Meyer's Accent Contours Revisited

TMH-QPSR Vol. 44 – Fonetik 2002

Meyer’s accent contours revisited

Olle Engstrand1 and Gunnar Nyström2 1Department of Linguistics, Stockholm University 2Institute for Dialectology, Onomastics and Folklore Research (SOFI), Uppsala

Abstract Do E.A. Meyer’s tonal word accents contours from the Swedish dialects provide a reliable basis for quantitative analysis? Measurements made on acute and grave tone-peaks in a number of dialects spoken in the province of Dalarna suggested that the timing of grave tonal peaks tended to vary systematically from south-east to north-west. The former dialects had relatively late and the latter relatively early tone-peaks. This finding suggests that Meyer’s accent data may be sufficiently accurate to reflect systematic variation within broad dialect areas. Implications for the historical development of the Dalarna dialects are discussed.

Introduction

Several decades ago, E.A. Meyer compiled his pioneering survey of tonal word accent contours in 100 Swedish dialects (Meyer 1937, 1954). Pitch curves were automatically generated using a pitch meter that Meyer himself had invented. The original pitch curves are thus likely to be quite accurate. To enhance the dialect-specific tonal characteristics, the original contours were time-normalized, averaged and smoothed by Figure 1. Tonal contours representing the eye. These schematized contours were arranged Central Swedish dialects (Stockholm, upper by province and displayed on charts. panel) and the Dalarna dialects (Leksand, lower The Meyer contours have proved useful in panel). establishing accent-based dialect typologies In this paper, the question is raised whether the (Gårding 1977, Bruce & Gårding 1978, Öhman Gårding and Bruce scheme can be elaborated to 1967). In broad outline, the tonal typology describe continuous variation within the broad agrees fairly well with other dialect categories. This would provide a possibility to classifications (Wessén 1960). This suggests observe small but systematic accentual that even the simplified contours provide a variations between neighboring dialects and, reliable data base. In addition, later analyses complemented with recent data from the (e.g., Olander 2000) have demonstrated a close SWEDIA 2000 project (http://www.swedia.nu), agreement between Meyer’s contours and F - 0 to shed light on more recent phonetic data obtained using modern techniques. developments of the word accents. The Gårding and Bruce typology categorized Is it, then, possible to quantify Meyer’s tone dialects according to number and timing of tonal contours at an adequate level of detail? To peaks. For example, Central Swedish dialects conclude that it is, we would require that were characterized by two peaks in the grave measurements fall into linguistically reasonable accent contour, the one-peaked acute contour patterns that are likely to be non-random. For reaching a tonal maximum late in the stressed example, a gradual shift in some tonal parameter syllable or early in the post-stress syllable. In as we travel across a particular dialect region, or contrast, the Dalarna dialects typically have one- an areal covariation between tonal and other peaked grave as well as acute contours, the acute dialect features would strengthen our belief in peak consistently appearing prior to the grave the tonal analysis. As a preliminary test of these peak. These contours are exemplified in figure 1 ideas, we have performed a set of measurements with Central Swedish represented by Stockholm on Meyer’s grave and acute word accent and the Dalarna dialects represented by contours in dialects in and near Dalarna. Leksand.

