V

V. Plotkin

Consonantal duration in the Eastern Scandinavian phonological systems

First published as:

В. Я. Плоткин. Консонантная протяжённость в шведской фонологической системе.

Скандинавский сборник, 24, p. 17-22. Таллинн, 1979.

With summary in Swedish: Konsonantlängd i det svenska fonologiska systemet.

Skrifter om Skandinavien, 24, p. 22. Tallinn: Eesti raamat, 1979.

Determination of the phonological nature of sound duration in the Scandinavian languages faces peculiar difficulties and has long been debated. The difficulties stem from the close interdependence between the phonic durations of a vowel and the following consonant – if either of the two adjacent sounds is long, the other is short. In phonetic terms this is a clear case of mutual dissimilation in duration between adjacent units, described in Scandinavian phonetic tradition as quantitative balance. But the phonological aspect of the problem remains highly contentious, as the distinctive feature of duration has to be assigned to a certain phonemic unit – either the vowel or the consonant is a long or short phoneme, while the phonic duration of the other phoneme varies allophonically depending on the phonemic duration of its neighbour. It is commonly presumed, however, that there is no conclusive proof for either vocalic or consonantal duration.

Numerous ideas have been put forward for a way out of this quandary [2, p. 40; 3, p. 134-36; 6, p. 35-44]. Besides arguments for assigning phonemic duration to vowels or to consonants, it has been suggested that duration is a phonologically relevant feature standing apart as neither vocalic nor consonantal, or a feature characterizing the entire syllable and thus prosodic, not phonemic in its nature. The phonic effect of duration can also be described as the sum of the durations of two adjacent identical phonemes – vowels or consonants, which leads to the denial of duration as a phonologically relevant feature. The problem is sometimes cited as an instance of non-uniqueness in phonological solutions, which means in fact that it can simply be removed from the phonological agenda, but the admission of non-uniqueness is unacceptable as running counter to the general demand for adequacy in scientific descriptions [4, p. 13].

The difficulties in determining the phonological nature of duration in the Scandinavian languages are to some extent aggravated by a peculiar aspect of the traditional approach to the problem – as a rule, its consideration is mainly based on the typical sequences of a vowel and a consonant with quantitative balance, e.g. Sw. and Norw. hat, Icel. hatur with a long vowel followed by a short consonant as against the respective hatt, hattur with a short vowel before a long consonant. The vicious circle is evidently based on this massive factual evidence. And since the typical situation presented in this way is common to all the Scandinavian languages except Danish, phonological solutions suitable for one language are sometimes presumed to be applicable to the other languages as well [cf. 3, p. 137, footnote 8].

It follows from the above that solutions should be sought only for particular languages and based on a comprehensive analysis of all the pertinent facts, not confined to typical, widespread instances of the phenomenon under consideration, but including the scrutiny of exceptional cases that deviate from the general pattern. This approach has proved expedient in the search for solutions to the peculiar problem of duration in Icelandic [1; 5], where two phenomena have proved to be particularly notable in that respect. First, in words with postvocalic consonantal clusters, e.g. kast, mynd, the first consonant is not lengthened after the vowel, which is short before a cluster, and duration is thus absent from both phonemes. Secondly, numerous word forms with a morphemic border between two identical consonants, e.g. neut. hvít|t (cf. fem. hvít), demonstrate the morphemic divisibility and hence the bi-phonemic nature of Icelandic geminates, which are essentially similar to heterophonemic clusters. Taking these facts into account provided the grounds for denying the phonemic relevance of consonantal duration in Icelandic [1, p. 143]. Since Old Icelandic vocalic duration has been replaced by the qualitative opposition between the so-called ‘thin’ (grannir) and ‘broad’ (breiðir) vowels, the duration of which varies allophonically depending mainly on the number of consonants (identical or different) after the vowel, this Scandinavian language has evidently abolished duration altogether as a phonemic feature [5].

This conclusion is inapplicable to Swedish, which is close to Icelandic in the typical situation represented by above-cited respective pairs like hat – hatt and hatur – hattur, but differs from it radically in the characteristics of clusters on the borders between the stem and a suffix – whereas the phonic duration of the stem vowel in Icel. neut. fúl|t – fem. fúl varies, shortening before two consonants and lengthening before one, in the respective Sw. ful|t – ful the duration of the vowel is unaffected by the addition of the suffix. Hence the need to search for a solution different from the one for Icelandic. The common trend towards dissimilation in phonic duration between a vowel and the following consonant is obviously materialized differently in these two kindred languages with their specific phonological systems.

Among phenomena on the morphemic border between the stem and a suffix affecting the feature of duration in Swedish particular attention should be paid to the rule on shortening the stem-final vowel when followed by /t/ or /d/ in the suffixes for neut. adjectives, pret., 2nd part. and supine of verbs, e.g. in ny|tt, så|dde – så|dd – så|tt. This rule is hard to explain in the traditional terms of qualitative balance, since the same suffixes do not shorten the stem vowel when a cluster is created by the addition of the suffix, and it is indeed paradoxical that the vowel remains long even before two consonants, e.g. in supine läs|t and neut. vis|t, while in ny|tt, så|tt it is shortened before a single consonant in the suffix.

