Published online by Cambridge University Press: 22 April 2026
[Application of the diachronic method to a structural study. A discussion of the segmental phonemes of a Dutch dialect introduces a reconstruction of the development of the vowel system, on historical evidence supplemented by an interpretation of distributional peculiarities in the modern dialect. Structural influences of the standard language on the dialect are discussed, and certain future developments are outlined.]
1 The material serving as a basis for this paper was collected during a stay in the Netherlands on a Fulbright grant, 1955–56. Many problems of analysis have been discussed with Einar Haugen, William G. Moulton, and Lester W. J. Seifert, to whom I am indebted for their valuable suggestions.
2 Studies of the dialects of the eastern area and their historical and geographical relationships to each other are G. G. Kloeke, Het taallandschap van onze noordoostelijke provinciën (Verhandelingen der Koninklijke Nederlandse Akademie van Wetenschappen, Afd. Letterkunde, NR Vol. 62, No. 3); De taal van Overijsel, Overijsel (Deventer, 1931); K. Heeroma, Het West-Overijselse Taallandschap (Amsterdam, 1955). General, semipopular works on the dialects of the Netherlands are Jac. van Ginneken, De Studie der Nederlandse streektalen (Amsterdam, 1943), and A. Weijnen, De Nederlandse dialecten (Groningen, 1941).
3 Many speakers of standard Dutch make a distinction between voiced and voiceless velar spirants in initial position, e.g. in gas and chaos. In the speech of many others, however, especially those born and raised in the western provinces of North and South Holland, the voicing in the earlier [g] has vanished in all environments, so that it has merged completely with [x]. Probably there are other natives whose speech, like Zw., lacks [g] only in initial position.
4 The distribution discussed under (a) and (b) holds true only when the word is in isolation. In context all consonants with phonemic voicing enter into regular morphophonemic alternations, which are characteristic as well of many varieties of standard Dutch. Consonant clusters whose components have phonemic voicing are either entirely voiced or entirely unvoiced; final voiceless stops are replaced by the corresponding voiced stop or spirant when followed by voiced stop or vowel: /op/ +/du·n/ > /obdu·n/ ‘to store up’, /skrif/+/da·dələk/>/skrivda·dələk/ ‘write right away’, /zεs/ + /y·r/ > /zεzy·r/ ‘six o'clock’; initial voiced spirants are replaced by the corresponding voiceless spirant when preceded by a voiceless stop or spirant: /op/ + /zy·kŋ/ > /opsy·kŋ/ ‘to visit’, /aləs/ + /van/ > /aləsfan/ everything of', /tɔx/ + /zøt/ > /tɔxsøt/ ‘yet sees’.
5 The following minimal pairs will illustrate this in full: /lif/ ‘waist’ ~ /li·f/ ‘dear’, /bykin/ ‘little stomach’ ~ /by·kin/ ‘little book’, /buk/ ‘stomach’ ~ /bu·k/ ‘book’, /wet/ ‘white’ ~ /we·t/ ‘knows’, /bøtin/ ‘little bone’ ~ /bø·tin/ ‘little boat’, /bot/ ‘bone’ ~ /bo·t/ ‘boat’, /dεkŋ/ ‘to cover’ ~ /ete·kŋ/ ‘blanket’, /strœtin/ ‘little throat’ ~ /strœ·tin/ ‘little street’, /lɔt/ ‘lot’ ~ /lɔ·t/ ‘(you) let’, /bækər/ ‘baker’ ~ /bæ·kər/ ‘nurse’, /valə/ ‘(I) fall’ ~ /va·lə/ ‘veil’. This analysis (i.e. as /bu·k/) was considered preferable to several other possibilities: (1) Use of the glides /h j w/ (e.g. /buhk/), while probably the best description of standard Dutch, is difficult for Zw. because of the questionable phonemic value of /h/. (2) Analysis as geminate vowel (e.g. /buuk/) would not agree with that of the diphthongs, which have been analyzed not as vowel + unlike vowel but, partly for reasons of consonant distribution, as vowel + consonant. (3) Trubetzkoy's Silbenschnitt correlation was asserted by N. van Wijk to be the only satisfactory analysis of the vowel systems of the Netherlands area, where in many dialects length does not seem to be phonemically relevant (Onze Taaltuin 9.233). Following this interpretation, according to which all vowels which can occur only in closed syllable are ‘checked’ and all others ‘free’, Zw. would have as checked all the simple vowels except /i y u/ and as free all the vowels followed by /·/, as well as /i y u/. Since /i· y· u·/ and /i y u/ are all free and contrast only in length, it would still be necessary to introduce a length opposition within the checked– free scheme and thus, by attempting to resolve a distributional irregularity at the phonemic level, create an additional complication.
