Authors: Kellen Parker van Dam
Acknowledgements: David Inman, Natalia Chousou-Polydouri, Kellen Parker van Dam, Marine Vuillermet, Anna Graff
This feature set targets two components of autosegmental phonology: the presence and type of a tone system and the presence and type of a prominence (or accent) system.
Tone refers to any system in which non-metrical contrastive pitch is used to mark distinctions in the lexicon or between grammatical forms. Accent refers to any system of prominence marking within a prosodic word, typically primarily correlated with amplitude (often referred to as stress) or pitch. Many languages have predictable accent rules, such as Italian, where the stress-accent is generally on the penultimate syllable. English also has a stress-accent system, but with less regularity in the system. Accent may also rely on pitch rather than stress, and in some cases, both equally. In many introductory sources, accent is referred to simply as "stress". For this feature set, all systems of word-level prominence marking are considered as accent systems, regardless of the acoustic correlates.
Pitch-accent systems and tone systems may not always be easily distinguished from one another. Despite this difficulty, there are some ways to make the distinction. Pitch accents are generally characterized by culminativity of pitch (only one H tone is permitted per word) or positional restrictions, such as those seen in stress-accent systems (e.g., accent occurring primarily on the penult or ultimate syllable). Hyman (2009) makes the case that no such distinction between pitch-accent and tone can be made. However, a few languages can be analyzed as having both pitch-accent and lexical tone, such as Cherokee [cher1273] (Iroquoian; US) (Uchihara 2016). Therefore, for the purposes of this typological study, the two uses of contrastive pitch (pitch-accent and tone) are distinguished. A more in-depth account of the pitch-accent and tone distinction is explored further below (see also Downing 2004, Kubozono 2012, Jun et al 2006).
A typology of tone and accent systems is made difficult by the lack of a shared definition of canonical systems. Contrary to Hyman (Hyman 2009), whose canonical definition of tone includes “pitch-accent languages”, Ito and Kenstowicz (2017) state that "pitch-accent languages stand between stress languages … and tone languages … and share properties of both".
This feature set takes the approach that pitch-accent systems exist and can be identified by certain criteria, and that they exist on a continuum of possible acoustic correlates with stress-accent systems. On one end, amplitude is the sole salient feature of the accentual system, and on the other, pitch. Rarely, the salient feature of an accent system can be duration. In most cases, languages use both amplitude and pitch to mark prominence, but with only one of these two being the most salient or primary phonemic marker. Even languages which are often treated as exemplars of one system will in fact have multiple acoustic correlates of the other. Stress-accent languages tend to have higher pitch on accented syllables, as in Arabic [arab1395] (Afro-Asiatic; world) (De Jong and Zawaydeh 1999), and pitch-accent languages often have greater amplitude on accented syllables, but secondary to the primary acoustic correlate of accent. A sketch of the differences and similarities between tone and prominence systems can be seen in Table 1.
| Where | How | Distinctions | Contrastiveness | |
|---|---|---|---|---|
| Tone | syllable | pitch, sometimes phonation | several levels of pitch are not uncommon | Almost always |
| Prominence | word | pitch, amplitude duration (often a mix of 2 or 3) | Binary, sometimes ternary (primary vs secondary stress vs none) | Sometimes |
Although tone and prominence are sometimes treated separately, a number of pitch-accent languages are described as having “emergent” or “privative” tone systems (Hyman 2001). Likewise some typical tonal languages such as Shanghainese [shan1293] (Sino-Tibetan; China) have been described as having a system similar to pitch-accent, once tone-spreading sandhi is taken into account (Roberts 2019). While this survey makes the assumption that prominence and tone systems are distinguishable, by tracking the primary (and secondary) phonetic correlates of stress, we encode sufficient detail about the overlap between the two systems (i.e., both can make use of pitch). Thus, languages coded as having tone systems and languages coded as having pitch-accent systems can be grouped together by their shared contrastive use of autosegmental pitch, an analysis in line with Hyman (Hyman 2009), in which any contrastive use of pitch at the lexical level falls under the category of “tone”. However, pitch-accent languages can also be separated from tone (and grouped with stress-accent languages) according to the presence of a prominence system (regardless of its primary acoustical correlate). This detailed coding schema allows some of the blurriness between tone and pitch-accent to be captured. Cross-cutting between the categories of tone and prominence also creates a formal way to capture the hypothesis that tonal languages may influence the phonetics of stress systems by moving them in the direction of using pitch.
