Corpus of Estonian dialects and the Estonian vowel system/[TEXT NOT REPRODUCIBLE IN ASCII].
The territory where Estonian is spoken is quite small but there are large differences between traditional dialects. Researches of Estonian dialects have classified at least eight main dialects and over hundred sub-dialects or parish dialects (see Pajusalu, Hennoste, Niit, Pall, Viikberg 2002; Pajusalu 2003).
The traditions of Estonian dialectology are also rather long. Andrus Saareste introduced dialect geography to Estonia already in the 1920s and compiled several Estonian dialect atlases starting from the late thirties. These were based on huge dialect data collections. Up to now, the Estonian dialect archive at the Institute of Estonian Language in Tallinn contains more than two million data units of dialect words, over 2900 hours of sound recordings with examples of each Estonian sub-dialect, and several thousand pages of transcribed texts. There are additional collections of Estonian dialect data at the universities of Tartu and Tallinn.
The amount and scope of comparative studies on Estonian dialect phonology and grammar has, however, been rather limited (an outstanding exception is Tauli 1956) because of the lack of a united data source for such kind of analysis. For facilitating such studies, the University of Tartu and the Institute of Estonian Language in Tallinn started a joint project for compiling an electronic corpus of Estonian dialects in 1998 (see also Lindstrom, Lonn, Mets, Pajusalu, Teras, Veismann, Velsker, Viikberg 2001). The main aim of this corpus is to enable the study the phonological and grammatical structure of Estonian dialects by means of electronic data processing. The corpus is planned to contain digitized sound recordings and electronic text versions of recordings from all Estonian dialects and main sub-dialect groups within the dialects.
1. Current state of the corpus
1.1. Classification of Estonian dialects in the corpus
We have followed the most detailed classification of Estonian dialects according to which the North Estonian dialect group includes four dialects --Insular, Western, Mid, and Eastern dialects; the South Estonian dialect group consists of three dialects--Mulgi, Tartu, and Vbru, including Setu dialect; the North-Eastern Coastal Estonian group includes the Coastal and North-Eastern dialects (see figure 1). Because of the exceptional position of Setu we have treated it as the fourth main dialect of South Estonian and thus there are ten dialects distinguished in the corpus.
[FIGURE 1 OMITTED]
1.2. Basic statistics of the current state of the corpus
Within the ten dialects we have determined the main sub-dialect groups and from each of these we have tried to choose examples of central parish dialects. In October 2003, there were about 456 000 text words in the corpus and the work is still in progress. Our aim is to compile a collection of at least half a million text words of archaic dialects for the first stage of the corpus (by the end of 2003). The second stage of the corpus should include more archaic parish dialects (theoretical maximum is about 2 million words) and newer data from Estonian vernaculars.
At the moment the corpus includes archaic dialect data from all the dialects and 34 sub-dialects. Table 1 shows the number of text words for each dialect in the corpus.
Current data of the corpus is based on the oldest sound recordings of the dialects. These are interviews on various topics. The first records originate from 1938. The largest amount of text was recorded in the 1960s and 1970s (see table 2). Most of the speakers were born in the second half of the 19th century (see table 3).
The corpus is in fact a text collection of spontaneous spoken language. We have taken into account special features of speech and transliterated all discourse particles, word repetitions, corrections, pause-fillers, and so on. The interviewer's text has also been transliterated.
The texts are presented in two versions. At first, there are texts transliterated in the standard Finno-Ugric phonetic transcription. The reason why we have used the Finno-Ugric transcription, which is unknown for most researchers of other language families, and not the International Phonetic Alphabet (IPA), lies in the tradition of Estonian dialectology. All the old texts of Estonian dialects based on sound recordings were transcribed in the Finno-Ugric transcription. In the future, it will be possible to modify the texts additionally into an IPA version because the current version of Finno-Ugric transcription is sufficiently precise for that.
It is already possible to carry out various types of phonetic research on this version of the corpus. For example, Karl Pajusalu, Merike Parve, and Pire Teras have studied the prosody and vowel system of Southern Estonian dialects on the basis of the corpus (Pajusalu, Parve, Teras 2001; Parve 2003; Teras 2003).
The second version of the texts is available in simplified transcription. For that the phonetic texts have been converted automatically into simple txt-format. The version is aimed to be a basis for studying grammar but marks also several features of spoken language, such as pauses and co-articulations. Compound words, interviewer's text and commentaries are marked with special symbols.
1.3. Morphosyntactic tagging in the corpus
The tagging of morphosyntactic categories has already started. The dialect corpus is a multi-lingual corpus by its nature, or more precisely, it is a corpus of languages without existing standards and complete knowledge about the linguistic structures. For this reason the tagging of parts of speech must be relatively open. It has to be possible to make corrections and introduce new categories.
We have started working out principles of morphological tagging following the example of the corpus of Finno-Ugric languages compiled at the University of Helsinki (Suihkonen 1998). In comparison with Helsinki corpus, we have added some parts of speech, e.g. discourse particles and onomatopoetic words, and we have determined several sub-classes, such as pro-adjectives and pro-adverbs.
