Grammatical change in Old English strong verbs: early traces of elimination.

ABSTRACT

The original Proto-Germanic consonantal alternations of voiceless and voiced fricatives, generated by the operation of Verner's Law, though slightly modified, were relatively well attested in Old English. They were most regularly preserved in the strong verb paradigm where they emerged as: [thorn] ~ d, h ~ g/w, h ~ ng, s ~ r, as in sni[thorn]an: sna[thorn]: snidon: sniden, ceosan: ceas: curon: coren, teon: teah: tugon: togen. The focus of the present paper is the process of gradual elimination of the effects of Vernerian voicing from the Old English strong verb system. While a wholesale process of elimination must be dated no earlier than the (Early) Middle English period, available data indicate clearly that the tendencies towards the decay of Vernerian alternations can be traced back to Old English. A close examination of the Old English textual evidence is intended to capture and reveal peculiarities and tendencies which charasterised this very early stage of elimination.

1. Introductory remarks

The original Proto-Germanic consonantal alternations of voiceless and voiced fricatives, induced by the operation of Verner's Law, had been considerably obliterated by the time of the earliest attestations of Old English, yet their modified reflexes were relatively well preserved and systematically displayed in the Old English strong verb paradigm. (1) The modifications, taking place on the way from Proto-Germanic to Old English, involved various phonological developments pertinent to different stages within this period. They included rhotacism, characteristic of the whole Northwest Germanic subbranch, the West Germanic voicing of the dental fricative *[theta] and its subsequent occlusion in clusters with a nasal and a liquid (l[theta] > ld, n[theta] > nd), as well as voicing of medial voiceless fricatives (f, [theta], s), and finally, loss of the voiceless velar fricative *[chi] in medial position. All of these processes occasioned the emergence of a new pattern of alternations in Old English with the following alternating pairs: [thorn] ~ d, h/o ~ g, h/o ~g/w, h/o ~ ng and s ~ r. The ensuing pattern, viewed against the original Germanic proto-alternants, is presented in Table 1.

The emergence of these alternations in the shape in which they surface in Old English entailed the following phonological developments:

a) [thorn], [eth] ~ d

The change *[eth] > d took place already in the West Germanic stage and was unconditioned; the fricative in the cluster *l[theta] was very early occluded (West Germanic) and the sequence appeared as ld in the very early stage, e.g., PGmc. *fal[thorn]an- (Go. fal[thorn]an) > OE faldan, fealdan 'fold', OE wilde 'wild' (Go. wil[thorn]eis). Probably the earliest shift, dated as early as the Proto-Germanic period, was the occlusion of *[eth] in the neighbourhood of a nasal *n[eth] > nd, a process which affected also other voiced fricatives ([beta], [gamma], [[gamma].sup.w]) in medial position after the corresponding nasal, e.g., OE findan vs. Go. fin[thorn]an, OE (Go., OS) bindan 'bind' vs. Skt. bandhanam 'a binding' (< *bhendh-). The voiceless *[theta] remained a fricative and, just like the other voiceless fricatives in Old English was subject to voicing in voiced environment /[theta]/ > [[eth]]. Finally, in word-final position the voiced fricative underwent devoicing.

b) h/[phi] ~ g

Although the reflex of the original Proto-Germanic voiced fricative *[gamma] was Old English <g>, it was used to represent either a fricative [gamma], continuing the Proto-Germanic value (e.g, slogun, flogon, belagen, getogen) or a palatal fricative, occasionally made distinct in spelling and marked as <g> (e.g., betigen, [thorn]igen, [thorn]wegen, forsleginum). The voiceless velar fricative /[chi]/ was lost in a few contexts, among others, in the intervocalic position, relevant hem, which yielded the alternation with zero (e.g., *flea[chi]an > flean 'flay'). (2) This velar alternation is present in all verba contracta where the infinitive no longer preserves the voiceless fricative, e.g., slean < */sla[chi]an/, [thorn]eon < */[theta]i[chi]an/, [thorn]wean < */[theta]wa[chi]an/.

c) h/[phi] ~ ng

The nasal (phonetically [[eta]]) preceding the voiceless spirant in the original alternation which surfaced as Proto-Germanic *n[chi] ~ *n[gamma] was lost probably still in Proto-Germanic stage and is no longer evident in Old English. The voiced velar fricative /[gamma]/, when following a nasal, was occluded already in Proto-Germanic, hence OE ng in place of PGmc. *n[gamma]. Traces of the original alternation are preserved in a few Old English contracted verbs, e.g., [thorn]eon (<*/[theta]i[chi]an/ < */[theta]i[eta][chi]an/) : [thorn]ungon : [thorn]ungen, befon : befengon : befangen/befengen.

d) h/[phi] ~ g/w

The voiceless labiovelar fricative *[chi]w lost its rounding in root final position at some point, probably as early as the Northwest Germanic stage, and appeared in Old English as h or was eventually dropped. The voiced labial alternant was reflected as labial approximant w or as unrounded velar fricative [gamma], depending on the following vowel: [gamma] before PGmc. u and w before i, which yielded two new alternating sequences: h ~ [gamma] and h ~ w respectively. The only (strong) verb in Old English which preserved traces of the labiovelar alternation is seon (< *se[[chi].sup.w]an) and its related prefixed verbs with the forms siwen, sigen (< se[[gamma].sup.w]-ana-). (3)

e) s, z ~ r

The Proto-Germanic alternation *s ~ *z was very early (NWGmc.) subject to rhotacism whereby *z became r medially (but was dropped in word-final position) and was preserved in Old English as the alternation of s ~ r. (4) Affected by voicing in medial position, OE s surfaced as [z] in voiced environment, which yielded, in a sense, a triple alternation between s/z and r.

f) *f ~ *b

The labial series *f ~ *b became very early obliterated and was no longer distinct in the Old English strong verb paradigm. Given the evidence provided by the other Germanic dialects (cf. OS (af)heffian : hof: ho[beta]un : giha[beta]an, OHG heffen, hevan : huob : huobum : (ir-)haban), the alternation is to be expected at least in hebban 'heave' (PGmc. *[chi]a[beta]io > OE hebbu, hebbe). Old English hebban, however, is preserved with the reversed distribution of alternants: the voiceless consonant generalised in the preterite and the voiced one in the present system, consequently without a trace of the original alternation: hebban : hof : hofon : hafen, hoefen. Further development of the sequence involved voicing of PGmc. *f (< PIE *'-p-) in voiced environment, yielding a voiced labial fricative [beta] which did not develop into a stop but remained a fricative and as such was soon to merge with PGmc. *[beta] in this position. At the same time PGmc. *[beta] (< PIE *bh or -p-') was subject to devoicing in word-final position and appeared as f. When the voiced labiovelar fricative [v], produced by medial voicing, merged with *[[beta]] (as [v]), a new phonemic contrast arose, namely /f/ vs. /v/.

To some extent, the opposition was preserved in the early Mercian material, in Epinal and Erfurt Glossaries (dated no later than 700) and Corpus Glossary (late 8th century) (5) where the voiceless and voiced variants were made distinct in spelling: <f> was used to represent [v] (in voiced environment) and <b> represented [[beta]], the former--a labiodental allophone of /f/, the latter--an allophone of /b/ which only later developed into a labiodental fricative [v], e.g., fifadae 'butterfly', geroefa(n) 'reeve', uulfes 'wolf' vs. bebr, later beofor 'beaver' (< PGmc. *bebru- [be[beta]ru-]), obaer (later ofer) 'over' (< PGmc. *uberi [u[beta]efi], fibulae (Li. fibula > fifele) 'handle', gebuli (geabuli) 'tribute', halbae 'halves', gibaen 'given'. Already at that time a tendency to use <f> in all positions became steady. Levelling in favour of <b> took place only sporadically, e.g., glo(o)b 'glove', raebsid 'reproved' (Hogg 1992: 283).

The differentiation between PGmc. *f (< *'-p-) and *[beta] (< *-p-', *bh) is evident in consonantal clusters in Old English where *[beta] developed into a stop b already in PGmc. or West Germanic period, bringing about alternations of PGmc. *-mf- : *-mb- > WGmc. *-mf- : *-mb- > OE -Vf- : -mb- (fif vs. climban); and similarly, PGmc. *-f(i)j- > *-[beta](i)j- > WGmc. *-ffj- : *-bbj- > OE -ff- : -bb- where geminated ff derives from PGmc. *f (<*'-p-) and geminated bb from PGmc. *[beta] (< *bh) (Go. hafjan vs. OE hebban 'heave') (6) (Brunner 1942: 156, 160).

2. Verner's Law and the Old English strong verb paradigm

The distribution of the vestiges of the original Vernerian voiceless and voiced alternants in Old English strong verb paradigm entailed the presence of the old voiceless fricative in the infinitive and 1, 3sg. preterite ind. and the reflex of the voiced fricative in the rest of the preterite system, i.e. 2sg. preterite ind., preterite ind. plural, preterite subjunctive singular and plural, and past participle. Such pattern of distribution of Vernerian alternants emerged as a result of the accentual conditions characteristic of Proto-Germanic, whereby forms of the infinitive and preterite singular displayed root accentuation, whereas forms of the preterite plural, past participle and subjunctive preterite received suffixal accent. Noteworthy is the presence of the vestiges of Verner's Law in forms of the 2sg. preterite ind. which, apparently, alongside the root vowel of the preterite pl., adopted its consonantism. A similar situation obtains for subjunctive preterite which was formed by employing the stem of preterite plural. Hence, the voiced alternants induced by Verner's Law may be expected in both subjunctive pret. singular and plural. (7)

The infinitive of all strong verbs containing an originally (Proto-Germanic) voiceless fricative reflects the early Old English process of voicing of intervocalic fricative following a stressed vowel, whereby medial /f, [theta], s/ developed into [v, [eth], z] in voiced environment. The change did not affect the velar fricative /[chi]/ which by that time was no longer preserved in this position. The relevant examples are: li[thorn]an, sni[thorn]an, seo[thorn]an [[eth]], dreosan, hreosan [z], etc.

In terms of class division, grammatical change is present in all classes of Old English strong verbs, with the exception of class IV where none of the verbs contained stem-final fricative which could potentially undergo voicing by Verner's Law. It is worth noticing that the first three classes are characterised by remarkable regularity and discipline with respect to grammatical change. This marked consistency detected in these classes is attributed to accentual pattern which must have been stable at the time of the operation of Verner's Law (D'Alquen 1988: 90). In the remaining classes (i.e. V, VI and VII) grammatical change is not accent-dependent but tan be viewed as a result of "analogical transfer" from the other classes, hence is less regular and sporadic (Prokosch 1939: 64). Accordingly, forms such as waeron, cwaedon, slogon, fengon, etc. can be viewed as analogical rather than originally induced by the accentuation pattern. (8)

3. Elimination of the effects of Verner's Law

3.1. The nature of the process

The original consonantal alternations tended to be obliterated through the influence of analogical levelling--a process of morphological simplification, which worked towards introduction of one single root consonant in all forms within the strong verb paradigm. As a result of such generalisation, the allomorphy rendered by the operation of Verner's Law was being gradually removed. (9)

The interpretation of the data does, in fact, depend to a large extent on how the term is understood. In the present analysis it is used in line with the above definition to mean restoration of the original Proto-Germanic voiceless variant and consequent generalisation of this variant to all forms which displayed the effects of the operation of Verner's Law. The definition of elimination can be extended however to include some less regular cases to the effect that forms such as Anglian past participles in classes VI and VII (such as befoen, geseen) can be interpreted as instances where Vernerian alternations were lost. Such interpretation is justified on the assumption that the mere absence of the voiced alternant in these forms is enough to view it as a case of elimination. In the present analysis these cases will be treated separately and will not be counted as instances of elimination.

