Paradigm preservation of long vowels in Norwegian verbs

[Magb]

This phenomenon probably also exists in at least Swedish and maybe Danish and even Icelandic/Faroese, but in an attempt to avoid spreading misinformation I'll limit this post to my own idiolect of Norwegian.

A little background knowledge: vowel length is mostly predictable in Norwegian. If a vowel is followed by two or more consonants (including a single geminated consonant), it's short; if it's followed by one or zero consonants, it's long. There are some exceptions to this rule, but it describes most cases.

One of the aforementioned exceptions comes from paradigm preservation in verbs. In Norwegian, the verbal suffix -te (also -de or -dde, depending on what precedes it) is used to form the preterite for many verbs. Many verbs that take this suffix have the stem structure (C)*V:C, e.g. lime, stole, mene, etc., and in the majority of these cases the long vowel is preserved when the suffix is added, giving [li:mte], [stu:lte], [me:nte], and the like. There are also verbs such as kjenne which belong to the group, and in these verbs the short vowel is of course maintained.

However, there's a small group of verbs where the vowel is long in the infinitive (and the present tense), but short in the preterite (and the supine/past participle) -- sometimes optionally so, but usually obligatorily. One example of a verb where the shortening is obligatory is like "to like". This should demonstrate the difference:

Code: Select all

Word    Inf.    Pret.   meaning
like    li:ke   likte   "like"
vise    vi:se   vi:ste  "show"
spille  spil:e  spilte  "play"

Here's a list of some of the verbs in question. There are no doubt more, but these are the ones I can think of right now. Verbs marked with * are ones where I sometimes hear people preserve the long vowel, and the same may apply to some of the other verbs. Personally I always use the short vowel in all of these.

Code: Select all

Norw.     Old Norse   Eng.
like      líka        "like"
leke*     leika       "play"
tape      tapa        "lose"
rope      hrópa(?)    "shout"
lage[1]   laga        "make"
løpe      hlaupa      "run"
døpe*     deypa       "baptize"
kjøpe     kaupa       "buy"
spise*    (loanword)  "eat"
sage*     saga        "saw"

[1] normally this verb belongs to the -a group, but using -de is also possible, and in those cases the pronunciation is always [lagde]

I've given the Old Norse forms to show that this phenomenon doesn't seem to have anything to do with the syllable weight of the ancestor word in Old Norse (not directly anyway). I've noticed that many of the verbs I've given contain a monopththongized historical /au/, /øy/ or /ei/, but I can't guarantee that the list I've given is representative.

A final note: in verbs belonging to the group with no consonants in the coda of the stem, e.g. flå, fri, spre, etc., the vowel is always long in the infinitive, and always short in the preterite. These are the verbs where the suffix is -dde.

My questions are:
1. Does anyone know how this phenomenon has arisen in Norwegian?
2. How would you analyze the difference between e.g. like and vise. Would you say that:
- a) The stems are underlyingly /lik/ and /vi:s/, with the length of the vowel of /lik/ being determined by the consonant(s) following it; or
- b) The vowel shortening is the effect of two different forms of the -te suffix?

[Åge Kruger]

Some interesting possibilities in this direction can be found in Einar Haugen's The Scandinavian Languages, §11.3.9 and §11.4.12.

Consonant clusters which were not geminates varied in their relations to preceding vowel length. ...the general tendency was for vowels to be short before clusters. Those clusters which contained a morpheme boundary, however, were often treated differently, since the analogy with a long base form could maintain length even before clusters.

§11.4.12 talks about strong verbs becoming weak (leika is specifically mentioned), with -er and -te being added by analogy.

I think §11.3.9 is probably the most interesting part here.

[Magb]

Thanks Åge, your post has led me to the answer to the first question. Out of the verbs I've listed, only kaupa took the -ti suffix in Old Norse. The others were either strong verbs, or they were regular first conjugation verbs taking -aði in the preterite. So this group of verbs appears to be made up of verbs where the -te suffix has only been employed at a quite late stage, probably after quantity loss took place.

I'm still not sure about my second question though. Either of the solutions I proposed would introduce a rule of a type that doesn't exist anywhere else in Norwegian, as far as I can tell. I think I'm leaning towards (b), but maybe there's something else I haven't thought of.

[Åge Kruger]

I'm still not sure about my second question though. Either of the solutions I proposed would introduce a rule of a type that doesn't exist anywhere else in Norwegian, as far as I can tell. I think I'm leaning towards (b), but maybe there's something else I haven't thought of.

I'm afraid I can't help you there. I don't really understand the whole point of having "underlying forms", especially when cases like this come up. Maybe a trained linguist will pop her head in and give you some good advice.

[Magb]

It occurred to me that Kristoffersen talks about this in The Phonology of Norwegian (chapter 8.2.2). As I've been suspecting, he concludes that the difference between the verbs where shortening occurs and the ones where it doesn't is unpredictable, and must therefore be lexically encoded. The specific solution he offers for that is something to do with "Mora Insertion" that's a bit beyond my abilities as a phonologist.

He notes that there are patterns to when shortening happens based on what consonant appears in the coda of the verb, but he doesn't mention my theory about the -te suffix only having been applied fairly recently (which I still think has some merit). He also points out that that there's an analogous phenomenon with the neuter suffix -t for adjectives and adverb formation, but in that case the shortening is phonetically predictable based on the final consonant of the adjective.

[Qalool]

This phenomenon probably also exists in at least Swedish and maybe Danish and even Icelandic/Faroese, but in an attempt to avoid spreading misinformation I'll limit this post to my own idiolect of Norwegian.

In standard Swedish, the only word from the 2nd conjugation (with preterite in -de or -te) that behaves like that, apart from words that have a stem consonant that itself is either t or d, like sköt|-a -te, is the verb köp|-a -te "buy" [ˈɕøːpa, ˈɕœptə].

[Ulrike Meinhof]

This phenomenon probably also exists in at least Swedish and maybe Danish and even Icelandic/Faroese, but in an attempt to avoid spreading misinformation I'll limit this post to my own idiolect of Norwegian.

In standard Swedish, the only word from the 2nd conjugation (with preterite in -de or -te) that behaves like that, apart from words that have a stem consonant that itself is either t or d, like sköt|-a -te, is the verb köp|-a -te "buy" [ˈɕøːpa, ˈɕœptə].

I think some dialects have /OktE/ for "åkte" too.