A more appropriate example of morphological tagging for our purposes is being compiled at the Institute of Estonian Language in Tallinn where a morphological data base of Southern Estonian morphology is being put together (see http://www.eki.ee/dict/hargla/). Our aim was to make it possible to join the two databases in the future and therefore most of the tags are the same in these two databases. We have added some categories which are not used in Southern Estonian (e.g. possessive suffixes) and some categories specific to spontaneous speech (e.g. discourse particles). We have been able to distinguish between interrogative words and adverbs and relative-interrogative pronouns (where they traditionally belong) and we have separated pro-adjectives and pro-substantives.
For all the text words the following information is given (see table 4).
Each text file is marked by the following information:
1) dialect; 2) sub-dialect; 3) village; 4) informant's name and age or date of birth; 5) date of recording; 6) interviewer's name.
The main grammatical categories presented are as follows:
I. Nominals (substantive, pro-substantive, adjective, pro-adjective, proper name, numeral, relative and interrogative pronouns):
1) number (sg, pl)
2) declension (15)
3) possessive suffix
II. Verbs (verbs, auxiliary verbs):
1) category for infinite form: infinitive, gerund, supine, participle
2) voice: personal, impersonal passive, personal passive
3) mood: indicative, conditional, imperative, jussive, quotative, potential
4) tempus: present, preterite, perfect, pluperfect
5) number: singular, plural
6) person: 1, 2, 3
7) affirmative or negative form
III. Uninflected words: adverbs, pro-adverbs, auxiliaries (of compound predicate), postpositions, prepositions, interrogatives, discourse markers, onomatopoetic words, conjunctions, negations, comparative words.
We have used a special program called Mark (written by Karlis Goba) to facilitate tagging and to avoid mistakes. The tagged text is in xml-format. In October 2003, about 60 000 words were morphologically tagged. Currently an Internet based search engine for the use of tagged texts is being developed.
1.4. Studies of dialect vocabulary using the corpus
A diagnostic study of the most frequent vocabulary of Estonian dialects has been carried out up to now (see also Lindstrom, Lonn, Mets, Pajusalu, Teras, Veismann, Velsker, Viikberg 2001). In this study one hundred most frequent words of three geographically and linguistically distinct dialects were compared. These dialects were the Western Estonian dialect, the North-Eastern dialect and the South-Eastern Voru dialect. Among the frequent words adverbs, conjunctions and discourse particles were the most numerous groups. It is usual that some words function in the text at the same time as discourse particles and adverbs, or discourse particles and conjunctions. The occurrence of such words in texts is relatively high and was of particular interest.
It appeared that the majority of these particles are common, however, for all the three dialects. From 24 analyzed particles 15 occurred in each dialect. Their phonological form may be quite different but they have the same stem (e.g. Voru iks ~ iks, Western and North-Eastern ikke 'still; certainly, surely'; Voru ka ~ kahh, Western koa ~ kaa, North-Eastern kaa 'also'). Five particles occurred in two dialects and were missing in one dialect.
Only four particles were present only in one dialect among 100 of the most frequent words. These four particles are as follows. Firstly, naet 'to see', originally 'you (sg.) see' in Voru, which indicates that it is a verb form that has grammaticalized as a particle in South-Eastern Estonian. In the Western dialect the word was in the 269th place by its frequency and in the 2074th place in the North-Eastern dialect data. According to these text frequencies it is possible to suggest that the form is acquiring the status of a particle in the Western dialect as well but is not used in this meaning in the North-Eastern dialect. Secondly, vot 'oh' in Voru that was in the 242nd place in the North-Eastern dialect and is missing in the data of the Western dialect. The same particle is typical of Russian and therefore has been mentioned among Russian loans in these eastern dialects (Must 2000 481). Thirdly, ninda 'so' that was common for the North-Eastern dialect but occurred only once in Voru and the Western dialect. And fourthly, naa 'so' that was characteristic of the Western dialect but did not occur in the other dialects.
We can conclude that such analysis of frequent words shows local developments as well as dialect and language contacts. Additionally, it is possible to detect the age and way of spreading of loan words. An apparent loan particle from Russian no/nu 'so what' occurs in all the three dialects, but was rare in the Western dialect and frequent in the eastern dialects. Vot 'oh' was unknown in the Western dialect and rare in the North-Eastern dialect. Also, the discourse marker a 'but' occurred only in the South-Eastern Voru dialect, and did not appear in the North-Eastern and Western dialect data. Thus, the South-Eastern dialect is most deeply influenced by Russian according to the use of particles; further the different stages of contacts became evident.
In Estonian dialectometrics in the 1980s and 1990s, the lexical relationships between Estonian dialects were calculated only according to the occurrence of words in the dialect (see Murumets 1982-1983; Krikmann, Pajusalu 2000) but in the future it will also be possible to count the occurrence frequencies on the basis of the data of this corpus.
1.5. Availability of the corpus
It is possible to use the sound recordings and transliterated texts in Finno-Ugric transcriptions in the form of Word-files and in simplified transcription of txt-files. At the moment there is not yet an open Internet access to the corpus but it is possible to access the database with a personal user name and password. In order to obtain these, please contact the corpus manager Liina Lindstrom by e-mail: firstname.lastname@example.org.
2. Basic characteristics of Estonian dialect vowel systems
The corpus has so far been used mostly for phonological studies. Recently, a statistical survey of Estonian vowel systems was carried out, the results of which will be presented here. In this study we will calculate vowel frequencies for stressed and unstressed syllables in all the Estonian main dialects in order to detect general changes in the development of vowel systems. Many, although not all of them, are related to changes in vowel harmony.