3.2. The database and data analysis

The data for the analysis of the material come from The dictionary of Old English electronic corpus (known as Toronto Corpus), a collection spanning the period between 600 and 1150, considered to be a complete record of surviving Old English, with the exception of a few manuscripts of individual texts. An Anglo-Saxon Dictionary by Bosworth and Toller (1898) served as the main source for identifying the principal parts of strong verbs liable to voicing by Verner's Law. Forms of the prefixed verbs were drawn from the Dictionary as well and, if not specified there, were systematically built on the pattern of the simplex verbs. In this way a list of 211 verbs which could potentially display Vernerian alternations was compiled. In the effect of search procedure, the list was narrowed down to 149 verbs since some of the prefixed verbs were not attested in the corpus in forms which could testify to the earlier operation of Verner's Law. (10) The etymological dictionary of Germanic strong verbs by Seebold (1970) (Vergleichendes und etymologisches Worterbuch der germanischen starken Verben) was consulted to verify the Proto-Germanic root forms which were susceptible to the operation of Verner's Law and could be reflected in Old English.

3.2.1. Strong verbs Class I

In verbs belonging to Class I the alternations induced by Verner's Law are well preserved and very regular. All alternating pairs are represented, including the sequence h/[phi] ~ ng, preserved in forms of wreon and [thorn]eon (wrungen, [thorn]ungen), modelled after S[V.sub.2]. In a number of Old English strong verbs elimination of the effects of Verner's Law took place very early, before the date of their earliest attestation. They form a separate group of verbs in the sense that they never showed the morphophonemic alternations induced by Verner's Law. All of these generalised the old voiceless fricative, extending it to all forms which would have otherwise displayed effects of voicing. They are scattered across various classes but are most numerous in Class I, including: mi[thorn]an, bemi[thorn]an 'conceal, avoid', risan 'rise', arisan 'arise', wri[thorn]an, gewri[thorn]an 'twist' and aetcli[thorn]an 'adhere'. The principal parts of these verbs in Old English as well as examples of contexts in which they were used are presented below:

   (aet-)cli[thorn]an : -- : -- : --
   mi[thorn]an : ma[thorn] : mi[thorn]on : mi[thorn]en
   bemi[thorn]an : bema[thorn] : bemi[thorn]on : bemi[thorn]en
   risan : ras : rison : risen
   arisan : aras : arison : arisen
   gerisan : geras : gerison : gerisen
   wri[thorn]an : wra[thorn] : wri[thorn]on : wri[thorn]en
   gewri[thorn]an : gewra[thorn] : gewri[thorn]on : gewri[thorn]en

The following sentences illustrate the use of these levelled forms in Old English:

1) bemi[thorn]an

Ic his word oncneow, [thorn]eh he his maegwlite bemi[eth]en haefde. (And: 855)

2) mi[thorn]an

   Da hie [thorn]a us gesawon hie selfe sona in heora husum deagollice
   hie mipan [eth]a wilnade ic [thorn]ara monna onsyne to geseonne,
   [thorn]aet hie us fersc waeter & swete getaehton. (Alex: 15.5)

3) arisan

   et quando exortus est sol exaestuauit eo quod non haberet radicem
   exaruit & [eth]a arisen waes 7 [eth]a uppeode waes sunna gedrugade 7
   forbernde for[eth]on naefde wyrtruma gedrugade. (MkGl (Li): 4.6)

   and paer laegen swylce we deade waeren, and we geherdan [thorn]one
   aengel cwe[eth]en to [thorn]an wifen [thorn]aet Godes sune waere of
   dea[eth]e arisen. (Nic (C): 46)

   [thorn]u gedydest me under[thorn]eodde [thorn]a [thorn]e wi[eth] me
   upp arison, and minra feonda baec [thorn]u onwendest to me, and me
   hine gesealdest, and [thorn]u tostenctest [thorn]a [thorn]e
   me hatedon. (PPs (prose): 17.38)

4) gerisan

   [THORN]a waes he semninga from deofle gerisen, & ongon cleopian &
   hlydan & mid his to[eth]um gristbitian, & [thorn]a faam of his
   mu[eth]e ut eode, & he missenlecum styrenessum ongon his leomu
   [thorn]raestan. (Bede 3: 9.184.24)

5) risan

   On [thorn]is kinges time wes al unfri[eth] & yfel & raeflac. for
   agenes him risen sona [thorn]a rice men [thorn]e waeron swikes.
   (ChronE (Plummer): 1135.18)

6) gewri[thorn]on

   [THORN]a deoflu [eth]a gecyrdon ongean to [eth]am drye. and
   gewri[eth]on his handa to his hricge, and swa laeddon to [eth]am
   apostole [thorn]us cwe[eth]ende; (AECHom II, 31-32:
   243.55)

7) wri[thorn]an

   Waeter hine ond eor[thorn]e feddan faegre, o[thorn][thorn]aet he
   frod dagum on o[thorn]rum wear[eth] aglachade deope gedolgod, dumb
   in bendum, wri[thorn]en ofer wunda, wonnum hyrstum foran gefraetwed.
   (Rid 53: 3)

Discernible traces of elimination of Vernerian alternations appear in five verbs: forli[thorn]an, li[thorn]an, scri[thorn]an, sni[thorn]an and ofsni[thorn]an, where levelling affected primarily preterite plural and past participle forms, extending the voiced fricative /[eth]/, reminiscent of the original voiceless fricative, to these forms. In all mentioned verbs the process affected the dental alternation [eth] ~ d which is definitely the most frequent alternating set in this class. Other present alternations remained intact, staying resistant to the encroaching working of levelling processes.

Interestingly, a few verbs belonging to this class in some forms eliminated the effects of Verner's Law completely. These are: forli[thorn]an, li[thorn]an and scri[thorn]an. The only attested preterite forms of li[thorn]an are li[thorn]ion (possibly li[thorn]eon), appearing in place of the expected lidon. A similar situation obtains for forli[thorn]an which is attested without traces of Verner's Law not only in preterite plural but also in subjunctive plural (in both cases the attested form is forli[thorn]an, instead of the expected forliden). Another case is the past participle form of scri[thorn]an, attested only as scri[thorn]en in place of the expected scriden. All of these forms indicate straightforwardly that the levelling process must have been in progress. The expected past participle form of li[thorn]an, *liwen, with traces of the labiovelar alternation, is not attested in the investigated material; neither are any of the related prefixed forms such as beliwen, forliwen, the expected past participle forms of beli[thorn]an, forli[thorn]an.

Table 2 (facing page) demonstrates distribution of the effects of Verner's Law in verbs belonging to Class I. The symbols P and L stand for preserved and levelled and refer to the number of occurrences and percentage of forms in which Verner's Law was preserved and eliminated respectively.

Instances of forms which eliminated the effects of Verner's Law in Class I were round in the following contexts in the investigated material.

8) li[thorn]an

   [ETH]a aet nehstan se foresprecena cyning self, & se halga biscop
   Trumwine mid him & monige o[thorn]re aefeste weras & rite li[eth]on
   on [eth]aet ealond. (Bede 4: 29.368.9)

   [eth]a we [eth]a waeron on midre [eth]aere sae, [eth]a waes somninga
   hiofones smyltnes tosliten, [eth]aere [eth]e we aer lio[eth]on uut;
   ond swae micel winter us onhreas. (11) (Bede 5: 1.384.18)

[THORN]a eodon hi ealle gesunde of [thorn]am scipe, [thorn]a [thorn]e li[thorn]on & foron mid [thorn]am forecwedenan Maximiane [thorn]am arwyr[eth]an were. (GDPref and 3 (C): 36.249.13)

9) oferli[thorn]an

   Ac [thorn]ylaes ic lenge [thorn]one [thorn]anc hefige [thorn]ara
   leornendra mid gesegenum [thorn]ara fremdra taelnysse, swa swa <ic>
   strange sae and mycele oferli[eth]e, and nu becume to [thorn]aere
   smyltestan hy[eth]e Gudlaces lifes. (LS 10.1 (Guth): 0.31)

10) forli[thorn]an

   And se ilce Nathan waes forli[eth]en, [thorn]aet he waes gefaren
   fram aelcen lande to odren and fram sae to sae, swa [thorn]aet he
   haefde ealle eor[eth]e gemaeren [thorn]urhfaren.
   (VSal 2 (Ass 17): 9)

   naufragauerunt forli[thorn]an 7 forferdan. (AldV 1 (Goossens): 4368)
   naufragarent forli[thorn]an 7 forferdan. (AldV 1 (Goossens): 4501)

   naufragauerant forli[thorn]an, forferdon. (AldV 13.1 (Nap): 4490)
   naufragarent forli[thorn]an, forferdan. (AldV 13.1 (Nap): 4621)

11) ofsni[thorn]an

   [THORN]a namon hig an tiecen & ofsni[eth]on hit & bedipton his
   tunecan on [thorn]am blode. (Gen (Ker): 37.31)

   & willa[eth] mid [eth]y gedon [eth]aet hie mon hlige wisdomes, mid
   [eth]y [eth]e hie ofsni[eth]en mid [eth]y seaxe hefiglices gedwolan
   [eth]a unborenan bearn, [eth]e [eth]onne fur[eth]um beo[eth] mid
   wordum geeacnode on geleaffullra mode; (CP: 48.367.14)

12) scri[thorn]an

   [THORN]onne dogor beo[eth] on moldwege min for[eth] scri[thorn]en,
   sorg <geswe[eth]rad>, ond ic si[thorn][thorn]an mot fore meotudes
   cneowum meorda hleotan, gingra geafena, ond godes lomber in
   sindreamum si[thorn][thorn]an awo for[eth] folgian; (Guthlac: 1038)

13) sni[thorn]an

And hundeahtatig [thorn]usenda hine sni[eth]on, & feidon. (Notes 12.1 (Nap): 14)

Figures in Table 3 show overall distribution of forms which preserved and eliminated grammatical change in Class I with respect to particular inflectional categories. The figures indicate that the tendency towards levelling of the Vernerian alternations is most pronounced in forms of subjunctive pret. pl. and preterite ind. pl. Slight hesitancy can be detected in past participle forms attested in two verbs without the voiced alternant. Most conservative, accordingly, are the forms of preterite sg. and subjunctive pret. sg., where none of the verbs shows signs of levelling.

The forms identified as displaying traces of elimination are scattered across various texts belonging to West Saxon dialect. Most of them come from texts dated to the 10th c. and 11th c., especially Glosses on works by Aldhelm, Vindicta Salvatoris (11th C.), and earlier traces in Guthlac, Pastoral Care and Bede's History of the English Church and nation.

3.2.2. Strong verbs Class II

In Class II, as could be expected, grammatical change was characterised by a considerable degree of regularity. The effects of Verner's Law are no longer seen in breo[thorn]an 'ruin, destroy' and abreo[thorn]an 'ruin, perish' which generalised the voiceless fricative in all forms:

   breo[eth]an : bread : bru[eth]on : bro[eth]en
   abreo[thorn]an : abru[thorn]on : abro[thorn]en

The sentences below illustrate the use of forms in which the effects of Verner's Law had been eliminated before the attestation date.

