Old Norse/Icelandic: -a- (in compounds)

Discussion in 'Etymology, History of languages, and Linguistics (EHL)' started by Gavril, Dec 9, 2012.

  1. Gavril Senior Member

    English, USA

    In present-day Icelandic, compounds between nouns (or between nouns and adjectives) are often formed with a connective element -a-:

    orðabók "dictionary" < orð "word" + bók "book"
    áhugaverður "interesting" < áhugi "interest" + verður "worthy"

    The genitive plural ending for most nouns is also -a (manna "of men", þjóða "of countries", etc.), and the -a- seen in compounds usually seems to be identical to this ending.

    For example, the few nouns that have a genitive plural ending -na rather than -a can be seen to retain this ending in compounds:

    augnablik "moment" < auga "eye" (gen. pl. augna) + blik "flash"

    Similarly, when an adjective is the first member of a compound, the adjectival gen. pl. ending -ra appears after it:

    sjúkrahús "hospital" < sjúk- "sick" (gen. pl. sjúkra) + hús "house"

    However, many (if not most) nouns with nominative singular ending in -a (whose genitive plural ending is usually also -a) fail to use the gen. pl. ending in compounds: instead, they use -u- (which is also the genitive singular ending for this stem class):

    tunga "tongue" (gen. pl. tungna) > tungumál "language"
    stjarna "star" (g.pl. stjarna) > stjörnuþoka "galaxy, stjörnufræði "astronomy", etc.
    "week" (g.pl. vikna) > miðvikudagur "mid-week day, Wednesday"
    "gathering" (g.pl. samkom(n)a) > samkomulag "agreement"

    The word auga seems to be an exception to this pattern (as it becomes augna- in compounds), but it differs from the above nouns in other respects as well (for example, its gen. singular is auga not augu).

    Given all this, I wonder, how far back is the connective -a- in Icelandic compounds though to date? For example, was it used in Old Norse, and was it identical to the genitive plural in that period as well?

    Also, what's the general theory as to why so many nouns with nom. sg. -a (tunga, stjarna etc.) don't use the genitive plural ending in compounds, and what is thought to be the origin of the -u- that these nouns use instead?

    Thanks for any help
    Last edited: Dec 9, 2012
  2. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    You can make compounds in a variety of ways. The way you mentioned is what is called eignarfallssamsetningin, but there is also stofnsettningin which connects the stems of words and doesn't have a genitive ending (vitlaus, bílstjóri, snjóhús etc.). There are different types of compounds and when you have a typical root and an affix this is called an afleitt orð (a derivative) which connects together and doesn't have a genitive ending. The initial element in stjörnufræði is not to be confused with any sort of nominal declension but simply the forliður stjörnu- combined with -fræði (just like you'd use aðal- in other types of compounds) and the same things go for the other examples that behave like this in your question. These are derived from the oblique case (singular) of the related noun form.

    It's not always the genitive plural that you use to make the plural, only where this would make sense so in the example samkomulag that you mentioned, that actually is eignarfallssamsett in the sense that it's using a genitive ending and isn't any affix or form used in a compound word. It's supposed to be clear which form actually makes the most sense. There is an interesting case of ofvöndun (hypercorrection) where people add in singular genitive endings in compound words that actually require plural (because the plural one is less in form, people think when they say it that it should be the typically longer (in some genders) singular form). An example is mánaðamót - which means the beginning of the month, but as you can see it's based on the "meeting of the months" and it makes no sense to use the singular form - but people do put it back in when hypercorrecting and say/write mánaðarmót :cross:. So, it's linked to genitive but it doesn't always have to be plural.

    I can't help you with how it worked in Old Norse though. :(
    Last edited: Dec 10, 2012
  3. fdb Senior Member

    Cambridge, UK
    French (France)
    In ancient Indo-European languages (e.g. Greek or Sanskrit) the usual procedure in compounds is that the first component (“Vorderglied”) is an undeclined stem, and only the second component (“Hinterglied”) is declined for number, gender and case. Thus, you have to deduce the syntactical relationship of the two in each case. Still in English we have tatpuruṣa compounds like “stationmaster” (master of the station, quasi genitive), but “bathtub” (tub for a bath, quasi dative), alongside bahuvrihi compounds like “redhead” (whose head is red). But in many IE languages (occasionally already in Sanskrit) we also have compounds with a declined Vorderglied, like German “Ratskeller” (genitive Vorderglied) alongside “Rathaus” (undeclined Vorderglied).
  4. Gavril Senior Member

    English, USA
    Hi Alxmrphi,

    So, if I understand right, samkomulag contains the genitive sg. of samkoma, whereas the stjörnu- seen in stjörnufræði isn't (historically) a genitive at all? What's the evidence (in past or present Icelandic) for this distinction?