In addition to tone, amplitude based prominence marking (stress-accent) is also included in this study. Pitch-driven accent systems may correlate with a wider tonal region, as described above, and the presence of a prominence system may itself be areally significant, as well as the contrastiveness or non-contrastiveness of such a system. Since prominence systems must be investigated to check for the presence of pitch, prominence systems as a whole were cataloged and typologized according to their presence and their primary phonetic correlates.
This project immediately uncovered a descriptive problem, as many language varieties have not had their prominence systems instrumentally measured. Rather, these are often impressionistically documented by the linguist undertaking the study, and very often without the acoustic correlates of stress directly mentioned. In cases where this was the only data, such impressionistic descriptions were used in coding. In a few cases, more detailed phonetic studies were published. In such cases, these took priority. Three main prominence correlates were found: amplitude, pitch, and (much more rarely) duration. The effects of accent placement on vowel quality were not addressed in this study.
It is sometimes difficult to draw a clear line between different types of accent systems — i.e. whether they are of the stress-accent variety (distinguished primarily by amplitude) or the pitch-accent variety (distinguished primarily by pitch). In some cases, both correlates were coded as primary.
Accent systems can be either contrastive or non-contrastive. Non-contrastive accent systems are ones in which accent occurs in a systematic manner in generally consistent locations within the word. An example of such a system is penultimate stress in Italian, where accent is predictable. See below for more information. Contrastive accent systems are ones in which accent is used contrastively to distinguish between parts of speech e.g. English óbject (n) vs objéct (v), meanings e.g. Japanese [nucl1643] (Japonic; Japan) 箸 háshi "chopsticks" versus 橋 hashí "bridge" (Kubozono 2017), or other differences. Such systems will generally have unmarked default patterns, with major contrasts being marked. See below for more information.
An issue that comes up in coding is whether a language's accentual system should be called contrastive or not. In many languages, the placement is generally rule-governed, but will have many exceptions. These are the cases where the decision can be more difficult as to whether or not it should be coded as contrastive or not.
The general rules that have been followed in coding this distinction is to code as non-contrastive if the following are true:
Most languages will be addressable by the first two points alone. The third requirement may be the most important in edge cases, but should not be the primary factor if requirements in points 1 and 2 are addressed in the description.
There will undoubtedly be cases where the coding of contrastiveness may not perfectly align with this. Contrastiveness of accent placement is not necessarily a binary distinction. I have done my best to approach coding the feature in a principled way, but there have been cases where it was still not an easy call to make even with the above rules, in large part due to the descriptions given for a feature which often does not get much attention in grammars. An example of a language with exceptions but which is still coded as non-contrastive is Tuscarora [tusc1257] (Iroquoian; US, Canada) (Williams 1974: 280f) which has an overall rule-governed system of accent placement, and in which the exceptions are minimal with clearly explained origins. On the other hand, Panare [enap1235] (Cariban; Venezuela) (Payne and Payne 2013: 43) is described as having a contrastive accent system, despite the patterns being mostly predictable, due to the high number of exceptional patterns.
A language which has entirely predictable accent placement but with which the placement consistently distinguishes between parts of speech – as is the case with English noun-verb stress placement on homographs – would still be considered contrastive, since prominence in such a system would serve to contrast between parts of speech (in addition to syntactic cues).