2.1. Establishing the phonemic status of the vowels
The first task is the specification of the phonemic status of the vowels. In Estonian written language, traditionally a distinction is made between nine monophthongs (see Table 5).
The most problematic sound in the phonological description of Estonian monophthongs is o. In traditional descriptions of the language and in textbooks, o is treated as a mid-high unrounded vowel, i.e. similar to schwa (see e.g. Ariste 1984 : 74). Still, according to its phonetic characteristics o can be classified rather as a back vowel (Eek, Meister 1994 : 409 ff), and it also has characteristics of a high vowel (see also Viitso 1981 : 68).
All short monophthongs occur in Estonian written language in the stressed first syllable; unstressed syllables contain only the so called primary vowels a, o, u, e and i, while o occurs only in the unstressed syllables in newer loan words and names. Estonian dialects, on the other hand, contain several specific monophthongs such as for example the long open C in the stressed first syllable in West-Saaremaa, which is between a and e in its quality, and the open o instead of o (see ibid.). In the non-initial syllables, the Western dialects of South Estonian have a reduced central vowel but the Eastern dialects, a high central vowel.
A common characteristic of Estonian dialects is the coarticulatory fronting or backing of vowels, as well as a certain raising or lowering of the sounds. In the following statistical analysis we have replaced the less common specific vowels with their nearest phonemes in the vowel system: e > e, [??] > e and [??] > o. As the only exception, we have retained the South Estonian high unrounded central vowel i (i.e. back i; for the description of the vowel see Viitso 1990 : 163; Parve 2000; Teras 2003). Thus, in the following, we differentiate maximally between ten vowel phonemes.
All short monophthongs have their long Standard Estonian counterparts which occur only in the primarily stressed syllables of a word. This applies also as a rule to dialects although in some dialects, long vowels can also occur in unstressed syllables. Estonian long monophthongs have been interpreted as a sequence of two short vowels (see Hint 1997: 42-43) or as separate monophthongs (Viitso 1981). In our analysis we have first treated long monophthongs as a sequence of two vowels and then as one monophthong. Long monophthongs with the secondary quality change have been treated as the same with their closest phonetic equivalents; e.g. the overlong raised equivalents of the South Estonian long mid-high vowels have been grouped together with respective high vowels (e.g. u > uu).
Estonian is rich in sequences of two different vowels; the standard contains at least 26 different types (see Viitso 1981 : 64-67; Hint 1998 : 113-115). In addition to the old diphthongs which end in an i or u there are several newer diphthongs that have developed as a result of diphthongization of long vowels, and sequences of vowels that have appeared due to the loss of a consonant. In some instances, also triphthongs are possible. The situation in Estonian dialects is extremely varied and therefore their treatment is beyond the scope of the present article. In the following statistical analysis, all sequences of vowels are separated into monophthongs.
Tables 6 and 7 give an overview of the occurrence of vowels in Estonian dialects. The treatment of long monophthongs as a unit of one or two vowels has a very small effect on the results. The number of long vowels is on the whole relatively small, the most frequent long vowel being ii (see table 8).
In all Estonian dialects the most common vowels are a, i and e. In the South Estonian dialect of Mulgi e is more common than i. The percentage of a ranges between 28 in North Estonian Mid dialect and 22 in the Mulgi dialect. The most frequent occurrence of a in the central dialects coincides with the most restricted occurrence of the a-harmony in the same area, i.e. a can occur also in non-initial syllables of the words containing front vowels (for a detailed account see the following treatment of front and back vowels). The smallest percentage of a in South Estonian Mulgi dialect can be explained by the reduction of low vowels and their replacement by their mid-high unrounded equivalents starting from the third syllable (a, a > [??], e). This explains also the frequent occurrence of e in the Mulgi dialect.
The percentage of i ranges between 27 and 20 whereas in the North-Eastern dialect group and in the Tartu dialect i is even slightly more common than a. There is, however, no easy explanation for the higher percentage of i in the North-Eastern dialect group as even e is common in these dialects. But noteworthy is the relatively stable occurrence of the historic i also in the diphthongs ending in an i and the vowel sequences that have developed due to the loss of a consonant (see also the treatment of high vowels). In the Tartu dialect, on the other hand, there is a low percentage of a. The reasons for this are similar to those for the Mulgi dialect.
The third most common vowel in all dialects is e (except Mulgi where it is on the second place). At the same time, the percentage of e fluctuates more between dialects than that of a and i, being for instance 21.8% in the Coastal dialect and only 10.8% in the Setu dialect. In the North Estonian and North-Eastern coastal dialects, the occurrence of e is even (between 21.8% and 19.2%) whereas the situation is different in the South Estonian dialects where the percentages in dialects are more uneven: 21.4% in Mulgi, 15.3% in Tartu, 12.8% in Voru and even less in the Setu dialect. In Mulgi, the frequent occurrence of e is linked to the extensive change in non-initial syllables: a, a > e. But in other parts of South Estonia, it is its less frequent occurrence that is connected to non-initial syllables. These dialects have o-harmony which means that the equivalent of e in non-initial syllables of the words containing back vowels is o. Therefore o occurs in Setu almost as much as e (10%) and is in non-initial syllables even more common than e (see the following treatment of o).
