1 Vincent van Heuven: Acoustic Correlates and Perceptual Cues of Word and Sentence Stress: Towards a Cross-linguistic Perspective

In this chapter the focus of interest is the phonetic realisation of stress at the word and sentence level. While not dealing with the physiological basis of stress (but see van Heuven and Sluijter Reference Heuven, Sluijter, Goedemans, van der Hulst and Visch1996 for more discussion), it concentrates on the acoustic consequences of increased versus decreased effort and asks (i) what acoustic correlates can be found for the difference between a stressed syllable and its unstressed counterpart, and (ii) what the relative importance is of each acoustic correlate in the marking of stress. At the same time, the author considers the question of which acoustic properties are used by human listeners and to what extent these are used to decide whether or not a syllable is stressed. Van Heuven makes a strict terminological distinction between acoustic correlates of stress (which can be used, for instance, to identify a stressed syllable by some computer algorithm) and the perceptual cues used by the human listener, showing that some acoustic correlates, notably the (peak) intensity of a syllable, allow excellent separation of stressed from unstressed syllables but are hardly used by the human listener.

2 Larry Hyman: Positional Prominence versus Word Accent: Is There a Difference?

Hyman addresses one of the major unresolved issues in the study of word-accentual systems, which is determining what exactly counts as accent, a problem which is further complicated in languages with tone or so-called pitch-accent (see the above remarks and the introductory chapter in van der Hulst Reference Hulst2014c for a general discussion of such matters). In this chapter, Hyman analyses three African cases, each of which display diverse positional prominence effects that are clearly word level, reasonably subject to a metrical (accentual) interpretation, but which do not consistently coincide. In Ibibio, a Cross-River language spoken in Nigeria, greater consonant and vowel contrasts suggest that the initial stem syllable is the accented head of a trochaic foot, whose required shape varies, however, by construction. In Punu, a Bantu language spoken in Gabon, tone suggests that the word-penultimate syllable is accented, while vowel length suggests that both stem-initial and word-penultimate syllables are accented. In Lulamogi, a small, understudied Bantu language spoken in Uganda, vowel length suggests that all stem (versus prefix) and word-penultimate syllables are accented, while tone suggests it is the penult. While some, or all, of these instances of positional prominence resemble what is found in stress-accent systems, Hyman concludes that we should focus more on the specific properties of prominent positions and less on what we call them.

3 Anya Lunden: Explaining Word-Final Stress Lapse

Focusing on rhythmic stress, this chapter proposes and examines evidence for a motivation behind the well-known asymmetrical tolerance for a stress lapse word-initially versus word-finally across languages. While many binary-stress languages tolerate a stress lapse at the right edge of the word, very few tolerate a stress lapse at the left edge (see Gordon Reference Gordon2002; van der Hulst 2014b). Lunden proposes that, in the languages that tolerate a final stress lapse, there is nevertheless a rhythmic alternation present at the right edge of the word, due to the phonetic effect of word-level final lengthening. However, if final lengthening were able to perceptually contribute to a word’s rhythm in cases of final lapse, we would expect this to only be possible when stress in the language has duration as a stress correlate. Evidence from two different sources is shown to support the connection between languages’ tolerance of final stress lapse and their use of duration in stress. Drawing on a database of stress correlates that the author made for this research project, it is shown that languages which tolerate a final stress lapse are indeed extremely likely to have duration as a stress correlate, whereas no such correlation exists for final lapse and the stress correlates of pitch or intensity. Several perception experiments also support this connection: final lengthening was found to be confusable with stress only if stressed syllables included increased duration. Finally, an account is sketched of how this could be captured in an Optimality Theoretic (OT) analysis.

