In this series of posts I will put forward the following set of ideas. They are not entirely new or original, but I think that there is a strong philological case to support them.
In southern Britain in the fifth century there was a significant Latin-speaking Christian British population. This is evidenced by pre-550 Latin loanwords in Old English and in particular by Christian terminology. The loanwords are most likely to be due to this population adopting Old English as their first language. This may be why there is also an important group of religious terms coined in Old English rather than borrowed from Latin. And it also explains why there are virtually no British loanwords in Old English.
V. Other lexis of Latin origin from this period.
There are a number of other late Latin loanwords which seem also to have been borrowed into Old English between the settlement and the Augustinian mission. I give a selection below. It has been noted above that the variety of late Latin from which primitive Old English derived its loanwords presents some important phonological differences from the nearest recorded form of late Latin, namely Gallo-Romance, and that to explain some forms we have to assume that they go back to variants that differ from the recorded equivalents among the Romance languages. For example, in the case of biscop we may have to posit absence of the usual lowering of Latin /i/ to /e/ and in the case of preost we need to assume a divergent form */presbuter/. In the case presented below there are a number of other systematic and individual divergences. This constitutes the common thread that I believe guards against treating individual cases as special pleading.
Campbell §544(3) seems to assume that Old English ælmesse and Old High German alamuosan are cognate, i.e. that they descend from a common West Germanic ancestor. But there is no good reason why they should. ‘Alms’ is another Christian concept for which pagans could have no possible use, embedded as it is in a specific religious context. Additionally, there is no reason to assume a long vowel in the primitive Old English form of the second syllable of ælmesse as the Old High German form implies. The form borrowed into Primitive Old English could as well have been derived from a late Latin form similar to that underlying Old French aumozne (with /ɔ/) in the second syllable): something like *al(i)mosina. However, this does not fully account for the OE form; as Campbell points out, the double –ss- needs explanation. Clearly something odd has happened in the transmission. A late Latin form */alimosnja/ might account for it:
*/alimosnja/ > */alꞌmossja/
borrowed as */ꞌælmossjæ/
by i-umlaut */ꞌælmøssæ/
*/alimosnja/ > */aliꞌmossja/
borrowed as */ꞌalimossjæ/
by i-umlaut */ꞌæl(i)møss(j)æ/
The heaviness of the third syllable in the late Latin etymon makes it likely that Old English -messe did have i-umlaut and if so the word is a pre-550 loanword.
OED (first edition, not yet revised), at chalice n., says: ‘L. calix, calic-em cup, has appeared in Eng. in various forms. (1) Early OE. celic, genitive celces, corresp. to OS. kelik (MDu. kelec, kelc, Du. kelk), OHG. kelihh, chelih (MHG. and mod.G. kelch):—WGer. *kalik, an early (pre-Christian) adoption of L. calic-em. (2) The Latin word was re-adopted in later OE., in Christian use, as calic, cælic, cælc, whence early ME. calc, calch (cf. ON. kalk-r).’
The transmission implied in (1) cannot be correct. A West Germanic */kalik-/ would (by A1a, fronting) become Primitive Old English */kælik-/, then by A3, palatalization) become */kjælikj-/, then (by A4, palatal diphthongization) *kjealikj-/, and finally (by a5, i-umlaut) **/kjiel(i)kj-/ **ciel(i)c-. Therefore the borrowing cannot have been into West Germanic, and the borrowing into Old English must have been independent from that into Old Saxon, Old High Germanic, etc.
This word has numerous variant spellings in Old English.
A common one is calic-, without syncope in the inflected forms and evidently without i-umlaut or palatalization. This is presumably a relatively late borrowing directly from Latin with minimal change, but possibly also representing a refashioning of earlier forms of the word.
There is also an Anglian form cælc. This accords with (southern) early Middle English calch, calche, seen in:
If hit may so beo. Of þis ilche calche. nv forbere þu me. (Jesus College (Oxford) MS I. Arch. I. 29, f. 219r, the Passion of Our Lord, line 158 in R. Morris, Old English Miscellany (1872));
Ne mot ich nouht drynke peter vor þe. þene calch þat my vader. haueþ y-yeue me (Ibid., f.220r, line 202);
boc oðer belle calch oðer messe-ref. (Trinity Cambridge B. 14. 52, Sermo in Ps. cxix. 110, f. 152 in R. Morris, Old English Homilies of the Twelfth Century, Second Series, 1873, p. 215).