The three vowels with average percentage of occurrence in Estonian dialects are u, o and a. By its occurrence, u is the fourth most common in all the Estonian dialects except Setu where it is surpassed by o. The per centage of u is relatively even being highest in the Coastal dialect (11%), and lowest in the Mid dialect (8.8%). There is no explicit reason for these small differences.
The fifth vowel by its occurrence is o in six dialects, and a in four dialects. This statistics means that o is the primary vowel that has the most restricted occurrence in Estonian dialects. There are large differences in its occurrence: 9.5% in Voru and only half as much (4.9%) in the Eastern dialect. The less frequent occurrence of o in several dialects is linked to the change of o into u in non-initial syllables, and in the Eastern dialect, to the unrounding of the o in the first syllable, i.e. the change o > o. The percentage of a is high in all South Estonian dialects (10.1-9.2%), considerably lower in North Estonian dialects (from 7.3% in the Eastern dialect to 4.7% in the Central dialect) and in the North-Eastern Coastal dialect group (5.3% in the Coastal dialect and 6.9% in the North-Eastern dialect). The high frequence of a in the dialects of South Estonia and its low percentage in North Estonian dialects is above all caused by the larger productivity of the a-harmony in South Estonia. Still, this should not be the reason for the low percentage of a in the North-Eastern coastal dialects because the a-harmony occurs there as well (cf Wiik 1988 : 82). It is possible, however, that here the reason lies in the diphthongization of the long aa into is in the North-Eastern dialects.
The vowels with the most restricted occurrence in Estonian dialects are o, u, o and i (i.e. back i). The percentage of o differs largely in different dialects ranging from the Setu 10.1% to the Insular 0.01%. Therefore, the occurrence of this vowel will be treated separately in the following subsection of the article. The vowel u does not have a high rate but at the same time it appears relatively evenly in all dialects. Its highest percentage (3.3-3.6%) is in South Estonia where there is the u-harmony and lowest in the North-Eastern dialect (2.3%). The vowel o is most common in the Insular dialect (2.3%) where the central vowel o has undergone rounding and turned into o. In other North Estonian dialects and in the North-Eastern dialect group the percentage of o is 0.3-0.5%. This figure is even smaller in the South Estonian dialects (0.13-0.17%).
2.2. o in Estonian dialects
One well-known difference between the Estonian and Finnish vowel systems is the occurrence of unrounded central vowel o in Estonian. But o is not equally common for all Estonian dialects (see figure 2). We can see that o is more atypical for the Insular and Coastal dialects and most frequent in the Eastern and South Estonian dialects.
In the case of Insular and Coastal dialects, texts from only those areas were analyzed where o does not occur as a rule. In the Insular dialectal area, o has undergone rounding and changed into o, and in the Coastal dialects, similarly to Northern Finnic dialects, o has never occurred. The fact that the dialect corpus for these dialects contains any occurrences of o at all points to the beginning of levelling of these dialects with Standard Estonian.
It is, however, noteworthy that the percentage of o increases gradually in the Estonian dialectal area from the North West to the South East. The percentage of o is also relatively small in the North Estonian Western and Mid dialects that are the historic foundations of Standard Estonian. The percentage of o is similar in the Eastern and North-Eastern dialects and in the Mulgi dialect in the Western part of South Estonia, and this in spite of the fact that none of these dialects have the o-harmony and that o only occurs in the first syllable.
[FIGURE 2 OMITTED]
The percentage of o increases sharply in the South-Eastern dialects of South Estonia. The o-harmony is characteristic of the dialects of Tartu, Voru and Setu but the percentage of o is larger in Voru than in Tartu and in Setu considerably larger than in Voru. Do the differences in the spread of o point to a more general tendency of rounding in the western dialects of Estonian and unrounding in the eastern dialects? This question will be addressed in the following analysis.
2.3. Rounded vowels in Estonian dialects
Figure 3 presents the percentage of all rounded vowels in Estonian dialects, i.e. the overall occurrence of u, o, o and u. As can be seen, the hypothesis presented above is only partly valid. The percentage of rounded vowels is indeed highest in the island dialects but in addition to the expected high occurrence in these dialects, unrounded vowels are also common in South Estonian dialects where a low percentage was predicted.
The general increase in the percentage of rounded vowels in the West and the decrease in the East is valid only in the case of the North Estonian dialects where we can maintain that the change o > o in the Insular dialects is merely part of the more general tendency to vowel rounding. As an example of this tendency is also the rounding of the a in the same area, as in sauna 'sauna' > sauna, in Hiiumaa a > a, as in kaks 'two' > kaks (see Tauli 1956 : 174), and i > u, e > o next to a labial consonant as in mitu 'several' > mutu, levad 'loafs of bread' > lovad. In non-initial syllables, the rounding occurs in the Coastal dialect: e.g. in Vaivara tulo < tule 'come (Imperative)', tulemo < tuleme 'we come'. As an example of unrounding is the change o > o in the Eastern dialects, as in e.g. koht 'place' > koht and oli 'was' > oli, and the change of non-initial syllables in Votic: o > (> o) > a, as in ainogo ~ ainago 'the only one' (see Pajusalu 2000).