17 Speech, Music and Hearing

Areal word accent variation in dialects having a relatively early grave tone- peak and ending with those having a relatively and near Dalarna late peak. As can be seen, the range of variation Digitized versions of the Meyer tone curves in between the first dialect in the list (Mora) and disyllabic words were used to evaluate a number the last dialect (Norrbärke) is rather large (47 of parameters for the grave and acute contours. time units). For the acute accent, the picture is This paper reports data on the location in time of similar with a range of variation of 50 time acute and grave tonal peaks relative to the VC units. boundary as indicated in Meyer’s figures There is at least one rather clear discontinuity (shown by vertical lines as exemplified in figure in the list of grave peak times. That 1 above). Time units are arbitrary since Meyer’s discontinuity falls between the Grangärde and figures do not provide a time scale. For Ål dialects as marked with a double line in the comparability, only dialects with a one-peaked table. On the map, the discontinuity corresponds grave contour have been considered. to a geographical boundary (indicated by tic marks) that separates a south-eastern dialect area Table 1. Timing of tone-peaks in grave and characterized by relatively late grave tone-peaks acute accent words in Dalarna dialects with from an area to the north-west with earlier peaks one-peaked grave contours. Negative values (Floda, Boda, Malung, Rättvik, Leksand, represent peaks appearing prior to the VC Siljansnäs, Grangärde). This intermediate area, boundary and vice versa. Further explanation in text. Based on Meyer (1937, 1954). in turn, has been tentatively separated on the map (again using tic marks) from the area Peak time containing the Mora and Orsa dialects, which Dialect (arbitrary units) both display particularly early grave tone-peaks. Grave Acute However, the latter discontinuity is not quite as Mora -21 -49 sharp. Finally, a third tic-marked boundary sets Orsa -19 -71 an upper left area off from the remaining Floda -13 -56 dialects under discussion. That area contains Boda -11 -52 dialects that are known to have two-peaked Malung -6 -32 grave accent contours (Meyer’s survey includes Rättvik -2 -44 Transtrand and Älvdalen but not Lima). Leksand -1 -38 In summmary, grave tone-peaks tend to Siljansnäs -1 -28 appear later the further south by south-east that Grangärde 3 -68 we move across the one-peaked dialects on the map. Some transitions were found to be Ål 15 -49 relatively abrupt. In particular, Ål and Djura Djura 22 -35 agreed better with the south-eastern dialects Grytnäs 22 -18 (Grytnäs, Norrbärke) than with the more Norrbärke 26 -36 neighboring dialects to the north and west, e.g., Leksand, Siljansnäs and Floda). It can be noted, Dialects and measurement values are presented in addition, that the pattern with relatively late in table 1, and the geographical distribution of grave tone-peaks continues even further to the the dialects is shown on the map in figure 2 south-east including dialects spoken in eastern comprising most of the traditional Dalarna Västmanland (Skultuna) and western Uppland province. The dialects appearing in table 1 are (Norrby and Simtuna; the Simtuna dialect printed in boldface on the map. (The contour appears to have lost the word accent contrast). appearing in the middle of the map with Mora to Thus, table 2 shows that the grave peak values the north and Siljansnäs to the south is Lake found for those dialects are in fair agreement Siljan.) with those pertaining to the four bottom Dalarna In the table, data pertaining to dialects dialects in table 1. represented by more than one speaker have been Sorting the dialects by the timing of acute averaged and rounded to the nearest integer. A tone-peaks does not give the same neat negative number for a tonal peak indicates that geographical distribution as that based on the the peak falls to the left of the VC boundary and grave peaks. In the Dalarna dialects, however, vice versa. The rows are sorted by the ‘Grave’ the grave and acute peak times display a fair column in ascending order, thus starting with statistical correlation (r=0.55) suggesting a weak

18 TMH-QPSR Vol. 44 – Fonetik 2002

1 2 3 4 5 Älvdalen

Transtrand Våmhus Ore Orsa Boda Venjan Mora Lima Rättvik Sollerön Siljansnäs Malung Bjursås Leksand Ål Järna Djura Floda Gagnef Nås Äppelbo

Grangärde

Norrbärke Grytnäs

Figure 2. Map of the Dalarna dialect area. The dialects in table 1 are printed in boldface; Table 2. Timing of tone-peaks in grave and dialect areas not discussed here have been left acute words in three one-peaked dialects spoken out. Tic-marked boundaries are tentatively south-east of Dalarna. Based on Meyer (1937, drawn on the basis of the timing of grave tone- 1954). peaks. Lines 1-5, essentially based on Levander Peak time (1925, 1928) and SOFId (see references), Region Dialect (arbitrary units) indicate linguistic boundaries (later Grave Acute developments to the SE, conservative features Västmanland Skultuna 9 -28 to the NW, as discussed in the next section): 1. Uppland Simtuna 13 13 North: four-way case inflection preserved; 2. Uppland Norrby 37 -29 NW: vowel balance preserved ([a] in suffixes following historically long syllables, [o] in tendency to a relatively stable time distance suffixes following historically short syllables; between the grave and acute tone-peaks. see Bergfors 1961, p. 15 ff.); 3. northern [s] At the moment, we have no explanation of as opposed to southern [] for old [sk] before high front vowels; 4. NW: use of the dative in the less systematic behavior of the acute definite form of nouns preserved; 5. SE: contours. The possibly greater variability of feminine forms of pronouns ending with –um acute peak times should be tested using data and –om (innovation from about 1300 AD, also from other dialect regions with one-peaked occurring in parts of Uppland, Västmanland grave and acute accent contours. and Gästrikland; see Fries 1958.)