14) abreo[thorn]an

   An her ys OMNIS GENERIS, [thorn]aet is aelces cynnes: hic et haec
   et hoc nugas, [thorn]aet is abro[eth]en on englisc and ungebigendlic
   on declinunge. (AE Gram: 51.4)

   Eac hwilon by[eth] geset NOMINATIVVS for VOCATIVVM, swaswa
   LVCANVS cwae[eth]: <degener> o populus eala [eth]u abro[eth]ene
   folc. (AE Gram: 32.8)

   [THORN]a Ulfkytel [thorn]aet undergeat, [thorn]a sende he [thorn]aet
   mon sceolde [thorn]a scipo toheawan, ac hi abru[eth]on [thorn]e de
   he to [thorn]ohte, and he [eth]a gaderede his fyrde digolice swa
   he swy[eth]ost mihte. (ChronD (Classen-Harm): 1004.8)

15) breo[thorn]an

   [THORN]a efston lat[thorn]eowas & fromrincas gegrap hy fyrhto
   bru[eth]un ealle eardigende Tunc festinauerunt duces edom et
   principes mohabitarum adprehendit eos tremor, Tabuerunt omnes
   inhabitantes chanaan. (PsCaD: 5(4).15)

No occurrences of forms which could testify to the earlier operation of Verner's Law in the non-prefixed freosan 'freeze' were round in the analysed material. The same situation obtains for the forms of begreosan, attested only in the present. The single past participle form of this verb, found in Table 4, was provided by Bosworth and Toller's Dictionary (1898) in the following context:

16)

   Atole gastas sfisle begro-rene [MS. begrorenne] the horrid spirits
   fearfully overwhelmed with torment. (Cd. 214; Th.268: 9)

The single occurrence of onahruron (12) was counted together with onhreosan 'assail, attack' and the doubly attested inhruron together with hreosan 'fall'. The past participle forms with the prefix i-: icoren, icorene and icorenae were analysed and treated together with the past participle forms of ceosan. The two past participle forms ofroren and ofrorene, attested in AIdV, were eventually recognised as belonging to ofhreosan 'overwhelm'. (13)

Instances of elimination of Vernerian alternations are limited to three verbs only: ceosan, geceosan, beseo[thorn]an and appear in all forms in which they could be expected except for subjunctive pret. sg. The verb beseo[thorn]an is attested once without traces of Verner's Law in past participle, next to the regular besoden. In ceosan elimination affected forms of preterite plural and past participle, where, along with the expected curon and coren, cuson and cosan are attested. Apparent traces of levelling are present also in geceosan in Northumbrian texts (Lindisfarne Gospels) where the preterite plural form geceason and pret. subjunctive singular gecease (occurring along with gecuron, gecure) are attested. Characteristic of these forms is also extension of the vowel proper to 1, 3 sg. pret. in eoand u- presents which in this particular case may have been partly responsible for the extension of consonantism as well. It may have been the case that the loss of Vernerian alternantions in these forms was to some extent triggered by the developments in the vocalic system of this verb, i.e., once the vowel from the 1, 3 sg. preterite indicative was extended, the voiceless consonant followed suit. Table 4 (next two pages) presents distribution of the effects of Verner's Law in verbs belonging to Class II.

The following sentences illustrate the use of forms which levelled the effects of Verner's Law in Old English:

17) ceosan

   [ETH]a cusen [thorn]a munecas to abbot Brand prouost. for[eth]an
   [thorn]aet he woes swi[eth]e god man & swi[eth]e wis. & senden him
   [thorn]a to AEdgar ae[eth]eling. (ChronE: 1066.35)

   & [eth]es o[eth]er daeies aefter Sancti Iohannes messedaei. cusen
   [thorn]a muneces abbot of hem self and brohten him into cyrce mid
   processionem. sungen Te Deum Laudamus. ringden [thorn]a belle.
   (ChronE: 1131.20)

   [ETH]a cusen hi an clerc Willelm of Curboil woes gehaten, he was
   canonie of an mynstre Cicc hatte. (ChronE: 1123.32)

   & te munekes innen daeis cusen o[thorn]er of heom saelf. (ChronE:
   1154.12)

   & swa swa hi haefden cosen aercebiscop aeror in here capitele
   aefter. (ChronE: 1123.41)

18) geceosan

   non uos me elegistis sed ego elegi uos et posui uos ut eatis et
   fructum adferatis ne gie mec geceaso ah ic geceas iuh & ic gesette
   iuih [thorn]aette gie geongae & waestem gie gebrenga. (JnGl
   (Li): 15.16)

   Dicebat autem et ad inuitatos parabolam intendens quomodo primos
   accubitos eligerent dicens ad illos cuoe[eth] [eth]a & to [eth]aem
   la[eth]endum l [thorn]aet bisen beheald huu [eth]a formo hraesto hia
   geceason cuoe[eth] to him. (LkGl (Li): 14.7)

19) beseo[thorn]an

   and hi him [thorn]a to comon aerest mid medemum fultume, ac
   si[eth][eth]an hy wiston hu hit baer besu[eth]an woes.
   (ChronD: 1052.1.25)

Figures in Table 5 show overall distribution of forms which preserved and eliminated grammatical change in Class II with respect to particular inflectional categories.

The figures indicate that the process of analogical levelling affected a handful of forms. In terms of individual tokens, the tendency is most pronounced in preterite plural where six occurrences have been recognised as forms which dispensed with the voiced alternant. In terms of percentage value however, preterite, sg. seems to be most susceptible to the influence of analogy, though only two occurrences were attested without Verner's Law. Unaffected by the working of levelling remains subjunctive preterite plural where no levelled forms have been found.

All the forms which testify to the operation of analogical levelling in Class II derive from Late West-Saxon texts (most of them appear in late manuscripts of Anglo-Saxon Chronicle (Chron E), dated to the first half of the 12th century, but also late Northumbrian (Lindisfarne Gospels), from the latter half of the 10th century.

3.2.3 Strong verbs Class III

Although the effects of Verner's Law were displayed with remarkable rigour and systemacity in Class III, they can be identified in very few verbs: the scarcely attested feolan and its prefixed forms, and the very well attested weor[thorn]an with its derivatives where grammatical change was abundantly preserved.

The effects of Verner's Law were levelled very early in findan (*fin[theta]-a-n), i.e. before the time of its attestation. The expected regular forms would have been fi[thorn]an :fo[thorn] :fundon :funden, yet the present and preterite sing. (fand) were created by analogy with verbs like bindan and the cluster nd of the preterite pl. and past participle was extended to the present and preterite singular.

The earliest traces of the demise of grammatical change in Class III appear in weor[thorn]an, geweor[thorn]an and forweor[thorn]an, primarily in past participle but also preterite plural. Weor[thorn]an is one of the few verbs where elimination of the effects of Verner's Law is relatively advanced and can be clearly identified.

The distribution of the effects of Verner's Law in verbs belonging to Class III is presented in Table 6 (next page). The presentation of contexts illustrating early traces of elimination in this class follows.

20) weor[thorn]an

   Hit seg[eth] [thorn]aet hi wur[eth]on raedlice afyrhte [thorn]a ure
   drihten com on [thorn]as niht to [thorn]aere hellegatum & [thorn]a
   locu toburston. (Nic (E): II)

   [THORN]a twegen kyngas Willelm & Swaegn wur[eth]on saehtlod.
   (ChronE: 1070.45)

   [THORN]a wundreden heo ealle and wur[eth]en afyrhte. (Nic (C): 43)

   [THORN]a wur[eth]en heo ealle swy[eth]e sarige and afyrhte. (Nic
   ((2): 93)

   And on [thorn]ine dea[eth]e ealle tunglen and gesceafle wur[eth]en
   gestyrede. (Nic (C): 318)

   [ETH]a wur[eth]en [thorn]a twegen cnapan sona on slaepe, & se
   [thorn]ridda wacode <swi[eth]or> for ege [thorn]onne for his
   gebedum. (Leof: 45)

   Ond <[thorn]us> hit is iwriten on holie wisdome, Fiat, et facta sunt
   omnia, he seide, <Iwur[thorn]e>, <ond> alle bing iwor[thorn]en.
   (HomU 5.6 (Buch F): 43)

   [THORN]us mid one worde al hit was iwur[thorn]en. HomU 5.6
   (Buch F): 46)

   [THORN]a hi ne leng ne muhten [thorn]olen [thorn]a stali hi ut &
   flugen. & hi wurthen war widuten & folecheden heom. (ChronE:
   1140.31)

   & se eorl com mid him. & wur[eth]on [thorn]a alswa gode freond swa
   hi waeron aeror feond. (ChronE (Plummer): 1129.5)

21) forweo[thorn]an

   We wiscea[eth] [eth]aet we on Egypta lande waeron aer deade & na on
   [eth]isum westene, & we wiscea[eth] swy[eth]or [thorn]aet we
   forwur[eth]on her & us Drihten ne laede in to [eth]am lande,
   [thorn]aet we [eth]aer licgon ofslagene & ure wif & cyld wur[eth]on
   gehergode; (Num: 14.1)

   Sethes sunes yherden adames wytegunge be twam domon & [thorn]aet
   [thorn]a yfudonne creftes ne forwur[thorn]on writen hi on twam
   columban; (HeptNotes: 15)

   Hu gewordene beo[eth] on forletnesse ferlice hie terogoden &
   forwur[eth]on fore unrihtwisnesse here Quomodo facti sunt in
   desolationem subito defecerunt et perierunt propter iniquitates
   suas. (PsGIE: 72.19)

22) geweorpan

   Ne ondred [eth]u [eth]onne weolig gewor[eth]en bi[eth] mon &
   [eth]one gemonigfaldad bi[eth] wuldur buses his Ne timueris cum
   diues factus fuerit homo et cum multiplicata fuerit gloria domus
   eius. (PsGIA: 48.16)

   & na ic ondraede forpon streng[eth] min & lof min drihten &
   gewor[eth]en is me on haelo et non timebo quia fortitudo mea et laus
   mea dominus et factus est mihi in salutem. (PsCaF: 2(1).2)

   Ealla onhyldon somed unnytte gewor[eth]ene syndon ne his se [thorn]e
   de[thorn] god ne his o[eth][thorn]e on aenne Omnes declinaverunt
   simul inutiles facti sunt non est qui faciat bonum non est usque ad
   unum. (PsGIE: 52.4)

3.2.3.1 The case of feolan (< * fel[chi]an)

The preterite plural form fulgon was in fact rare, instead in West Saxon the form faelon tended to spread, and so did fulon and -felun (attested only twice as aetfelun (VP)). (14) Both -felun and the past participle folen (also folgen) were formed on the pattern of S[V.sub.4]. The past participle forms -fulen, folen, faelon, scattered across various texts, are the only forms of this verb, which exhibit no traces of Vernerian alternations. In fact, the status of these forms is quite ambiguous: although they show no alternations, they certainly cannot be attributed to analogy, working towards generalisation of one variant and restoration of the voiceless fricative. They can be viewed instead as effects of some individual development, entailing liability to adopting a different ablaut pattern, in this particular case, the pattern of S[V.sub.4] The following forms are attested without Vemerian alternants:

23) aetfelun (2 occurrences)

   Unsce[eth][eth]ende & rehtwise oetfelun me for [eth]on ic arefnde
   [eth]e dryhten Innocentes et recti adheserunt mihi, quoniam
   sustinui, te domine. (PsGIA: 24.20)

   ... & adrugao heorte min for oon offergeotul le eam eotan hlaf mine
   from stefne geamrunge mime oetfelun 7 aetclofodon ban rein flaesce
   minum ... et aruit cor meum quia oblitus sum manducare panem meum a
   uoce gemitus mei adheserunt ossa mea carni meae. (PsGIA: 101.4)

24) bi/befolen (5 occurrences)

   Waes him niwe gefea befolen in fyrh[eth]e, woes him frofra maest end
   <hyhta> nihst heofonrices weard. (El: 195)

   Him [eth]a gleawhydig ludas onewae[eth], haele[eth] hildedeor, him
   woes halig gast befolen faeste, fyrhat lufu ... (El: 934)

   Se maeg eal fela singan ond secgan [thorn]am bi[eth] snyttru craeft
   bifolen on fer[eth]e. (Christ A, B, C: 666)

   Ic [thorn]aere sawle ma geornor gyme ymb [thorn]aes gaestes forwyrd
   [thorn]onne [thorn]aes lichoman, se [thorn]e on legre sceal
   weor[eth]an in worulde wyrme to hro[thorn]or, bifolen in foldan.
   (Jul: 413)

Table 7 presents overall distribution of Vernerian alternations in Class III.

Figures in Table 7 illustrate the pattern of distribution of forms which eliminated and preserved Vernerian alternations in this class. The pattern indicates that the forms of subjunctive pret. p1. and preterite ind. plural were most susceptible to the operation of analogical levelling. Affected were also forms of past participle, whereas no, even slight traces of levelling were round in subjunctive plural and 2sg. preterite ind.