[FIGURE 3 OMITTED]
The more frequent occurrence of rounded vowels in South Estonia is most probably connected to their greater stability in non-initial syllables, as e.g. the u-harmony that has prevented the change characteristic of the non-initial syllables of the North Estonian dialects: u > i, as in kusu > kusi 'ask for'.
2.4. High and low vowels in Estonian dialects
There are also systematic differences between dialects in the percentage of high and low vowels. But here the main direction of change is not from North West to South East as in the case of the occurrence of o, or from East to West as in the case of vowel rounding, but instead from South West to North East. The percentage of high vowels (i, u, u, and in South Estonia also i) is smallest in the Western dialects and largest in the North Eastern dialects (see Figure 4). High vowels occur most of all in the North-Eastern Coastal dialect group. Among the dialects of the North Estonian dialect group, high vowels occur most in Eastern dialect although the difference is small as compared to the Mid and Insular dialects. In the South Estonian dialect group there are on average slightly more vowels than in the northern central Estonian dialects; the highest percentage is in the Tartu dialect. The lowest percentage of high vowels can be found in the South Estonian Mulgi dialect which is bordering with the Western dialect. The main reason for the differences lies most probably in the reduction and lowering of high vowels in South-Western dialects of Estonia as in non-initial syllables: i > e, e.g. pulmalised 'wedding guests' > pulmalest, suureline 'great' > suurelene, suuresti 'greatly' > suureste, (ei) tohi 'may (not)' > tohe but also in the lowering of high vowels in the first syllable in certain environments as in uheksa 'nine' > oheksa, mitte 'not' > Mette. Such changes occur least in the North-Eastern Coastal dialect.
[FIGURE 4 OMITTED]
The percentage of low vowels (a, a) decreases also in the direction from South West to North East if we discount the South Estonian dialects (see figure 5). The highest percentage of low vowels can be found in the Western dialect and the lowest in the dialects of the North-Eastern Coastal dialect group. Again, it is probably the Western dialect that is more innovative as here the high vowels as a rule change into mid-high, as we saw above, and mid-high vowels change into low vowels. This tendency is strongest in the southern group of the Western dialect, e.g. in non-initial syllables: e > a as in Tahkuranna mere 'of sea' > mera, rehe 'of drying barn' > riha. More common is the change in first syllables: e > a e.g. enam 'any more' > anam, vedama 'to drag' > vadama, erk 'perky, alert' > ark. In northern dialects, on the other hand, it is more common for the long dd and as to diphthongize as in paaseb 'escapes' > peaseb ~ piaseb, maa 'land' > moa ~ mua.
[FIGURE 5 OMITTED]
South Estonian dialects, however, do not follow the general trend of change of the low vowels. Here the percentage of low vowels is smallest in the Mulgi dialect in the west, and largest in the Setu dialect in the east. This is caused by opposite changes taking place on the different edges of the dialect continuum. The western dialects of South Estonia can be characterized by the reduction of a and a and their changing into e (Pajusalu 1998) as e.g. in Karksi armastama 'to love' > armasteme. On the other hand, in the eastern dialects of South Estonia, it is common for the mid-high vowels to lower in different groups (see Pajusalu 2000) as in *taloille > tal(l)alo 'to the farms'. Additionally, it is characteristic of the eastern South Estonian dialects to lower the e in the first syllable in some words as in lesk 'widow' > lask, seitse 'seven' > saitse, and in Baltic loans, to have ai instead of ei as in hain 'hay', saivas 'pole', saista 'to stand'.
2.5. Front and back vowels in Estonian dialects
In order to clarify the distribution of front and back vowels we divided all the vowels into two groups so that the front vowels would include i, e, a, u, o and the back vowels a, o, u as well as o and i because the latter two behave similar to back vowels from the point of view of vowel harmony (although i can sometimes also occur in the words containing front vowels, see Parve 2000). Figure 6 presents the percentages of these back vowels.
[FIGURE 6 OMITTED]
It can be seen that the percentage of back vowels is relatively smallest in the North-Eastern Coastal dialect group and largest in the eastern dialects of South Estonia. This is the result of the o-harmony in these dialects: where in words containing back vowels e has been replaced by o the general percentage of back vowels is also slightly larger than that of the front vowels. The dialects where the neutral e is counted as a front vowel because of its phonetic characteristics exhibit a slightly larger number of front vowels. But if we discount the neutral e and i all dialects have considerably more back vowels than front vowels. It is apparent that in Estonian dialects the backness of vowels is the primary unmarked feature which is retained even after the loss of the o-harmony.
2.6. Vowels in the fourth syllable
So far we have looked at the general tendencies of the vowel systems in Estonian dialects disregarding the syllables where the vowels occur. A closer look, however, shows that that there are large differences in the vowels of different syllables. We will present here as a separate example the statistics about the fourth syllable that is always unstressed (see table 9). As i did not occur in the fourth syllable it has been left out of the table.
The fourth syllable which in Estonian dialects is always part of a suffix is characterized in all dialects by the restricted number of vowels. The largest number (9 vowels) appears in the South Estonian Voru dialect but as already mentioned this dialect lacks i in the fourth syllable and therefore does not contain the full set of vowels as well. In addition to i, Setu lacks o, Tartu additionally o, and Mulgi o and u. The North-Eastern Coastal dialect group has six vowels in the fourth syllable: a, e, i, u, a and u, whereas the North Estonian dialects have only five: a, e, i, u and a, and the Western dialect also has additionally o. Thus, the fourth syllable in all North Estonian dialects contains only primary vowels: a, e, i and u, and additionally, because of the wide-spread a-harmony, a in words with front vowels.
