (pitch accents vs. phrase accents vs. boundary tones), the nature of the post-focal (and possibly pre-focal) pitch compression, or the issue of ‘destressing’ given constituents (complete loss of prosodic prominence vs. loss of greatest prominence). Finally, it remains to be seen if any specific pitch accent or tonal event is associated with topic – be it aboutness or contrastive topic. The investigations are scarce and their results not very satisfactory.
