XVII. Special aspects of Nordic language history I : Typology
202. The typological development of the Nordic languages I:
1.
2.
3.
4. 5.
6.
7.
8.
9.
10.
11.
12.
Phonology
Proto-Nordic Common Old Nordic Old East Nordic (1100-1500) Old West Nordic
Modern Danish Modern Swedish Modern Norwegian Modern Farnese Modern Icelandic Pervasive tendencies Survey
Literature (a selection)
1. Proto-Nordic
1.1. Phonological units
Vowels: The Proto-Germanic vowel system underwent three essential changes from Indo- European: a and i5 merged to i5 (cf. ON brboir vs. Lat. frater 'brother'), a and o merged to a (cf. PN ahtau vs. Lat. octo 'eight'), and ei and i merged to i (cf. ON stiga vs. Gr. steikho 'rise'). However, at the following Common Old Nordic stage
e
1 changed to a (cf. ON mani vs. Lat. mensis 'month'), and thereby a system of five long units was restored (Haugen 1984, 131). The point of departure for a separate Proto-Nordic language was, thus, a vowel sys- tem with only four short and five long vowels (and three diphthongs). During the Proto- Nordic period, a short o developed from u by a-umlaut, and the short and long systems were again parallel. The two latter changes (i\ >a, u > o) occurred in both Proto-Nordic and
West Germanic, a fact that indicates close con- tacts between these two branches of the Ger- manic language community, as opposed to East Germanic, in which the two vowels re- tained their ludo-European quality (Antonsen 1965; Haugen 1984, 139f.). Proto-Nordic is considered a separate language from about 200 A.D.
The Proto-Nordic vowels at about the year 200 can be visualized as a double system:
Long Short
~
. u
--- e (o)
a
Diphthongs: ai, au, eu.
Consonants: The Proto-Nordic consonant phonemes were:
Labial Apico- Velar Glottal alveolar/
palatal Stop p/b [~] t/d [c'i] k/g [y]
Fricative <I>
e
hSibilant s I R W
Nasal m n
Liquid 1
Vibrant r
Glide w J
The voiced palatal sibilant R originated from the Proto-Germanic z by an innovation com- mon to Proto-Nordic and West Germanic, cf.
maiza > maiRa 'more'. The voiced stops b, d, g were probably fricatives in medial and fi- nal positions (except for the position after n and l; Krause 1971, 38).
1.2. Syntagmatic structures
1.2.1. Distribution
Vowels: During the Proto-Nordic period, there was a considerable reduction in the number of vowels in unstressed position: o
202. The typological development of the Nordic languages I: Phonology 1853 preceding m changed to u (ON berum), and
final o, ai, ei, eu changed to u, e, T, iu (cf.
PN worou, dage, bru/JT, magiu). Proto-Nordic and West Germanic shared these vowel changes.
This weakening process of the vowel quality in unstressed position continued with some monophthongizations: ai, au, iu > e, i5,
z ,
then a subsequent shortening:e , z
> e, i, and some mergers: ii > e, i5 > a, e > i and a > u (Haugen 1984, 191). The result was that, by the end of the Proto-Nordic period, unstressed syllables contained only a subset of three vowels: a, i, u. The changes can be illustrated with the following examples:kurnai > kurne > kurne > kurni mageu > magiu > magT > magi sunauR > suni5R > sunar
In stressed position some vowels underwent changes which were dependent on the follow- ing consonants: the diphthong ai was mono- phthongized to ii before h and R, and au to i5 before h, cf. faihioi5 > fahioi5 ( > ON faoa) 'painted' and pauh > poh ( > ON pb) 'though'.
Before ht, the vowels 1, i,
u ,
u were lowered to e, e, i5, o (wihtiR > wehtiR (> ON vettr) 'creature', jJuhti5 > jJi5hti5 (> ON pbtta) 'thought' past). These mergers represented phonotactic restrictions for the phonemes in question.In Proto-Nordic the two geminate glides ii and ww were sharpened into ggj and ggw (cf.
ON byggja 'build, live' and tryggva 'true' (masc.pi.ace.)). This is one of the very few in- novations common to Proto-Nordic and Gothic.
Consonants: After 600, the initial glides were lost (woroa > ON oro, Jiira > ON ar), with the exception of w when preceding a non- rounded vowel or r
+
non-rounded vowel (wi- tan 'know', wripan 'writhe'). In medial posi- tion, j was in general lost after long syllables, and after short syllables when preceding a front vowel (domijan > ON d@ma 'judge' vs.wiljan > wilja 'will'). The consequence of this redistribution was that the glides from then on were mere non-syllabic variants of u and i. Thus, the diphthongal stress shift on ia turned i into j in a rising dipthong: hiartan >
ON hjarta 'heart'.
Geminate spirants became stops: muppan >
ON motti 'moth' and glaoioo > glaooi > ON gladdi 'made happy', and p was assimilated after land n: guljJa > ON gull 'gold',finpan
> ON finna 'find'. Unvoiced fricatives in me-
dial or final position became voiced in voiced
environments. This new distributional pattern represented a phonemic reorganization since the new voiced fricatives merged with the fricative variants of b, d and g in correspond- ing environments. Consequently, voiced frica- tives had an archiphonemic status, and un- voiced fricatives were restricted to initial po- sition and before another unvoiced conson- ant: ON finna [qiin:a] 'find' and ljuft [lju:qit]
adj.neut. 'pleasant' vs. gaf [ga~] 'gave', and pakka [Saka] 'thank' vs. gbp [go:5] 'good'.
Correspondingly, by the end of the Proto- Nordic period h was restricted to stem-initial position.
Thus, a drift towards a more bound (unfree) distributional pattern had started with regard to both vowels and consonants.
1.2.2. Stress and quantity
Unstressed Germanic prefixes were eliminated during the Proto-Nordic period with the effect that all words had the main stress on the stem syllable: ga-sinjJa > ON sinni 'companion'.
During the transition from Proto-Nordic to Common Old Nordic, many consonants were dropped, and some of them were compensated for by lengthening of the preceding vowel.
This was systematic as far as medial and final h, p and nasals were concerned: rehtaR >
rettR 'right', ansuR > tisR (> tiss) 'god'.
However, it caused no principal changes in the distributional possibilities.