The differences in percentages are very large. The percentage of rounded vowels is the smallest, with only u occurring regularly. The only frequently occurring high vowel is i and back vowel a. The most common vowel in the nine dialects is the mid-high e, and in Setu, its back equivalent o which is also widespread in Voru. If we disregard the neutral i and e it appears that back vowels are in all dialects many times more frequent than the front vowels.
Several previous studies have investigated the vowel systems and sound changes of Estonian dialects but the statistical analysis presented above is the first of its kind. The results of the analysis show that several changes of single sounds are connected to more general tendencies of change in the vowel system. A more thorough study of the causes of these wide ranging changes in the vowel systems remains to be carried out in the broader context of areal linguistics. For instance, the above-described tendency to vowel rounding in West Estonia can be explained with language contacts with Swedish, and the changing of the rounded vowels into unrounded vowels in East Estonia can be due to the Slavic influence. It is apparent that the characteristic traits of the vowel system of Estonian dialects reflect often even broader characteristics of the Baltic Sea language area.
In addition to language contacts, the study of the general characteristics of the vowel systems is important from the point of view of establishing the internal rules of the systems including the markedness of the vowels. The statistical analysis showed that the primary vowels a, i, e, u and o are as a rule most frequent in Estonian dialects. But in four dialects a is more common than o, and in the Setu dialect also o is more frequent. In non-initial syllables of Setu, o is even more common than its front equivalent e, which raises doubts about the markedness of o in the Setu vowel harmony. The present study implies clearly that the comparison of the most general statistical characteristics of dialectal vowels enables to pinpoint several broader traits of the sound systems and their dynamics.
[TEXT NOT REPRODUCIBLE IN ASCII].
Ariste, P. 1984, Eesti keele foneetika I, Tartu.
Eek, A., Meister, E. 1994, Eesti vokaalide sihtvaartused haaldus- ja tajuruumis.--KK, 404-413, 476-483, 548-553.
Hint, M. 1997, Eesti keele astmevahelduse ja prosoodiasusteemi tupoloogilised probleemid, Tallinn--Helsinki.
--1998, Haalikutest sonadeni. Eesti keele haalikususteem uldkeeleteaduslikul taustal. Teine, umbertootatud trukk, Tallinn.
Krikmann, A., Pajusalu, K. 2000, Kus on keskmurde keskpunkt.--Inter dialectos nominaque. Puhendusteos Mari Mustale 11. novembril 2000, Tallinn (Eesti Keele Instituudi toimetised 7), 131-172.
Lindstrom, L., Lonn, V., Mets, M., Pajusalu, K., Teras, P., Veismann, A., Velsker, E., Viikberg, J. 2001, Eesti murrete korpus ja kolme murde sagedasema sonavara vordlus.--Keele kannul. Puhendusteos Mati Erelti 60. sunnipaevaks, Tartu (Tartu Ulikooli eesti keele oppetooli toimetised 17), 186-211.
Murumets, S. 1982-1983, Eesti keeleala murdelisest liigendusest "Vaikese murdesonastiku" pohjal 1-2.--KK 1982, 11-17; 1983, 615-623.
Must, M. 2000, Vene laensonad eesti murretes, Tallinn.
Pajusalu, K. 1998, Vowel Reduction in South Estonian.--LU XXXIV, 234-240.
--2000, Alternation of e and a, a in Non-Initial Syllables in the Southern Group of the Finnic Languages.--Facing Finnic. Some Challenges to Historical and Contact Linguistics, Helsinki (Castrenianumin toimitteita 59), 156-167.
--2003, Estonian Dialects.--Estonian Language, Tallinn (Linguistica Uralica. Supplementary Series 1) (In press).
Pajusalu, K., Parve, M., Teras, P. 2001, On the Main Characteristics of the Prosody of South Estonian Dialects.--CIFU IX. Pars VI, 9-13.
Pajusalu, K., Hennoste, T., Niit, E., Pall, P., Viikberg, J. 2002, Eesti murded ja kohanimed, Tallinn.
Parve, M. 2000, Voru luhikeste monoftongide akustikast.--K. Pajusalu, M. Parve, P. Teras, S. Iva, Voru vokaalid I, Tartu (Tartu Ulikooli eesti keele oppetooli toimetised 13).
-- 2003, Valted lounaeesti murretes, Tartu (Dissertationes philologiae estonicae Universitatis Tartuensis 12).
Suihkonen, P. 1998, Documentation of the Computer Corpora of Uralic Languages at the University of Helsinki. TR-2, Helsinki.
Tauli, V. 1956, Phonological Tendencies in Estonian, Kobenhavn.
Teras, P. 2003, Lounaeesti vokaalisusteem. Voru pikkade vokaalide kvaliteedi muutumine, Tartu (Dissertationes philologiae estonicae Universitatis Tartuensis 11).
Viitso, T.-R. 1981, Laanemeresoome fonoloogia kusimusi, Tallinn.
--1990, Vowels and Consonants in North Setu (South Estonian).--LU XXVI, 161-172.
Wiik, K. 1988, Viron vokaalisointu, Helsinki (Suomi 140).