It is highly significant for the correct assessment of the phonological implications of this rule that this is by no means a marginal phenomenon. Established in the 14th century, it has been spreading since then. True, the traditional 3rd conjugation where the rule operates includes no more than 30 verbs, but apocope widespread in everyday speech expands the action of the rule to the 2nd conjugation of weak verbs, e.g. bre(da), klä(da), rå(da), trä(da), and many frequently used strong verbs build supine forms by this pattern, e.g. få|tt, gå|tt, stå|tt, se|tt, le|tt, dö|tt. Due to the joint impact of apocope in inf. and syncope in supine this pattern replaces older forms, e.g. ge – ge|tt, be – be|tt besides the respective standard forms giva – givit, bedja – bedit. The stem vowel in pret. hade of ha(va) is shortened without it being reflected in the spelling [6, p. 98-102]. The adjectives that build neut. by this pattern are few, but their list is not closed and can be joined by recent borrowings, e.g. disträ|tt.

The expansion of this morphonological trend is even wider in Norwegian, where weaker standardization contributes to the spread of apocope and syncope. The pattern used to build supine forms like Sw. få|tt, gå|tt is productively followed in 2nd part. forms like Norw. bli|tt, ta|tt, dra|tt, gi|tt, la|tt, slå|tt. Some strong verbs with apocope in inf., e.g. by(de), ri(de), gli(de), build new pret. forms on the analogy of such 2nd part. forms. The verb ha builds pret. hadde and 2nd part. hatt by this pattern. The use of adjectives following the analogous pattern in building neut. forms is widened by the apocope of stem-final /d/, as in bre(d) – bred|t, rø(d) – rød|t.

It should be noted that in Icelandic, unlike Swedish and Norwegian, the shortening of the stem-final vowel before /t/ in the suffix is restricted to only two forms – nom. and accus. sing. neut. of the strong adjectival declension (e.g. ny|tt), where it affects only some of the vowels. The phenomenon does not affect verbs. The pattern is thus non-productive, and its retention for a small closed list of adjectives may be due to analogy with the overwhelming majority of Icelandic adjectives, whose stem vowel is regularly shortened before a cluster on the morphemic border.

There is an apparent connection between the above-considered phenomenon and another, also expanding in use, which affects the morphonological situation in neut. nouns with stem-final vowels. The suffixed definite article can be used with them in two variants – the full one -et and the elided -t, e.g. knä|et ~ knä|t, bi|et ~ bi|t, frö|et ~ frö|t, of which the latter is gaining ground in everyday speech. There is, however, a crucial difference between the two phenomena – the consonant /t/ in the article, unlike its counterpart in the forms of verbs and adjectives, has no impact on the phonic duration of the preceding stem-final vowel. As a result, minimal pairs appear with the same morphonological structure, e.g. rå|tt (neut. of rå ‘raw’ or supine of rå ‘to dominate’) – rå|t (def. of rå ‘wafer’), gry|tt (supine of gry ‘to dawn’) – gry|t (def. of gry ‘grit’), fly|tt (supine of fly ‘to flee’) – fly|t (def. of fly ‘bog’). Such pairs can also appear when the noun and the verb or adjective are derived from the same stem, e.g. strö|tt (supine of strö ‘to strew’) – strö|t (def. of strö ‘bedding’), ny|tt (neut. of ny ‘new’) – ny|t (def. of ny ‘new moon’).

Thus both the short consonant /t/ and its long counterpart occur in identical phonemic and morphonological contexts, which is conclusive proof that short /t/ in the suffixed neut. def. article and long /t:/ in the suffixes of verbs and adjectives are two consonant phonemes different in the presence or absence of the feature of duration. This phonemically distinctive feature is therefore consonantal in Swedish and Norwegian. As for the phonic duration of vowels, it varies allophonically before these two consonant phonemes in suffixes in full accordance with the rule for quantitative balance – vowels are long before /t/ and short before /t:/. It is reasonable to conclude that what has been proved for these two consonants can be ascribed to all the other consonants in the absence of facts that refute this conclusion. The widening use of suffixes with short and long dental consonants /t/, /t:/, /d/, /d: / in Swedish and Norwegian testifies to the growing functional load on the phonemic feature of consonantal duration.

There are fewer long consonants than short ones in Swedish, as /v/, /j/, /h/, /ć/ have no long correlates. The long /t:/, /d:/ in the suffixes of verbs and adjectives retain their duration only immediately after stressed stem-final vowels, but are shortened after stem-final consonants, e.g. in läs|te, and after unstressed vowels, e.g. in kalla|de. Within the stem consonantal duration is clearly distinct in postvocalic and intervocalic positions, e.g. kal – kall, vina – vinna. Before another consonant the duration distinction is retained if the cluster results from productive suffixation, which occurs in two cases – when the stem-final consonant is followed by a consonant in the suffix, e.g. in kal|t – kall|t, kal|na – kall|na, or the vowel between two consonants in the stem is elided, e.g. in kakl|et – kackl|et, kakl|a – kackl|a derived from kakel – kackel. Consonantal duration is not distinct before stable clusters within the stem, e.g. in kasta, folk, as well as clusters resulting from non-productive suffixation, e.g. in ren|sa, vid|ga, vak|t, söt|ma).

References

1. BENEDIKTSSON, H. The non-uniqueness of phonemic solutions: Quantity and stress in Icelandic. Phonetica, 1963, 10.

2.      ELERT, C.-C. Phonologic studies of quantity in Swedish. Stockholm: Språkförlaget Skriptor, 1964.

3. ELIASSON, S., LA PELLE, N. Generativa regler för svenskans kvantitet. Arkiv för Nordisk filologi, 1973, 88.

4. MARTINET, A. Indétermination phonologique et diachronie. Phonetica, 1965, 12.

5. PLOTKIN, V. Y. Is length phonologically relevant in Icelandic? Phonetica, 1974, 30, 1.

6. WITTING, C. Studies in Swedish generative phonology. Uppsala: Almqvist & Wiksell, 1977.