The presence of lexical tone is described as an areal feature for the Pueblo region (Bereznak 1995, Campbell 1997), Mesoamerica (Maddieson 2013), Southern and North-Western Amazonia (Rondônia and Mato Grosso, and along the Vaupés river, from Brazil into Peru; Aikhenvald 1999: 10) and sometimes for part of the Pacific Northwest (Campbell 1997), although there it is suspected that that is limited to a small subset (Tlingit, Haida, and Athabaskan languages). It is also well described as an areal feature for regions outside the Americas, such as in South and East Asia (Kirby and Brunelle 2017) as well as multiple regions in Africa (Hyman et al 2015). Within Asia, the presence of a tonal sprachbund has been argued as the source for the pitch-accent system found in some modern Korean dialects (Ito and Kenstowicz 2017). It is therefore possible that such a tone-and-pitch area could be found elsewhere, and is therefore worth investigating across the Americas.
{ yes | no }
This may also be described as a stress system, an accent system, including metrical- and/or stress-accent. It is some form of syllabic prominence within a word which can be based on pitch, amplitude (loudness) or duration. Different types are exemplified in the definitions above. This can be distinct from phonemic vowel-length distinctions which may exist outside of a tone/accent system, unless such vowel length distinctions are inherently tied to the metrical system and syllable prominence.
In Ese Ejja, there is a well-described non-contrastive pitch-accent system, where the accent consistently falls in a predictable location within the prosodic word (Vuillermet 2012).
In Iu Mien, a tonal language of Southeast Asia, there is a system of lexical tone with high functional load. It is described as syllable-timed — i.e. each syllable is described as taking equal weight —and has no word-level marking for prominence (Arisawa 2016).
While evidence of isochrony in human languages is not well established despite many attempts to do so (Russo and Barry 2008), and the notion of a language being syllable-timed or stress-timed is often a case of impressionistic judgment, it is still helpful in coding data for this feature as it gives further insight into what the linguist writing the description may have been hearing.
Languages described as stress-timed are ones in which the accented/stressed syllables occur at approximately regular intervals, with unstressed syllables varying in tempo as needed to fill in the gaps. American English is one such example.
{ NA | ; separated list of [ amplitude & pitch & duration ] }
Yuchi has an accent system in which amplitude (loudness) is the only required acoustical correlate of prominence. Duration and pitch are secondary, and optional (Linn 2001: 83f).
Northwest Maidu has a system which uses both equally. Neither are secondary to the other in salience or significance (Ultan 1967: 24f).
Dukha is a rare case where two accentual systems coexist (Ragagnin 2011: 28). One uses pitch as the marker of prominence, and the other amplitude. The amplitude-based (stress-accent) system is contrastive and can result in a difference in meaning of the word, while the pitch-accent system always falls on the last syllable of those able to be accented based on prosodic rules, regardless of the actual placement of the stress-accent.
{ NA | yes | no }
Yuchi relies on amplitude to mark word-level prominence. Pitch, as is often the case, is a secondary phonetic correlate of amplitude in the accent system, but is not necessary or consistently produced (Linn 2001: 83f).
Dukha has one system which uses pitch, as well as having another system which relies on amplitude (Ragagnin 2011: 28).
Tokyo Japanese accent is marked by pitch, without any major additional loudness or duration effects. Pitch serves to distinguish between lexical items, e.g. 箸 háshi "chopsticks" vs 橋 hashí "bridge". Duration is highly limited by the mora-based system of timing, and the presence of amplitude is negligible.
{ NA | contrastive | non-contrastive }
This feature asks if the pitch-accent system follows predictable and generally consistent rules. If the system is described as having prominence in a particular place within a word and that the prominent syllable is predictably in that location for the entire lexicon or part of speech, then we code it as non-contrastive. A contrastive accent system would be something like English, in which stress is regularly used to distinguish parts of speech (e.g. object as a noun vs a verb), despite otherwise having a more general pattern of where stress falls.
