10 March 2013

To Be or to Be: Aspects of Existence (Act III)

Here is the present-tense paradigm of ‘to be’ in five Germanic languages. The figure below is a cladogram showing how those languages are related to one another.

Old Norse
Old Frisian
Old English

The Gothic and Old Norse forms are all derivable (albeit with some complications) from the inherited paradigm of PIE *h₁es-/*h₁s-. The Old Norse root is er- (rather than es-) probably because the Germanic forms of ‘to be’ for the most part reflect unstressed, enclitic variants of the verb. If Proto-Germanic *s was intervocalic and preceded by an unstressed vowel, it became *z by so-called Verner’s Law. In North and West Germanic languages this *z became /r/ (there was no such change in Gothic). The third-person forms er and  eru were innovative, created analogically to match the rest of the paradigm.

More curious is the occurrence of an initial /b/ in some of the West Germanic forms (see the blue-shaded cells in the table). In Old High German, we find verb forms which look like the expected reflexes of *h₁es-/*h₁s-, except that a /b/ is prefixed to the verb in the first and second persons, singular as well as plural. Old Frisian exhibits the same oddity, but only in the singular, because the Anglo-Frisian languages had given up distinguishing grammatical persons in the plural and generalised the 3pl. form. Note that there are no “b-forms” whatsoever in Gothic and Old Norse, so it looks as if the use of /b/ in the first and second persons were a distinctly West Germanic innovation.

And yet, most strangely of all, English does not follow the same pattern. This would be less surprising if English were a “basal” West Germanic language loosely related to the rest of the group. But no, English is nested rather deep inside one of the subgroupings of West Germanic. Still, even today, English has am, are (older art), is as opposed to Modern German bin, bist, ist. Instead of isolated appearances of an initial /b/, Old English had a full set of b-forms as an alternative present-tense paradigm beside the rather faithfully retained reflexes of *h₁es-/*h₁s-. In no way can the Old English pattern be explained as a local innovation. It must be an archaism: OE preserved two “alleles” of ‘to be’ for each person, whereas both its sister language, Old Frisian, and its slightly more distant cousin, Old High German, had eliminated one of the variants and fixed the other. The fact that in both languages the b-forms show a similar distribution is therefore the result of convergence, not a shared West Germanic feature. Also the total absence of b-forms in Gothic and Old Norse must be due to the parallel elimination of a pattern present already in Proto-Germanic. The strange root *bi-, which seems to underly the Old English b-forms, is unique to that language; Old Frisian and OHG have forms contaminated with it rather than its direct reflexes. Nevertheless, this contamination is indirect evidence of the presence of *bi- in Proto-West Germanic. Since outside Germanic PIE *bʰuH- has provided verb forms with the meaning of ‘to be’ in several branches (Slavic *byti, Lat. fuī ‘I was’, etc.), the archaic status of the alternative b-paradigm in Old English is quite clear. Note the importance of the outgroup data, overriding the negative evidence of Gothic and Old Norse. Apparently the insular isolation of English since the Anglo-Saxon conquest of Britain in the fifth century helped it to preserve a curious lexical relict.

If we were to reconstruct the early Germanic prototypes of the OE b-forms, they would have to look more or less like this:
1sg. *bi-ō, 2sg. *bi-si, 3sg. *bi-þi, 3pl. *bi-anþi
Such a reconstruction is hardly satisfactory, since the endings look partly like those of the thematic conjugation (1sg.), partly athematic (2-3sg.), and partly ambiguous (3pl.). The pattern must have been reworked in some way, but how? Let us note first that the root *bʰuH- occurs in Germanic outside of the paradigm of ‘to be’. One old imperfective stem derived from the root aorist, *bʰuH-jé/ó- yielded Germanic *bū-i/a- (OE būan, ON búa ‘dwell, inhabit’), but it doesn’t match the OE b-forms either formally or semantically. Secondly, although the residue of the b-conjugation was completely integrated with  the *h₁es-/*h₁s- pattern in continental West Germanic, Old English had two semi-independent sets of forms, preserving subtle semantic differences which may offer a clue to understanding the origin of *bi-.

But I suppose this is more than enough for one blog post. The investigation will be continued in the next.


  1. Is old English's preservation of two "to be" verbs perhaps due to a Celtic substrate?

    The Celtic languages generally have had a stative/essive distinction in their to be verbs; I think it's pretty widely understood to be the reason for the developement of Portuguese's (Port of the Gauls-ese-'s) and Spanish's esse/esta(r) distinction and I believe a historical but now eroded away distinction in French. I could certainly be wrong; I'm just a hobbyist with no citations, but it seems likely to due to the Welsh-like substratum in English leaving things behind like a progressive present, an over-eager (relatively) to do, a somewhat different set of idioms from the other Germanic languages, etc.

    Evidence for this I believe would come out of a distinction between beo and eom (beon and senan? wesan?). I know next to nothing about Old English grammar but if I'm remembering right it did in fact have something of an essive/stative distinction but it got conflated with the temporal-aspect issues one would expect from a speaker of a language with very weak distinctions between the two.

    This is important, I think, because it undermines the terminology - and the underlying picture the terminology represents. Namely- if the pressure for a beo/eom distinction came from the Celts, it isn't unreasonable to envision an already defective (hypothetical) pair of verbs getting new derivitives uniquely in OE. From what I know, the beo conjugation looks a lot like a weak verb, which is what I would expect from a Germanic language creating new verb forms through analysis.

    Of course, that's not much of a good explaination for OF and OHG. They undoubtly interacted with Celtic groups before arriving in their current lands but either not remotely like English or they were particularly oddball from a Celtic typology perspective to have blended with the Germanic languages so well (or maybe that's irrelevant?). So maybe I just walked into a Chicken or the Egg problem.

    But still, to my uneducated-probably-doesn't-grasp-anything-he-speaks-of-hobbyist eyes, it seems to me like beo is "newer" than early Germanic.

  2. Brittonic Celtic influence as the source of the double paradigm in Old English has been proposed by several authors. While the retention of the b-conjugation in English is a possible case of areal convergence, I don't think Celtic parallels alone can account for its rise in English. The b-forms do not look like derivatives of anything post-Proto-Germanic. If anything, they are simpler that any other Germanic stem related to PIE **bʰuH-. They do not display any characteristic features of the weak declension either (unlike būan, which had a weak preterite and an two alternative past participles, weak and strong). The participle (ge-)bēon is of course strong. Finally, the existence of a mixed paradigm everywhere in West Germanic presupposes something like the OE situation already in West Germanic.

    1. Interestingly AAVE has an is/be distinction as in "He doin' it again." vs. "He be doin' that all the time." although this appears to be due to contact with either Irish or Hiberno-English speakers in the early days of slavery rather than some dialect survival in English.

      Imagine trying to sort this out as a historical linguist a thusand years from now with most of the evidence erased by tiem. What kind of mistaken conclusions might one come to?

  3. Good point, good point. Obviously I lacked the empirical end of the argument there. An eyeing is no substitute for an analysis, afterall. Thanks for the reply; I'm off to find some more specifics.

  4. But you raised a very good point yourself. Cogent arguments in favour of Old Welsh influence have been made e.g. by Ilse Wischer (2010). It's impossible to address all such issues (and discuss competing hypotheses on the origin of *bi-) in a blog post, but of course I'll have to do so in a more formal way if I decide to publish my views.