4 Natalia Kuznetsova: What Danish and Estonian Can Show to a Modern Word-Prosodic Typology

The typology of word prosody is still a subject of hot debate (see van der Hulst Reference Hulst and van der Hulst2014a). Kuznetsova asserts that tone and stress-accent remain the central units of classification, but shows that there is no established consensus about a definition of these notions. In this chapter, she focuses on two specific word-prosodic units with a non-pitch-based primary phonetic exponent: prosodic quantity in Standard Estonian and prosodic laryngealization in Copenhagen Danish. Kuznetsova summarizes their main phonetic and functional features. She also compares these prosodic units with functionally similar cases of pitch-based word prosody in other languages within what she calls the framework of mainstream word-prosodic typology. Both cases are challenging for the typology, as they do not qualify either as tone or as stress. In the end, she proposes a view on the word-prosodic typology which incorporates a clear separation between the variable of location and ways in which the location is realized or cued (see also van der Hulst Reference Hulst and van der Hulst1999a, Reference Hulst, van der Hulst, Goedemans and van Zanten2010, 2014). At the word level, we find three logically possible location values: (i) ‘no prosody at the word level’, (ii) ‘prosodic marks on some syllables of a word’, (iii) ‘prosodic marks on every syllable of a word’. Case (i) would imply that no phonetic cues have phonological relevance at the word level, the tentative term for this could be ‘non-accentual prosody’. This is the case, for example, in French, where all prosody can be defined at the post-lexical levels. Case (iii) is what is prototypically called ‘tone’. Case (ii) is what Hulst (Reference Hulst, van Oostendorp, Even, Hume and Rice2011) proposes to call ‘accent’. She concludes that an extensive typology of all the accent varieties should be based on accurate descriptions of the word prosody of particular languages.

5 René Kager and Violeta Martínez-Paricio: Mora and Syllable Accentuation: Typology and Representation

This chapter has two goals. First, the authors argue that metrical feet can immediately dominate morae. This situation occurs only under duress of metrical foot form constraints, which impose strict requirements on the number of morae in the head and dependent positions in metrical feet. The authors propose to encode this situation in terms of the internally layered (IL) foot, a minimally recursive metrical foot (Martínez-Paricio Reference Martínez-Paricio, Arnett and Bennett2012, Reference 12Martínez-Paricio2013, Martínez-Paricio and Kager Reference Martínez-Paricio and Kager2015). They support this claim with data from Gilbertese, which exhibits a metrical distribution of high pitch and stress that disrespects syllable integrity. They show that this pattern can be analysed straightforwardly using IL feet. A second goal of this chapter is to show that IL feet offer an insightful account of ‘mora-counting’ metrical patterns in which prominence is located on the syllable that contains the antepenultimate mora, in particular in Tokyo Japanese loanword accentuation and Dihovo Macedonian stress. These patterns pose serious problems for a standard moraic trochee analysis, due to the fact that the weight sequences … LH# and … HL# behave similarly as trimoraic units in locating prominence on the penultimate syllable. Using IL feet, Kager and Martínez-Paricio propose analysing these systems as ‘mixed binary/ternary’: IL feet occur in sequences ending in light syllables, non-IL feet elsewhere. Additional evidence for this analysis comes from an innovative pattern of loanword accentuation of Tokyo Japanese, which differs from the conservative pattern in shifting the accent to the antepenultimate syllable precisely in … LH# sequences. On this analysis, this pattern has IL feet in all forms. The fact that Tokyo loanword accentuation shows signs of accentual instability and is moving in the direction of the Latin stress pattern can be interpreted as movement towards consistency in foot parsing.

6 Hisao Tokizaki: Word Stress, Pitch Accent, and Word Order Typology with Special Reference to Altaic

It has been claimed that in some languages the location of word stress correlates with the word order of a syntactic head and its complement – for example, a verb and its object (Donegan and Stamp Reference Donegan and Stampe1983, Plank Reference Plank1988) – that is, languages with left-hand stress have head-final order while languages with right-hand stress have head-initial order. Based on an analysis of the data in Haspelmath and Dryer (2005), Tokizaki shows that this correlation generally holds in the world’s languages. However, potential counterexamples to this generalization are Altaic languages, a large number of which have been reported to have right-hand stress and head-final order (see Goedemans, Heinz and van der Hulst Reference Goedemans, Heinz and van der Hulst2014 and data in Haspelmath et al. Reference Haspelmath, Dryer, Gil and Comrie2005). It is argued in this chapter that Altaic languages in fact have word-initial stress as well as right-hand pitch accent. Thus, the general correlation between stress and word order also holds for Altaic languages.