This appears to show (in reverse sequence):
The operation of A.5, i-umlaut, by which unfronted /a/ becomes /æ/
The operation of A.3, palatalization of final /k/, after /i/
Absence of A.3, palatalization of initial /k/, before the root vowel
Absence of A.1a, fronting of /a/, in the root vowel
It has a by-form celc. There’s no reason to think that this had palatalization of the first /k/ so the /e/ may represent either a borrowing of late Latin /kalike-/ as */kælik-/, becoming */kelik-/ by i-umlaut, or further raising of the /æ/ resulting from i-umlaut to /e/ which normally occurred before nasals.
This is another word with Christian associations that appears to have been borrowed into Old English between the settlement and 550.
The OED (unrevised) remarks:
‘not known in the other Germanic languages. By Grimm thought to be derived < Latin columba; but even if we take culufre as an earlier form (in which we are hardly justified), it is not easy to connect this phonetically with the Latin word. The thoroughly popular standing of the name is also against its adoption from Latin.’
The last sentence here begs the whole question. I contend that a significant part of the Latin loanword lexicon in early Old English came from a community that spoke late Latin as a vernacular, so we would expect such words to have ‘popular standing’.
It is indeed not easy to connect cul(u)fre with the Latin word, but it is hard not to believe, nonetheless, that they are connected.
Onions’s Oxford Dictionary of Etymology gives the columba etymology without apology, and MED gives the etymon as *columbula without comment, and this seems promising, with dissimilation of the second l, assuming
*/kolumba/ > */kolumbula/ > * /kolumbra/
and implying borrowing as
(i) */ꞌkulumbræ/ (with C4, raising of /o/ to /u/)
(ii) > */kulmbræ/
(iii) > (with elimination of the middle consonant of the group */kulbræ/)
(iv) > (with substitution of phonologically regular /v/ for medial /b/) */kulvræ/
(ii) > */ꞌkulubræ/ (with loss of the nasal element under low stress or by dissimilation)
(iii) > (with substitution of phonologically regular /v/ for medial /b/) */kuluvræ/
This sequence would make the loss of the /m/ occur on the ‘Germanic’ side of the loan transaction.
Could the anomalous loss of /m/ be attributed to the underlying Latin dialect?
We might explain it in terms of assimilation of –mb- to –bb- followed by simplification of –bb- to –b-. We might postulate simple loss of –m- in the polysyllabic environment of */kolumbula/. We might postulate loss of /m/ the consonant cluster in */kolumbra/, though that is not in any way an unusual Latin sound sequence.
Might some confusion have occurred between Latin columba ‘dove’ (or late Latin *columbula, *columbra) and coluber ‘snake’ (late Latin colubra, cf. Spanish culebra), by which the sense of the former took on the form of the latter? (The Vulgate of Matthew 10:16 (‘be ye wise as serpents, and innocent as doves’), however, uses columbae and serpentes.)
cylen and cealc
Old English cylen is a Latin loan word, but there are no cognates in Germanic, except for Scandinavian words which are arguably loans from English.
Latin culina does not continue into French or other Romance languages in any sense. coquina (late Latin cocina) replaced culina for ‘cooking place’ in Latin, and its reflexes were adopted in all Romance languages, as also in West Germanic whence it came into Old English as cycene.
As far as I know, culina is not recorded meaning ‘furnace’ or ‘kiln’ in Latin, and in all Romance languages a reflex of furnus is used for ‘kiln’(note for example the French place name Forcalquier (Occitan Fourcauquié) < Furnus calcarius, where the kiln is specified as used for lime-making; cf. discussion below).
Therefore some group using the reflex of culina originally in the sense ‘cooking-place’ shifted its sense to ‘furnace (either for burning lime or for firing pottery and tiles or for drying grain)’. It seems unlikely that Germanic speakers would have borrowed this word in its original sense (since for the same concept they borrowed the then current cocina) and then shifted its meaning. It is arguably more likely that a Latin-speaking group who, like all other Latin-speaking groups, had transferred ‘cooking place’ to cocina, retained culina, unlike other groups, but shifted it to the ‘furnace’ sense; the word and sense being then adopted into Old English.