19 Speech, Music and Hearing

Relationship with other dialect 1977, Bruce & Gårding 1978). It was found that the timing of the tonal peak in one-peaked grave features accent contours tended to vary systematically From a linguistic point of view, Dalarna is more across the Dalarna dialects with late tone-peaks variable than most other Swedish provinces. A in the south-east and early tone-peaks towards pre-medieval boundary cuts across the northern the north-west. These findings are compatible part of the province such that the far north- with the hypothesis that Meyer’s data reflect a western dialects (with, e.g., i-Umlaut in the historical spread from south-east to north-west present tense of strong verbs) belong to the West which agrees in many ways with independent Scandinavian language area, whereas the historical evidence. Clearly, however, these remaining dialects are clearly linked to Eastern tentative conclusions have to be confirmed by Scandinavia. Of these, the dialects north of Lake measurements on older recordings available in Siljan are the most distinctive with a variety of the dialect archives. For the time being, archaisms in combination with radical however, the results strengthen our belief that it innovations, e.g., diphthongization of long Old is possible to use the Meyer accent material as a Scandinavian vowels. basis for quantitative analysis. A typical feature of the dialectal scenario in Dalarna is a flow of later south-eastern Acknowledgments influences that have originated in central parts of Thanks to Östen Dahl for sharing the digitized Meyer Svealand and spread to the north-west to various curves and to Birgit Eaker for preparing the map. degrees. In many cases, the spread has reached Work supported by the Bank of Sweden no longer than to the district south of Lake Tercentenary Foundation, grant 1997-5066 to the project Phonetics and phonology of the Swedish Siljan (Levander 1925, p. 17 ff.). Some dialects around the year 2000 (SWEDIA 2000). examples are given by lines 1-5 on the map and summarized in the figure legend. The lines indicate the northern and/or western boundaries References of the respective developments. Thus, whereas Bergfors E O (1961). Tilljämning a > å i dalmål. Skrifter utg. av Inst. för nordiska språk vid the dialects north of a given boundary represent Uppsala universitet 10. Uppsala: Almqvist & an older situation, those south of the boundary Wiksell. have been affected by historical change. Bruce G & Gårding E (1978). A prosodic typology Meyer’s word accent survey suggest that the for Swedish dialects. In E. Gårding, G. Bruce & R. phonetic implementation of the Dalarna word Bannert (eds.), Nordic Prosody, Travaux de l’Institut de linguistique de Lund 13. Lund : accents may also be a result of influences from Gleerup, 219-228. the south-east. In the case of the accents, Fries S (1958). Slutljudande m i stället för n i however, the historical development appears to fornsvenskan och i uppsvenska dialekter. Svenska have gained in strength from south-east to north- landsmål och svenskt folkliv 1958, 149-169. west. In particular, the relatively late grave tone- Stockholm: Norstedts. peaks in the Ål and Djura dialects seem to agree Gårding E (1977). The Scandinavian word accents. Travaux de l’Institut de linguistique de Lund 11. with the southern Dalarna as well as some Lund : Gleerup. Västmanland and Uppland dialects. In contrast, Levander L (1925, 1928). Dalmålet. Beskrivning och the most far-reaching tonal development seems historia I-II. Uppsala. to have taken place in Mora and Orsa, just north Meyer E A (1937, 1954). Die Intonation im of Lake Siljan. This agrees with the fact that, Schwedischen I-II. Studies Scand. Philol. 10-11, based on other linguistic criteria, these dialects Stockholm University. occupy a place apart from the surrounding Olander E (2000). Word accents in today’s Orsa dialect: preliminary F0 measurements. Proceedings dialects. Finally, the north-western dialects have of the 13th Swedish Phonetics Conference (Fonetik preserved the two-peaked grave contour along 2000), Skövde, pp. 117-120. with other archaic features. SOFId=The dialect collections at the Institute for Dialectology, Onomastics and Folklore Research (SOFI), Uppsala. Summary and conclusions Wessén E (1969/1935). Våra folkmål. Malmö: We have considered the possibility that the word Fritzes (6th edition). accent contours published by Meyer (1937, Öhman S (1967). Word and sentence intonation: a quantitative model. STL-QPSR 2-3. 1954) may reveal systematic tonal variation within broad typological categories (Gårding