6 This distribution illustrates the historical consequence of the adoption of the uvular r in Zw. It had a lowering effect on /e/ and a fronting effect on /a/: /ært/ ‘hard’ < *hardu-and ‘hart’ < *heruta-. Although it seems probable that /o/ and /ɔ/ (earlier /u/ and /o/) had at an early date fallen together before /r/ in Old Saxon (Gallée, Altsächsische Grammatik §77) and in Middle Low German (Lasch §61), the later Zw. uvular /r/ nevertheless had a fronting effect which made /ɔ/ identical with the regular umlaut of the same vowel: /bœrgər/ ‘citizen’ < *burgâri- and /dœrəp/ ‘village’ < *þurpa-. Zw. /ɔ·/ was likewise fronted until it merged with the umlaut reflex of the same vowel: /jœ·r/ ‘year’ < *jâra-and /əsprœ·krj/ ‘spoken’ < ?*gisprokina-.
7 Note that after labials, dentals, and velars we have syllabic /m n ŋ/ rather than /əm ən əŋ/. In order to avoid a distributional irregularity of syllabic nasals one could analyze /lipm ærtn wε·gŋ/ as showing /ə/ + nasal, but the fact remains that even in careful pronunciation and in isolation these words are spoken without any perceptible vocalic element, and words like /ny·mm/ ‘to name’ are pronounced with a geminate nasal, never as *[ny·-məm] or [ny·mən].
8 The phonemic status of the anaptyctic vowel in Dutch in words like kalf, elk, arm, and werk and the strong social resistance to it are discussed by P. C. Paardekooper, De foneem-waarde van de svarabhakti-vocaal, Nieuwe Taalgids 42.74–84. Thanks to the absence of a written tradition in Zw., the resistance to it encouraged in Dutch by the orthography is lacking.
9 The evidence from which we build up our diachronic scheme is as follows. (1) Handbooks of Proto-Germanic, e.g., Prokosch's Á comparative Germanic grammar (Philadelphia, 1939) and Hirt's Handbuch des Urgermanischen (Heidelberg, 1931–4), as well as W. F. Twaddell's article The prehistoric Germanic short syllabics, Lg. 24.139–51 (1948). (2) Old Saxon documents, the forms being quoted from E. H. Sehrt, Vollständiges Wörterbuch zum Heliand (Göttingen, 1925); J. H. Gallée, Altsächsische Grammatik (Halle, 1910). (3) Middle Low German evidence, with forms quoted from A. Lasch, Mittelniederdeutsche Grammatik (Halle, 1914), hereafter Lasch; this evidence is used sparingly because of the large dialect area represented. The first medieval texts which can with some certainty be localized in Zwolle are the Stadboeken van Zwolle of about 1570, edited by A. Telting in Overijsselsche Stad-Dijk- en Markeregten, Vol. 1, Part 12 (Werken van de Vereniging tot Beoefening van Overijssels Regt en Geschiedenis, 1897) hereafter Stadboek or, accompanied by quoted forms, Stb. (4) W. Kloeke's Zwolsche Sketsies (Haarlem, 1931) is a valuable collection of anecdotes written by a native speaker and representing Zw. of the middle of the 19th century. Two other samples of Zw. published in collections of dialect texts are unfortunately not reliable: J. Winkler, Algemeen Nederduitsch en Friesch Dialecticon 1.376–80 ('s Gravenhage, 1874), and J. A. and L. Leopold, Van de Schelde tot de Weichsel 1.568–72 (Groningen, 1882). Lastly, the modern dialect as recorded by the writer from the speech of J. H. de Groot, aged 69, and of several other informants of various ages.
9a But see now Marchand, Lg. 33.346–54 (1957).
10 The period referred to as Old Low German, although represented by some texts normally called Old Saxon, is so named here simply to suggest the historical connection with Middle Low German. Note also that Zw. is included for historical purposes in the ‘Saxon’ dialects of Middle Low German (Lasch §§11 ff.) rather than in the ‘Franconian’ dialects of Middle Dutch.
11 L. van Wijk, Tijdschrift voor Nederlandse Taal- en Letterkunde 67.161. Haudricourt and Juilland, Essai pour une histoire structurale du phonétisme français (Paris, 1949), give the same explanation for an analogous phenomenon in Latin.
12 Gallée, Altsächsische Grammatik §78; Prokosch, CGG §112.
13 According to H. L. Bezoen, Klank- en Vormleer van het Dialect der Gemeente Enschede §2 (Leiden, 1938), this dialect showed two distinct vowels, a closer and an opener /e/, until rather recently. Other dialects which distinguish the two—or did until recently—are discussed by N. van Wijk, De Umlaut van a in de ripuarisch en salies-frankiese dialecten van België en Holland, Tijdschrift 33.203 ff.
14 This lowering took place throughout the Netherlands area and consequently is found in standard Dutch as well. Compare the vowels of Dutch kind, stuk, mens, and kop with those of German Kind, Stuck, Mensch, and Kopf.