This still leaves some explaining to be done: clearly a cooking-place would usually contain an oven, and a kiln is a kind of oven, but this does not give us the full picture: why did this postulated culina-using community cease to use furnus? West Germanic (or North Sea Germanic), of course, had the word oast (OE ast) for ‘furnace for drying’ (only specialized to hops or malt at a later date). However, like other words discussed here it points to a form of late Latin that has taken a divergent path from other known late Latin dialects.
It seems suggestive that the same community that alone preserved culina as their word for ‘kiln’, an oven in which lime, among other things, is made, also alone shifted the meaning of calx from ‘lime’ to ‘chalk’.
The reflex of Latin calx, which in all Romance and Germanic languages means ‘lime’, underwent a semantic shift at some point before literary Old English to ‘chalk’ (Latin creta); i.e. the word for the product was transferred to the raw material. (The word ‘lime’ which replaced calx originally meant any sticky substance, then (still in the earliest Old English) ‘mortar’, and then (again transferring the word for the product to the antecedent material) ‘lime’ (from which mortar is made).) One would expect such a shift to have happened among people who are familiar with the process of using chalk to make lime, but this shift in the meaning of calx, calcem could have happened among a Latin-speaking group. Since there is no reason to dispute that cealc is a common West Germanic word, and therefore already part of the Old English vocabulary, we would have to assume that the change of meaning occurred as a result of semantic borrowing from British Latin. In other words, primitive Old English */kjealk/ ‘lime’ was sematically influenced by British Latin */kalkje/ * ‘chalk’
Was lime-burning known in sub-Roman Britain? There was a British place name Calcaria ‘of limestone, lime works’ apparently denoting Tadcaster (D. N. Parsons ‘Sabrina in the Thorns’, TPS 109 122; compare the French place name above). If so, the two loanwords reflect a change in the application of two Latin words not paralleled in other Latin-using communities.
The supply of lime is the subject of Vindolanda tablet 314: ‘…missi quae calcem peteren[t / quam nobis commodasti / quas rogo continuo / iubeas onerari ut prim[o / mane nobis item…’ (‘I have sent…to get the lime which you have provided for us. I ask you to order them to be loaded without pause so that…to us early in the morning in turn (?)…’).
The OED (not revised) says:
‘The sense suggests formation on Romance *laigo < ecclesiastical Latin lāicus (see LAY adj.) with suffix -ede -ED suffix2; but it is not easy to see the phonological possibility of this. The attempt to trace the word to a late Latin type *lāicātus (u stem) is still more open to objection. It has been proposed to obviate the phonetic difficulties by assuming influence from the verb læwan to betray; but the sense is too remote, and læwede is not participial in form.’
It seems a pity to give up on a Latin etymology for this word.
1. Old English læwede is another stand-alone Christian Old English word. The other Germanic reflexes of laːicus, Middle Dutch leec, Old High German leigo have come from a form like */laigu/, or perhaps */laiigu/, which was received into the host languages with a diphthong /ai/ in the first syllable, with preservation of the velar of the Latin suffix, and without any ending corresponding to -ede, and they both mean ‘layman’, i.e. they are nouns not adjectives. There does not appear to be a common West Germanic etymon. In any case, what could such a word possibly mean to a pagan? Like my opening eight words, it belongs to a Christian milieu.
2. The Old English word contains æ2, as the Middle English forms show: leawede, Ormulum læwedd, and northern and Scottish lawed, laued, etc. Therefore the æ represents the i-umlaut of /aː/, which is the reflex of /ai/ in inherited Germanic words, but this change (A.1b) happened before or very early in the settlement period.
3. The /i/ which caused the umlaut cannot be represented directly by the first /e/ of the suffix -ede, since a short /i/ in this position would have suffered syncope. The suffix itself must be the usual adjective-forming suffix, primitive Old English */-oːdi-/.
4. We then have to explain two things:
(1) the shape of the suffix, and
(2) the appearance of the –w-.
Let us take as a starting point a late Latin */lajigu/. We could hypothesize that the transfer of the word into primitive Old English occurred among bilingual speakers who were accustomed both to Latin /-igu/, and to primitive Old English /-oːdi/, as adjective-forming suffixes, and were also accustomed to code-switching between late Latin and primitive Old English.
We could hypothesize that they replaced the suffix /-igu/ with the primitive Old English suffix /oːdi/, giving /lajoːdi/. For other words showing suffix substitution see munuc above (in which a similar element /-igu/ was replaced, in this case by /-uk/) and miltestre below. It would be easy to understand this substitution if we assume that the well established formula leornede ond læwede goes back to very early times.
