Symposium B4: Vowel harmony

Organisers: László Fejes, Péter Rebrus, Péter Szigetvári, Miklós Törkenczy

Wednesday, August 24, Department of African studies (Afrikanistik)

(Original symposium call)


András Czentnár:

A typological approach to discover vowel harmony 

Vowel harmony is frequently mentioned when Uralic and Altaic languages are compared. However, vowel harmony itself is not enough to prove any kind of genetic relationship between languages, moreover, it is also a vague statement to claim that two languages are typologically similar, just because they have vowel harmony. In fact, vowel harmony can appear throughout a diverse range of possibilities, thus one has to consider and determine patterns and types of harmony beforehand in order to establish a framework to compare languages.

As a survey to acquire this, eight steps will be applied to the languages that are being examined: a) assessing the vowel inventory of the language and giving a sketch of the system by specifying the qualities of the vowels (height, backness, roundedness, etc.); b) investigating the phonotactic vowel restrictions of the first syllable; c) investigating the occurrences of vowels outside of the first syllable with special attention to last syllables (if necessary); d) setting the harmonic oppositions: any kind of harmony makes opposing groups among the vowels, and each vowel involving in the same harmony has its opposing pair(s) from the other group(s); e) deducing the dimension and hence the type of the harmony (height, palatal, labial, etc.) based on the occurrences of the vowels in various positions and the sets of harmonic oppositions; f) outlining the extent of the harmony in the language (radical and/or affixal harmony), g) determining the number of possible allomorphs of affixes taking vowel quality into account (vowel quality of the affixes is taken into account); h) finding and characterising neutral vowels in the inventory (transparent, opaque).

With the help of this typological approach, the similarities as well as the differences of the vowel harmony in various Uralic languages can be discovered. Also, Uralic languages can be compared to Turkic, other Altaic, or any other languages of the world. This paper will scrutinise and compare some languages using this framework.




László Fejes:

Erzya stem-internal vowel-consonant harmony: a new approach

The descriptions of Erzya morphophonology deal with harmony only in suffixation. There is no research made on harmony inside stems. If grammars mention stem-internal harmony (Keresztes 1990: 37), they just state that there are both harmonic stems (with front or back vowels) and there are also ‘‘mixed stems’’ (containing both front and back vowels). Descriptions of Erzya (Bartens 1999: 66–67; Bondarko & Polyakova 1993: 94–95; Keresztes 1990: 37; idem 2011: 22–23) write about vowel harmony, mentioning that consonants also can play some role in it. However, suffix alternations triggered by the stem show that Erzya has a vowel-consonant harmony: palatalized stem-final dentals trigger fronting of suffix vowels (/kɑl-so/ ‘fish-INE’ : /kɑlj-se/ ‘willow-INE’) and front stem vowels trigger palatalization of suffix-initial dentals (/kudo-t/ ‘house-PL’ : /velje-tj/ ‘village-PL’). However, speaking about stem-internal harmony, the role of dentals is not considered: one of Keresztes’ examples, /kosjke/ ‘dry’ shows a case when harmony is ruined by a palatalized dental.

Only /e/ and /o/ are harmonic vowels, since only they alternate due to harmony. Back vowels /a/ and /u/ (the latter is rare in non-initial syllables) are opaque neutral vowels in the sense that they appear after any vowel or consonant, but non-palatalized dentals, neutral vowels and /o/ of the harmonic vowels tend to follow them. Similarly, front /i/ is an opaque neutral vowel in the sense that it can stand after any vowel or consonant, but only palatalized dentals, neutral vowels and /e/ of the harmonic vowels can follow them.

The aim of the paper is to examine, based on statistics made on dictionary material of Standard Erzya, to what extent ·disharmony is typical for monomorphemic stems;

  • word initial coronals determine the frontness/backness of (harmonic and neutral) vowels;
  • vowels of the initial syllable determine the palatalization of consonants following them;
  • front vowels are followed by back ones and vice versa in bisyllabic monomorphemic stems and among them:
    • both are neutral vowels;
    • neutral vowels stand after harmonic vowels;
    • harmonic vowels stand after neutral vowels;
    • both are harmonic vowels;
    • whether the switch of harmonic classes happens in the consonant (group) between them.