15 Schönfeld's Historische Grammatica van het Nederlands §74 bis, ed. A. van Loey (Zutphen, 1954).
16 For the details see Lasch §217.
17 The time can also be inferred from the dating of the lengthening in Middle Dutch, which probably took place even earlier and had run its course by the time of the earliest written records in the 12th century. The Stadboek, on the other hand, shows many forms with a short vowel where in Middle Dutch we would expect a lengthened vowel, e.g. anne ‘at’, betteren ‘better’, botter ‘butter’, gewest ‘been’, gestollen ‘stolen’, hefft ‘has’. Modern Zw. still shows short vowels in some forms, e.g. /an əwεst εf/ (Dutch aan, geweest, heeft).
18 While fourteen long vowel phonemes are by no means distinguishable with any certainty in the spellings of the Stadboek texts, which like all medieval documents of the area followed rigid scribal traditions with little regard to phonetics, some evidence at least points in this direction. /i·/ bewysen ‘to prove’, bliuen ‘to stay’, /y·/ dueren ‘to last’, /u·/ bruken ‘to use’, buyten ‘outside’; the orthography only rarely gives indication of front-rounded vowels; a distinction of this kind can be assumed here only because these last two phenomes are distinct in the modern dialect, /é·/ breuen ‘letters’, ement yement ‘someone’, /ø·/ genoemet ‘named’, kuelen ‘cool’, /ó·/ boeck ‘book’, to ‘at’, /e·/ alleene ‘alone’, brede ‘wide’, /ø·/ hoerde ‘heard’, oeren ‘their’, voele ‘many’ (for these last two, cf. OHG iro, filu; here the stressed vowel has been lengthened to [e·] and rounded by the following vowel); /o·/ koop ‘purchase’, bodem ‘bottom’, /ε·/ anspreken ‘to address’, rakenen ‘to reckon’ (the second spelling occurs only once and may be an error; the presence of this phoneme behind inevitably noncommittal spellings must be assumed because the modern /ε·/ is distributed along historical lines); /œ·/ possibly in angesproken ‘addressed’, gebroken ‘broken’ because of modern Zw. /əsprœ·kŋ/ etc., /ɔ·/ ouer auer ‘over’, /æ·/ gemaent ‘warned’, makene ‘to make’, /a·/ auent ‘evening’, strate ‘street’. The last-named vowel is written with the same symbols as the preceding but the two cannot have been identical at this time—Gmc. /a·/ in auent, strate later fell together with /ɔ·/, whereas lengthened Gmc. /a/ remained /æ·/. This orthographical identity is probably an influence of the already dominant dialect of Holland, where the two vowels may already have fallen together (Dutch avond, straat, gemaand, maken); the fact that in the Stadboek the letter a in auer could be used for what must have been [ɔ·] indicates that /a·/ in all likelihood already inclined toward this sound.
19 Bezoen, Dialect der Gemeente Enschede; P. Fijn van Draat, Klankleer van den Tongval der stad Deventer, Tijdschrift 42.194 ff.
20 At some time impossible to fix, the fronting of the reflex of Gmc. /u·/ brought about rearrangements in the frequencies of Zw. /y·/ and /y/, though if it is assumed that the umlaut of /u·/ occurred earlier this would not have altered the phonemic pattern. This fronting had probably already taken place by the time the system showed the configuration of §2.4, since it affected the members of the phoneme /u·/ in the modern dialect which go back to Gmc. /u·/ but never those which go back to Gmc. /o·/. This fronting has greatly increased the frequency of Zw. /y/, and replacements of /u/ (and of /u·/ before /r v z g/) are still occurring at the present time under the influence of the expanding fronting area. See Bloomfield's discussion of this, the so-called huis-muis problem (Language 328–31), which is based on G. G. Kloeke's detailed study De Hollandsche expansie in de zestiende en zeventiende eeuw en haar weerspiegeling in de hedendaagsche Nederlandsche dialecten ('s Gravenhage, 1927).
21 See the discussion by H. Penzl, Lg. 25.223–40 (1949).
22 G. G. Kloeke, Ontlening der Hollandse a, meer in 't bizonder in het Zwolse dialect, Verzamelde Opstellen 60–8 (Assen, 1952).
23 The Dutch ui is only secondarily from such obscure words and a French diphthong; primarily it is the result of the fronting and subsequent diphthongization of Gmc /u·/.
24 It can be argued that a similar movement of earlier /é· ø· ó·/ or perhaps diphthongs into the place occupied by /i· y- u·/ was responsible for the diphthongization of at least the old /i·/ and /y·/ to [εj] and [œy] in standard Dutch (e.g. dijk ‘dike’, tuin ‘yard‘).
25 This outside influence on the city undoubtedly increased the latter's cultural and commercial prestige, and probably explains the adoption of the uvular r, which to this day has barely extended beyond the city's limits.
26 Kloeke, op.cit. 61–2.