1.3. Synchronic phonogical processes Umlaut: The very simple Proto-Nordic vowel system probably disguises extensive allo- phonic splits that were a result of a vowel har- mony process by which the unstressed vowels i, a, u influenced the quality of the preceding stressed vowel. These were the phonetic um- laut processes, which were not phonemicized until the syncope period after 500. Except for the u-umlaut, these processes were partly shared with West Germanic, but not with Gothic. The Proto-Nordic processes had the following phonetic results: i-umlaut fronted the back vowels (u, o, u, o) and the low (unrounded) vowels (ii, a): komiR [k0miZ]
'comes', jungiB_e [jyngiie:] 'younger', miilia [m:e:lia] 'speech', gastiR [gcstiZ] 'guest'; a-um- laut lowered the high vowels (i, u): hurna [horna]; and u-umlaut rounded the preceding vowels (i,
z ,
e, e, a, ii): trigguR [tryg:uZ] 'safe', teguR [t0guZ] 'ten', dagum [d;:igum] 'days'.(;:, will hereafter be rendered as Q except in pho-1854 XVII. Special aspects of Nordic language history I: Typology netic transcription). The diphthongs were ex-
posed to an assimilation parallel to umlaut:
ai, au > :ei, QU, but eu became iu, which dem- onstrates a raising of the first segment.
A similar harmony process was breaking, changing a short e into ia (>Ja) when it pre- ceded a or u (eona > iaon), except when the e followed w, l, r, as in wer]Jan > veroa 'be- come'. When the triggering vowel was u, the outcome was }Q, which can be explained as u-umlaut of ia; cf. erou > iarou ( > ON }Qro) 'earth'. An effect of breaking was the rein- troduction of initial j-: er]Ju > }Qro, see sect.
1.2.1.
Verner's law: In Proto-Germanic, voiceless fricatives became voiced if the Indo-European accent did not immediately precede them, a process causing e. g. stem-final h to become g [y] (i.e. a voiced fricative velar) when the suffix was stressed, cf. past sg. sloh ( > ON sl6) vs.
past pl. slogum 'hit'. West-Germanic par- ticipated in this process, but it is disputed whether there were similar developments in Gothic. When word stress later was fixed to the stem in Germanic, this process was no longer phonological, but rather morpho-pho- nological.
Devoicing: In Proto-Nordic, voiced stops were devoiced when occurring in final posi- tion, cf. bindan (inf.) vs. bant (past) 'tie'. The devoicing rule disappeared during the syncope period, when several more instances of final stops arose, cf. PN kwelda > ON kveld 'even- ing'.
1.4. The relationship between phonology and morphology
Proto-Nordic had a rather agglutinative mor- phology, cf. bot+iJ+an 'mend' (stem+ con- jugational suffix + infinitive suffix) and kun+ing+a+R (stem + derivational suffix + declension suffix + nominative suffix).
Both the starting segment and the final seg- ment of each suffix can be identified, and, thus, the morphology can be characterized as pho- nologically transparent. However, we know little about assimilation processes across the morpheme boundary except for the umlauts.
2. Common Old Nordic
It is practical to describe the period from 700 to 1100 as Common Old Nordic, although, from this still pre-literary period, there are relatively few reliable data which point to how
common the various characteristics actually were. To a large extent it is a matter of recon- struction. Some of the changes leading to the characteristics which differentiate Common Old Nordic from Proto-Nordic are attested in the 6th c., which is an argument for setting the beginning of the new period at about 700 A.D. There was, however, a long transitional period between Proto-Nordic and Old Nordic, which is reasonable when taking into account the vast area in which the language was used.
Some of the Proto-Nordic changes mentioned above may well have taken place somewhere in the Nordic area during the Common Old Nordic period, just as changes mentioned be- low may have started during the Proto-Nordic period.
2.1. Phonological units
Vowels: The transition from Proto-Nordic to Common Old Nordic is characterized by the loss of short unstressed vowels except when followed by a final m, n, r: horna > horn 'horn', dagaR > dagR 'day', katilaR > katilR 'kettle' (but daganR > daga pi.ace. 'days');
long unstressed vowels were shortened: dagoR
> dagaR pl.nom. 'days'). After this syncopa- tion, which occurred in the period 500-700 A.D., the umlaut vowels were phonemicized since the vowels which were dropped were the conditioning factors, and the Common Old Nordic vowel system of about 700 thus had increased its number of units:
y
e
011 6
a
Q[;,:Je
y u
0 0
a Q [;,]
As nasals were deleted under certain condi- tions, nasality was transferred to the preceding vowel, and thus nasal vowels became pho- nemic: ansuR > asR 'god', fimflaR > ff;{l 'monster'. This caused the establishment of a long nasal vowel system parallel to the long oral one, with the effect that the total vowel system comprised 27 phonemic contrasts (cf.
Benediktsson 1972, 128-146). (The feature long is consequently redundant in asR and
Jf;fl).
Consonants: As a consequence of the changes mentioned above, Common Old Nor- dic had 17 units in about 700 A.D.:
202. The typological development of the Nordic languages I: Phonology 1855
Labial Apico- Palato- Glottal alveolar/ velar
palatal
Stop p/b t Id k / g [y]
Fricative f [~] e [ol h
Sibilant s / R W
Nasal m n
Liquid 1
Vibrant r
Glide w j
The apico-palatal sibilant R was during this period assimilated to a preceding n, s, r, l (e.g.
stainaR > steinn 'stone'), and by 1100 other instances of R had merged with r, and thus the system lost one unit. During the Common Old Nordic period, the intervocalic variant [I3]
changed to a labio-dental [v] in most areas, and the glide w merged with it. Considering later changes in the Nordic language, it seems reasonable that l was velarized ('dark') in some Central Scandinavian dialects in non-al- veolar environments (Kock 1883).
2.2. Syntagmatic structures
2.2.1. Distribution
Initial h-: After the Proto-Nordic develop- ments, h was restricted to initial position:
hestr, hnjbsa, hlutr, hringr, hjarta, hvitr 'horse', 'sneeze', 'part', 'ring', 'heart', 'white'. During the Common Old Nordic period, the use of h was restricted even further, viz. to initial po- sition before a vowel: hestr, but njbsa, lutr, ringr,jarta. This new pattern for initial h orig- inated in Denmark in the 9th c. and was es- tablished by 1400 (Haugen 1984, 264f.). Ice- landic retained all these clusters - and still does so in the form of an unvoiced pronunciation ofn, l, r, andj. In Faroese, most of Norwegian and Finland Swedish hv-merged with kv-dur- ing the same period, in most of Danish and Swedish with v-, and Icelandic still has both kv-and hw- [xw].
Intrusion: As a consequence of syncope, Common Old Nordic was characterized by complex consonant clusters, many of which were word-final, e. g. vetr, valskr, blbmstr 'win-
ter', 'Welsh', 'flower'. As early as about 1067-
69, however, an Old Norwegian inscription
contains an intrusive (svarabhakti) vowel pre- ceding the final r, certainly in order to facilitate thepronunciation(Skard 1976, 94). The intru- sive vowel obtained the status of a segment in almost all Nordic dialects and has charac- teristics of a suffix vowel, with regard both to quality (varying geographically between a, u and e) and to its appearance only when no other suffix vowel follows the r: akur - akrar 'field - fields'.