To account for the /w/, we could either postulate (1) that a glide /w/ is generated before the rounded suffix /oːdi/ before or when it loses its rounding: /lajwoːdi/. The i-umlaut of /aj/ gives /æː/ (compare æce < */ajuki/), hence after umlaut we have /læːwøːdi/. Alternatively, the /w/ could be explained as a glide generated by the transition between the unrounded front vowel /æː/ and the rounded front vowel /øː/.
Or (2) pace OED, we could assume that an Old English element læw- influenced the formation. There are two candidates: (a) The root meaning ‘betray’. We should note that originally there would have been variation in the verb læwan between forms retaining w and those which lost it when original i followed (Campbell §406); hence the North. form bilede (< *læːwi-) cited by Campbell, but w was frequently restored by analogy, hence belæwde ‘betrayed’. Possibly some such alternation had the effect of introducing w into læwede. (b) Or, perhaps semantically more likely, it was the morpheme seen as the second element of limlæw ‘mutilation’, limlæwa, limlæweo ‘mutilated’, and in the rare word læw, lew ‘injury, weakening’: a folk-etymological idea that lay status incorporates some kind of deficiency (in holiness, presumably) seems quite possible (compare the later development of lewd with implications of lack of learning). It is true that the root vowel of these words is æ1 not æ2 < /aː/, and that therefore the falling together of the forms would only occur in West Saxon after the merger of the two æ’s, but the regional dialect in which the form emerged is most likely to have been a southern one.
The features of this hypothesis are similar to those seen in other etymologies proposed here:
- The word represents an isolated borrowing, not in West Germanic, but earlier than i-umlaut
- The word underwent some adaptations that are best explained by a bilingual or code-switching environment.
mattoc and adesa
Old English mattoc (variants meottoc, mettic, etc.) has no cognates in Germanic. OED3 suggests derivation from Vulgar Latin *matteuca club, cudgel (which is the likely etymon, via */maꞌttjuga/, of OFr. maçue, massue, etc.). The e variants, which point to i-umlaut, and therefore borrowing before circa 550, are found in Epinal and Corpus (for example) but do not seem to survive into ME. We might expect /maꞌttju:ka/ > mettuc with i-umlaut, but the a-form clearly predominated. To explain this we could assume late Latin */maꞌtteuka/ > */maꞌtteuga/ > primitive Old English */ꞌmatteok/ (with Old English /eu/ > /eo/) > /ꞌmattuk/ with reduction of low-stressed /eo/ to /u/ or /o/ as seen in weofod < weohbeod.
Like ‘kiln’ and ‘chalk’, this is a term of working life and not religion, again pointing to a context of borrowing between communities living on a similar, possibly shared, cultural level.
The etymology of Old English adesa is normally regarded as unknown; it has no Germanic cognates.
However, it seems significant that there exist Romance derivatives of Latin ascia ‘axe’ which refer to similar tools such as hoes and contain a suffix including a dental consonant: for example Spanish azada, Portuguese enxada, Catalan, Galician aixada ‘hoe’ from an underlying late Latin *asciata; meanwhile Italian ascia, Catalan aixa, Spanish azuela, Galician aixola all mean ‘adze’.
In order to connect the OE word with these we would need to posit consonantal metathesis at some stage in the transmission, perhaps in late Latin:
*/aꞌskjaːta/ > */aꞌksjaːta/ > */aꞌxsjaːda/ > */aꞌdaːxsja/ > */aꞌdaːsja/
or */aꞌskjaːta/ > */aꞌtsjaːda/ (cf. Older Spanish açada with /ts/) > */aꞌdaːtsja/ > */aꞌdaːsja/.
The exact fate of the second consonant cluster is unclear, but under low stress simplification to /-aːsja/ does not seem impossible: primitive Old English */ꞌadaːsja/ would regularly yield adesa. For the consonantal metathesis, compare Latin /akeːtum/ > */ateːkum/ > Old High German ezzih.
Given the existence of ‘mattock’ as an isolated loan from Latin in Old English, it is arguable that ‘adze’ could be, as it were, a stablemate. In this word again, an aberrant late Latin form seem to be implied.
This is generally regarded as a borrowing of Latin meretricem, but it is difficult to derive miltestre from meretricem by a direct route.