3.2.4. Strong verbs Class V

Regular grammatical change in Class V can be identified in four simplex verbs: seon, feon, wesan and cwe[thorn]an. Out of these, in turn, the tendency towards levelling of Vernerian alternations is most pronounced in cwe[thorn]an. Pre-Old English levelling affected four verbs in this class: fnesan 'sneeze' (no attestations found in the corpus), lesan 'collect, lease', alesan 'choose', (ge)nesan 'survive', in all of which the voiceless fricative was generalised:

   genesan : genaes : genaeson : gensen
   lesan : laes : laeson : lesen
   alesan : alaes : alaeson : alesen

Following are contexts in which they were round in the investigated material:

25) genesan

   Swa he ni[eth]a gehwane genesen haefde, sli[eth]ra <geslyhta>, sunu
   Ecg[eth]iowes, ellenweorca, o[eth] [eth]one anne daeg [thorn]e he
   wi[eth] [thorn]am wyrme gewegan sceolde. (Beo: 2397)

   Of [thorn]am him aweaxeo wynsum gefea, [thorn]onne hi [thorn]aet
   yfel geseo[eth] o[eth]re  dreogan, [thorn]aet hy [thorn]urh miltse
   meotudes genaeson. (Christ A, B, C:
   1252)

   AEfter [thorn]aem for Hannibal ofer Bardan [thorn]one beorg,
   [thorn]eh [thorn]e ymb [thorn]one tieman waeren swa micel
   snawgebland swa [thorn]aette aeg[thorn]er ge [thorn]ara horsa
   fela forwurdon ge [thorn]a elpendas ealle buton anum, ge [thorn]a
   men selfe unea[eth]e [thorn]one ciele genoeson. (Or: 4 8.100.9)

26) lesan

   and hig foron towerd Sandwic, and loeson aefre for[eth] mid heom
   ealle [thorn]a butsecarlas [thorn]e heo gemetton, and comon [thorn]a
   to Sandwic mid geotendan here. (ChronC (Rositzke): 1052.16)

27) alesan

   Haefde him alesen leoda duge[eth]e tireadigra twa [thorn]usendo,
   [thorn]aet waeron cyningas and cneowmagas, on [thorn]aet <ealde>
   riht, ae[eth]elum deore. (Ex:
   183)

   ... vsig on dea[eth] rod' lvstvme gisaldest gilef allvm vs rehtlice
   gilefendvm [thorn]aette [eth]erh [thorn]aet ilca dea[eth] from deaoe
   ecvm ve sic alesen. (DurRitG1 1: (94.10)

Class V is most numerous with respect to forms which levelled effects of Verner's Law and at the same time not much diversified in the sense that in fact all of the forms affected by levelling belong to one simplex cwe[eth]an and its prefixed derivatives. The data allow for identification of 39 forms of preterite plural which no longer preserve the voiced alternant induced by the operation of Verner's Law. Single past participle and preterite subjunetive sg. forms without Verner's Law can also be identified. Liable to the working of levelling was also Gecwe[eth]an which showed discernible traces of the process in past participle and sporadically in preterite plural and subjunctive pret. singular. Very sparse traces of levelling are present in acwe[eth]an, becwe[eth]an and tocwe[eth]an which levelled Verner's Law in single forms of past participle and preterite pl. (acwe[eth]on, Becwe[eth]on, tocwe[thorn]on). Finally, slight tendency towards elimination can also be detected in wi[thorn]cwe[thorn]an where the voiceless fricative is extended to forms of past participle (wi[thorn]cwe[thorn]en).

Given that seon is the only verb in Old English, which testifies to the labiovelar alternation of *[chi]w ~ *[gamma]w, as well as given the diversity of forms in which the verb was attested, it deserves a brief comment. The West Saxon regular preterite plural form of seon was sawon and in Anglian segun, -on, the latter often "West-Saxonized" as saegon ([Ru.sup.1] -segon, -saegon, -sagun). The past participle form sawen adopted a from preterite plural; sewen, an alternative past participle, remains in accordance with the regular vocalism of S[V.sub.5]. The Anglian past participle form was gesegen with the consonantism adopted from the preterite and consistently the form of Anglian subjunctive pret. sg. was sege. (15) The loss of medial g is evident in some Northumbrian texts, namely in forms such as gesene ([Ru.sup.1], [Ru.sup.2], Li.) (in fact an adjective employed often as past participle), geseanae, gesaenae (Rut), geseen (Li.). Again these forms preserve no voiced alternant reminiscent of the earlier operation of Verner's Law; still they cannot be viewed as the effect of analogical levelling but rather as products of dialect-specific phonological development (cf. section 3.3).

The distribution of the effects of Verner's Law in verbs belonging to Class V is demonstrated in Table 8 (facing page), followed by the presentation of contexts in which forms without Vernerian alternants were found.

Following are contexts illustrating forms which eliminated the allomorphy brought about by the operation of Verner's Law:

28) acwe[thorn]an

   Waerun wuldurlice wi[eth] [thorn]e wel acwe[eth]ene, [thorn]aet pu
   si cymast ceastra drihtnes; eac le gemyndige [thorn]a maeran Raab
   and Babilonis begea gehwae[eth]eres. (PPs A5: 86.2) intermediate
   saxon

29) cwe[thorn]an

   [ETH]a cwae[eth]on his apostolas, drihten, geic urne geleafan.
   (Lk (WSCp): 17.5)

   [THORN]a halgan sawla [thorn]a mid unasecggendlicum gefean cleopodan
   to Drihtne, & [thorn]us cwoe[thorn]on. (HomS 26 (BIHom 7): 83)

   And syd[eth]on [thorn]aet fers & geseald [thorn]are abbodesse
   bletsunge, beon geared  o[thorn]ere feower raedunge of [thorn]are
   niwe cy[eth]nesse [thorn]are aenbyrdnesse, [thorn]e we before
   cwe[rth]on. (BenRW: 11.47.8)

   Se ste[eth]e is ycwae[eth]en si denae of taeran besyde hebron
   [thorn]e adam & eue wypen. (HeptNoles: 8)

   [THORN]a wuldorfestan cwe[thorn]ene sien be [eth]e ceaster godes
   Gloriosa dicta sunt de te civitas dei. (PsGIE: 86.3)

   et respondentes dicunt iesu nescimus respondens iesus ait illis &
   ondueardon cuoe[eth]on [eth]aem haelende neutu woe geonduearde se
   haelend cuoe[eth] to him (MkGI (Li): 11.33)

   And [thorn]aeraeter hine God geuferade [thorn]aet he wear[eth] prior
   & faeder [thorn]aes bufan cwe[eth]enan mynstres. (StWulf: 18)

30) becwe[thorn]an

   of ealles cynnes [eth]ingae swa full 7 swa for[eth] 7 swa freo swa
   Tosti eorll 7 Leofrun his wif min fostermoder hit firmest ahten/7
   [eth]ider inn becwae[eth]en on ece yrf[eth]e. (Ch 1137 (Harm 93): 4)

   Ealle [eth]a. mynstra and in cyrican waeron givene and becwe[eth]ene
   Gode. (Chr. 694; Th. 66.6, note 2: Th. Diplm. A. D.830; 465,1)

   Aend ic cype eow paet ic hebbe bicwe[eth]en Portland 7 eall
   [thorn]aet [eth]erto bily[eth] in to Gealden Mynstre on Wyncheastre
   Gode to lofe 7 Sancte Petre 7 Sancte Suunthune [thorn]am monekan to
   scrudan ... (Ch 1154: 2)

31) gecwe[eth]an

   et ait illi iesus uide nemini dixeris sed uade ostende te sacerdoti
   & cue[eth] him haelend loca geseh [thorn]aet [eth]u aenigum menn
   [eth]u gecuoe[eth]a gesaecga ah gaa aedeau [eth]ec [eth]aem
   meassepreost. (MtGI (Li): 8.4)

   Her is geswutelod an [eth]is gewrite hu AElfheah ealdorman his
   cwidae Gecwae[eth]en haef[eth], be his cynchlafordaes
   ge[thorn]afuncge. (Ch 1485 Whitelock 9): 1)

   & aelc [thorn]aera [thorn]inga forgifan beon sceolde [thorn]e <him>
   gedon o[eth][eth]e gecwe[eth]en waere. (ChronE: 1014.8)

   Seo feor[eth]e mihte is Pacientia, [thorn]oet is ge[eth]yld
   gecwe[eth]en. (AEAbusWarn: 256)

   Rex we cwe[eth]e[eth] cyng, [thorn]aet is gecwe[eth]en Wissigend,
   for[thorn]an [thorn]e he sceal wissigen mid wisedome his folc, &
   unriht alecgen. (EAbusWarn: 129)

   Ic blitsige on [thorn]ysum [thorn]ae gecwe[eth]ene syndon to me on
   huse drihtnes we ga[eth] Letatus sure in his quae dicta sunt michi
   in domum domini ibimus. (PsGIE: 121.1)

   Swa hit gecwe[eth]en is. [eth]onne se heretoga waca[eth] [thorn]onne
   bi[eth] eall se here swi[eth]e gehindred. (ChronE: 1003.10)

32) tocweapon

   & ordal & a[eth]as syndan tocwe[eth]en freolsdagum & rihtymbrendagum
   & fram Adventum Domini o[eth] octabas Epiphani // & fram
   Septuagessiman o[eth] XV niht ofer eastran. (LawVAtr: 18)

   et si quis uobis aliquid dixerit dicite quia dominus his opus habet
   et confestim demittet uobis & gif hwa eow // inc awiht
   tocwae[thorn]e saecga[thorn] [thorn]aet dryhten heora [eth]earf &
   sonae forlete[eth] heo. (MtGI (Ru): 21.3)

33) wi[eth]cwe[eth]an

   Him wi[eth]cwae[eth]en muneces & eorles & [thorn]eignes ealle mest
   [thorn]e [thorn]oer waeron. (ChronE: 1123.30) late saxon

   & ealle [thorn]a o[eth]re [thorn]e [eth]aer waeron munechades men
   hit wi[eth]cwae[eth]en fulle twa dagas. (ChronE: 1123.28)

   Ne [eth]a opre ongean [eth]aet nan [eth]incg wi[eth]cwe[eth]on.
   (BenRWells: 64.118.1)

Table 9 presents overall number of occurences and percentage of forms which lost Vernerian alternations in this class, taking into account the distribution of these in particular inflectional categories (without wesan (16)).

The figures indicate that the form of subjunctive sg. was evidently most innovative, showing traces of elimination in close to 4% of all attested forms. Analogical levelling operated also on forms of preterite pl. and past participle, affecting 1.66% of attested forms in the former and 1.59% in the latter. The alternations are best preserved in 2sg. preterite indicative and subjunctive preterite plural which show either no or very scarce traces of elimination.

The single occurrences of levelled forms attested in this class are scattered throughout various texts. Most of the forms are round in texts of West-Saxon provenience, especially Anglo-Saxon Chronicle, Benedictine Rule (11th c.), Cura Pastoralis, various Charters, Homilies, West-Saxon Gospels. Quite frequent occurrences of the levelled forms tan also be traced to Northumbrian texts, here especially Lindisfarne Gospels but also Rushworth Gospels and Durham Ritual, all dating back to the 10th century as well as Coedmon's Hymn, dated to early 8th century (c. 730), or Mercian Vespasian Hymns.

3.2.5. Strong verbs Class V1

Verbs belonging to Class VI manifest a different pattern as far as preservation of the effects of Verner's Law is concerned, displaying the alternations with relative regularity at the same time. The voiced fricative was apparently extended to the form of 3sg. preterite ind. in verbs whose stems ended in a velar fricative (log, flog, hlog, slog, [thorn]wog). Later forms of preterite singular, however, contain already the voiceless fricative, hence loh, floh, hloh, sloh, [thorn]woh and their corresponding prefixed forms. Frequent spellings with -h in the 3sg. preterite ind. tan be explained by the process of unvoicing of final voiced fricatives, which took place shortly after the voicing of medial spirants (Campbell 1959:180).