2.2.2. Stress, quantity and pitch accent The characteristic Nordic pitch accent must have developed no later than the Common Old Nordic period, as suffixes surviving syn- cope were included in the basis of the accent, i.e. they were counted when disyllabic words - and not monosyllabic words - received ac- cent 2, e.g. the plural dag+ar 'days' with ac- cent 2. On the other hand, the definite article, which was not suffixed until late in this period (Indreb01951, 88), did not influence the ac- cent: dag+ inn 'the day' (with accent 1) vs.
han+i+ (i)nn 'the cock' (with accent 2 be- cause -i-was a Proto-Nordic suffix).
Danish developed a different pronunci- ation, i.e. the glottalization ("st0d"), which in principle has the same phonological dis- tribution as pitch accent 1. It is a matter of debate whether Danish glottalization arose during the same period or is a later develop- ment (Skautrup 1944, 238 ff.; Liberman 1982).
The pitch accent contrast is still retained in Norwegian and Swedish, but has disappeared from Icelandic, Faroese and most of Finland Swedish.
Quantity: At the Common Old Nordic stage, the possibility of stressed short final vowels disappeared, and thus the vowels of such one-syllable words were lengthened (phonemically the lengthening was redun- dant!): pa > pa 'the' (masc.pl.acc.), su > su 'the' (fem.sg.nom) (Benediktsson 1968).
2.3. Synchronic phonological processes Umlaut: The Proto-Nordic i- and a-umlauts disappeared as phonological processes as a re- sult of syncope, and their effects were trans- ferred to morphology (as stem variants) or the lexicon (new phonemes in the stem). This meant that new unstressed i did not trigger umlaut: landi ( < PN lande) 'land' (dat.sg.).
A-umlaut was more consistently established in West Nordic than in East Nordic, cf. OWN boo, golf 'message', 'floor' vs. Old EN buo,
1856 XVII. Special aspects of Nordic language history I: Typology
gulv. U-umlaut seems to have finished func- tioning in East Nordic before the Common Old Nordic period as there are almost no traces of it. In West Nordic u-urnlaut (i.e.
"younger u-umlaut") still functioned: lQndum 'lands' dat.pl.
Palatalization of velars: At an early stage, the Nordic languages palatalized the velar stops, k and g, when the two consonants pre- ceded a front vowel, as in ON girr, gyss, kenna, k@la, kleti 'desire', 'squirts' (pres.), 'know', 'cool', 'joy'. This process took place before both stressed and unstressed vowels, and the phonetic result appears to have been a palatal glide between the velar and the vowel: /gji:r:, gjy:s:/ etc., which would mean that e. g. the infinitive gjbsa 'squirt' would have the same two initial segments as the present gyss. (Pre- viously kj and gj had been restricted to posi- tions where the palatal glide was not a product of this phonological process, i. e. before non- front vowels). This process is reflected in 12thc. spelling in all the Nordic languages.
2.4. The relationship between phonology and morphology
A-umlaut caused a vowel shift in the stem: PN horna 'horn' (nom.sg.) vs. hurne (dat.sg.). The vowel shift was levelled, and the a-umlauted vowel became generalized in the stem. Thus, a-umlaut had mainly lexical effects in Com- mon Old Nordic. The i- and u-urnlauts, on the other hand, played important roles in the morphological system of Common Old Nor- dic, as the shift in a stem vowel throughout a paradigm was in some cases the only mor- phological signal, e.g. land - lQnd 'land - lands'. In other paradigms, the shift in the stem vowel was accompanied by specific suf- fixes: vellir - valla - VQllum 'fields' (masc.nom.
-gen. -dat.pl.). In general, the stem contained more information in Common Old Nordic than in Proto-Nordic, and the language had moved away from an agglutinative language type, as the stem included or anticipated mor- phological information.
The assimilation of -R (cf. sect. 2.1.) caused a morphological complication as the R, when it was e. g. the masc.nom.sg. suffix, no longer was transparent as a suffix: stblR > stall 'chair' (following a long stressed vowel), ketilR > ketill 'kettle' (following a short un- stressed vowel). However, the first structure, long vowel + long consonant, was not possi- ble unless it contained a morpheme boundary either before or within the long consonant,
and therefore the structure itself contained some morphological information. On the other hand, the structure unstressed vowel + long consonant was not constrained in a par- allel way, and was therefore by no means mor- phologically transparent.
3. Old East Nordic (1100-1500)
3.1. Introduction
The Nordic area has been a dialect area where changes spread in different directions at vary- ing rates. Therefore, a distinction between East Nordic and West Nordic should not be interpreted as more than a labelling of tenden- cies. The first detectable split between East Nordic and West Nordic can be traced back to the 7thc. with the assimilation of nt, nk,
mp > tt, kk, pp (Moberg 1944). This assimi-
lation is reflected all over the Nordic area, as in drekka/drikka < PN drinkan 'drink'; how- ever, the innovation was implemented more consistently in the west, cf. ekkja, brattr, vetr 'widow', 'steep', 'winter' in the west vs. enkja, branter, vinter in the east. As mentioned in sect. 2.3., this was true of a-umlaut as well.
East Nordic contained, on the other hand, more words with breaking (cf.jak < eka 'I').
/-umlaut in the present tense of strong verbs (taka - tekr 'take - takes') cannot be traced in the East Nordic area, which instead has the forms taka - takr. (The dialect of Jutland is, however, an exception, cf. Wessen 1965a, 33).
This may reflect a difference dating back to Proto-Nordic, or it at least demonstrates the first systematic difference between the two branches of Nordic.
An interesting illustration of how language development should not be described as a splitting up into independent entities like a branching tree but rather as a contact phe- nomenon in the continuous Central Scandina- vian dialect area covering parts of both East Nordic and West Nordic (Finland Swedish, northern and central Sweden+ East Norway and Trnndelag), as will be demonstrated in sect. 3.2.-3.5. This has been an area with many common innovations. East Nordic appears to have been rather homogeneous until the 13thc., after which Danish showed rapid in- novations (Haugen 1984, 259).
3.2. Phonological units
Vowels: There are very few traces of u-umlaut in East Nordic. (In Danish, one of the few is
202. The typological development of the Nordic languages I: Phonology 1857
bQrn > b@rn 'children'). Therefore, Q became very marginal and merged with o (but with@
before r) at an early stage (Skautrup 1944, 47).
The vowels e and~ also merged early on. Na- sal and oral vowels probably merged during the 11 th c. Thus, a typical East Nordic vowel system from about 1100 was like the following:
y
e
0u
6
a
i y u
e 0 0
a
The main difference between east and west was monophthongization, which started in Den- mark in about 900 and spread during the next century to Swedish and some neighbouring
Norwegian dialects in 0sterdalen and
Trnndelag by the end of the Middle Ages. East Nordic monophthongization eradicated the three Germanic diphthongs from the vowel in- ventory: ei > e and au, @y > @. Thus, the diphthongs merged with previously existing monophthongs. In the dialects of Gotland, Finland Swedish and North Swedish there was no monophthongization.