A non-contrastive pitch-accent system is one where pitch is the primary realization of accent, and the accentual system is governed by systematic rules without significant exceptions, and if exceptions exist, they are rare.
Cavineña is described in Guillaume (Guillaume 2008) as having a non-contrastive pitch-accent system. It is addressed in Campbell's Indigenous Languages of South America (Campbell 2012) as having suprasegmentals which are called “stress”, referring to the same thing Guillaume called pitch accent, but again as non-contrastive. It predictably falls on the penultimate syllable and denotes prosodic word boundaries rather than anything tone-like.
{ NA | yes | no }
An example of a more constrained accent system is Northern Emberá, which has a non-contrastive stress-accent system.
[Northern Emberá] stress tends to be ultimate on words in isolation but penultimate on suffixed words. This is an area for further study in the NE languages.(Mortensen 1999: 7)
The accentual system found in Emberá is a stress-accent system, with amplitude as the primary feature of prominence. While the accent system is generally predictable, marked exceptions do exist for a small set of verbs. Only a few of these are given in the grammar (Mortensen 1999), shown below:
| úsʰia | ‘planted’ | usʰía | ‘swatted’ |
|---|---|---|---|
| kʰásʰia | ‘bit’ | kʰasʰía | ‘wove’ |
| awásʰia | ‘carried it on his back’ | awasʰía | ‘was given to doing that’ |
Due to the very few exceptions to the rules governing stress placement as outlined in the grammar, Emberá is coded as non-contrastive. This is keeping in line with the guidance for coding in such cases given above, namely that the system is overwhelmingly predictable, and exceptions are explicitly stated as uncommon. Compare this to a system such as Japanese, where accent placement is described as lexically specified, and attempts to determine rules governing placement of accent can only account for some cases of placement but not for the system as a whole (Kawahara 2015).
{ NA | contrastive | non-contrastive }
{ NA | yes | no }
If yes, this may be what has been described as metrical-accent, which is equivalent to being stress-timed, and may still also be a pitch-accent or stress-accent system. An accent system that is primarily distinguished by duration is rare compared to pitch and accent, but examples which could not be categorized as either pitch-based or amplitude-based were encountered in the process of encoding languages.
Savosavo has a system of predictable accent placement from the end of the prosodic word, wherein the primary feature of the accented syllable is longer and clearer production than that of unaccented syllables (Wegener 2012: 23f).
{ NA | contrastive | non-contrastive }
{ yes | no }
This feature encompasses both lexical and grammatical tone systems, as given in the definition list. If a system is described as having a tone system at all in a grammar, the state is <yes>.
In prototypical tone languages such as Mandarin, nearly all morphemes fall into one of the set of tone categories. When a description refers to a language having for example "four tones", that is referring to the four tone categories or tonemes. As with phonemes having allophones, there are multiple surface realizations of tonemes (frequently called allotones), often the result of things like sandhi patterns or assimilation effects. This can be likened to the phoneme /p/ as distinct from /b/; In English /p/ may be realized a number of ways, as can tonemes in some languages, but the range of allophonic (allotonic) variation is still restricted for the underlying phoneme. Thus when one speaks of the size of the tonal inventory of a language, this is analogous to speaking of the size of a consonantal inventory. It is the number of underlying tonal phonemes.
As mentioned above, these are also referred to as tone categories, and refer to a closed set of contrastive features, generally based around pitch but often including other features such as phonation or duration.
Grammatical tone refers to a tone system where phonemic pitch marks a difference in grammatical structure, generally affecting TAME or case marking, or part of speech information. Such systems are found in many African tonal languages, but not all (Green 1949).
This feature does not take into account boundary tones, which are sentence-level or utterance-level intonational pitch targets, and not tonemes as defined for this typological survey. Despite the name, they are not a feature exclusive to tonal languages, but rather are a concept used to describe any sort of intonational target at different places within an utterance. In many English dialects, questions have a high final boundary tone, i.e. rising intonation at the end of the sentence. Languages described as having tone/pitch only in the context of boundary tones are not considered as tone languages for the purposes of this study.
