Hostname: page-component-74656f9884-npj9w Total loading time: 0 Render date: 2026-11-09T22:41:06.807Z Has data issue: false hasContentIssue false
  • English
  • Français

An explanation for the early phonemicisation of a voice contrast in English fricatives1

Published online by Cambridge University Press:  01 July 2009

STEPHEN LAKER
STEPHEN LAKER*
Affiliation:
Linguistics and English Language, University of Manchester, Manchester M13 9PL, Englandstephen.laker@manchester.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

Most handbooks and grammars contend that in Old English the voiced fricatives [v, ð, z] were merely allophones of /f, θ, s/ in sonorous environments. How these voiced fricatives became phonemes is debated among scholars. In this article, all previous accounts are critically reviewed. A new proposal is then presented, which explains the facts in a more direct way than previous theses. I argue that phonemicisation of a previous allophonic voice alternation in fricatives had already taken place in many areas of Anglo-Saxon England through language contact with Brittonic. Voiceless as well as voiced fricative phonemes existed in Brittonic at the time of contact, and language shift would have led directly to the phonemicisation of the previous allophonic variation found in early Old English.

Information

Type
Research Article
Information
English Language & Linguistics , Volume 13 , Issue 2: Re-evaluating the Celtic hypothesis , July 2009 , pp. 213 - 226
Copyright
Copyright © Cambridge University Press 2009

1 Introduction

Students of Old English must learn in which phonetic environments the fricatives f, þ, s have voiceless and voiced variants, since the distinction is not expressed orthographically.Footnote 2 As a rule of thumb, they are voiceless in initial and final positions, as well as in medial position when next to a voiceless consonant, yet they are voiced between vowels and between a vowel and a voiced consonant. A summary of the rules, with illustrative examples, is presented in table 1.Footnote 3

Table 1.

Distribution of voiceless and voiced fricatives in Old EnglishFootnote 4

Once the rules are learned, the variation does not cause too much difficulty; however, a number of exceptions to the general rules do exist. In particular, fricatives in prefixes and suffixes at morpheme boundaries are not voiced, e.g. þrym-fæst ‘glorious’, of-āscian ‘enquire’, o þ-īewan ‘show’, lang-sum ‘tedious’. Nor are they voiced in verbal roots after unstressed prefixes, e.g. ge-findan ‘find’, a-þenċan ‘devise’, be-singan ‘sing’. The status of the feminine abstract noun suffix -þu and the ordinal suffix -þa is especially debated. Bammesberger (Reference Bammesberger, Kastovsky and Bauer1988: 124) and Fulk (Reference Fulk2001: 70; Reference Fulk2002: 95) prefer not to explain voiceless fricatives in these suffixes by morpheme juncture, but as resulting from a rule, first suggested by Luick (Reference Luick1914–40: 845), that fricative voicing did not take place after unstressed vowels, e.g. streng-þ(u) ‘strength’ (< *strangiþu), fyl-þ ‘filth’ (< *fūliþu), seofo-þa ‘seventh’, eahto-þa ‘eighth’. By contrast, Dietz (Reference Dietz1997: 168–74) maintains that these two suffixes also fall under the morpheme juncture rule. He regards the productive suffixes -þu and -þa as lexical, thus akin to -sum in lang-sum and -fæst in þrym-fæst, contrasting with inflectional suffixes which do not inhibit voicing, e.g. hūs-es ‘houses’, luf-ode ‘loved’. He also argues that his account can explain a number of apparent exceptions to Luick's rule, which states that there was no voicing after unstressed vowels, such as the (assumed historical) voiced fricatives in forms like Old English adesa ‘adze’ and Temese ‘Thames’.Footnote 5 Be that as it may, a phonemic distinction /s/ vs /z/ has also been argued for on the basis of forms such as cærse (< *cræsse) ‘watercress’, hyrse (< *hruss(i)j-) ‘mare’ and Horsa (< *hrossa = a personal name), which show simplification of -ss- following r-metathesis but, apparently, no subsequent voicing of s in accordance with the rules set out in table 1.Footnote 6 Likewise, Late West Saxon blosm ‘blossom’ (< blostm) shows syncope of t but no subsequent voicing of s either. Although it is not possible to find minimal pairs to finalise the issue of phonemicisation for all fricatives, Bammesberger (Reference Bammesberger, Kastovsky and Bauer1988: 125) calls into question the allophonic status of Old English voiced fricatives and, more forthrightly, Fulk (Reference Fulk2001: 61) concludes that ‘the distribution of voiced and voiceless varieties cannot be determined on a purely phonological basis, and thus voiced and voiceless fricatives cannot have been allophones in Old English’.Footnote 7