Old French miautrice shows that the first half of the word in Gallo-Romance or a dialect similar to it could have been the source of the OE word:
*/mereꞌtrikje-/ > (by dissimilation) */meleꞌtrikje-/
(by syncope) > */melꞌtrikje/-
borrowed (early, with raising: compare C.4) as */ꞌmiltrikje/
But the second half, the feminine suffix –trice-, which would in late Latin have been /tri(ː)kje/, is not easy to convert into */istræ/, and if we assume association of the /t/ with the first syllable the remnant /-riːkje/ is even harder to convert. (The Index Probi form menetris shows that the word could take anomalous forms in late Latin, but does not contribute a solution.)
I suggest that at transfer from late Latin to primitive Old English
(a) the word was analysed, at or soon after the point of borrowing into prOE, as *melt-trice; and
(b) the transferring speakers, being bilingual, knew that the feminine agent suffix, late Latin -trice, was the equivalent of the primitive Old English feminine agent suffix –istræ, and substituted the latter suffix
(c) moreover, raising of /e/ to /i/ probably occurred after borrowing and perhaps after suffix substitution
(d) and perhaps milt- was interpreted as related to mild.
The involvement of suffix substitution implies a code-switching community using both late Latin and primitive Old English.
Group 2 (I place less reliance on these, though I think they are good supporting evidence.)
The OED is dubious about the apparent Germanic cognates being descended from a common ancestor, and rightly so. If Latin balteum had been borrowed into West Germanic it would have had the form *baltia- which should give Old English (West Saxon) **bielte, **bylte, (non-West Saxon) **belte: compare (for first syllable) mieltan and (for second syllable) cyse. Presumably it was borrowed with /a/ later than the routine change of Germanic /a/ to /æ/; if so it could have been borrowed before Breaking, which would have affected /æ/ but did not affect /a/.
This is another word probably borrowed after the settlement and before i-umlaut, pointing to the existence of a Latin-speaking British population in contact with the settlers.
Old English cerfille / cerfelle. It is assumed that this is based on late Latin cerfolium (recorded in a gloss) < caerefolium (Pliny) < chaerephylum. Old High German has kervil(a) and kervol(l)a (German Kerbel). The Old High German kervil(a) does not look a possible cognate for the Old English since the latter has double l. The Old High German and Old English forms could really only be descendants of a West Germanic ancestor if they showed reflexes of –ulya-. Neither the Old High German nor the Old English forms satisfy this requirement. The Old High German kervol(l)a could be a reflex of a Latin form in –olja- if borrowing occurred later than raising of /o/ to /u/ before /i,j/. We can therefore treat the OHG and OE forms as independent borrowings.
As regards the first half of Old English cerfelle, by the time of the borrowing late Latin palatalization of */ker-/ > */kjer-/ must have taken place. This is reflected in the English form. There are no variants with cio- or ceo-, which points to the loan occurring later than the period of breaking. There is no variant with cie-, which would show that Palatal Diphthongization had occurred, but there is a variant with cy-, which could imply it; though it could also be a late development of e between a palatal and r. The variant with cæ- is a little difficult to explain. Could the word have come into Old English with long open /ɛː/, variously shortened as /e/ or /æ/? Was the Latin vowel, unlike in the gloss cited above a long mid vowel /ɛː/? Or was it a short vowel but especially open before r?
The second half implies */-voljæ/ with unraised /o/ before /j/ and so it belongs to the same stratum of loans as ele.
The double l, however, makes the word look as if it has undergone West Germanic doubling of single consonants before /j/. To explain the doubling we would need to assume forward assimilation of /lj/ to /ll/ some time later than I-umlaut. This assumption could be extended to native words such as sellan and would explain the absence of breaking in such words. If sellan remained *saljan (with a single l) until after the incidence of I-umlaut, just as nerian remained *narjan, this would explain why it did not undergo breaking, in contrast to fiellan which was already *fallian (with a double l) at the time of breaking. We would only need to postulate the same preservation in primitive Old English of /j/ after /l/ that is already postulated for /j/ after /r/ in words like nergan / nerian (the difference being that there was no forward assimilation in the latter).