* This study has been carried out as part of the projects No. 4192 and 4404 of the Estonian Science Foundation.
LIINA LINDSTROM (Tartu), KARL PAJUSALU (Tartu-Helsinki)
Table 1 The number of words in the corpus (October 2003) Dialect Sub-dialects Number of words Coastal Joelahtme, Kuusalu 43 905 North-Eastern Johvi, Luganuse 36 550 Insular Kihelkonna, Kihnu, Kaina, Mustjala, Puhalepa 88 764 Western Haademeeste, Mihkli, Varbla 40 625 Mid Juuru, Juri, Keila, Pilistvere, Viru-Jaagupi, Vaike-Maarja 45 179 Eastern Kodavere, Torma 20 499 Mulgi Halliste, Karksi, Tarvastu 23 903 Tartu Kambja, Noo, Otepaa, Rongu, Vonnu 67 682 Voru Hargla, Polva, Rapina, Urvaste, Vastseliina 46 574 Setu Northern Setu, Western Setu 42 219 Total 455 900 Table 2 Data collection periods Year Number of recordings 1938 5 1957-1959 17 1960-1969 60 1970-1979 31 1980-1986 8 unknown 1 Total 122 Table 3 Years of birth of the informants Year Number of informants 1865-1869 7 1870-1879 40 1880-1889 44 1890-1899 20 1900-1909 13 1910-1919 7 Total 131 Table 4 Information fields *SNE: precise phonological shape of the word FRA: phrase where the word appears *MSN: a lemma or base form for the word (in most cases it is an entry for the word given in the Dictionary of Estonian Dialects) TAH: the meaning (if it is different for the standard meaning of the entry) *SLK: part of speech *MRF: morphological categories *-- filling is obligatory Table 5 Vowels of Standard Estonian Vowel a o u e i o a o u High - - + - + (+) - - + Low + - - - - - + - - Rounded - + + - - - - + + Back + + + - - (+) - - - Front - - - + + - + + + Table 6 Number of occurrence and percentage of vowels in different Estonian dialects (Long monophthongs are grouped together with short monophthongs) a e Mid 16878 28.04% 12660 21.03% Western 13571 27.69% 10209 20.83% Insular 14064 26.77% 10757 20.47% Eastern 5888 25.88% 4676 20.55% Coastal 14005 24.43% 12519 21.84% North-Eastern 11240 23.90% 9036 19.21% Tartu 17347 22.89% 11561 15.25% Mulgi 7392 22.25% 7119 21.43% Voru 11289 24.06% 6033 12.86% Setu 11654 24.40% 5179 10.84% i o Mid 14482 24.06% 4640 7.71% Western 10588 21.61% 3947 8.05% Insular 11923 22.69% 4768 9.07% Eastern 5469 24.04% 1122 4.93% Coastal 14053 24.52% 5181 9.04% North-Eastern 12666 26.93% 2778 5.91% Tartu 17597 23.22% 6549 8.64% Mulgi 6599 19.86% 2340 7.04% Voru 10077 21.47% 4458 9.50% Setu 10267 21.50% 4455 9.33% u o Mid 5286 8.78% 1602 2.66% Western 4648 9.49% 1349 2.75% Insular 5444 10.36% 4 0.01% Eastern 2071 9.10% 1209 5.31% Coastal 6274 10.95% 357 0.62% North-Eastern 4460 9.48% 2367 5.03% Tartu 7701 10.16% 4949 6.53% Mulgi 3608 10.86% 1745 5.25% Voru 5020 10.70% 3638 7.75% Setu 4361 9.13% 4807 10.07% a o Mid 2850 4.73% 186 0.31% Western 3361 6.86% 212 0.43% Insular 2958 5.63% 1282 2.44% Eastern 1660 7.30% 80 0.35% Coastal 3047 5.32% 273 0.48% North-Eastern 3251 6.91% 167 0.36% Tartu 7041 9.29% 110 0.15% Mulgi 3198 9.62% 43 0.13% Voru 4298 9.16% 78 0.17% Setu 4808 10.07% 74 0.15% u i Mid 1613 2.68% 0 0.00% Western 1118 2.28% 0 0.00% Insular 1342 2.55% 0 0.00% Eastern 576 2.53% 0 0.00% Coastal 1610 2.81% 0 0.00% North-Eastern 1070 2.27% 0 0.00% Tartu 2617 3.45% 315 0.42% Mulgi 1182 3.56% 1 0.00% Voru 1646 3.51% 388 0.83% Setu 1565 3.28% 585 1.23% Table 7 Number of occurrence and percentage of vowels in different Estonian dialects (Long monophthongs are counted as two vowels) a e Mid 17804 27.44% 13291 20.48% Western 14593 26.93% 11397 21.03% Insular 15345 26.25% 11977 20.49% Eastern 6227 25.43% 4931 