13 March 2013

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

There is a subtle difference between the is-forms and the biþ-forms in Old English. The reflexes of   *h₁es-/*h₁s- typically refer to ongoing temporary states, which is not surprising, given the imperfective value of the PIE root verb. The b-forms, on the other hand, usually refer to future states or to general, timeless  truths:

  • Iċ bēo ġearo sōna ‘I shall be ready soon’
  • On twelf mōnðum bið þrēo hund daga & V & syxtiġ daga ‘In twelve months there are 365 days’

This contrast, unique to Old English among the Germanic languages, shows that the root *bi- retained something of the perfective value of *bʰuH-. Altough it can’t reflect the unmodified root aorist, it seems reasonable to look for its origin among formations genetically connected with the aorist, and preserving an “aoristic” lexical aspect – verbs that refer to actions or states having a terminal point, evolving towards completion, or treated as a unitary whole without internal temporal structure. As mentioned in a previous post, “present” stems with a zero grade of the root and an accented *-e/o- suffix, derived from root aorists (e.g. *gʷr̥h₃-é/ó- ‘devour’ from *gʷérh₃-/*gʷr̥h₃- ‘swallow up’), fit the bill. Let us therefore suppose that *bʰuH- also generated a present stem of that type, namely *bʰuH-é/ó-, formally imperfective but with the approximate meaning ‘is going to be/become’ or ‘is (at any time, by definition)’. Such a verb is indeed attested outside Germanic, cf. Greek pʰúomai ‘grow, spring up, become’. The expected development of such a stem from PIE to PGmc. would be as follows:

1sg. *bʰuH-ṓ > *buō
2sg. *bʰuH-é-si > *buisi
3sg. *bʰuH-é-ti > *buiþi
3pl. *bʰuH-ó-nti > *buanþi

Let us suppose that the unaccented short high vowel in hiatus (resulting from the loss of the laryngeal) became a glide in pre-Germanic times, and that the resulting exotic cluster *bw- was simplified to *b already in Proto-Germanic. The predicted outcome would then be something like this:

1sg. * > OE * (or *?)
2sg. *bisi > OE bis(t)
3sg. *biþi > OE biþ
3pl. *banþi > OE *bœ̄þ

The first and last forms do not resemble the attested ones. But the Proto-Germanic pattern would have been so unlike anything else in the verb system that speakers could be expected to favour any “mutation” reducing its irregularity. For example, the reanalysis of *bisi and *biþi as *bi-si and *bi-þi (rather than *b-isi, *b-iþi with an aberrant asyllabic root) would have motivated the introduction of the neo-root *bi- also in the first person and the plural:

1sg. *b-ō → *bí-ō > OE bēo
3pl. *b-anþi → *bí-anþi > OE bēoþ

This restructuring could have happened at any time between late Common Germanic and archaic Old English, yielding the same outcome.

This little essay cannot exhaust the secrets of ‘to be’ in Old English. One vexing problem is the OE 2sg. eart (also Anglian (e)arþ) and the alternative plural found in the northern dialects (Mercian earun, Northumbrian aron, hence Modern English are). Their vocalism (*a, as if from pre-Gmc *rather than *e) is puzzling and makes comparison with Old Norse ert, eru problematic. They may represent the lingering traces of another verb which has contributed to the paradigm of ‘to be’, possibly PIE *h₁er- ‘arrive’. PGmc. 2sg. *arþ(a) and 3pl. *arun(d) would make sense as reflexes of the IE perfect built to that root, retaining their stative value and interpreted as present-tense forms (‘to have arrived’  ‘to be there’).

Another crazy quilt (Hat tip: The Dotty One)
I have not discussed the past tense (1/3sg. wæs, 2sg. wǣre, pl. wǣron < PGmc. *was-/*wēz-) because its origin is not controversial. Since both *h₁es- and *bʰuH- seem to have been reluctant to form perfects of their own, and since the PIE perfect was the foundation on which the new past tense of basic (non-derived) verbs was built, the missing past-tense forms were co-opted from the conjugation of the near-synonymous stem *wis-i-/*wes-a- (Gothic wisan; Old English wesan, an alternative infinitive beside bēon ‘to be’), continuing the PIE root *h₂wes- ‘stay, remain, spend the night’. The perfect of that verb became utilised as the past of ‘to be’, while the present forms just withered away (they occur sporadically in OE texts, but are extremely rare). This symbiotic relationship goes back to Proto-Germanic times, and the suppletive use of the ex-perfect stem *was-/*wēz- as ‘was/were’ occurs throughout Germanic. In Old English, we also find forms of wesan as alternative imperatives (wes/bēo ‘be’) and present participles (wesende/bēonde ‘being’). To sum up, the English paradigm of ‘to be’ is a Frankenstein monster sewn together from pieces of defective paradigms and involving perhaps as many as four different PIE roots (*h₁es- [am, is], *bʰuH- [be, being, been], *h₂wes- [was, were], and possibly *h₁er- [art, are]). A really impressive case of lexical symbiosis!


  1. As I recall, Stefan Schumacher (building on older research) has recently argued that the similarity between the (arguably) West-Germanic and the Celtic split in the present tense of the verb 'to be' was due to Celtic-Germanic language contact on the continent, i.e. before the settlement in England. Would that be possible in your framework?


  2. Hi, Anders! This kind of mutual reinforcement or areal "synergy" between languages in contact is of course possible and indeed quite common. In such cases it may be difficult to determine the source and original direction of influence; I wonder if, and to what extent, Proto-Italic participated in this partnership -- there is surely no shortage of f-forms there (including, beside the sum/fui suppletive complex, also the use of fio as the passive of facio).