The result will show how and in which cases vowel-consonant harmony is active inside stems.


Bartens, Raija. 1999. Mordvalaiskielten rakenne ja kehitys. Helsinki: Suomalais-ugrilainen Seura.

Bondarko & Polyakova. 1993.: Современные мордовские языки. Фонетика. Составление и общая редакция Л. В. Бондарко, О. Е. Полякова. Саранск: Мордовское книжное издательство.

Keresztes László. 1990. Chrestomathia morduinica. Budapest: Tankönyvkiadó.

Keresztes László. 2011. Bevezetés a mordvin nyelvészetbe. Debrecen: Debreceni Egyetemi Kiadó.


Katalin Mády & Valéria Krepsz:

Variation in the Hungarian vowel /E/ and its consequences for vacillating stems 

Hungarian is a language with front/back vowel harmony [3]. Most monomorphemic stemscontain either front or back vowels, and most suffixes have a front and a back alternate (Kabul+ban ‘in Kabul’, Memphis+ben ‘in Memphis’). In words with mixed harmony, front vowels are transparent meaning that these words tend to take back suffixes. The degree of transparency depends on vowel height [2]: /i/ stems nearly always take backsuffixes, /e:/ stems show some vacillation (Athén+ban/ben, ‘in Athens’), while many /E/ stems only appear with front suffixes or are vacillating.

A common observation is that the vowel /E/ is produced more open by young, especially female, speakers of Hungarian. No systematic investigation of these vowels have been carried out. In this study, the impact of the following consonant was investigated, along with the impact of word- and sentence-level prominence.

474 occurrences of /E/ were analysed in a spontaneous speech corpus [1], produced by young (< 30 years) and older (> 50 years) speakers. Three potential triggering factors were investigated: the impact of /l/ that tends to be vocalised in young speakers’ speech,word-level stress (word-initial vs. non-initial syllables) and sentence-level prominence (whether the target word itself was accented). Vowels were analysed before /l/ and the obstruents /t d s z/. The first vowel formant F1 was measured 5 ms before the vowel offset to compensate for coarticulatory effects of the preceding consonant. Since the dataset was not balanced across conditions, analysis was based on Mann-Whitney tests.

Results show that /E/ was produced more open before /l/ by young female speakers in stressed (p = 0.031), but not in unstressed position. Male speakers show a similar, but non-significant tendency. Interestingly, young female speaker produced /E/ more closed in stressed position before obstruents (p = 0.030), showing centralisation and thus a tongue position due to coarticulation. Sentence-level prominence did not influence vowel realisations.

Future work will focus on the co-occurrence of lowering of /E/ vowels in the speech of young Hungarian speakers and their preferred choice of suffixes in vacillating stems. Data collection will be extended to speakers of the Western Hungarian dialect that contains two, phonologically marginally distinctive front mid unrounded vowels, one being more close and another one being more open than /E/ as specified in the IPA system. According to the height effect, it is expected that speakers using a close /e/ in vacillating stems will prefer back suffixes with these words compared to non-dialectal, especially young speakers.

[1] M. Gósy. Magyar spontánbeszéd-adatbázis – BEA. Beszédkutatás, pages 194–207,2008.
[2] B. Hayes and Z. Cziráky Londe. Stochastic phonological knowledge: the case of Hungarian vowel harmony. Phonology, 23, 2006.
[3] P. Siptár and M. Törkenczy. The Phonology of Hungarian. University Press, Oxford, 2000.


Péter Szigetvári, Péter Rebrus & Miklós Törkenczy:

A gap in Hungarian possessives

The Hungarian 3p possessor and/or pl possessee suffix has [j]-initial (yodful) and vowel-initial (yodless) allomorphs (the Y-suffixes). Some nouns take the yodful allomorphs only (paːr-jɑ ‘her/his pair’), some only the yodless ones (kaːr-ɑ ‘her/his damage’) and some take both (tor-ɑ % tor-jɑ ‘her/his wake’), the latter are in the zone of variation with respect to yodfulness. These suffixes also alternate harmonically: with low ɑ~ɛ (kaːr-ɑ vs. bøːr-ɛ ‘her/his skin’, and paːr-jɑ vs. zyːr-jɛ ‘her/his mess’) high u~y (kaːr-uk ‘their damage’ vs bøːr-yk ‘their skin’, and paːr-juk ‘their pair’ vs zyːr-jyk ‘their mess’). Bɛ stems, eg hotɛl ‘id’, take both F and B suffixes: hotɛl-ɛm % hotɛl-om ‘my hotel’, they are in the zone of variation with respect to B/F harmony.