Old English cellendre. It is assumed that this is based on late Latin coliandrum < Latin coriandrus, -um. Old High German has cullantar, chullentar with raising of /o/ > /u/; this is evidently not a direct cognate of the Old English word, which, to share a common ancestor, would have to have the form **cyllendre. The Old English word shows lack of vowel harmony like the two previous words, but like them must have been borrowed before umlaut. Doubling of l in the first syllable can be explained in the same way as doubling in the second element of cerfille: /kolja-/ > /kølja-/ > /kølla-/ (by assimilation) > celle-.
The names of the above two culinary and medicinal herbs might have been expected to be borrowed into Old English within the context of monastic cultivation, and hence belong to the post-Augustinian period, but if the dating followed here is correct, they would have already been in use among a Latin-speaking community in Britain and have been transferred into Old English earlier than this.
This is a very odd reflex for Latin cochlear(e). If it were a later ecclesiastical Latin loanword one would expect Latin cochlear(e) to be represented by something like **cocliar / coclier. Campbell (§518 f.n. 1) says that Latin o was normally developed to u before j in hiatus and that ‘the native suffix –ǣri- replaced the sounds developed from Lat. –ear before mutation took place.’ This implies an early loan, when vowel harmony was still operating. Medial –c- has been retained, or rather, Latin /k/ must have > /g/ but the latter must have been replaced by /k/ in Primitive Old English. Latin cochlea(ː)r(e) > *koklja(ː)r > *kogljar > prOE *kukliær? > *kuklæːri. None of this explains the second –u- in the Old English form. Is an epethetic –u- between an obstruent and l expected in Old English, or is it more likely to have happened in late Latin? Is it linked in any way to the absence of the –i- expected after a consonant cluster in primitive Old English? Could the suffix alteration have occurred in late Latin? In the Index Probi, there is an entry ‘cochleare non cocliarium’. It would not be unreasonable to assume the latter as the starting-point. Could the Old English form go back to late Latin *cocularium, i.e. with a further (and dialectally idiosyncratic) alteration of the middle two syllables? (This form does seem to occur in very much later documents.)
Eced is probably not a West Germanic loanword. German eßig cannot be a cognate as its underlying form (before the High German consonant shift) must have been *atik- (with metathesis) < *akit-, without intervocalic voicing of t. Before the Old English word was borrowed the /t/ of Latin acetum must have become /d/ by intervocalic voicing (end of the fourth century).
By the time this change had occurred, Latin /k/ before front vowels had already been palatalized (third century). In the Gallo-Romance ancestor of Central French, palatalized c had already reached the condition of the affricate /ts/ by the end of the same century, so clearly if the loan were from Gallo-Romance, it could not have come from a central dialect. It could have come from a northern form of Gallo-Romance, such as the ancestor of Picard, in which the first palatalization was to /t∫/ rather than /ts/, or indeed from a variety of late Latin with /t∫/ or even a less advanced palatalization.
For I-umlaut to have occurred, the second vowel of the word must have been /i/, or more likely, /iː/. The simplest assumption is that (1) late Latin /eː/ was still intact and that primitive Old English, lacking a comparable phoneme, substituted /iː/. Possibly (2) /eː/ had lost its length and become /e/; the identification of this with prOE /i/ rather than /e/ is more problematic. Or as (3), it could be assumed that diphthongization, as in Gallo-Romance (where it is dated to the fifth century) had produced /ei/ which was identified in PrOE with /iː/.
Old English ele is based on a late Latin form such as *oli. It must have been borrowed at a stage when vowel harmony, causing raising of mid vowels in Latin loanwords, no longer operated, since it becomes Old English */øle/ > ele, whereas earlier the form would have been *yle (<*uli). Vowel harmony seems to have been operative still during the period when some of the other loans discussed above occurred, so this has to be a relatively late borrowing; but as it shows I-umlaut, it belongs, on our dating, to a period before the mid sixth century.
The question arises whether this should be regarded as originally part of the Christian vocabulary: i.e. whether to envisage the liturgical use of oil as the rationale for the loan or simply oil as a general-purpose commodity.
The OED, under ELECAMPANE, observes: ‘In Old English inula was (corruptly) adopted as eolone ( < earlier *iluna).’ This form, with metathesis of n and l, might conceivably have been influenced by Latin helenium, but the vocalism (with /una/ as against /eniu/) is rather against this. There is no Germanic parallel to the form eolone, whereas the metathesis is well supported in Romance. Once again, there is a Latin loanword in Old English in a form not paralleled in the cognate languages but showing a sporadic sound change likely to go back to late Latin.