{ NA | lexical | grammatical | both }
Lexical tone and grammatical tone refer to tone being used to convey different functions in a language, and are defined more precisely in the accompanying definitions list above.
Mandarin has a system of lexical tone. Tonal differences on otherwise homophonous morphemes serve only to distinguish between concepts and do not provide grammatical information.
Krongo has a system of lexical tonal contrasts, as well as grammatical tone on verb stems which distinguishes between perfective and imperfective aspect (Reh 1985: 75).
{ NA | _number_ }
<NA> is the appropriate state if a language has no tonal distinctions, so that the state to this feature is never <0> or <1>.
If there is only one marked tone, such as in Ingush (Nichols 2011) or many Athabaskan languages (Lehman 2018), and everything else is unspecified for tone, then the state to this feature is <2>, as there is a 2-way contrast (despite it being a privative system) which contrasts H (or L) with ∅. While many languages can be analyzed both as a 2-tone system or a privative 1-tone system, we are interested in the number of toneme contrasts.
Hadza is described as only contrasting high and low tonemes, a typical 2-tone language (Sands 2013: 89f).
Lao has a more complex inventory with a total of 6 tonemes including contour tones (Enfield 2007: 35f).
Ingush has a system where a high tone contrasts with no tone. Not all syllables are marked for tone, and it is likely that the high tone is a reflex of an earlier non-initial stress-accent marker. Ingush is described in Nichols (Nichols 2011: 108) as a one-tone language, as most analyses would likely conclude. However, for the purposes of this feature, we code it as 2 levels of contrast since what we are really capturing are degrees of contrast, based on the assumption that the areality of tone systems may end up showing a high-to-low inventory size between source language and recipient language.
{ NA | yes | no }
This feature is a <yes> if even one of the tonal categories has creakiness as a secondary feature not found in other categories.
In Southeast Asia, many tone systems have developed from earlier phonation distinctions, such as with Vietnamese (Alves 1995). These have usually retained some features of the phonation system, such as creak or duration on at least one of the categories. In Tai languages including Standard Thai, the C category (A B C and D are used to label tone categories) developed from *creaky, and in many modern Tai varieties, so-called "C" tone words still retain some degree of creaky phonation (Sagart et al 1989).
In Krongo, as in many other African languages, there is no difference in phonation or duration between the different tone categories, which are based only on pitch (Reh 1985: 75).
{ yes | no }
| yes | if ToneProm-03 is <yes> OR ToneProm-09 is <yes> |
| no | if (ToneProm-03 is <no> OR ToneProm-03 is <NA>) AND ToneProm-09 is <no> |
{ yes | no }
| yes | if ToneProm-03 is <yes> AND ToneProm-04 is <contrastive> OR ToneProm-09 is <yes> |
| no | if (ToneProm-03 is <no> OR ToneProm-03 is <NA> OR (ToneProm-03 is <yes> AND ToneProm-04 is <non-contrastive>)) AND ToneProm-09 is <no> |
{ NA | contrastive | non-contrastive }
| contrastive | if ToneProm-04 is <contrastive> OR ToneProm-06 is <contrastive> OR ToneProm-08 is <contrastive> |
| non-contrastive | if ToneProm-04 is <non-contrastive> OR ToneProm-06 is <non-contrastive> OR ToneProm-08 is <non-contrastive> AND ToneProm-04 is not <contrastive> AND ToneProm-06 is not <contrastive> AND ToneProm-08 is not <contrastive> |
| NA | if ToneProm-01 is <no> |
The presence of some prominence system (ToneProm-01) is overwhelmingly the most common across the world, with noticeable exceptions being Africa, Southeast Asia, and parts of the western Amazon. Not coincidentally, these are the regions in which tone is most common (ToneProm-09). Accent systems are overwhelmingly realized with amplitude as their primary phonetic correlate, with pitch the second most common correlate (ToneProm-02, ToneProm-03, ToneProm-05, ToneProm-07). No very clear large-scale clustering of these correlates emerged from our data.