Variation in yodfulness and B/F harmony are independent of each other. Stems that vary both ways are expected to take four allomorphs: hotɛl-yk % hotɛl-uk % hotɛl-jyk % hotɛl-juk ‘their hotel’. With the 3sg poss suffix the yodless B allomorph is systematically missing: hotɛl-ɛ % hotɛl-jɛ % hotɛl-jɑ, but *hotɛl-ɑ ‘his/her hotel’. The existence of this gap is corroborated by extensive web searches for all (100+) harmonically variable Bɛ stems. *hotɛl-ɑ is not due to a general ban on this allomorph (cf notɛs-ɛ % notɛs-ɑ ‘his/her notebook’). It is also not a dispreference for having “too many” allomorphs, since all are available in ‘their hotel’. This is unexpected under standard assumptions on how independent (variable) phonological phenomena interact and seems an entirely arbitrary restriction.

The gap and the asymmetry in the behaviour of high vs low vowel possessive suffixes is related to the quality of suffix-initial vowels elsewhere within the possessive subparadigm of nouns, which are high u~y and o~ɛ. The gap and the asymmetry is due to the fact that u, y, and ɛ occur both in Y-suffixes and elsewhere in the possessive subparadigm but ɑ only occurs in Y-suffixes and o only occurs elsewhere.

In short, *hotɛl-ɑ is ungrammatical, because it is not supported by other forms of the paradigm of hotɛl. The F variant hotɛl-ɛ on the other hand is supported by other poss forms like hotɛl-ɛm, hotɛl-ɛd, hotɛl-ɛtɛk ‘my, your-sg/pl hotel’. The 3pl poss forms hotɛl-uk is supported by the 1pl form hotɛl-unk ‘our hotel’.


Tuomas Koukkari:

Vowel Harmony and Related Phenomena in Inari Saami 

This presentation explores the topic of progressive vowel harmony in Inari Saami. First-syllable vowels in the language are analyzed as being divided into two groups, front and back, and which group a vowel belongs to as having systematic effects upon subsequent syllables.

Most studies of vowel alternations in Inari Saami (and Saamic in general) have focused mainly on regressive metaphony, with vowel harmony, in the narrow sense of progressive assimilatory effects between vowels, receiving relatively little attention. In a recent study, however, Morottaja (2016) has devoted a section to describing progressive effects which he explicitly terms vowel harmony. This presentation aims to expand upon Morottaja's findings and deliver a systematic overview of the first-syllable vowel groups and their morphophonological effects.

The clearest and most prototypical example of a front-back harmonic pair are the low vowels /a/ and /ɑ/. These can generally not appear together in the same word, and an /ɑ/ in the first syllable will usually cause subsequent low vowels to also emerge as /ɑ/ in contexts where /a/ would otherwise be expected. A similar relationship exists between the mid vowels /e/ and /ә/, although it may be possible to explain these purely under regressive rules. In limited contexts in loanwords, /y/ may also occasionally be found as the front counterpart of /u/. More broadly, first-syllable vowels of the front group (excluding /i/) trigger certain subsequent vowels to emerge as higher variants, specifically as /i/ and /u/ instead of, respectively, /e/ and /o/. Other effects include ones which do not strictly speaking fall under vowel harmony, such as determining the declension or conjugation type of a word. These are outside the main scope of this presentation, but share enough of the same morphophonological rules to deserve some mention in a synchronic discussion of vowel harmony in Inari Saami.

In addition, various restrictions and exceptions to the system are considered. Long vowels in non-initial syllables generally do not harmonize, and certain types of polysyllabic words may display disharmonic sequences. Finally, a diachronic perspective is offered: how this system of partial harmony emerged and how, due to certain ongoing developments, it may currently be in the process of being lost.