By the end of the Middle Ages, spellings indicate the development of secondary diph- thongs from old long vowels - e.g.
y
> @yin Jutland and 6 > oi on Gotland (Haugen 1984, 322 f.). In about 1400,a
changed to[;,:] in East Nordic, except in Gotland.Consonants:
Labial Apico- Retro- Palato- Glottal alveo- flex velar
lar
Stop p/b t / d k/g
Fricative, f/v s J h
sibilant
Nasal m n
Liquid 1
Vibrant, r
flap (
p
merged with t by 1400 in East Nordic, first in Denmark, anda
was dropped in most dia- lects, except for some minor areas where it merged with d.Thick l, a retroflex flap [i:], seems to have emerged in the Central Scandinavian area
during the 12th c. The sound spread to eastern Norway, Trnndelag, the southern part of northern Norway, and northern and central Sweden. It developed partly from a dark velar
l (a variant of l used postvocalically) and part-
ly from an assimilation product of the cluster ro: gala > [gai:a] 'crow', garo > [gai:] 'farm' (Storm 1908, 106ff.). The change l > rwas constrained from the position after front high vowels, and thick l does not appear in word- initial position. (In Ovan-Siljan (Dalarna, Sweden), however, thick l occurs even in initial position). At an early stage, this sound was a mere variant of l; however, other later changes led to it becoming a phoneme, e. g. as
o
was dropped and saoill 'saddle' developed to [sal] with an alveolar [l], in contrast to [sai:]'sale'.
3.3. Syntagmatic structures 3.3.1. Distribution
Weakening: The common Nordic vowels in suffixes were -i, -a, -u. In Danish, these were reduced during the 12thc. to one: -E. This weakening (reduction), which was a continu- ation of the Proto-Nordic tendency to differ- entiate the distribution of vowels in stressed and unstressed position, had of course tremen- dous effects for the morphological system as many contrasting suffixes now merged, e. g.
stbri haninn 'the big cock' nom.sg. vs. stbra hanann acc.sg. > stbrE ha~n, and stbrir solar 'big soles' masc.pl. vs. storar solir 'big suns' fem.pl. > stbrcer sblcer. This innovation had later parallels in some areas in Norway and Sweden.
Progressive palatal umlaut: Prevocalicjper- haps spread its palatal feature to the vowel during the 11 the., as in hjarta > hjerta. After 1200, some instances are also found in Nor- wegian manuscripts.
Palatalization of alveolars: Before 1300, long alveolars were palatalized in Danish:
mann > [maJl] 'man', all> [a,(] 'all'.
Assimilations: East Nordic has been the centre of consonant assimilations. Some of them, e.g. mb, ng, nd, ld > [m:, IJ:, n:, l:] orig- inated in Jutland by 1300 and spread later to southern and central Sweden, eastern Nor- way, Trnndelag and northern Norway: lamb, tung, sand, kveld > [lam:, tu!]:, san:, kvd:]
'lamb', 'heavy', 'sand', 'evening'. Gotland, Dalarna and the Finland Swedish area are relic areas in this respect, and even N orrbotten
1858 XVII. Special aspects of Nordic language history I: Typology and a part of central Sweden with regard to
mb, nd, ng.
Lenition: Postvocalic voiceless stops be- came voiced in about 1200 in Danish (includ- ing Scania and Halland): gata > gak. This innovation spread later to the southern coast of Norway. In Danish, these voiced stops weakened further to fricatives during the next century, cf. gak > gaoce.
Lowering: Towards the end of the Middle Ages there was a tendency for short vowels to be lowered in certain positions in most East Nordic dialects. The pattern differed from dia- lect to dialect and depended to a large extent on the consonantal environment. The lower- ing was more widespread in short syllables than in long ones, cf. vika > Sw. vecka 'week' vs. fisk 'fish'. The lowered vowel usually merged with the vowel one step down in the vowel system.
3.3.2. Stress and quantity
From the 13thc., the Scandinavian countries were under heavy North German (Hanseatic) influence, which led to an immense influx of German loanwords, many of which contained unstressed German prefixes (see art. 135).
Thus, stress again was not consistently on the first syllable of a word: bedraga [be1dra:ga]
'deceive'.
Vowel balance: An innovation in the central Scandinavian area (Finland Swedish, North Swedish, East Norwegian and Trnndelag) was vowel balance, which has been interpreted as a combination of a low tone in accent 2 and short syllables (Torp 1983). Through this the group of disyllabic words with a short first syl- lable (hana, viku 'cock', 'week') formed a dis- tinct group with a characteristic pronunci- ation. This group was the basis for other pho- nological changes as there appears to have been an equilibrium of stress between the first and the second syllable. Therefore, the suffix vowel was protected from weakening in these words in Central Scandinavian.
The great quantity shift: In Proto-Nordic and Common Old Nordic, quantity was in- dependent of stress, and it was inherent in each segment as it was not affected by the quantity of the following consonant struc- ture, cf. meta - metti 'meet - met'. The first change in this free distribution of quantity arose in unstressed position, where the con- trast long - short disappeared, e. g. the defi- nite suffix -inn in the masc.nom. vs. -in in the fem.nom. (s6linn 'the sole' vs. s6lin 'the
sun') or the geminate -rr in the masc.nom.
vs. -r in the masc.ace. (hamarr vs. hamar 'hammer').
Quantity shift in stressed syllables started in Denmark possibly during the 12thc.
(Skautrup 1944, 237), and spread to all the Scandinavian languages by the end of the Middle Ages. The first step was that long vowels preceding consonant clusters or a long consonant were shortened (natl > natt).
Quantity shift in words with a short stem syl- lable started in Denmark at about 1250 and was carried through in several different ways in the Nordic dialect area. The Danish change differentiated between monosyllabic and di- syllabic words: in disyllabic words, the stem vowel was lengthened: (gata >) gad;e >
ga:k, (taka >) tagce > ta:gce 'street', 'take'.
In monosyllabic words, the short vowels re- mained unchanged, e. g. had 'hate'. As the cor- responding disyllabic vb. had a long vowel (ha:de), this shift could be regarded as a syn- chronic process in which the underlying short stem vowel a was lengthened in open syllables. Short vowels before long consonants or a cluster did in principle remain short. How- ever, at some later stage in Danish, vowels pre- ceding st, sk, bl were lengthened (cf. epli >
ce:ble).