Accent systems are more commonly contrastive than not (ToneProm-01a, ToneProm-04, ToneProm-06, ToneProm-08). The distribution of non-contrastive systems is fairly scattered, but it is possible that with a denser sample some regions will have a strong preference for non-contrastiveness, such as northwest South America or the interior Pacific Northwest.
Tone was overwhelmingly lexical, with grammatical tone most common in Africa (ToneProm-10) and most languages having 2 or occasionally 3 phonemic tones (ToneProm-11). Tone linked with phonation types was rare (ToneProm-12). The results of our tone typology should however not be overinterpreted, as two of the densest concentrations of phonemic tonal contrasts, Africa and Southeast Asia, were very sparsely sampled.
Conceptualization: Kellen Parker van Dam
Data collection: Natalia Chousou-Polydouri, Kellen Parker van Dam, Anna Graff, David Inman, Marine Vuillermet
Supervision of data collection: Kellen Parker van Dam
Aikhenvald, Alexandra Y. 1999. Areal Diffusion and Language Contact in the Içana-Vaupés Basin, North-West Amazonia. In Dixon, R. M. W. and Aikhenvald, A. Y. (eds.), The Amazonian Languages, 384–416. Cambridge: Cambridge University Press.
Alves, Mark. 1995. Tonal features and the development of Vietnamese tones. Working Papers in Linguistics 27. 1–13.
Arisawa, Tatsuro Daniel. 2016. An Iu Mien grammar: a tool for language documentation and revitalisation. Melbourne: La Trobe University dissertation.
Bereznak, Catherine. 1995. The Pueblo Region As A Linguistic Area: Diffusion Among the Indigenous Languages of the Southwest United States. Louisiana State University and Agricultural & Mechanical College dissertation. (223pp.)
Campbell, Lyle. 1997. American Indian Languages: the Historical Linguistics of Native America. (Oxford Studies in Anthropological Linguistics, 4.) Oxford: Oxford UP: Oxford University Press. xi+512pp.
Campbell, Lyle. 2012. Typological characteristics of South American indigenous languages. In Lyle Campbell and Verónica Grondona (eds.), The Indigenous Languages of South America: A Comprehensive Guide, 259–330. Berlin, Boston: De Gruyter Mouton.
De Jong, Kenneth & Bushra Adnan Zawaydeh. 1999. Stress, duration, and intonation in Arabic word-level prosody. Journal of Phonetics 27(1). 3–22. Elsevier.
Downing, Laura J. 2004. What African languages tell us about accent typology. ZAS papers in linguistics 37. 101–136.
Enfield, Nick J. 2007. A grammar of Lao. (Mouton grammar library, 38.) New York: Mouton de Gruyter. xxvii+567pp.
Green, MM. 1949. The classification of West African tone languages: Igbo and Efik. Africa 19(3). 213–219. Cambridge University Press.
Guillaume, Antoine. 2008. A grammar of Cavineña. (Mouton Grammar Library, 44.) Berlin: Mouton de Gruyter. xxiv+900pp.
Hyman, Larry M. 2001. Privative tone in Bantu. Cross-linguistic studies of tonal phenomena. 237–257.
Hyman, Larry M. 2009. How (not) to do phonological typology: the case of pitch-accent. Language sciences 31(2-3). 213–238. Elsevier.
Hyman, Larry M, Peter SE Jenks, Geoffrey Bacon, Nicolas Baier, Emily Clem, Matthew Faytak, Spencer Lamoureux, Florian Lionnet, John Merrill, Nicholas Rolle & others. 2015. Areal features and linguistic reconstruction in Africa. In Workshop 3: Areal Phenomena in Northern Sub-Saharan Africa.