On the superficial level, it seems, the forms of preterite singular must have been affected by Vernerian voicing since the voiced fricative present here corresponds to those induced by Verner's Law in preterite plural, subjunctive and past participle. In fact, the status of these preterite singular forms is ambiguous. To view the alternation as original, i.e. generated by the operation of Verner's Law, one would have to assume a different pattern of accentuation for the form of preterite singular, with root accentuation in the present and suffixal accent in all the other forms, including preterite singular. Such assumption is not uncontroversial however. According to standard grammars no such special accentuation pattern need be postulated. Campbell (1959) claims that the preterite singular form, in contrast to the present system, received root rather than suffixal accentuation, and the voiced fricative was simply transferred from the past participle form (Campbell 1959: 305). Luick (1921), Wright (1925), Brunner (1902), much in the same line, maintain that the extended voiced fricative appeared under the influence of the preterite plural. Finally, as suggested by Wright (1925), the alternation should be viewed as orthographie rather than phonological. The exact conditions of the variation can be defined as follows:

   When Germanie 3 came to stand finally in OE., it is probable that it
   became a voiceless spirant (x) just as in Goth. OS., and prehistoric
   O.Icel., but that the g (= 3) was mostly restored again owing to the
   influence of the inflected forms. After liquids and guttural vowels
   the restoration of the g was merely orthographical, but the further
   history of the sound in OE. shows that after palatal vowels it was
   mostly restored in pronunciation as well ...
   (Wright 1925: 169)

Elimination of the effects of Verner's Law in this class is evident in the verb sce[thorn][thorn]an only, which in the past participle form preserved the levelled form scea[eth]en. The verb evidently favoured the fricative and extended it very early to past participle; hence frequent citations of scea[eth]en as the basic form of past participle in major Old English grammars (Brunner 1942, 1902; Campbell 1959; Wrigth 1925; Bosworth and Toller's 1898). sce[thorn][thorn]an corresponds to the Gothic verb ska[thorn]jan : sko[thorn]; the strong present scea[thorn]an is attested only in poetry. The verb already in Old English tended to shift to weak declension and the weak preterite forms sce[thorn], sce[thorn]edon, sce[thorn]eden (corresponding to Olcel. Ske[eth]ja/ skaddi) are attested in West Saxon. (17) The following sentences illustrate the use of the levelled forms of sce[eth][eth]an:

  Scyldfull mine scea[eth]en is me sare, frecne on ferhoe; ne dear nu
  for[eth] gan for [eth]e andweardne. (Gen A, B: 869)

  Nu earttu <scea[eth]ana> <sum>, in fyrlocan feste gebunden. (Sat: 57)

Table 10 (next page) presents the distribution of the vestiges of Verner's Law in Class VI.

In other verbs belonging to this class grammatical change is preserved intact. The forms slohge (attested once; subj. pret. sg.) and sloghen (attested once; past participle) indicate no more than spelling variation rather than levelling. They were identified in the following contexts:

   & hine heton burh bismornysse witegian hwa [thorn]aet waere pe hine
   [eth]onne slohge. (HomS 24.2: 197)

   & flemden be king aet te Standard. & sloghen suithe micel of his
   genge. (ChronE: 1138.2)

Needless to say, any conclusions based on the type of data availed by Class VI (with one verb displaying traces of elimination in single attested forms) inspire little confidence and must be treated with a considerable degree of criticism.

3.2.6. Strong verbs Class VII

Class VII, combining several accent types, is very irregular with respect to Verner's Law. In fact, most verbs belonging to this class display no grammatical change at all, and in those which do, preterite singular and preterite plural have the same foot consonant. The reason for the irregularity may be sought in the tact that the past participle form had full grade vocalism rather than zero or reduced grade, which was the case in verbs showing relatively regular grammatical change. The presence of the same vocalism in the prescrit and past participle may be seen as a factor encouraging levelling of the effects of Verner's Law in this class. Irregular traces of Verner's Law tan be identified in two verbs only hon and fon and their prefixed forms, where the preterite plural and singular have the same consonantism. (18)

Alongside the regular forms containing the ng alternation, a set of forms without traces of Vernerian alterations was attested. All of them are of Anglian provenience and the attested past participle forms of-fon and -hon are -foen and -hoen. (19) This evident lack of Vernerian alternations in all of these forms however cannot be attributed to the operation of analogical levelling which, in line with the adopted definition of elimination, involved generalisation of one consonantal variant. The forms could be subsumed under a common heading of Anglian past participles (cf. section 3.3). Table 11 (next page) presents the distribution of the vestiges of Verner's Law in Class VII, taking into account the less regular cases of elimination to be discussed and illustrated in the following section.

Table 12 illustrates overall distribution of forms which eliminated and preserved effects of Verner's Law in verbs belonging to Class VII:

Table 12. The percentage of forms which preserved and eliminated
effects of Verner's Law

CATEGORY                 PRESERVED           ELIMINATED

                          #         %         #       %

PRETERITE SG.             93    100%          0      0%
PRETERITE PLURAL         628    100%          0      0%
PAST PARTICIPLE          570     95.64%      26      4.36%
SUBJUNCTIVE PRET. SG.    146    100%          0      0%
SUBJUNCTIVE PRET. PL.     17     80.90%       4     19.05%

3.3. Anglian forms

A group of verbs and their past participle forms due to their individual pattern of development deserve special attention as their status with respect to Verner's Law is quite ambiguous. When subject to closer scrutiny, it can be noticed that all of them are of Anglian provenance and come from three classes: ahon, gehon, befon, onfon, gefon, (gifon), belonging to Class VII; [thorn]wean, ge[thorn]wean, belonging to Class VI and geseon, assigned to Class V. Their past participle forms attested in Anglian show no traces of Verner's Law on the surface level. Accordingly, in Northumbrian the past participle forms of [thorn]wean are [eth]uaen (20) ([eth]waen), -[eth]uenum (dat.pl.), -[eth]ueanum (for [eth]uaenum), [eth]weanum, attested alongside more regular [eth]uegenum, [eth]uegnum. The attested past participle forms in Li. are ge[eth]uaen, un[eth]uen, un[eth]uenum (for *ge[eth]waegn, [eth]wegn). (21) Similarly, the mentioned Northumbrian past participle forms of-fon and -hon are -foen and -hoen. (22)

This evident lack of Vernerian alternations in all of these forms however cannot be attributed to the operation of analogical levelling which, in line with the adopted definition of elimination, would involve generalisation of one consonantal variant. Instead, it should be viewed as the effect of an individual, Anglian development which involved the loss of medial g and contraction of the vowels.

Following are some instances of these "ambiguous" past participles and subjunctive plural forms found in the corpus:

34) ge[thorn]wean (the only form round in the corpus)

   dicit ei iesus qui locutus est non indiget ut lauet sed est mundus
   totus et uos mundi estis sed non omnes cuoe[eth] him to se haelend
   se[eth]e ge[eth]uoen is l sprec uaes ne [eth]orfae[eth] ?
   [thorn]aette a[eth]oa hine ah is claene all & gie claeno aron ah ne
   alle. (JnGl (Li): 13.10)

35) [thorn]wean (only forms with the prefix un- are attested)

   Quaerentibus de non lotis discipulorum manibus ea dicit hominem quae
   de corde exeunt inquinare soecendum of un[eth]weanum [eth]ara
   [eth]eigna hondum [eth]a ilca cwoe[eth] done monno [eth]a of heorta
   gaes unclaensia. (MtHeadGl (Li): 53)

   haec sunt quae coincinant hominem non lotis autem manibus manducare
   non coincinant hominem [eth]as aron [eth]a widlas done monno
   un[eth]uenum uutedlice hondum eatta ne widlas [eth]one monno.
   (MtGl (Li): 15.20)

36) ahon (altogether 17 past participle forms found in the corpus)

   At illi instabant uocibus magnis postulantes ut crucifigeretur et
   inuallescebant uoces eorum so[eth] hia onstodon stefnum miclum
   gebedon [thorn]aette ahoen woere & ontrymmedon l stefno hiora.
   (LkGl (Li): 23.23)

   Fratres hoc scientes quia uetus homo noster simul crucifixus est
   ... bro' [eth]is witendo f'[eth]on se alda //monn// vser gilic
   ahoen is ... (DurRitGl 1 (Thomp-Lind): 26.10)

37) gehon (2 past participle forms attested in the corpus)

   ... et tradidit iesum flagellis caesum ut crucifigeretur & salde
   [eth]one haelend mi[eth] suuippum to ge[eth]earscanne [thorn]aette
   were gehoen. (MkGl (Li): 15.15)

   scitis quia post biduum paselm fiet et filius hominis tradetur ut
   crucifigatur wutas gie for[eth]on aefter tuaem dogrum l dagum eastro
   bia[eth] & sunu monnes gesald bi[eth] [thorn]aette he se gehoen l
   ahongen. (MtGl (Li): 26.2)

38) befon (one form only attested in the corpus)

   adducunt autem scribae et pharisaei mulierem in adulterio
   deprehensam et statuerunt eam in medio laeddon l brohton uutedlice
   [eth]a wu[eth]uotto & aelaruas uif in [eth]ernelegerscip ofnumen l
   befoen & aseton [eth]ailca l hia on middum. (JnGl (Li): 8.3)

39) onfon (2 past participle forms and 3 subj. pret. pl. forms found in the corpus)

   In [eth]on [eth]onne bio[eth] gesomnade aile somud wi[eth] me
   [eth]aet hie onfoen sawle mine ge[eth]aehtende werun In eo dura
   congregarentur otaries simul aduersum me ut acciperent animam meam
   consiliati sunt. (PsGlA (Kuhn): 30.16)

   Tunc duo erunt in agro unus adsumetur et unus relinquetur [eth]a
   tuoege bi[eth]on on lond an ondfoende bi[eth] l him bi[eth] onfoen
   genumen bi[eth] & an bi[eth] forleten. (MtGl (Li): 24.40)

40) gefon (4 past participle forms altogether)

   et farissei mulierem in aduherio depraehensam et statuerunt eam in
   medio & aes larwas wif in dernegilegerscipe ginumen gifoen & asettum
   [eth]a ilco on middum. (JnGl (Ru): 8.3)

   at illi accepta pecunia fecerunt sicut erant docti so[eth] hia
   gefoen haefdon feh dedon suae weron gelaered. (MtGl (Li): 28.15)

The exceptional contracted past participle form biseon, belonging to S[V.sub.1] and attested once in Christ A, B, C can also be included here. It appears in the following eontext:

41) biseon

   Ne bi[eth] him to are [thorn]aet [thorn]aer fore ellpeodum usses
   dryhtnes rod ondweard stonde[eth], beacna beorhtast, blode bistemed,
   heofoncyninges hlutran dreore, biseon mid swate [thorn]aet ofer side
   gesceaft scire scine[eth]. (Christ A, B, C: 1083)

4. Concluding remarks

Elimination of the effects of Verner's Law started very early in some verbs, even before the time of their earliest attestation. Such was the case with mipan, bemipan 'conceal, avoid', risan 'rise', arisan 'arise', wripan, gewripan 'twist', aetclipan 'adhere" (Class I), abreopan, breoban (Class II), lesan 'collect, lease', alesan 'choose' (ge)nesan 'survive' (Class V), all of which were attested in Old English texts with the voiced fricative, reminiscent of the PGmc.original voiceless variant, with the exception of findan (Class III) which generalised the voiced stop in a cluster with a nasal. Levelling of the effects of Verner's Law in verbs which evince Vernerian alternations more or less regularly in historic Old English begins towards the end of the Old English period and is a gradual process. It must be noticed that none of the verbs eliminated the voiced alternant completely, instead they tend to vacillate between the two forms: the archaic Vernerian one and the innovative one without Verner's Law.

All instances which display the levelling tendencies derive from late West Saxon and late Anglian texts, and they are best represented in Anglo-Saxon Chronicle and Lindisfarne and Rushworth Gospels. In fact, most of them come from the later manuscript of the Chronicle--MS. Laud. 636 (noted in the corpus as Chron. E) referred to as the Peterborough Chronicle. The manuscript, dated to the first half of the 12th century, has been frequently classified as belonging to the Early Middle English period or the intermediate stage between Old and Middle English known as the transition period. It comes as no surprise then that forms with both alternants, the voiceless and the voiced one are round alongside each other, indicating the imminent process of elimination. Traces of earlier elimination appear sporadically in earlier West Saxon texts: Pastoral Care and Bede's History of the English Church and nation, dating back to the 10th and 11th century.

The present analysis of the Old English material took into consideration the behaviour of Vernerian alternations, viewed form the perspective of their loss, with respect to a number of factors (variables). These included the relevant inflectional categories, alternation sets and class membership.

Tables 13 and 14 present the overall percentage of forms which eliminated the alternations by Verner's Law with respect to two variables, namely class membership and inflectional category.