    1. PS I was naturally thinking of a deeper chronology -- (pre)Proto-Germanic rather than West Germanic, as in Schumacher's scenario. That would mean assuming the existence of b-paradigms in early Germanic and their elimination from Gothic and Scandinavian. Their retention in West Germanic would then be attributable to prolonged contact with Celtic.

  3. Here's Tolkien on this very subject in his 1955 O'Donnell lecture, later published as "English and Welsh". I've replaced the macrons with acutes for ease of typing, and I'm posting Tolkien's footnotes in a separate comment.

    As an example of a curious parallelism I will mention a peculiar feature of the Old English substantive verb, the modern 'be'. This had two distinct forms of the 'present': A, used only of the actual present, and B, used only as a future or consuetudinal. The B functions were expressed by forms beginning with b-, which did not appear in the true present: thus, bío, bist, biþ pl. bíoþ. The meaning of biþ was 'is (naturally, always, or habitually)' or 'will be'.

    Now this system is peculiar to Old English. It is not found in any other Germanic language, not even in those most closely related to English. The association with the b-forms of two different functions that have no necessary logical connexion is also notable. But I mention this feature of Old English morphology here only because the same distinction of functions is associated with similar phonetic forms in Welsh.

    In Welsh one finds a true present without b-forms, and a tense with a b-stem used both as a future and a consuetudinal[21]. The 3 sg. of the latter tense is bydd from earlier *biþ[22]. The resemblance between this and the OE form is perhaps made more remarkable if we observe that the short vowel of OE is difficult to explain and cannot be a regular development from earlier Germanic, whereas in Welsh it is regularly derived.

    This similarity may be dismissed as accidental. The peculiarity of OE may be held to depend simply on preservation in the English dialect of a feature later lost in others; the anomalous short vowel of bist and biþ may be explained as analogical[23]. The OE verb is in any case peculiar in other ways not paralleled by Welsh (the 2 sg. of the true present earþ, later eart, is not found outside English). It will still remain notable, none the less, that this preservation occurred in Britain and in a point in which the usage of the native language agreed. It will be a morphological parallel to the phonetic agreement, noted above, seen in the English preservation of þ and w.

    But this is not the full story. The Northumbrian dialect of Old English uses as the plural of tense B; the form biþun, bioþun. Now this must be an innovation developed on British soil. Its invention was strictly unnecessary (since the older plural remained sufficiently distinct from the singular), and its method of formation was, from the point of view of English morphology, wholly anomalous[24]. Its similarity (especially in apparent relation to the 3 sg.) to Welsh byddant is obvious. (The still closer Welsh 1 pl. byddwn would not have had, probably, this inflexion in Old Welsh.)

  4. Tolkien's footnotes:

    [21] The association of these two dissimilar functions is again notable. Old Irish uses b-forms in these two functions, but distinguishes between future and consuetudinal in inflexion. The Welsh tense (byddaf &c.) as a whole blends the two functions, though the older language had also a form of the 3sg bid (bit) limited to consuetudinal use. The difference of functions is not yet fully realized by Anglo-Saxon scholars. The older dictionaries and grammars ignore it, and even in recent grammars it is not clearly stated; the consuetudinal is usually overlooked, though traces of it survive in English as late as the language of Chaucer (in beth as consuetudinal sg. and pl.)

    [22] The Irish, Welsh, and English forms relate to older bí, bij (cf. Latin fís, fit, &c.). The development from bij to bið in Welsh is due to a consonantal strengthening of j which began far back in British. When ij reached the stage is not known, but a date about A.D. 500 seems probable.

    [23] The influence of the short i in the forms of the true present might be held responsible. In a pre-English stage these would have been , is, ist (is).

    [24] The addition of a plural ending (normally belonging to the past tense) to an inflected form of the 3sg. In this way biðun differs from the extended form sindum made from the old pl. sind. The latter was already pl. and its ending -nd could not be recognized as an inflexion, whereas the -ið of bið was the normal ending of the 3sg.

  5. Some comments on the quotetion from Tolkien:

    One could perhaps just as well make a case for Old English influence on Welsh, as regards the merger of the future of 'to be' with the habitual ("consuetudinal") present. Middle Welsh byd was cognate to OIr. biid, also habitual; both are usually derived from *bʰuH-jé/ó-, which does have Germanic cognates (OE būan 'inhabit') as well as Italic ones (Lat. fīō, Oscan fiíed). If the merger of the two functions in Welsh is relatively recent, could it really have influenced Old English? One also wonders why such a powerful effect (the duplication of the whole present-tense paradigm of the basic substantive verb) should have been caused by contact with a language which otherwise did not influence Old English significantly (judging e.g. from the low count of lexical borrowings).

    As regards Anglian bi(o)þun, building a new plural stem of a highly irregula verb on the basis of the 3sg. (the most frequent form) is nothing unusual. Why, we have it in Polish: (jesteśmy 'we are' comes from jest 'he is' + śmy (the ending of the 1pl. preterite, but etymologically a free-standing auxiliary, which was itself a reflex of the original 'we are'). Bioþun looks simply like sindon with biþ substituted for the unanalysable part.

    I don't think bist, biþ are analogical. Analogical to what, especially in the 2sg.? Unattested hypothetical pre-OE forms? And why don't we get 3sg. *bis(t)? Tolkien also fails to address the larger West Germanic question. If there was no alternative present there in Proto-West Germanic, what are the b-initials doing in the 1-2sg. of 'to be'?