Ito, Chiyuki & Michael Kenstowicz. 2017. Pitch accent in Korean. Oxford University Press.
Jun, Jongho, Jungsun Kim, Hayoung Lee & Sun-Ah Jun. 2006. The prosodic structure and pitch accent of Northern Kyungsang Korean. Journal of East Asian Linguistics 15(4). 289–317. Springer.
Kawahara, Shigeto. 2015. The phonology of Japanese accent. The handbook of Japanese phonetics and phonology. 445–492. Citeseer.
Kirby, James & Marc Brunelle. 2017. Southeast Asian tone in areal perspective. The Cambridge handbook of areal linguistics. 703–731. Cambridge University Press Cambridge.
Kubozono, Haruo. 2012. Varieties of pitch accent systems in Japanese. Lingua 122(13). 1395–1414. Elsevier.
Kubozono, Haruo. 2017. Accent in Japanese phonology. In Oxford Research Encyclopedia of Linguistics.
Lehman, Blake. 2018. Tone-prominence interaction in Hän (Athabaskan).
Linn, Mary Sarah. 2001. A grammar of Euchee (Yuchi). University of Kansas dissertation. (xvii+547pp.)
Maddieson, Ian. 2013. Consonant Inventories. In Matthew S. Dryer and Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology.
Mortensen, Charles Arthur. 1999. A Reference Grammar of Northern Embera Languages. (Studies in the Languages of Colombia, 7.) Dallas: SIL: SIL International and University of Texas at Arlington. xiv+194pp.
Nichols, Johanna. 2011. Ingush Grammar. (University of California Publications Series.) University of California Press. xviii+806pp.
Payne, Thomas E. & Doris L. Payne. 2013. A Typological Grammar of Panare: A Cariban Language of Venezuela. (Brill's Studies in the Indigenous Languages of the Americas.) Leiden: Brill. 486pp.
Ragagnin, Elisabetta. 2011. Dukhan, a Turkic variety of Northern Mongolia: Description and analysis. Wiesbaden: Harrassowitz. xiii+290pp.
Reh, Mechthild. 1985. Die Krongo-Sprache (Nìino Mó-dì): Beschreibung, Texte, Wörterverzeichnis. (Kölner Beiträge zur Afrikanistik, 12.) Berlin: Dietrich Reimer. xv+490pp.
Roberts, Brice D. 2019. An intonational model of Shanghai Wu: A pitch-accent analysis. The Journal of the Acoustical Society of America 146(4). 3012–3013. Acoustical Society of America.
Russo, Michela & William J Barry. 2008. Isochrony reconsidered. Objectifying relations between rhythm measures and speech tempo. In Proceedings of Speech Prosody, 419–422.
Sagart, Laurent & others. 1989. Glottalised Tones in China and South-East Asia. In Prosodic analysis and Asian linguistics: to honour RK Sprigg. Pacific Linguistics.
Sands, Bonny. 2013. Hadza. In Rainer Vossen (ed.), The Khoesan languages, 38–42, 89–90, 107–124, 265–274. London & New York: Routledge.
Uchihara, Hiroto. 2016. Tone and accent in Oklahoma Cherokee. (3.) Oxford University Press.
Ultan, Russell. 1967. Konkow Grammar. University of California at Berkeley dissertation. (198pp.)
Vuillermet, Marine. 2012. A Grammar of Ese Ejja, a Takanan language of the Bolivian Amazon. Université Lumière Lyon 2 dissertation. (xvi+736pp.)
Wegener, Claudia U. 2012. A grammar of Savosavo. (Mouton Grammar Library, 61.) Berlin, Boston: Mouton de Gruyter. xvi+400pp.
Williams, Marianne Mithun. 1974. A Grammar of Tuscarora. New Haven: Yale University dissertation. (xviii+315pp.)