3.4. Synchronic phonological processes Weakening of suffix vowels: Within the Cen- tral Scandinavian dialect area the weakening of suffix vowels depended on a preceding long root syllable. Accordingly, the reduction was a synchronic phonological process by which the old suffixes -a/-i/-u and -ar/-ir/-ur showed up as reduced variants in this particular en- vironment: bakkce, visce, bakkcer, viscer < ON bakki/a 'hill' masc.sg., visa/u fem.sg. 'song', bakkar masc.pl.nom. 'hills', visur 'songs' fem.pl.nom./acc. Corresponding words of the same declensions with a short root syllable re- tained the original suffix vowel: hana, viku, hanar, vikur 'cock', 'week', 'cocks', 'weeks'. In principle this process took place in all word categories; however, since some declensions and conjugations were dominated by words of a specific quantity, the activation of the pro- cess differed quite a lot. It is most typical of the infinitive, weak masc. and weak fem., cf.
leva - visce 'live', 'show', bakkce - hana 'hill', 'cock', viku - visce 'week', 'song'. This result of vowel balance is still typical of dialects in eastern Norway, Trnndelag and northern Sweden.
202. The typological development of the Nordic languages I: Phonology 1859 Vowel harmony: In Old Nordic, there was
assimilation of a vowel feature from the stressed to the unstressed vowel: mid-high root vowels triggered the suffix vowels e and o, while high vowels took the corresponding suffix vowels i and u: sender, erom, viku 'sends', 'are', 'week'. The low vowels demonstrated a more complex pattern: re triggered i and o/u, whereas a and Q were followed by e and u.
This harmony process arose in Scania about 1100 and spread to Swedish (with the excep- tion of Gotlandic) and Norwegian (Hessel- man 1948-53, 280). After 1300, the weakening process affecting suffix vowels, as described above, became dominant and vowel harmony disappeared before the end of the Middle Ages (Brnndum-Nielsen 1927; Ragland 1978).
Retroflexion: The emergence of the retro- flex flap (thick l) in the Central Scandinavian area gave rise to an extensive phonological as- similation process in which the retroflex ar- ticulation spread to the following alveolar stops and continuants t, d, s, n, !, r, and the flap was dropped. This rightward process works across both morpheme boundaries (marked
+
below) and word boundaries and is, thus, rather frequent and characteristic of the language: gul+
t, kj@l+
d, tel sakte, stol+
n, ma!+ l@s, gul+
rot [gli:t, i;:0q, ti:'~akte, stu:ri., m;:i:l0s, gli:-tut] 'yellow', 'cold', 'count slowly', 'the chair', 'speechless', 'carrot'.For some phonetic reason which has not yet been satisfactorily explained, the correspond- ing clusters of r
+
alveolar stops or continu- ants merged with the thick l clusters, thus e. g.gult 'yellow' and surt 'sour' rhyme. This co- alescence caused hypercorrect spellings that provide us with historical data about the pro- cess, which appears to have been in operation during the 13thc., as in e.g. altio in stead of artio 'requiem' (Seip 1955, 177).
On the periphery of the Swedish and Nor- wegian retroflex area there is still a difference in pronunciation between the corresponding flap clusters and r clusters, as the flap clusters are pronounced as retroflexes, and the r clus- ters as post-alveolars: f@lgt [fo:t] part. 'fol- lowed' vs. f@rt [fo:!] part. 'carried'.
3.5. The relationship between phonology and morphology
The process of suffix weakening in the Central Scandinavian area did not lead to a mor- phological restructuring as long as the weakening was a phonological process: [bi:ta]
> [bi:te] 'bite', caused by the long root syl-
lable, vs. vera 'be' which was unaffected be- cause of the short root vowel. However, the later quantity shift (e. g. [vera] > [ve:ra]) veiled the conditioning quantity distinction, with the effect that the reflexes of the process ([bi:te]
vs. [ve:ra]) caused grammatical reorganization as there was an increase in inflectional classes.
This morphological complexity was then fol- lowed by morphological levelling with many regional variations.
4. Old West Nordic 4.1. Phonological units
Vowels: As the Trnndelag area in Norway re- tained theshortreuntilabout 1200, this dialect had totally parallel short and long subsystems till then (cf. Benediktsson 1964).
The u-umlaut vowel Q remained in Icelandic until the 13th c., perhaps a century longer in Norwegian, and until the 16th c. in Farnese.
Therefore, the monophthong system in West Nordic had one more unit than East Nordic, and in about 1100 it was as given in the fol- lowing:
y
11 y u1 - - - - ; - - - 1 -- - - - -- > - - - - ; 1 - - - - 1 -- -----
e
0 6 e 0 0a
Q a QThe main difference was the existence of the three additional diphthongs QU, ei, ey (@y). Ac- cording to the First Grammarian (see Be- nediktsson 1972) the oral : nasal contrast still existed in the middle of the 12th c. for long vowels, but it certainly disappeared soon after.
In literary times, this nasal distinction has been attested only in Icelandic in the West Nordic area. (It still exists in Dalarna in Sweden).
Because of lack of congruence between the short and the long subsystems, there was a po- tential for the two subsystems to develop dif- ferently. In Icelandic, the short non-low vowels were generally lowered phonetically: i [1], e [i:], y [Y], @ [re], u [u], o [;:,]. Benediktsson (1968) has interpreted written data from the 12thc. as an indication of this with regard to i and u. Both the phonemic and the phonetic situations in today's Icelandic support this in- terpretation, and West Norwegian dialects seem to have undergone the same develop- ment. It is characteristic of this West Nordic
1860 XVII. Special aspects of Nordic language history I: Typology
lowering that it was not - or was only to a very little extent - regulated by the consonan- tal environment.
Before the end of the Middle Ages, the West Nordic vowel system started developing in dif- ferent directions in Icelandic, Faroese and Norwegian. The first and most radical changes occurred in Icelandic during the 13 th and 14 th centuries with the transitions Q > @,
e
>je,
@ > k and
a
> Q, which resulted in the fol-lowing vowel system:
y u
i y u6 e 0 0
Q a
The diphthongs were QU, ei, ey, and the num- ber increased during the late Middle Ages with the transition of Q to au and 6 to ou.
According to Benediktsson (1970), the fea- ture roundness was reduced in importance through this development - especially among the long vowels - and the subsequent delabial- ization tendency of front rounded vowels was a "natural" effect.
Data from Faroese exhibit these changes by 1300:
e
> k (ser > skr 'sees' pres.), and Q >6 (bQru > bbru 'carried'). Even in Faroese de- labialization occurred, but it was restricted to shorty > i. In the long vowel system, the de- velopment was the opposite: i >
y.
The two last mergers very probably took place rather late in this period. During the 14thc., QU was delabialized and fronted to ez, and Faroese emerged from the Middle Ages with this vowel system:y u
u0 6 e 0 0
------ --- t---
a
a QDiphthongs: ei, @y, ei.
The Norwegian vowel system seems to have been more conservative than the Icelandic and Faroese systems. During the 13thc., the co- alescence
a
> Q became apparent, and during the last two centuries of the Middle Ages, Qmerged with o. A "normal" Norwegian vowel system in about 1500 is as below:
y u
i y ue
0 6 e 0 0Q a
Consonants: No changes seem to have occur- red in the consonant inventory during the first centuries of the Old West Nordic period, ex- cept in the dialects of eastern Norway and of Trnndelag which developed the retroflex flap.