The data unerringly indicate that levelling tendencies are most pronounced in the form of subjunctive pret. pl. and preterite ind. plural. The forms of 2sg, preterite ind. seem to be most conservative, showing relatively little inclination towards levelling.

The noted frequency pattern indicates that there is a relatively high proportion of forms which lost Vernerian alternations in strong verbs of Classes I and III and VII. Classes VI and II are apparently most archaic in that very few verbs display tendency towards levelling of the effect of Verner's Law. In fact, in Class II the number of occurrences is relatively high (12 occurrences altogether), but given the ample attestation of verbs belonging to this class, this does hot constitute more than 0.63% of all occurrences. In Class VI traces of levelling are present in one verb only (sce[thorn][thorn]an).

The picture emerging form the analysed material with respect to particular alternations shows that the dental alternation was most innovative and most readily subject to the working of analogy. A similar tendency is present in the alternation of sibilant/r where traces of elimination are attested in ceosan and its prefixed forms. Intact remain all other alternations which do not show even slight tendency towards levelling.

Viewed against the complete corpus, without considering class membership and inflectional categories the overall percentage of forms which eliminated the effects of Verner's Law in Old English strong verbs relative to those which still preserved it looks as in Table 15:

Table 15. Overall percentage of strong verb forms which eliminated the
effects of the operation of Verner's Law (irrespective of class,
inflectional category, alternation or any other variable)

PRESERVED   ELIMINATED

  98.45%      1.55%

In fact, if it were not for the data provided by subsequent period in the history of English, one would be justified in treating such percentage distribution, with 1.55% of forms where levelling operated, as a case of some sporadic, ad hoc development; an irregularity rather than a trace of a consistent process working towards regularisation. Given the knowledge availed by the Middle English data however as well as the nature of the process of Vernerian voicing, one must view these figures as testifying to the presence of some development encouraging elimination of the Vernerian alternants. Accordingly, the figures indicate that the effects of Verner's Law were still firmly preserved in Old English strong verbs. The process of elimination, very irregular and sporadic affected only a handful of Old English verbs and must be viewed as no more than a marginal process.

In a number of verbs the tendency to extend the voiced alternant to forms of the present or preterite singular could be observed. Such was the case, for instance, with wreon where the voiced fricative was extended to 1, 3 pret. sg. (wrag) and the present system (3sg. pres. ind. wriga[thorn]), or [thorn]eon ([thorn]ag, ge[thorn]ong). Such an unexpected extension can be viewed as a development in the opposite direction in the sense that the voiced rather than the voiceless alternant is generalised.

The behaviour of three verbs li[thorn]an, forli[thorn]an and sce[thorn][thorn]an deserves some attention. These three verbs show traces of elimination, in forms of preterite plural (li[thorn]an, forli[thorn]an) and in past participle (sce[thorn][thorn]an). Noteworthy is the fact that they eliminate the working of Verner's Law completely in the attested occurrences in the respective forms, i.e. li[thorn]on and forli[thorn]an are the only attested preterite plural forms of li[thorn]an and forli[thorn]an, respectively, whereas scea[thorn]en is the only attested past participle form of sce[thorn][thorn]an.

Finally, the pattern of distribution of the levelled forms in the Old English material indicates an evident correlation between the process of levelling and the frequency of occurrence of particular verbs. It could be observed that verbs which were attested more frequently in the corpus (e.g., cwe[thorn]an, weor[thorn]an) tended to be more liable to the operation of the levelling process.

Concluding, the forms which show traces of elimination of Vernerian alternations are rather sporadic and scantily evidenced, their number being very low throughout the Old English period. None of the classes and none of the inflectional forms testify to a radical or rapid removal of Vernerian alternations from the paradigm of Old English strong verb. The slight increase in the number of forms without Vernerian alternants can be dated to the end of the 10th century, and later. This slight frequency increase towards the end of the Old English period indicate that some linguistic development must have been in progress. Certainly, the relatively small number of these forms does hot allow no view them and the process as actual spread of the tendency at this early stage. The development will continue into the Early Middle English period and will be characterised by the increasing popularity of the forms without traces of Verner's Law at the expense of those still preserving in.

Table 1. Vernerian alternations in PGmc. and their OE reflexes

      PROTO-GERMANIC                 OLD ENGLISH

       *f ~ *[beta]                      --
     *[thorn] ~ *[eth]           [thorn], [eth] ~ d
    *[chi]) ~ *[gamma]                 h/o ~ g
*[eta][chi] ~ *[eta][gamma]    o (+ vowel length) ~ ng
    *[chi]w ~ *[gamma]w               h/o ~ g/w
          *s ~ *z                       s ~ r

Table 2. Distribution of the effects of Verner's Law in Class I

CATEGORY       PRET. SG.         PRET. PL.      PAST PTCP.

             P      L      P       L          P      L
INFINITIVE
             #    #   %    #    #     %       #    #     %

onleon        0   0         0   0             1    0    0%
lipan         0   0         0   3    100%     3    0    0%
belipan       0   0         0   0             3    0    0%
forlipan      0   0         0   2    100%    14    1   6.67%
gelipan       0   0         1   0     0%      5    0    0%
scripan       0   0         2   0     0%      0    1   100%
seon          5   0   0%   13   0     0%      8    0    0%
aseon         0   0         1   0     0%     18    0    0%
geseon        1   0   0%    0   0            23    0    0%
snipan        0   0         5   1   16-67%    1    0    0%
asnipan       0   0         0   0             2    0    0%
gesnipan      0   0         0   0             4    0    0%
ofsnipan      0   0         1   1    50%      1    0    0%
tosnipan      0   0         0   0             2    0    0%
ymbsnipan     0   0         0   0             8    0    0%
peon          0   0         3   0     0%     45    0    0%
gepeon        1   0   0%    8   0     0%     93    0    0%
oferpeon      0   0         0   0             5    0    0%
teon          0   0         5   0     0%      0    0
beteon        0   0   0%    0   0            12    0    0%
wreon         3   0   0%    6   0     0%      1    0    0%
awreon        1   0   0%    3   0     0%     11    0    0%
bewreon       6   0         4   0     0%     53    0    0%
gewreon       0   0         0   0             6    0    0%
oferwreon    10   0   0%    5   0     0%     59    0    0%
unwreon       0   0         2   0     0%     11    0    0%
onwreon       2   0   0%    2   0     0%     27    0    0%

TOTAL        29   0   0%   61   7   10.29%   415   2   0.48%

CATEGORY       SUBJ. PRET SG     SUBJ. PRET. PL.

             P       L          P      L
INFINITIVE
             #    #   %         #   #    %

onleon        0   0             0   0
liban         0   0             0   0
beliban       0   0             0   0
forliban      0   0             0   2   100%
geliban       0   0             0   0
scriban       0   0             0   0
seon          1   0   0%        1   0    0%
aseon         0   0             0   0
geseon        3   0   0%        0   0
sniban        1   0   0%        0   0
asniban       0   0             0   0
gesniban      0   0             0   0
ofsniban      0   0             0   0
tosniban      0   0             0   0
ymbsniban     0   0             0   0
beon          2   0   0%        0   0
gebeon        2   0   0%        0   0
oferbeon      3   0   0%        0   0
teon          2   0   0%        0   0
beteon        0   0             0   0
wreon         0   0             0   0
awreon        0   0             0   0
bewreon       2   0   0%        0   0
gewreon       0   0             0   0
oferwreon     0   0             1   0    0%
unwreon       0   0             0   0
onwreon       4   0   0%        0   0

TOTAL        20   0   0%        2   2    50%

Table 3. The percentage of forms which eliminated the alternations by
Verner's Law in Class I with respect to particular categories

CATEGORY                PRESERVED     ELIMINATED

                        #      %      #     %

PRETERITE SG.           29     100%   0       0%
PRETERITE PLURAL        61   89.71%   7   10.29%
PAST PARTICIPLE        415    99.5%   2    0.50%
SUBJUNCTIVE PRET. SG.   20     100%   0       0%
SUBJUNCTIVE PRET. PL.    2      50%   2      50%

Table 4. Distribution of the effects of Verner's Law in strong verbs
Class II

CATEGORY         PRET. SG.         PRET. PL.         PAST PTCP.

               P      L          P       L          P       L
INFINITIVE
               #    #     %      #     #     %      #     #     %

ceosan         2    2    50%     10    4   28.6%     12   2   14.29%
aceosan        0    0             1    0    0%       10   0     0%
geceosan      14    0    0%      54    2   3.6%     722   0     0%
wipceosan      0    0             9    0    0%        5   0     0%
wiperceosan    0    0             0    0             18   0     0%
dreosan        0    0             2    0    0%        0   0
adreosan       0    0             0    0              1   0     0%
bedreosan      0    0             0    0              9   0     0%
gedreosan      0    0             1    0    0%        4   0     0%
todreosan      0    0             0    0              1   0     0%
fleon         14    0    0%     120    0    0%        0   0
afleon         0    0             1    0    0%        4   0     0%
aefleon        0    0            13    0    0%        0   0
befleon        0    0             1    0    0%        3   0     0%
forfleon       0    0             1    0    0%        1   0     0%
gefleon        0    0             7    0    0%        2   0     0%
opfleon        0    0            14    0    0%        2   0     0%
tofleon        0    0             0    0              1   0     0%
utfleon        0    0             1    0    0%        0   0
leosan         0    0             0    0              2   0     0%
beleosan       0    0             0    0              5   0     0%
forleosan     20    0    0%       9    0    0%       96   0     0%
gefreosan      0    0             4    0    0%        2   0     0%
oferfreosan    0    0             0    0              2   0     0%
begreosan      0    0             0    0              1   0     0%
hreosan        1    0    0%      20    0    0%        2   0     0%
ahreosan       0    0             3    0    0%        0   0
behreosan      0    0             0    0              2   0     0%
gehreosan      0    0             5    0    0%        6   0     0%
ofhreosan      0    0             0    0             12   0     0%
onhreosan      0    0             8    0    0%        0   0
tohreosan      0    0             9    0    0%        0   0
seopan         1    0    0%       2    0    0%       14   0     0%
aseopan        1    0    0%       0    0             14   0     0%
beseopan       0    0             0    0              3   1    25%
geseopan       0    0             0    0            115   0     0%
toseopan       0    0             0    0              1   0     0%
teon           1    0    0%      50    0    0%       58   0     0%
ateon         12    0    0%      16    0    0%       47   0     0%
beteon         0    0             0    0              4   0     0%
forpateon      1    0    0%       0    0              2   0     0%
forteon        0    0             0    0              4   0     0%
geteon         3    0    0%      29    0    0%       64   0     0%
toteon         0    0             1    0    0%        4   0     0%
onteon         0    0             0    0              0   0
ofteon         2    0    0%       3    0    0%       11   0     0%
oferteon       0    0             0    0              3   0     0%
opteon         0    0             0    0              1   0     0%
purhteon       1    0    0%      10    0    0%       50   0     0%
wibteon        0    0             2    0    0%        5   0     0%
upateon        0    0             0    0              1   0     0%
forweosan                                             4   0     0%
TOTAL         73    2   2.67%   406    6   1.46%   1330   3   0.23%

CATEGORY       SUBJ. PRET. SG.   SUBJ. PRET. PL.