20.14% Coastal 15379 24.87% 12623 20.41% North-Eastern 12090 23.90% 9076 17.94% Tartu 18916 22.68% 12035 14.43% Mulgi 8028 22.31% 7236 20.11% Voru 12167 23.15% 6312 12.01% Setu 12587 23.92% 5536 10.52% i o Mid 16166 24.92% 5025 7.74% Western 11822 21.82% 4568 8.43% Insular 13413 22.95% 5316 9.09% Eastern 6059 24.74% 1276 5.21% Coastal 15955 25.80% 5300 8.57% North-Eastern 14335 28.34% 2888 5.71% Tartu 19393 23.25% 6996 8.39% Mulgi 7405 20.58% 2471 6.87% Voru 11446 21.78% 4819 9.17% Setu 11257 21.39% 4771 9.07% u o Mid 5862 9.03% 1656 2.55% Western 5224 9.64% 1411 2.60% Insular 5841 9.99% 5 0.01% Eastern 2260 9.23% 1217 4.97% Coastal 6759 10.93% 359 0.58% North-Eastern 4800 9.49% 2381 4.71% Tartu 8956 10.74% 5249 6.29% Mulgi 3951 10.98% 1767 4.91% Voru 6413 12.20% 3768 7.17% Setu 5273 10.02% 5085 9.66% a o Mid 3159 4.87% 259 0.40% Western 3712 6.85% 319 0.59% Insular 3729 6.38% 1467 2.51% Eastern 1770 7.23% 126 0.51% Coastal 3420 5.53% 275 0.44% North-Eastern 3672 7.26% 168 0.33% Tartu 8346 10.01% 175 0.21% Mulgi 3709 10.31% 74 0.21% Voru 5161 9.82% 103 0.20% Setu 5602 10.65% 108 0.21% u [??] Mid 1660 2.56% 0 0.00% Western 1137 2.10% 0 0.00% Insular 1360 2.33% 0 0.00% Eastern 623 2.54% 0 0.00% Coastal 1770 2.86% 0 0.00% North-Eastern 1174 2.32% 0 0.00% Tartu 3012 3.61% 324 0.39% Mulgi 1334 3.71% 1 0.00% Voru 1961 3.73% 407 0.77% Setu 1808 3.44% 594 1.13% Table 8 Number of occurrence and percentage of long vowels in different Estonian dialects aa ee Mid 926 1.54% 631 1.05% Western 1022 2.09% 1188 2.42% Insular 1281 2.44% 1220 2.32% Eastern 339 1.49% 255 1.12% Coastal 1374 2.40% 104 0.18% North-Eastern 850 1.81% 40 0.09% Tartu 1569 2.07% 474 0.63% Mulgi 636 1.91% 117 0.35% Voru 878 1.87% 279 0.59% Setu 933 1.95% 357 0.75% ii oo Mid 1684 2.80% 385 0.64% Western 1234 2.52% 621 1.27% Insular 1490 2.84% 548 1.04% Eastern 590 2.59% 154 0.68% Coastal 1902 3.32% 119 0.21% North-Eastern 1669 3.55% 110 0.23% Tartu 1796 2.37% 447 0.59% Mulgi 806 2.43% 131 0.39% Voru 1369 2.92% 361 0.77% Setu 990 2.07% 316 0.66% uu oo Mid 576 0.96% 54 0.09% Western 576 1.18% 62 0.13% Insular 397 0.76% 1 0.00% Eastern 189 0.83% 8 0.04% Coastal 485 0.85% 2 0.00% North-Eastern 340 0.72% 14 0.03% Tartu 1255 1.66% 300 0.40% Mulgi 343 1.03% 22 0.07% Voru 1393 2.97% 130 0.28% Setu 912 1.91% 278 0.58% aa oo Mid 309 0.51% 73 0.12% Western 351 0.72% 107 0.22% Insular 771 1.47% 185 0.35% Eastern 110 0.48% 46 0.20% Coastal 373 0.65% 2 0.00% North-Eastern 421 0.90% 1 0.00% Tartu 1305 1.72% 65 0.09% Mulgi 511 1.54% 31 0.09% Voru 863 1.84% 25 0.05% Setu 794 1.66% 34 0.07% uu [??][??] Mid 47 0.08% 0 0.00% Western 19 0.04% 0 0.00% Insular 18 0.03% 0 0.00% Eastern 47 0.21% 0 0.00% Coastal 160 0.28% 0 0.00% North-Eastern 104 0.22% 0 0.00% Tartu 395 0.52% 9 0.01% Mulgi 152 0.46% 0 0.00% Voru 315 0.67% 19 0.04% Setu 243 0.51% 9 0.02% Table 9 Percentage of vowels in the fourth syllable No. a e i Mid 720 23.9 56.1 15.7 Western 343 27.1 45.2 24.2 Insular 441 27.2 48.5 22.0 Eastern 326 22.7 53.4 18.1 Coastal 613 21.2 51.1 19.1 North-Eastern 689 26.1 43.5 21.3 Tartu 793 26.2 38.2 12.9 Mulgi 196 5.1 72.7 15.3 Voru 578 24.2 26.1 17.7 Setu 637 14.9 23.9 15.5 o u o Mid - 3.9 - Western 0.6 2.3 - Insular - 2.3 - Eastern - 4.9 - Coastal - 6.7 - North-Eastern - 3.3 - Tartu - 3.8 13.5 Mulgi 0.5 4.6 - Voru 0.7 5.7 21.3 Setu 1.3 1.7 39.7 a o u Mid 0.4 - - Western 0.6 - - Insular - - - Eastern 0.9 - - Coastal 1.3 - 0.5 North-Eastern 5.5 - 0.2 Tartu 4.5 - 0.9 Mulgi 1.5 - - Voru 3.5 0.2 0.7 Setu 2.5 - 0,5