However, by the end of the Middle Ages sev- eral changes arose in Norwegian: the apico- alveolar fricatives disappeared from the inven- tory through the loss of
a
(raoa > raa 'ad- vise'), and p merged with t: pak > tak 'roof.The labial fricatives were transformed to la- bio-dentals. Thus, the Norwegian consonant system was very similar to the Old East Nordic system given in sect. 3.2. The palatalized velars [ki, gi] (cf. sect. 2.3.) seem to have merged with the clusters kj and gj at various points of time (about 1200 in Norwegian, about 1400 in Ice- landic and about 1500 in Faroese), and the mergers in Norwegian and Faroese have been changed into the palatal affricates c]: and :J], i.e. to a single segment.
In Faroese p merged with tor h during the 14thc. (cf. pak > tak 'roof, petta > hetta 'this'). Otherwise the Faroese consonant in- ventory was, by the end of the Middle Ages, like the Icelandic one, which showed very few changes from Common Old Nordic:
Labial Apico- Palato- Glottal alveolar/ velar
palatal
Stop p /b t / d k/g[g,y]
Fricative f j V
e [e,
5J j hSibilant s
Nasal m n
Liquid 1
Vibrant r
4.2. Syntagmatic structures
4.2.1. Distribution
Elimination of final consonants: Final -n and -t after unstressed vowels have been elimin- ated in Norwegian and most Swedish dialects.
202. The typological development of the Nordic languages I: Phonology 1861
The -t (as in huset) has been retained in south- ern Sweden, but the -n only in Scania and the adjacent area. Similarly, -m was dropped in most of the Norwegian area. Both -n and -m nasalized the preceding vowel, which caused both a lowering and a rounding, with geo- graphically very varying results, e. g. sblin >
sola, sole, sofa::, solo, sola.
Medial
o:
PNo
after n or l merged with d during the 13th and 14th centuries. This in- novation (PN talioe > taloi > taldi 'counted', PN wanioe > vanoi > vandi, 'accustomed') restricted the distribution ofo
to post-vocalic position and to the cluster ro.Palatalization of alveolars: In northern Norway, Trnndelag and northern parts of western and eastern Norway, long n and/ were palatalized (mann > [maJ1] 'man', vill > [vi,{J 'wild'); in most of the dialects this also affected the assimilated clusters nd > nn and Id > ll (sand > [saJ1] 'sand', kveld > [kve,{J 'even- ing'). Moreover, in the major part of the area the palatalization included long d and t (kladd
> [klaJ] 'draft', kvitt > [kvic] adj.neut.
'white'). This innovation is reflected in 14 th c.
spellings (Seip 1955, 266). The new palatal consonants never occur in initial position, and as their distributional pattern remained stable for a long time, they represent variants of the alveolars in much of the area.
Segmentations/differentiations: During the 13th and the 14th centuries, some continuant clusters and the geminates showed parallel de- velopments in Icelandic, Farnese and south- west Norwegian: the changes rn and nn > dn,
rl and ll > di arose in alle three languages,
and.fi,fn > bi, bn in Icelandic and southwest Norwegian. Geographically the nn and ll in- novations were almost complementary to the palatalization of the same consonants and the changes ll > di and ll > ,{ may, diachronically, have had the same function of transforming a quantity distinction into a quality distinc- tion. The two new clusters (dn and di) did not exist previously. (Cf. art. 207).
4.2.2. Stress
Both Farnese and Norwegian followed the Scandinavian pattern in accepting German loans with an unstressed prefix. Icelandic used the traditional first syllable stress even in such loans:forkelast [1forcelast] 'catch a cold'. This development has to some extent been realized in Central Norwegian dialects, as well.
Quantity: At the first stage of the great quantity shift, long vowels preceding a long
consonant or a consonant cluster were shortened: dbttir; heilsa, sbkn 'daughter', 'greet', 'parish'. At the next stage, short syl- lables were affected. Three patterns prevailed in the quantity changes in Norwegian: East Norwegian followed a Central Nordic pat- tern: vit > [vet:], vita > [ve,ta]. Trnndelag and northern Norway demonstrated a North Nor- dic pattern in which both monosyllabic and disyllabic words with a short syllable lengthened the postvocalic consonant: vit >
[vet:], vita > [vet:a]. In western Norway, a West Nordic pattern was implemented in which the vowel in both monosyllabics and disyllabics was lengthened: vit > [ve:t], vita
> [ve:ta].
4.3. Synchronic phonological processes Palatalization of velars: During the 14thc., some East Norwegian dialects restricted the palatalization process of velars to the position before high front vowels, since the change a
> a:: before palatal consonants did not trigger
palatalization: (kart >) [kak] > [bek] 'man' and [kor] > [k0r] 'coal' (Rindal 1981, 64ff.).
Words with palatalization from the previous process retained (and lexicalized) the pala- talization, as kj~ta, kj@fa 'joy', 'choke'. Dia- lects with the weakening of unstressed -a and -u to -e did not palatalize the velar preceding this new front vowel, a fact which illustrates a restriction on the process in unstressed po- sitions. A previously palatalized k/g and the clusters kj and gj obviously had the same pro- nunciation at that time, as indicated by spell- ings in about 1200 (Indreb0 1951, 230). Dur- ing the High Middle Ages, suffix vowels in all Norwegian dialects were exposed to some changes, either weakening (u, a > e) or na- salization
+
lowering (in > -i, -e, -a, -o), but palatalization remained in the same items as before. Thus the previous phonological pro- cess changed its grammatical status to a mor- phophonological process, cf. declensions like vik - viqo 'bay - the bay'.4.4. The relationship between phonology and morphology
Neither in Proto-Nordic nor in Old Nordic was the quantity of a root vowel affected by the morphological structure of the word;
therefore, e.g., the derived nounfbr+n 'sac- rifice' maintained its long vowel from the root for- (related to the vb. fiJra 'lead') in spite of the cluster r
+
the derivational suffix -n. The1862 XVII. Special aspects of Nordic language history I: Typology same applied to a root
+
an inflectional suffix:meta - m@t
+
ti 'meet - met'. Thus, the mor- phological structure of a word was transpar- ent (Sand0y 1995). The great quantity shift, however, caused vowel shortening and re- moved this derivational transparency. There- fore, several old derivational suffixes became unproductive, cf. -n, -d etc. In words with an inflectional suffix, the long vowel was still kept for some generations, e. g. pot+ ti 'seemed'.5. Modern Danish 5.1. Phonological units
Vowels: Modern Danish has been exposed to a long series of sound changes, especially with regard to vowels that have split into different variants according to specific consonantal en- vironments. Later these variants have often become contrasts because of changes in the consonants. The dominant standard language has two main geographic varieties today with their respective centres in Jutland and Copen- hagen, and they demonstrate considerable dif- ferences in their vowel systems (Braunmuller 1998, 70-72; Haberland 1994, 318ff.). The Jutland system contains ten vowel contrasts - both short and long:
i y u
e 0 0
£ ce ;:,
a
The vowel system of the Copenhagen standard comprises thirteen short contrasts and twelve long:
i: y: u: y u
e: 0: o: e 0 0
z: ce: ;:,: £ ce ;:,
- ---
-
a:: o: a A
o: 0 0
Because postvocalic v and g have been vocal- ized, Danish has developed a large set of diph-
thongs: three with i as the final component (ui, ;:,i, ai) and seven with u as the final com- ponent (au, ;:,u, ceu, @u, yu, eu, &u, iu), e.g. in saglig, vogn, st@vle, d@vstum, syv, rev,jcevn, liv- lig 'matter-of-fact', 'carriage', 'boots', 'deaf- and-dumb', 'seven', 'fox', 'even', 'vivacious' (Braunmuller 1998, 73). The very complex vowel system is unstable and new innovations, especially diphthongs, still seem to be devel- oping (Brink/Lund 1975; Kristensen/Brink 1986, 39).