7 Keren Rice: Persistence and Change in Stem Prominence in Dene (Athabaskan) Languages

The stem has long been identified as a domain of prominence in Dene (Athabaskan) languages. Given this, one might ask if, in addition to the importance of morphology in the placement of prominence, phonological factors also play a role. In this chapter, Rice examines the role of morphology and of phonological constraints in a number of Dene languages, addressing the pathways of change that can be identified in this family. Overall, she notes that there is conservatism in the prominence systems reported for Athabaskan languages, with the root attracting prominence, all other things being equal. The phonological factor that plays a role is the placement of a trochaic foot. While the core of the reconstructed system is maintained, the trochee may shift from syllabic to moraic, with different factors involved in determining weight. Word-level prominence may be either left or right oriented. Prefixes may involve trochees as well. What is overall resistant to change is the domain of root.

8 Iggy Roca: Spanish Word Stress: An Updated Multidimensional Account

Generative Phonology research output on Spanish word stress spans just over half a century at the time of writing (1965–2016) and is substantial. However, no unanimity of analysis has as yet been achieved. This chapter provides both a précis of the historical peaks of the still ongoing endeavour and a proposal further elaborating the lines in Roca’s (Reference Roca, Martínez-Gil and Colina2006, Reference Roca2014, Reference Roca and Nuñez Cedeño2016) recent contributions. For reasons of space, Roca’s analysis is restricted to non-verbs. The proposed grammar uses OT as an analytical tool. It hinges on interactions between foot shape and size, on the one hand, and alignment relations between the metrical and morphological structures, on the other. The account is multidimensional in as much as it provides (i) a full formalisation of Spanish (non-verb) stress, (ii) a typology of its various materialisations, (iii) a historical justification for their emergence, (iv) an evaluation of some previous alternatives and (v) a conclusion bringing together the various strands.

9 Björn Köhnlein: Metrically Conditioned Pitch Accent in Uspanteko

Uspanteko, a Mayan language spoken in Guatemala, shows a remarkably rich interaction between the location of stress, vowel quality, syllable weight and pitch accent. Commonly, it is assumed that the language has privative lexical tone. Counter to previous analyses of the facts, this chapter proposes that both tonal contrasts and other relevant interactions can be derived from an opposition between trochaic and iambic feet. No tonal information is stored in the lexicon. While improving the empirical coverage of previous analyses with lexical tone, the current analysis adds little additional machinery, since the general distinction between trochees and iambs in Uspanteko has already been motivated on independent grounds. From a broader theoretical perspective, the chapter contributes to ongoing discussions on the phonological nature of tone-accent systems, one of the key issues in debates on prosodic typology.

10 Haruo Kubozono: Focus Prosody in Kagoshima Japanese

This chapter discusses focus prosody in Kagoshima Japanese, a dialect spoken in the south of Japan with a lexical prosodic system remarkably distinct from that of standard Tokyo Japanese. Starting with the phenomenon of question particle incorporation in direct (matrix) Wh-questions, Kubozono considers several apparently different phenomena involving the incorporation of sentence-final particles into the sentence-final prosodic phrase. He proposes that all these phenomena can be generalized as manifestations of focus prosody whereby sentential particles are incorporated into the domain of prosodic phrase in post-focal positions. This process can be attributed to the Obligatory Contour Principle (OCP) whereby sequences of High tones are avoided by a dissimilatory process of H tone deletion and a subsequent process of prosodic rephrasing. This analysis nicely explains why particle incorporation fails to occur in a particular accent class of words as well as in yes/no questions and other syntactic constructions. The chapter also argues that post-focal prosodic incorporation is not an isolated phenomenon in Kagoshima Japanese, but rather, that similar prosodic (re)phrasing phenomena are found in other dialects and languages such as standard Tokyo Japanese, Fukuoka Japanese and the south Kyungsang dialect of Korean.