In terms of relative chronology, voicing must have occurred before syncope (c. sixth century, Luick Reference Luick1914–40: 321, 847–8), because the dental past-tense marker d in weak verbs should otherwise have devoiced to t, had it come into contact with a voiceless fricative, e.g. cydde ‘made known’ (< *cyþide) or rǣsde ‘rushed’ (< *rǣside), but cyste ‘kissed’ (< *kyssida) (see Campbell Reference Campbell1959: 179–80; Hogg Reference Hogg1992: 284). Hogg (Reference Hogg1992: 284) points out that medial voicing of fricatives occurred after the lenition of medial *[x] > [h], because [x] should otherwise have voiced and become [Ɣ], which would have remained as such in Old English. This inference is plausible since we can observe that when a voiced velar fricative [Ɣ], which had not been subject to palatalisation, comes to be in word-final position, it devoices to [x], e.g. burh (< *burg), especially in Late West Saxon (see Hogg Reference Hogg1992: 285). It is unclear, however, when in prehistory [x] was lenited to a glottal aspirate [h] medially; nor do we know whether [h] first became voiced [ɦ] before disappearing (see also Suzuki Reference Suzuki1994: 78–9; Fulk Reference Fulk, Rauch and Carr1997: 31–3).

2 Previous explanations for the phonemicisation of voiced fricatives

Despite arguments for a possible phonemic voice distinction among some fricatives, e.g. /s/ vs. /z/, in Old English, the situation is different from that of other Old Germanic languages, and no clear motivation for a general early phonemicisation of voice in fricatives has been provided; thus other factors have traditionally been proposed. Following suggestions by Jespersen (Reference Jespersen1928, Reference Jespersen1933a, Reference Jespersen1937), many scholars hold French responsible for the phonemicisation of a contrast in voice. According to this thesis, the distinction of voice in French was quickly understood by English speakers, who were very perceptive of the contrast, rarely putting a foot wrong when adopting French loans with either voiceless or voiced fricatives. Most significantly, French introduced a contrast of voice in initial position.Footnote 8 Northern and most Midlands dialects adopted French loans with initial voiced fricatives into their phonemic system, e.g. vine (< Old French vi(g)ne ‘vine’),Footnote 9 while southern dialects, with their voiced initial fricatives (see footnote 3), introduced initial voiceless fricatives, e.g. fine (< Old French fin ‘settlement, payment’). However, numerous cases of voicing of initial voiceless fricatives from French loans were found in modern dialects until quite recently, suggesting that such voicing was even more widespread in Middle and Early Modern English. Derek Britton, in a discussion of Nielsen's Reference Nielsen, Laing and Williamson1994 paper, offers the following remarks about this discrepancy (discussion in Nielsen Reference Nielsen, Laing and Williamson1994: 56):

In the South you have this tendency for French words with initial [f] to be anglicised to fit the phonotactic rules of English by becoming voiced. So you get [varmər] ultimately out of [farmər]. But this did not happen in all the items. Whereas, seemingly outside the South, French loanwords with initial <v> never appear to have been devoiced to fit in with the pattern word-initially.Footnote 10

While, as Britton notes, the southern voicing of initial voiceless fricatives in French loans can simply be viewed as a phonotactic change, it is curious that speakers of northern and Midlands dialects did not devoice voiced fricatives to fit their native consonantal distribution too; after all, this is what happens when French, Dutch or English words with initial voiced fricatives are borrowed into Modern Frisian (see Visser Reference Visser1997: 49–50). Apparently, at the time of borrowing, the voice contrast was better established in northern and Midlands dialects of English than in the South.

A second explanation for the phonemicisation of the voicing contrast was set out by Trnka (1934–5, Reference Trnka1936, Reference Trnka1938). Trnka accepted that French loanwords were borrowed into Middle English with contrasting voiced and voiceless fricatives, but thought that such loans remained marked non-native members of a voice correlation. He contended that the voice contrast was ‘felt as a characteristic feature of foreign words and both z and v continued to be secondary variants of the phonemes s and f in the phonological structure of Middle English’ (Reference Trnka1936: 62). According to Trnka, the loss of unstressed vowels in final syllables ultimately phonemicised voiced fricatives in Middle English. Apocope brought voiced fricatives into final position, and rather than these becoming voiceless in accordance with the phonotactics of Old English, voice was retained because English speakers – somewhat out of character – took a dislike to homonymy, e.g. leaf: leave, sooth: soothe, rice: rise, etc. (Trnka Reference Trnka, Jones and Fry1936: 63).Footnote 11 Trnka concludes that phonemicisation of voiced fricatives proceeded from north to south, following loss of final -e in early Northumbrian Middle English and its later apocope in central and southern dialects, such as in London at the turn of the fifteenth century (1934–5 [1982]: 213). Only then, according to Trnka, were French loans which contained contrasting voiced and voiceless fricatives assimilated to the native stock.