Consonants: The system of consonants dis- plays no considerable geographic difference in Danish:
Labial Apico- Palatal Velar/ Glottal
alveo- uvular
Jar
Stop p/b t Id k/g
Fricative f/v ~ j h
Sibilant s If
Nasal m n I)
Liquid I
Historically, Danish
o
is a product of inter- vocalic d and final -t in suffixes, cf. ga:dce >ga:oe 'street', ta:get > ta:ged > [1thz:y;;i5] /
[1thz:j;;i~ 'the roof. The pronunciation of
6
is today more like a semivowel or a sonorant (Haberland 1994, 320). Uvular K was intro- duced in the Copenhagen dialect in the latter half of the 18thc., and now represents the onlyK allophone in Danish. In Danish dialects, palatal n and / are still used.
5.2. Syntagmatic structures 5.2.1. Distribution
In contrast to the numerous vowels in stressed position, there is only one suffix vowel in Dan- ish: e [;;i] (varying with [B], see sect. 5.3.).
There are quite a few distributional con- straints for the consonants: hand the unvoiced stops occur only in initial position; therefore, after a short vowel the contrasts p-b and k-g are neutralized in voiced variants: lappen-lab- ben and rykke-rygge (Braunmuller 1998, 69).
However, in Jutland the unvoiced stops occur even in final position. In unstressed position the final -t is both voiced and fricativized in the Copenhagen variant, but only voiced in
202. The typological development of the Nordic languages I: Phonology 1863 the Jutland variant: kastet [kast:;iru and
[kast:;id].
Apocope of a final vowel is widespr~~d in Jutland as well as in northern Sweden, Oland and no;thern Norway. It is most general in infinitives: ON syngja > [s0IJ:] 'sing'.
5.2.2. Stress, quantity and pitch accent Glottalization: The Danish glottalization ("st0d") corresponding to accent 1 in Swedish and Norwegian occurs in words with a stressed long vowel or a short vowel
+
voiced postvocalic consonant. There are some geo- o-raphic variations in the distribution of the;t0d, and, moreover, this peculiar Danish pro- nunciation is not found in all the southern dia- lects.
5.3. Synchronic phonological processes Postvocalic ff: Post-vocalic ff is vocalized to [B]: dor [dce:B] 'door' (Braunmiiller 1998, 68).
On the whole, ff has triggered many innova- tions in the neighbouring vowels; however, not all of them are regular enough to be con- sidered phonological processes. (This proce~s has a parallel in the dialects of Aust-Agder m Norway and a part of southern Sweden).
Palatalization of velars: Palatalization be- fore stressed vowels disappeared in Zealand during the 19thc.: kjobe > kobe 'buy', how- ever, it still operates in some dialects. Bef?re unstressed vowels, there is no palatalizat10n in Danish.
5.4. The relationship between phonology and morphology
The long vowel of a verb root is shortened when it precedes a final root consonant fol- lowed by the past tense suffix -te [d:;i]: kobe - kobte [k0:b:;i - k0bd:;i] 'buy-bought', but not before a continuant +te: b!fes+te [blE:sd:;i].
However, by extending the use of -ede as the past suffix, several clusters are avoided (cf.
levede 'lived'), and thereby phonology and morphology are no longer so extensively in- terwoven.
6. Modern Swedish 6.1. Phonological units
Vowels: The normal Swedish vowel system consists of nine units (both long and short):
y u
1 - - - 1 - - - + - - - -1 - - - -
e 0 0
a
The above symbols indicate long pronunci- ation of the vowels. When short they have a lower (more lax) articulation.
Diphthongs (&i, ai, oi, oi) exist almost only in loanwords, and they can alternatively be analysed as &j etc. Diphthongs are, therefore, very marginal in the phonology of Standard Swedish; however, high long vowels, m par- ticular tend to have an offglide. Many Swedi;h dialects, on the other hand, have ex- tensive diphthongization: e. g. in South Swedish, ON
a
has changed to au and a pr~- viouse (
< ONe
and ei) to ai. In the Gotland1c dialect, most of the old long vowels have diph- thongized: i,y, u , e ,
6 > ::ei, oi, ::eu, ej, ou, and these are added to the inventory of the three Old Nordic diphthongs. Finland Swedish dia- lects, too, have extensive diphthongization:e, o ,
6,a
> ie, yo, ou, uo (Haugen 1984, 323).Raising: A development most typical of the Central Scandinavian area is the raising of old long back vowels in a kind of chain-shift: Q
> [o:], 6 > [u:],
u
> [t1:]. As demonstrated, this movement involves the fronting ofu.
Whether or not the movement is a drag-chain trio-o-ered by this fronting or a push-chain in-
= .
ditiated by Q is hard to prove. Wessen assume that this innovation started in about 1400 in Swedish (1965b, 76). East Norwegian par- ticipated in this development, and today most Swedish and Norwegian dialects have also done so.
Consonants:
Labial Apico-Retro-Palatal Velar Glottal alveo- flex
Jar
Stop p/b t / d k/g
Frica- f/v s j I r;
s
htive, sibi-
!ant
Nasal m n lJ
Li- I
quid
Vi- r (
brant
1864 XVII. Special aspects of Nordic language history I: Typology In Swedish, as in East Norwegian, the retro-
flex fricative represents a historical assimila- tion of several consonant clusters: skj, stj, [S,
rs > ~'
J,
and the unvoiced palatal fricative f is a merger of kj and tj. Finland Swedish has the affricate c] instead of f. This standard sys- tem exhibits some geographic variations: uvu- lar If has replaced r in South Swedish during the last two centuries. In a transitional area in Gotaland, both If and r are used as variants in complementary distribution.6.2. Syntagmatic structures 6.2.1. Distribution
Suffix vowels: Five vowels can be utilized in Swedish suffixes: a, o, u, e, i: flicka - flickor, varuhandel, Jotter, funnit 'girl - girls', 'com- modity trade', 'feet', 'found'. Though,inmany spoken varieties there are only three, i. e. a, e, o, in some even o has been eliminated. Only dialects in northern Sweden, in the area where vowel balance operates, still keep all four.