    It does make sense that a compound in -fræði wouldn't normally contain a genitive sg. form (see, e.g., stærðfræði instead of :cross:stærðarfræði). On the other hand, it still seems to me that weak fem. nouns like stjarna, kona, etc. behave differently (morphologically speaking) in compounds than many or most other stem classes, regardless of the intended semantics of the compound.

    Do you know of any compounds in which a weak fem. noun appears in its genitive plural form (stjarna-, kona-, tungna- etc.)?

    Thanks again for your help
    Last edited: Dec 10, 2012
  5. Gavril Senior Member

    English, USA
    What about the connective -o- seen in Greek (and perhaps other IE) -- is this thought to come from an earlier genitive form? For example, I think that all the words in -ology, -onomy, -ometry etc. are based on Greek words that contain this connective (e.g., geōmetria < "earth" + -o- + -metría "measurement").

    I don't think that the Old Norse/Icelandic -a- connective is of the same origin as this -o- connective, but I'm not completely sure they're unrelated either.
    Last edited: Dec 10, 2012
  6. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    I don't know about the evidence, I just read what the Icelandic linguists write about and don't see any reason to doubt them.
    But you'd be treating things like stjörnu- in compounds as nouns to give you that impression, which isn't actually the case (according to currently accepted opinion).
    What irregularities do you notice that are still unexplained given this different way of analysing these elements in a compound?
    Kvennaklósett? To be honest, I am struggling to find other ones. Maybe the practice is for these to remain mainly singular in nature to show they are part of a compound because otherwise, besides the -(n)a examples the forms would be identical in the nominative singular and genitive plural. I've not seen that explained anywhere but it does seem likely. Ironically the only one I could find easily was one of the exceptions that has an irregular genitive so does in fact distinguish itself from the related nominative form.
  7. Gavril Senior Member

    English, USA
    Since stjörnu (gen. sg. of stjarna) and the stjörnu- seen in compounds are formally identical, isn't it reasonable to ask what the basis is for claims that they are different at a (putative) deeper level?

    None that occur to me right now, but I still find this analysis very surprising. In almost all the -fræði nouns that I can think of, the initial compound member appears in a form that would be a "legitimate" case form in isolation:

    nom. sg.: líffræði / stærðfræði / tölfræði / etc.
    gen. sg.: stjörnufræði / eðlisfræði
    gen. pl.: efnafræði

    The only exception I can think of right now is hagfræði (where hag- is most likely not based on the accusative of hagur). Are examples like hagfræði the basis of the idea that -fræði compounds are not composed of [noun] + [noun]?

    That seems like likely to me as well, except that I'm not sure what you mean by "singular in nature" -- do you mean semantically singular?
  8. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    Well, the basis is exactly the reason they behave differently. It's the reason you're asking the question - they don't behave like you'd expect. Is that not basis enough? I think what I visualise as a useful classification in a synchronic analysis you are treating as fundamentally different to the core and that's not really what I'm talking about. Elements in compound words can be seen as nouns in some cases and compounding elements in others. To look at it at a level of fundamental existence would only lead to unanswerable questions when what is being presented is just a useful way to classify elements and how they work. Obviously you can see how usage blends into another and it's obvious that they're related, but I don't see why we're looking at what sort of level of grammaticalisation (if any) they would be at. What does that solve? Although I said "not as a noun" I'm just detailing a way that has been made sense of it in a classification and if you understood that claim to be that there is actually nothing nominal about it in behaviour then that's not what I've been discussing. How can you tell when forms are blended? You would need strings of psychological tests and we can't use typical tests for syntactic behaviour because where these forms exist are pretty rigidly fixed in compounds. When you have options to compound with affixes, over time you can reanalyse what was once a nominal form as more prefixal in nature, fossilised in a sense, that behaves in a steady way. Am I certain this is the case? Absolutely not. Is anyone fully sure? I doubt it.

    Icelandic is not a perfectly rule-based system (as I'm sure you know) and you have things happen in one case and behave differently at another. In all truth, it's messy and there are always exceptions and this could just be one of them. Call them nouns that exist in oblique case to be different from the nominative that occur in weak feminine nouns. You could look at it semantically and try to derive some better rules and that might be to call them prefix-like compound elements that you get with that class.

    "Awful" and "Awesome" entered English with a few hundred years between them and now the lack of an -e in one of them has clouded the relationship. Things happen that we just get used to and a quick look through the dictionary with a lot of these words you talk about are not labelled as nouns, but "FORLIÐUR". You can doubt it and view it as some weird noun-like behaviour but I just don't see to what effect. Like with the awful/awesome thing, foreigners will want to find some system that makes sense but you just need familiarity with the language to let you know the quirks and idiosyncratic twists and turns of the system.