Itkonen, Erkki 1939: Der ostlappische vokalismus vom qualitativen standpunkt aus mit besonderer berücksichtigung des Inari- und skoltlappischen. SUST 79. Helsinki.

Itkonen, Erkki 1971: Ehdotus inarinlapin fonemaattiseksi transkriptioksi. In Erkki Itkonen, Terho Itkonen, Mikko Korhonen & Pekka Sammallahti (eds.): Lapin murteiden fonologiaa, p. 43-67. Castreanumin toimitteita 1. University of Helsinki.

Morottaja, Petter 2016: Anarâškielâ vokalij kvalitetnubástusâi myensteristem Kielâtekno morfologisii analysaattorist. MA thesis. University of Oulu.

Morottaja, Petter & Olthuis, Marja-Liisa 2016: Inarinsaamen kielioppi. Unpublished manuscript.


Jukka Mettovaara:

Thoughts on Inari Saami metaphony and mid-central vowels

In this presentation I will examine the variation and status of the mid-central vowels in Inari Saami (henceforth IS) from both diachronic and synchronic perspective. The data comprises recordings from the 1950s to 1970s as well as modern recordings of my own. Much of our knowledge on IS phonology still rests on the early works of Äimä (1914) and Itkonen (1939). Modern research has concentrated on phonological quantity and prosody (e.g. Bye, Sagulin & Toivonen, 2009; Türk, Lippus, Pajusalu & Teras, 2019).

IS vowel inventory is otherwise unproblematic, but the question of the mid-central vowels deserves a closer look. In the grammar sketch by Valtonen, Ylikoski and Aikio (forthcoming), IS minimally has the vowel phonemes /i/, /u/, /e/, /o/, /æ/, /ɑ/, /ә/ and /y/, /ø/ in loanwords. Mid-central vowels are represented by two separate graphemes in the orthography, ‹e› and ‹â›, which occur in complementary distribution: ‹e› appears in the initial syllable and ‹â› in non-initial syllables. However, the matter is complicated by the fact that the initial syllable ‹e› can refer to both /e/ and /ә/: ‹eidu› /ejdu/, cf. ‹peenuv› /pәnː uv/.

Unlike Valtonen et al., P. Morottaja & Olthuis differentiate between /ә/ ‹â› and /әə / ‹e›, describing the latter as being phonetically close to /ә/ but more fronted. M. Morottaja (2007) also differentiates between the two, adding that the difference is clearer in older speakers but that nowadays the two sounds have merged.

So, the most economical solution is to consider there to be only one mid-central vowel phoneme, which is realized as [ә ~ ɤ ~ ɘ ~ ɜ ~ ɐ] (Valtonen et al.). However, all IS dialects and idiolects do not have an initial-syllable /ә/, and contemporary IS shows signs of losing the mid-central vowel altogether, especially among younger speakers who replace it with /ɑ/ in non-initial syllables and /e/ in initial syllables. Do we then conclude that the possible allophonic variation is even greater, yielding [ә ~ ɤ ~ ɘ ~ ɜ ~ ɐ ~ e ~ ɑ]? I argue that this is not the case and attempt to find a more efficient analysis to account for the attested variation.


Bye, P., Sagulin, E. & Toivonen, I. (2009). Phonetic Duration, Phonological Quantity and Prosodic Structure in Inari Saami. Phonetica Vol. 66 (pp. 199–221)

Itkonen, E. (1939). Der ostlappische Vokalismus vom qualitativen Standpunkt aus mit besonderer Berücksichtigung des Inari- und Skoltlappischen. Lautgeschichtliche Untersuchung

Morottaja, M. (2007). Anarâškielâ ravvuuh. Kotus

Morottaja, P., & Olthuis, M. (forthcoming). Inarinsaamen kielioppi

Valtonen, T., Ylikoski, J., & Aikio, A. (forthcoming). Aanaar (Inari) Saami. In M. Bakró-Nagy, J. Laakso & E. Skribnik (Eds.), The Oxford Guide to the Uralic Languages

Türk, H., Lippus, P., Pajusalu, K., Teras, P. (2019). The acoustic correlates of quantity in Inari Saami. Journal of Phonetics Vol. 72 (pp. 35–51)

Äimä, F. (1914). Phonetik und Lautlehre des Inarilappischen