               P      L          P      L
INFINITIVE
               #   #     %       #    #   %

ceosan         3   0   0%        5    0   0%
aceosan        0   0             0    0
geceosan       9   1   10%       2    0   0%
wipceosan      0   0             0    0
wiperceosan    0   0             0    0
dreosan        0   0             0    0
adreosan       0   0             0    0
bedreosan      0   0             0    0
gedreosan      0   0             0    0
todreosan      0   0             0    0
fleon          7   0   0%        8    0   0%
afleon         0   0             0    0
aefleon        0   0             2    0   0%
befleon        1   0   0%        1    0   0%
forfleon       1   0   0%        0    0
gefleon        1   0   0%        0    0
opfleon        0   0             0    0
tofleon        0   0             0    0
utfleon        0   0             0    0
leosan         0   0             0    0
beleosan       0   0             0    0
forleosan     20   0   0%        2    0   0%
gefreosan      0   0             0    0
oferfreosan    0   0             0    0
begreosan      0   0             0    0
hreosan        5   0   0%        0    0
ahreosan       0   0             0    0
behreosan      0   0             0    0
gehreosan      1   0   0%        0    0
ofhreosan      0   0             0    0
onhreosan      0   0             0    0
tohreosan      0   0             0    0
seopan         0   0             0    0
aseopan        1   0   0%        0    0
beseopan       0   0             0    0
geseopan       0   0             0    0
toseopan       0   0             0    0
teon           3   0   0%        2    0   0%
ateon          8   0   0%        0    0
beteon         0   0             0    0
forpateon      0   0             0    0
forteon        2   0   0%        0    0
geteon         2   0   0%        0    0
toteon         0   0             0    0
onteon         1   0   0%        0    0
ofteon         1   0   0%        0    0
oferteon       0   0             0    0
opteon         0   0             0    0
purhteon       6   0   0%        0    0
wibteon        0   0             0    0
upateon        1   0   0%        0    0
forweosan
TOTAL         73   1   1.35%    22    0   0%

Table 5. The percentage of forms which eliminated the alternations by
Verner's Law in Class II with respect to particular inflectional
categories

CATEGORY                   PRESERVED       ELIMINATED

                          #        %       #       %

PRETERITE SG.              73    97.33%    2     2.67%
PRETERITE PLURAL          406    98.54%    6     1.46%
PAST PARTICIPLE          1330    99.77%    3     0.23%
SUBJUNCTIVE PRET. SG.      73    98.65%    1     1.35%
SUBJUNCTIVE PRET. PL.      22      100%    0        0%

Table 6. Distribution of the effects of Verner's Law in Class III

CATEGORY       PRET. SG.       PRET. PL.          PAST PTCP.

              P     L       P       L            P       L
INFINITIVE
              #   #   %     #     #      %       #     #     %

feolan        0   0          4    0     0%        0    0
aetfeolan     0   0          2    2     50%       0    0
befeolan      0   0          3    0     0%        0    5   100%
weorban      22   0   0%   729   25   3.32%      53    4   8.62%
aweorpan      0   0          1    0     0%      188    0   0%
forweorpan    1   0   0%    71    3    4.05%     17    0   0%
geweorpan     3   0   0%    56    0     0%     1402   14   0.99%
TOTAL        26   0   0%   866   30   3.35%    1660   24   1.43%

CATEGORY       SUBJ. PRET. SG.      SUBJ. PRET. PL.

              P       L            P     L
INFINITIVE
              #    #      %        #   #   %

feolan         0    0      0       0   0   0%
aetfeolan      0    0              0   0
befeolan       0    0              0   0
weorban      197    2    1.01%    28   0   0%
aweorpan       0    0              0   0
forweorpan    26    1    3.70%     0   0
geweorpan     66    8    10.81%   17   0   0%
TOTAL        290   11   3.65%     46   0   0%

Table 7. The percentage of forms which eliminated the alternations by
Verner's Law in Class III with respect to particular inflectional
categories

CATEGORY                  PRESERVED      ELIMINATED

                          #      %        #     %

PRETERITE SG.             26     100%     0   0,00%
PRETERITE PLURAL         866   96.65%    30   3.35%
PAST PARTICIPLE         1660   98.57%    24   1.43%
SUBJUNCTIVE PRET. SG.    290   96.35%    11   3.65%
SUBJUNCTIVE PRET. PL.     46     100%     0   0.00%

Table 8. Distribution of the effects of Verner's Law in Class V

CATEGORY           PRET. SG.            PRET. PL.

                 P         L         P          L
INFINITIVE
                 #     #      %      #       #      %

acwepan          0     0              1      0     0%
becwepan         0     0              1      1    50%
cwepan          97     0     0%    1745     39     2.19%
forcwepan        0     0              2      0     0%
forecwepan       0     0              3      0     0%
gecwepan         3     0     0%      35      2     5.41%
oncwepan         0     0              3      0     0%
tocwepan         0     0              4      0     0%
wipcwepan        0     0             18      4    18.18%
wipercwepan      0     0              0      0
feon             0     0              1      0     0%
gefeon           0     0             24      0     0%
seon             2     0     0%      27      0     0%
beseon           1     0     0%       5      0     0%
forseon          6     0     0%      54      0
foreseon         3     0     0%       0      0
geseon         108     0     0%     795      0     0%
geondseon        0     0              0      0
oferseon         1     0     0%       6      0     0%
onseon           3     0     0%       3      0     0%
ymbseon          0     0              0      0
wesan          315     0     0%    5514      0     0%
gewesan          0     0              1      0     0%
aewesan          0     0              2      0     0%
TOTAL          539     0     0%    8244     46     0.55%

CATEGORY           PAST PTCP.            SUBJ. PRET. SG.

                 P        L             P         L
INFINITIVE
                 #     #       %        #     #       %

acwepan         30     1     3.23%      0     0
becwepan        12     2    14.29%      0     0
cwepan         123     6     4.65%     98    11    10.09%
forcwepan        4     0     0%         0     0
forecwepan      52     0     0.00%      2     0     0%
gecwepan       834    15     1.77%     23     1     4.17%
oncwepan         0     0                1     0     0%
tocwepan         4     1    20.00%      0     1   100%
wipcwepan        5     0     0%         3     0     0%
wipercwepan      1     0     0%         0     0
feon            12     0     0%         0     0
gefeon          15     0     0%         1     0     0%
seon            29     0     0%         3     0     0%
beseon           2     0     0%         5     0     0%
forseon         48     0     0%        11     0     0%
foreseon         6     0     0%         1     0     0%
geseon         364     0     0%       211     0     0%
geondseon        0     0                0     0
oferseon         2     0     0%         0     0
onseon           5     0     0%         1     0     0%
ymbseon          2     0     0%         1     0     0%
wesan            0     0             2996     0     0%
gewesan          0     0                0     0
aewesan          0     0                0     0
TOTAL         1550    25     1.59%   3357    13     0.39%

CATEGORY         SUBJ. PRET. PL.

                 P         L
INFINITIVE
                 #     #       %

acwepan          0     0
becwepan         0     0
cwepan           9     1    10.0%
forcwepan        1     0     0%
forecwepan       0     0
gecwepan         5     0     0%
oncwepan         0     0
tocwepan         0     0
wipcwepan        0     0
wipercwepan      0     0
feon             0     0     0%
gefeon           3     0     0%
seon             3     0
beseon           0     0     0%
forseon          3     0
foreseon         0     0     0%
geseon          32     0     0%
geondseon        1     0
oferseon         0     0
onseon           0     0
ymbseon          0     0
wesan          129     0     0%
gewesan          0     0
aewesan          0     0

TOTAL          186     1     0.53%

Table 9. The percentage of forms without Verner's Law in Class V
with respect to particular categories

CATEGORY                  PRESERVED        ELIMINATED

                          #       %        #       %

PRETERITE SG.             224     100%      0      0%
PRETERITE PLURAL         2727   98.34%     46   1.66%
PAST PARTICIPLE          1550   98.41%     25   1.59%
SUBJUNCTIVE PRET. SG      361   96.52%     13   3.48%
SUBJUNCTIVE PRET. PL.      57   99.47%      1   0.53%

Table 10. Distribution of the effects of Verner's Law in Class VI

CATEGORY       PRET. SG.       PRET. PL.        PAST PTCP.

                P     L        P      L       P        L
INFINITIVE
                #   #    %         #    %     #    #       %

flean           0   0          1   0   0%     0    0
aflean          0   0          0   0          3    0       0%
beflean         0   0          0   0          1    0       0%
hliehhan        3   0   0%    13   0   0%     0    0
lean            0   0          4   0   0%     1    0       0%
belean          0   0          1   0   0%     3    0       0%
forlean         0   0          0   0          9    0       0%
sceppan         0   0          3   0   0%     0    2     100%
gesceppan       0   0          1   0   0%     1    0       0%
slean          19   0   0%   107   0   0%    28    0       0%
aslean          0   0          2   0   0%    12    0       0%
beslean         0   0          0   0          6    0       0%
forslean        0   0          0   0         26    0       0%
geslean         0   0         23   0   0%    84    0       0%
ofaslean        0   0          1   0   0%    15    0       0%
ofslean        20   0   0%   255   0   0%   438    0       0%
oferslean       0   0          4   0   0%     2    0       0%
toslean         0   0          1   0   0%     7    0       0%
purhslean       0   0          1   0   0%     1    0       0%
pwean           0   0          7   0   0%     6    0       0%
apwean          0   0          4   0   0%    50    0       0%
gepwean         1   0   0%     2   0   0%    14    0       0%

TOTAL          43   0   0%   430   0   0%   707    2       0.28%

CATEGORY       SUBJ. PRET. SG    SUBJ. PRET. PL.

                P      L           P        L
INFINITIVE
                #   #    %         #     #      %

flean           0   0              0     0
aflean          0   0              0     0
beflean         0   0              0     0
hliehhan        1   0   0%         0     0
lean            4   0   0%         0     0
belean          1   0   0%         0     0
forlean         0   0              0     0
sceppan         1   0   0%         0     0
gesceppan       0   0              0     0
slean          18   0   0%         0     0
aslean          0   0              0     0
beslean         0   0              0     0
forslean        0   0              0     0
geslean         0   0              0     0
ofaslean        0   0              0     0
ofslean        26   0   0%         2     0     0%
oferslean       0   0              0     0
toslean         0   0              0     0
purhslean       0   0              0     0
pwean           1   0   0%         0     0
apwean          5   0   0%         0     0
gepwean         0   0              0     0

TOTAL          57   0   0%         2     0     0%

Table 11. Distribution of the effects of Verner's Law in Class VII

CATEGORY        PRET SG.        PRET PL.       PAST PTCP.

               P      L       P      L         P       L

INFINITIVE     #   #     %     #   #    %      #    #      %

fon            2   0    0%    65   0   0%      6    0    0%
afon          16   0    0%     9   0   0%      8    0    0%
aetfon         0   0           0   0           1    0    0%
anfon          7   0    0%    11   0   0%      7    0    0%
befon          1   0    0%    13   0   0%     85    1    1.16%
forfon         1   0    0%     0   0           5    0    0%
forefon        0   0           4   0   0%      0    0
oferfon        0   0           1   0   0%      8    0    0%
gefon          1   0    0%    58   0   0%     75    4    5.06%
onfon         40   0    0%   224   0   0%    106    2    1.85%
tofon          1   0    0%     0   0           0    0
underfon      20   0    0%   133   0   0%    102    0    0%
ymbfon         0   0           0   0           4    0    0%
hon            0   0          17   0   0%     13    0    0%
ahon           4   0    0%    91   0   0%    141   17   10.76%
behon          0   0           0   0           7    2    0%
gehon          0   0           2   0   0%      2    2   50%
TOTAL         93   0    0%   628   0   0%    570   26    4.36%

CATEGORY       SUBJ. PRET. SG.    SUBJ. PRET. PL.

               P       L           P       L

INFINITIVE     #    #    %         #    #    %

fon           27    0   0%         5    0    0%
afon           0    0              0    0
aetfon         2    0   0%         0    0
anfon          2    0   0%         3    0    0%
befon          0    0              0    0
forfon         1    0   0%         0    0
forefon        1    0   0%         0    0
oferfon        0    0              0    0
gefon          5    0   0%         5    0    0%
onfon         67    0   0%         4    4   50%
tofon          0    0              0    0
underfon      33    0   0%         0    0
ymbfon         1    0   0%         0    0
hon            1    0   0%         0    0
ahon           6    0   0%         0    0
behon          0    0              0    0
gehon          0    0              0    0
TOTAL        146    0   0%        17    4   19.05%

Table 13. The percentage of forms which eliminated the alternations
by Verner's Law in particular inflectional categories

CATEGORY                 ELIMINATED      PRESERVED

PRETERITE SG.               0.41%         99.59%
PRETERITE PLURAL            1.71%         98.29%
PAST PARTICIPLE             1.30%         98.70%
SUBJUNCTIVE PRET. SG.       2.57%         97.43%
SUBJUNCTIVE PRET. PL.       4.58%         95.42%

Table 14. The percentage of forms which eliminated the effects of
Verner's Law with respect to class membership

CLASS         ELIMINATED    ELIMINATED

CLASS I         2.04%         97.96%
CLASS II        0.63%         99.37%
CLASS III       2.17%         97.83%
CLASS V         1.70%         98.30%
CLASS VI        0.16%         99.84%
CLASS VII       2.02%         97.98%

(1) The original formulation of the law, as posited by its author, is as follows: "Indogermanische k, t, p gingen erst uberal in h, [thorn], fuber; die so entstanden tonlosen fricativa nebst der vom indogermanischen ererbten tonlosen fricativa s wurden weiter inlautend bei tonender nachbarschaft selbst tonend, erhielten sich aber als tonlose im nachlaute betonter silben" (Veiner 1876: 114).