    I'm not sure why you're singling out " -fræði compounds " here. To me, when you talk about professions, it's exactly expected that you would use combinatorial elements to form such words and all the listed elements are listed as such. You don't have a problem with hagfræði sticking out then because the related element is "hag-" (economic-). It's just stofnsamsetning, nothing more. :p
  9. Gavril Senior Member

    English, USA
    As far as I can see, we're comparing the following explanations for the distribution of stjörnu:

    1) a single form stjörnu is used in different contexts (i.e., in isolation and in compounds) for different reasons related to each context

    2) two synchronically separate (but formally identical) forms, stjörnu and stjörnu-, are used (respectively) in isolation and in compounds

    I think that only #1 accounts for the initial choice (somewhere in the history of Norse/Icelandic) to use stjörnu- in compounds rather than stjarna-/etc. Later on, either conception (#1 or #2), or some combination of both, may have operated when new compounds were formed with stjörnu-, or in speakers' analysis of existing compounds with stjörnu-.

    However, the term "forliður", as I (maybe wrongly) thought you were using it, suggests that only #2 is consistently operating in a speaker's mind when they hear (or form a compound with) stjörnu-.

    I could very well be misunderstanding you, but in the previous post, you said

    Why do you say "isn't actually the case" if we're only comparing equally plausible theoretical models?
    Last edited: Dec 10, 2012
  10. fdb Senior Member

    Cambridge, UK
    French (France)
    In the great majority of instances the Greek –o- is the thematic vowel of second-declension nouns (e.g. philo-sophy), but this often spreads by analogy to other types of stems, as in ge-o-metry. In neither case is it a case ending. But Greek does have a very small number of compounds with a case form as vorderglied, e.g. νεώσ-οικοι “wharfs” (lit. ship’s houses).
  11. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    ..."according to current accepted opinion."

    I may be wrong but it seems as if you're actively trying to misunderstand what I'm saying. Maybe the question needs to encompass why this is important at all? What are you hoping to find? I told you it depends on how you look at explaining what is seen with the data and mentioned that I've read that these are best viewed (by linguists working in the generative paradigm) to be a case of less-nominal behaviour. I don't know what other "equally plausible model" you're referencing because I didn't mention one. Maybe my view on how language should be analysed is just too fundamentally different to yours in how I've been taught to explain linguistic features. If that's the case then you can ask some linguistics who actively do work in psycholinguistics to see how these forms are represented in the mind but I guess at the end of the day I just don't see why it matters or why this hair-splitting is beneficial to any linguistic argument.
    This is probably where we diverge. I had no intention of talking about speakers' representations in the mind. I highly doubt that information is available given the size of the language and of the departments that actively do work on it, but I've read some studies on expecting case forms so it might not be impossible. I'm talking about describing and codifying the language with respect to the previous traditions of explaining it. There is no overlap with actual neurological structure. Even the most prominent generativists know and state, all the time, that their syntactic analyses and X-bar theory are theories on how to explain structure visually and are not to be taken as a blueprint for something that actually exists in the mind. That's where I think you're taking what I am saying to be something completely different.

    Do you see some huge irregularity that needs fixing in explanations of Icelandic grammar? Do you fundamentally disagree a dictionary should label these not as nouns but something else? I just don't see what you're trying to fix.
  12. Gavril Senior Member

    English, USA
    I'm sorry if I gave that impression. I know that you said "according to current accepted opinion", but it seemed to me as though you were using the acceptedness of this opinion to counter my suggestion about stjörnu- etc. (but now it looks like this was a wrong interpretation on my part).

    The other "model" here is my suggestion, namely that the stjörnu- seen in stjörnfræði, along with similar forms, doesn't clearly belong to a separate (non-genitive) grammatical category from the efna- of efnafræði, the eðlis- of eðlisfræði and so on.

    On the other hand, in compounds like Bjarney, tilviljanlegur, etc., the first element of the compound doesn't appear as an isolated case form of the corresponding noun (björn, tilviljun), so it does seem reasonable to place bjarn-, tilviljan- etc. in some kind of separate category (though not necessarily to deny that they're nouns to begin with).

    I'm skeptical of the focus on abstract structure to begin with (no offense meant in saying this). I try to explain linguistic phenomena in terms of semantics, pragmatics, learning processes, memory, etc., unless there's evidence that some other factor is overriding all these criteria.

    I don't know nearly enough about the dominant model(s) of Icelandic grammar to see huge irregularities in it. :) I just wonder why compound elements such as stjörnu- (as in stjörnufræði) would be categorized separately as "forliðir" while the compound elements efna- (efnafræði), samkomu- (samkomulag)etc. are categorized as nominal case forms -- thus far, I don't see any clear reason, semantic or morphological, for making this distinction.
    Last edited: Dec 11, 2012

Share This Page