Metaphony: In the heartland of the vowel balance area, a combined regressive and pro- gressive vowel harmony appeared - perhaps during the 15thc. (Grntvedt 1931). With metaphony one or more features from the suf- fix vowel spread to the stressed vowel in words with vowel balance (viku > vuku 'week', lifa
> lava 'live'), but in some dialects the stressed
vowel influenced the suffix vowel as well ( cf.
viku vs. hoso < hosu 'stocking'). This vowel levelling - or feature levelling - between the stressed and the unstressed vowel is most char- acteristic of the inner dialects of Trnndelag, eastern Norway, Dalarna, Harjedalen and Jamtland.
6.2.2. Stress and quantity
The great quantity shift reached Sweden be- fore 1500. Three main tendencies of shift can be discerned: (1) a North Nordic tendency in which consonants were lengthened in both short monosyllabic and disyllabic words: vit
> [vet:] 'sense', vika > [vek:a] 'week'; (2) a
West Nordic pattern, found even in southern Sweden, in which vowels were generally lengthened; and (3) a Central Nordic ten- dency, in which (unvoiced) consonants in short monosyllabic words were lengthened, whereas the vowels were lengthened in disyl- labic words with a short syllable: vit > [vet:], vita > [ve:ta] 'know'.
A vowel preceding the consonant clusters of stop
+
continuant has to some extent been lengthened, as inflagna, mogla 'peel', 'mould'. There are still relic dialects that maintain the old quantity system. Short syllables are still a characteristic of many dialects in e. g.Dalarna and northern Sweden. Some very conservative dialects of Dalarna and Fin- land Swedish even maintain all Old Nordic quantity patterns, i.e. short, long and over- long, as illustrated in [b;:,b, by:tum, byt:um, by:t:um] 'bake', 'we change', 'the pails' (dat.pl.), 'we changed' (Nystrom 2000). Most of the Finland Swedish area has lost the pitch accent.
6.3. Synchronic phonological processes The dialects of South Swedish and most of Finland Swedish have no retroflexion, al- though Standard Finland Swedish does. Pala- talization of velars before unstressed vowels has disappeared south ofVarmland and Upp- land, but still remains in dialects north of this border and is still a characteristic feature of most of the Finland Swedish dialects. Before stressed vowels, palatalization is no longer a productive process; however, the lexicalized results of historical palatalization still form a distinct pattern.
Devoicing: In the dialects of northern Sweden there is a devoicing process similar to the North Norwegian pattern, see sect. 7.3.
6.4. The relationship between phonology and morphology
To a large extent, word roots keep the long vowel in the various inflected forms even if the vowel precedes a cluster: [sa:del - sa:dlar, se:gel- se:glen, vi::va-vs:ft, fi:n-fi:nt] 'saddle - saddles', 'sail - the sails', 'weave - woven', 'fine - fine' (neut.). The long vowel indicates in such instances that there is a morpheme boundary before or within the following clus- ter. When there is a cluster of two stops, the preceding vowel is normally short in great parts of Sweden, irrespective of any mor- pheme boundary: drakt, kop
+
te 'costume','bought' (past), but lop+ te [10:pt;;i] 'ran'
(past).
202. The typological development of the Nordic languages I: Phonology 1865 7. Modern Norwegian
7.1. Phonological units
Vowels: The most widespread Norwegian vowel system consists of nine monophthongs and four diphthongs:
y u
e 0 0
a
Diphthongs: CH;t, @y, ei, ai. The ai is used in relatively new loanwords, cf. kai 'wharf.
The distinction between short e and short ce is marginal; in southwestern dialects es- pecially they tend to merge even in long con- texts. However, there are dialects with ample vowel contrasts. In Sunnm0re (western Nor- way), we encounter this "medieval" system, i. e. the dialect still keeps most of the Old Nor- die vowel contrasts but now as quality con- trasts:
y u
>----+-- ----!--- ---
y
e 0 0
> - - - - + - - - - I - ---- - -I - - - -
a
Diphthongs: ei, ou, ai, CH;t, @y, cei.
The Setesdalen dialect has also retained all old contrasts but with a more extensive diph- thongization of old long vowels: i,
y, u, e, a,
6 > ei, uy, eu, e;, @1, o".
Consonants: East Norwegian has in prin- ciple the "Swedish" consonant inventory, cf.
sect. 6.1. The retroflex[~] originated from the assimilation of
r
or r+
s, and was therefore always in postvocalic position. Later it merged with the assimilation result of sj and skj which were frequent as initial clusters, and ~ had from then no occurrence restrictions: sj@, vers [~0, vi:~] 'sea', 'stanza'. West Norwegian lacksthe retroflexes, but on the other hand, many
of these dialects still have both voiced and un- voiced affricates. In most of the area [IJ] is a variant of /n/. Southwest Norwegian has dur- ing the last generations exchanged the alveolar
r with the uvular K. A geographic pattern seems to have emerged where uvular 1, has
gained the upper hand in dialects that have no retroflexion (Torp 1997, 35 ff.).
Labial Apico-Palatal Velar Glottal alveo-
lar
Stop p/b t Id k/g Affricat C9 / J}
Fricative, f/v s j h
sibilant
Nasal m n
Vibrant
"
Liquid 1
North Norwegian dialects have a large con- sonant inventory as it comprises the palatals ,,{, {,Jl, c,J; thus it may contain 23 consonants, and would encompass another 5 post-al- veolars and 6 retroflexes if we did not analyse these sounds as products of phonological rules (Sand0y 1987, 29f.).
7.2. Syntagmatic structures 7.2.1. Distribution
Weakening: Southwest Norwegian has main- tained unstressed final -a, cf. the infinitives finna, vera 'find', 'be' and the weak feminine veka 'week'. (The -a in weak feminines is re- tained even in the northern part of northern Norway). In two "transitional" areas adjacent to the dialect areas with vowel balance, i.e. in the northern part of western Norway and in Aust-Agder, final -a has been weakened to -e irrespective of the quantity of the preceding syllable, cf.finne and vere 'find', 'be'. With the exception of a few minor aeras, all West Nor- wegian dialects have weakened even un- stressed -u to -e. In the dialects with vowel balance, neither -a nor -u have been weakened in words where an old short syllable preceded.
These dialects have, therefore, 3 or 4 suffix vowels.
Apocope: In Trnndelag, which belongs to the vowel balance area, final suffix vowels have been apocopated after a long syllable,
cf. !if a > lcevva 'live' vs. kasta > kast 'throw',
and in the southern part of northern Norway, which is outside the vowel balance area, final suffix vowels have been apocopated in general:
lifa > levv and kasta > kast.