(2) The voiceless fricative could also disappear in some other contexts: when doubled, when between a liquid and a following vowel (OE feolan 'hide' vs. Go. filhan), when between a vowel and a following liquid or nasal (/l, m, n, r/) as well as before s followed by another consonant. It was preserved in some consonantal clusters (before voiceless consonants) resulting from vowel syncope, and also in final position (Wright 1925: 171-2).

(3) In Old English medial and final *[gamma] remained spirants and initial [gamma] is viewed as aspirant in the early West Saxon texts. In the later period initial velar fricative became a stop before consonants and back vowels (where its articulation was velar), and was pronounced as a palatal spirant [j] before front vowels (only original front vowels: ae, e, i and j). The fricative is assumed to have gone through an intermediate stage of being a voiced palatal fricative, prior to losing friction.

(4) The development has traditionally been assumed to have consisted in two almost independent processes: voicing (*s > *z) and rhotacism proper (z > r). Rhotacism entailed a series of separate developments: /z/ > /z/, /z/ > /r/, /r/ > /r/, which can be reduced to /z/ > /r/ > /r/. The initial stage, i.e., the change from /z/ > /r/ presupposes the diminution of friction and increase in the degree of sonority. The second stage is the shift from /r/ > /r/, resulting eventually in the merger of old /z/ with /r/ (Smimitskij 1990: 197).

(5) The spellings with <b> can also be found in early Kentish documents coming from the first part of the 9th century (831-832), as in giban 'give' (WS giefan), gib (WS gif) 'if', geroebum (WS gerefa) 'earl' (dat. pl). Individual instances have been identified in CP naebre 'never' (for naefre), weobud 'altar' (later weofod); in poetry: Genesis and Exodus: tiber (dat. sg. tibre, later tifor, tifre, -um) 'victim' (Brunner 1962: 153).

(6) The fact that this alternation is underrepresented in Old English as well as in some other Germanic dialects is difficult to account for. The absence of this labial set as an alternating sequence in Old English may be attributed to the individual phonological development of this alternation, involving merger of the voiceless and voiced variants and consequent total obliteration of the original alternants, unparalleled in any other alternating set. Of some importance may be the chronological gap present in the development of the five sequences: the occlusion of bilabial stop must have been a later process since the bilabial stop was relatively rare when compared to other stops resulting from parallel occlusions (the medial and final [beta] remained aspirant for a relatively long time in other Germanic dialects as well (cf. van der Rhee 1995: 46)).

(7) An alternative view assumes that the 2sg. preterite ind. was formed directly from preterite subjunctive rather than preterite ind. plural (Wright 1925: 263). As far as the origin of this form is concerned, it has been viewed as an augmentless aorist (rather than an original perfect) with the -es ending (s voiced to z and lost in Proto-Germanic) and ablaut alternations according to the pattern of preterite ind. plural.

(8) The inherited mobility of accent was preserved in the early stage of Proto-Germanic and was still present at the time of the operation of Verner's Law. The period of mobile accent in Germanic is estimated to have lasted between several centuries and a millennium. The assumption is that the Germanic accent was still mobile at the rime when Verner's Law was operating; otherwise the process would not have operated at all. In the later stage of Proto-Germanic the accentual pattern changed and accent became confined to the foot or stem syllable (Bennett 1968: 220, 1972: 100).

(9) One of alternative theories, intended to account for the absence of Vernerian alternants in some forms, takes recourse to the existence of intervocalic laryngeal sounds in Proto-Germanic: "It is, of course, possible but unsatisfactory to invoke levelling to explain these forms, for it is difficult to see why these verbs should have been subjected to such early and widespread levelling, while other verbs occur principally or exclusively with regular GW [grammatischer Wechsel] until well past the time of the first attestations" (Conolly 1980: 97). Conolly's hypothesis involves the assumption that disyllabic roots of the structure CVXRC (with a medial non-syllabic laryngeal) would never exhibit grammatical change since the accent could never fall on the syllable preceding the medial voiceless fricative, irrespective of the type of accent (root or suffix) a given form carried (CVXRC and C(V)XRC-'). The author bas it that at the time when Verner's Law operated the original disyllabic roots must have been distinct from the diphthongal roots of the structure CVRC, where R represents a non-syllabic resonant. The conclusion is that the original diphthongal roots tend to have regular GW, whereas the disyllabic roots have either irregular GW or no change at all (the pattern refers to all classes with the exception of Class VII). Conolly attributes the irregularity in Class VII to the fact that the past participle form in this class had full grade vocalism, rather than the zero grade or reduced grade found in most other strong verb participles. The full grade was rot original but "of considerable antiquity". Levelling then in such circumstances could be triggered by the fact that the present and past participle displayed the same vocalism. The preterite forms, which were rot of IE origin but had instead analogical e-grade vocalism, adopted the consonantism from the past participle (Conolly 1980: 115, cf. Lehmann 1971).

(10) The tool used for the systematic search of the collection was Toronto Old English Database Search Form, designed specifically for data retrieval from the Toronto Corpus, available online at http://ets.lib.uchicago.edu/Databases/OldEnglish/. The software (Search Form) permitted a wide range of queries, including searches within particular Old English dialects, particular texts, phrase and co-occurrence searches, etc. For the present purpose the word search was used for retrieving frequencies of occurrences for particular forms and for identifying texts in which these forms appeared. Pattern matching options allowed for retrieving various inflectional endings or spelling variants.

(11) In the Anglo-Saxon Dictionary by Bosworth and Toller (1898) the form of leo[thorn]on with the diphthong eo is cited: "[THORN]a waes heofones smylmes tosliten, [thorn]aer [thorn]e we aer uton leo[thorn]on (li[thorn]on, v. 1.) interrupta est serenitas, qua uehebamur" (Bd. 5, 1).

(12) "For[thorn]an [thorn]e efne hig ongunnon hig begripon mine sawle onahruron on me [thorn]a strangan Quia ecce coeperunt animam meam irruerunt in me fortes" (Psalms 4137).

(13) The past participle forms of ofhreosan, which resulted form loss of/h/in hr- cluster, are attested in the following sentences with regular reflexes of Verner's Law:

   "obruti contriti tobrytte ofrorene". (AldV 1:4137)
   "obrutos ofsette ofroren". AldV 1 (2992)

(14) The regular ablaut pattern in S[V.sub.4], attested in Vespasian Psalter, was e--e--e--o (Brunner 1902: 298).

(15) According to Prokosch (1939), the y present in past participle is an expected development when the original IE ending was -ono-, but w when the original ending was -eno-. The Y round in preterite plural is a regularly expected form since it was followed originally by u (Prokosch 1939: 74).

(16) The verb wesan was hot included in the final eount since its frequency of occurrence is much higher than the frequency of other verbs in this class as well as verbs in other classes. The inclusion of the figures obtained from the analysis of wesan would be likely to level the final results and obliterate thus the tendencies observable in the above figures.

(17) Another related weak verb attested in Old English to which these forms can belong is scea[thorn]ian 'injure' (which corresponds to OIcel, ska[eth]a/ska[eth]a[eth]i, OHG scadon/scadota and sce[thorn][thorn]an/sce[thorn]ede (Bosworth and Toller 1898: 437).

(18) *fan[chi]an > * fan[chi]an (-an- > -a-) > *fan[chi]an (-a- > -o-) > *foan (loss of intervocalic [chi]) > OE fon and similarly: *han[chi]an > *ha[chi]an (-an- > -a-) > *ho[chi]an (-a- > -o-) > *hoan (loss of intervocalic [chi]) > OE hon. The Vernerian voiced alternant is extended to the form of preterite singular: feng and heng are new formations based on the corresponding plural fengon and hengon (Welna 1996: 73).

(19) They may go back to monosyllabic *foin and *hoin. According to Campbell (1959) these Northumbrian forms should be viewed as disyllabic, containing a sequence of o followed by e rather than monosyllabic oe (Campbell 1959: 234).

(20) [eth]uaen may be viewed as a misspelling of ae for ae (Brunner 1962: 290).

(21) The two regular forms with the prefix un-: un[eth]uegenum and un[eth]uegnum (occurring twice in the analysed collection) were treated together with [thorn]wean.

(22) They may go back to monosyllabic *foin and *hoin or can be treated as disyllabic forms with analogical ending. According to Campbell these Northumbrian forms should be viewed as disyllabic, containing a sequence of o followed by e rather than monosyllabic oe (Campbell 1959: 243).

REFERENCES

Bennett, William Holmes 

1968 "The operation and relative chronology of Verner's law", Language 44: 219-223.

1972 "Prosodic features in Proto-Germanic", in: Frans van Coetsem--Herbert Kufner (eds.), 99-116.

Brunner, Karl

1960-62 Die englische Sprache. Ihre geschichtliche Entwicklung. 2 vols. (2nd edition) Tuingen: Max Niemeyer.

1942 Altenglische Grammatik, nach der Angelsachsischen Grammatik von Eduard Sievers. Halle--Saal: Max Niemeyer Verlag.

Bosworth, Joseph--T. Northcote Toller

1898 An Anglo-Saxon Dictionary. Supplement (1921) by T. N. Toller. Oxford: Clarendon Press.

Campbell, Alistair

1959 Old English grammar. Oxford: Clarendon Press.

Coelsem, Frans van--Herbert L. Kufner (eds.)

1972 Toward a grammar of Proto-Germanic. Tubingen: Niemeyer.

Conolly, Leo

1980 "Grammatischer Wechsel and the Laryngeal Theory", Indogermanische Forschungen 85: 96-123.

D'Alquen, Richard

1973 "The Germanic sound shift and Verner's law: A synthesis", General Linguistics 13: 79-89.

1988 Germanic accent, grammatical change and the laws of unaccented syllables. New York--Bern--Frankfurt/M.: Peter Lang.

Healey, Antoinette di Paolo (ed.)

2000 The dictionary of Old English corpus in electronic form. Toronto: Toronto University Press.

Hogg, Richard M. 1992 A grammar of Old English: Vol I: Phonology. Oxford: Blackwell.

Lehmann, Winfred P. 1971 "Grammatischer Wechsel and current phonological discussion", in: Maria Tsiapera (ed.), 9-43.

Luick, Karl 1921 Historische Grammatik der englischen Sprache. Bd. II. Stuttgart: Tauschnitz.

Prokosch, Eduard

1939 A comparative Germanic grammar. Philadelphia--Baltimore: Linguistic Society of America.

Rhee, Florus van der

1995 "Entstehung und Verfall des grammatischen Wechsels", NOWELE 26: 43-55. Seebold, Elmar

1970 Vergleichendes und etymologisches Worterbuch der germanischen starken Verben. The Hague--Paris: Mouton.

Smirnitskij, A. I.

1990 "Rhotacism in Old English and West Germanic loss of *z", General Linguistics 30: 195-206.

Tsiapera Maria (ed.)

1971 Generative studies in historical linguistics. Edmonton: Linguistic Research.

Verner, Karl 1876 "Eine ausnahme der ersten lautverschiebung", Kuhn's Zeitschrift 23: 97-130.

Welna, Jerzy

1996 English historical morphology. Warszawa: Wydawnictwa Uniwersytetu Warszawskiego.

Wright Joseph

1925 Old English grammar. Oxford: Oxford University Press.

ELZBIETA ADAMCZYK

Adam Mickiewicz University, Poznan