11 Björn Köhnlein and Marc van Oostendorp : Where is the Dutch Stress System? Some New Data

There is an extensive body of theoretical work on the Dutch stress system, which is, however, mostly built on inspection of existing, sometimes rather exotic, words as found, for example, in dictionaries. In this chapter, Köhnlein and van Oostendorp confront these theories with new data from two online experiments in which participants had to indicate the most likely location for stress in biblical names or nonsense words respectively. The results partially confirm the claims already gathered in earlier work: there is a strong preference for stress on penultimate syllables and quantity plays a role in establishing the preferred location of stress. On the other hand, the way in which quantity works out is slightly different from what previous literature suggests. Furthermore, the authors report that they have not found strong evidence for the so-called three-syllable window.

12 Nicholas Rolle and Marine Vuillermet: Morphologically Assigned Accent and an Initial Three-Syllable Window in Ese’eja

In this chapter, Rolle and Vuillermet argue that Ese’eja demonstrates an unusual initial three-syllable window within which primary prominence must fall, a typologically rare type. Using a corpus of 2,000 elicited verb forms (Vuillermet Reference Vuillermet2012), the authors show that the position of prominence depends on syllable count and the type of morphologically assigned accent. They posit four types of this morphological accent: inherent transitive accent, dominant indexical accent, recessive accent with one set of tense/mood suffixes and rightmost-preserving accent with another. Further, tense/mood suffixes trigger the creation of iterative trochaic or iambic feet, which the authors capture using cophonology theory employing common OT constraints (Inkelas and Zoll Reference Inkelas and Zoll2007). The authors posit that iterative footing occurs with a leftmost constraint, resulting in primary accent falling on the first, second or third syllable, which is realized with primary prominence. Additionally, because iterative footing occurs prior to primary accent delegation, Ese’eja constitutes a true ‘count system’ challenging the Primary Accent First model (van der Hulst Reference Hulst, Goedemans, van der Hulst and Visch1996, Reference Hulst1997, Reference Hulst2012). Finally, they argue that when morphological accent in Ese’eja is assigned outside the metrical window, the position of primary prominence falls on a rhythmically dependent position, termed ‘rhythmic repair’. The authors contrast this to Kager’s (Reference Kager2012) typology, showing that under these circumstances primary prominence surfaces on a default position within the metrical window, termed ‘default repair’.

13 Alexandre Vaxman: A Scales-and-Parameters Approach to Morpheme-Specific Exceptions in Accent Assignment

This chapter addresses the long-standing problem of morphologically conditioned exceptions in accent assignment. Vaxman introduces a new approach, called the Scales-and-Parameters (S&P) theory, a new parametric, non-metrical theory of word accent, which takes as a point of departure the PAF theory of van der Hulst (Reference Hulst, Goedemans, van der Hulst and Visch1996, Reference Hulst1997, Reference Hulst2012, inter alia). The S&P theory is shown to uniformly capture regular and exceptional accent locations both within a given system and across different types of systems in terms of a single accentual grammar. As the author claims, the proposed grammar accurately derives accent location in lexical accent systems with dominant suffixes and in phonological weight-sensitive systems in which certain morphemes violate the accent rule. A core proposal of the theory is to extend the notion of ‘weight’ to morphemes by treating their accent-attracting ability as ‘diacritic weight’ (rather than lexical accent). Vaxman shows that since weight is a scalar variable, it allows for novel types of weight scales, that is, those containing diacritic and/or phonological weight. Reference to such scales allows the S&P parameter system to correctly assign word accent and to account for morpheme-specific exceptions, as illustrated here with detailed case studies of Central Selkup and Eastern Literary Mari.

We hope to have shown that the chapters in this volume cover a range of typological and theoretical issues in the study of stress/accent. On the typological side, various authors are concerned with the array of word-prosodic types and their distinguishing phonetic properties. Of specific theoretical concern is the question of foot structure and how it is manifested in different ways in different systems. In addition, several chapters explicitly discuss the role of morphological structure, as well as of lexical marking of accent. The 13 chapters collected here present a lively testimony of a field of inquiry that shows progress, while also identifying questions of ongoing concern.