The OED says that ‘Old English feorm strong feminine’ (which it derives < ‘prehistoric *fermâ’) is ‘not found outside English, and no satisfactory Germanic etymology has been proposed. On the assumption that the primary sense was “fixed portion of provisions, ration”, it would be admissible to regard the word as < late Latin firma , and so ultimately identical with FARM n.2 In Domesday Book firma unius noctis is equivalent to anes nihtes feorme of quot. c1122 below; and medieval Latin writers in England used firma in the sense of “banquet”. If the hypothesis of its Latin origin be correct, the word must have been adopted at a very early date: it occurs frequently in the oldest poetry. The derivative feormian to feed, is found in the Corpus Glossary a800 (‘fovet, feormat, broedeþ’, the corresponding Old High German gloss. ‘formot, fofet’ in St. Gall. MS. 913 may be derived from an Old English source, the verb being otherwise unknown in Old High German.’
This is another word which is (a) not found outside Old English, and therefore may not have been borrowed in continental West Germanic, and (b) must be of early date.
side ‘silk’ appears to date from the same time as eced, i.e. on the basis of –d- later than about 400. If this has /iː/ the probability of the retention of Latin /eː/ and its identification with prOE /iː/ (option (1) at eced) seems greater. (Contrast, for example, beta, which is probably a later loan from ecclesiastical Latin.)
ynce / yndse and pylece
OED (at INCH n., unrevised) states: ‘Old English ynce < *unkja, < Latin uncia twelfth part, inch (compare ounce n.1). A word of early adoption, not in the other Germanic languages.’
Nothing in the form of this word precludes it from being a West Germanic borrowing from Latin, except the absence of cognates in the other dialects, but this may be significant. OED’s ‘early adoption’ presumably refers to the presence of palatalization and I-umlaut. But what are we to make of the doublet yndse ‘ounce’, which shows the development of Latin /kj/ to /ts/ but also has I-umlaut? Does the presence of /ts/, which mostly seems to occur in late loanwords from ‘French’ (i.e. late Gallo-Romance), e.g. plæce), mean that I-umlaut has to be redated to a later time?
OED3 treats yndse as an early variant of OUNCE n. (revised): ‘In Old English < an unattested post-classical Latin variant (showing -ts- for -c- before i) of classical Latin uncia (see below)… Compare Old Frisian enze (which on formal grounds could be taken as cognate with Old English, indicating early borrowing), Old High German unza (Middle High German unze, German Unze)’.
Do we therefore have to treat yndse as later than ynce? Not necessarily. The High German form may not be relevant, as it lacks I-umlaut. The apparently cognate Old Frisian form could point to continental borrowing. We could assume that the postulated post-classical Latin */untsia/ was already available for borrowing into West Germanic and was continued in Old English. And we can assume that the more ‘primitive’ Latin form /uŋkia/ was the one still in use in Britain and borrowed into primitive Old English.
‘Pilch’, OE pylece, pylce, may support the foregoing argument. Like ynce, it seems to have no continental cognates, and like it also, the Latin termination –cia appears as –ce with palatal c: in the dialect of late Latin from which the word is borrowed, palatalization of –cia had not developed much further than -/kja/, and certainly not as far as /-ts(j)a/, which is evidenced as a further step in Old French pelice, pellice, Old Occitan pelissa (a1149), Catalan pellissa (14th cent.; 986 as pellicea), Spanish pelliza (1283 as peliza; 1207 in sobrepelliza; 1161 as belîssiya in a Mozarabic text); the only Romance form retaining the /t∫/ stage is Italian pelliccia (14th cent.; 12th cent. as pelixa denoting a pelt). The same applies to ounce: Portuguese onça unit of weight, coin (1114 as †unza), Spanish †onça (first half of the 13th cent.), onza (1250) unit of weight, gold coin, Italian oncia gold coin (1231–50 as †onza), very small amount (a1313), unit of weight (a1321), Catalan unça, †onça unit of weight, coin (13th cent.), Old Occitan onsa (a1240; Occitan onça) unit of weight, one-eighth of a minute.
Taken together, it could be deduced that in the dialect of Latin from which pilch and inch were borrowed the sequence in question had not advanced much further than /kja/, and that these are insular loans comparable to eced; yndse, which is arguably a continental loan, reflects a form of Latin close to that from which the major Romance languages descend, in which the change cia > /ts(j)a/ had occurred.