A third explanation for the phonemicisation of the voice contrast in fricatives – usually attributed to Hans Kurath (Reference Kurath1956) – is the argument that simplification of intervocalic geminate fricatives, which remained voiceless, brought about a phonemic distinction of voice in fricatives in medial positions.Footnote 12 Consequently, contrasting voiced and voiceless singleton fricatives arose in intervocalic position. However, there are several problems with this analysis. As James Sledd pointed out in Reference Sledd1958, the long voiceless fricatives occurred only after short vowels, while, after lengthening of short vowels in open syllables, the voiced fricatives occurred only after long vowels. Since the distribution of voiceless and voiced fricatives can be predicted according to their alternate phonetic environments, it is doubtful whether this explanation could accurately be described as phonemicisation.

Finally, it has been argued that the three explanations for the phonemicisation of the voicing contrast outlined above need not be mutually exclusive. For instance, Lass (Reference Lass and Blake1992: 57–61; Reference Lass, Hogg and Denison2006: 62) thinks that all three factors contributed to the characteristically stable voicing contrast of English fricatives in initial, medial, and final positions of words. Thus, a voice contrast was established in initial position through French loanwords, in medial position by degemination, and in final position as a result of apocope of unstressed vowels. In spite of this, it would be fair to say that French influence is usually considered the earliest and most decisive phonemicising factor among scholars. Even Trnka acknowledged that French loans were generally adopted without reversion to the Old English phonotactic rules set out in the introduction to this article. Other scholars have criticised the degemination theory, such as Nielsen, who concluded that ‘the influx of French loanwords into Middle English played a much greater role than anticipated by Kurath’, such that ‘the phonemic opposition between voiceless and voiced fricatives could simply have been imported along with the numerous loans’ (Reference Nielsen, Laing and Williamson1994: 24–5). Indeed, the notion that French gave rise to the voicing contrast in English fricatives is now so well established that it is sometimes given as an illustrative example of phonological borrowing in works on language contact; see Thomason & Kaufman (Reference Thomason and Kaufman1988: 124, 308), Thomason (Reference Thomason2001: 79), and Winford (Reference Winford2003: 54).

3 An explanation based on language contact and language shift

In this section, I shall explain how language contact and subsequent language shift of native Brittonic-speaking populations in many parts of Britain from about the mid fifth century onwards could have led directly to the phonemicisation of voiced fricative allophones at an early stage in the development of English.Footnote 13 However, before outlining the linguistic facts which would support such a thesis, it is necessary to say a few words about possible early Anglo-Saxon–British language contact and language shift, since most handbook accounts talk variously about native Romano-British populations being wiped out or, more often, driven into present-day Wales, Cornwall, and Scotland by Anglo-Saxons (see van Kemenade Reference Kemenade, König and van der Auwera1994: 110; Baugh & Cable Reference Baugh and Cable2002: 50; Crystal Reference Crystal2003: 8).

As far as I can gather, the view that Romano-British populations were exterminated or expelled seems to be founded on two inferences. First, Celtic languages are only attested in the western Celtic fringe of Britain in later medieval times. Second, there is a dearth of British loans in English. The first argument equates ethnicity with language alone, a very precarious approach, as should be clear from more recent cases of large-scale language shift, such as in Ireland in recent centuries. The second loanword-based argument equates lexical borrowing with intensity of contact; this is now considered a misguided notion by experts of language contact, since it is widely believed that ‘lexical interference may be negligible in cases of interference through shift’ (Thomason & Kaufman Reference Thomason and Kaufman1988: 60 and passim). In situations of language shift one expects to find signs of structural influence, such as morphosyntactical and phonological/phonetic influences. In the present study, we shall assume that the intensity of linguistic contact and the amount of language shifting varied throughout Britain, but that some degree of population continuity at least stands for many areas of Britain, and especially those areas of Britain which were anglicised late, e.g. much of the north and west of present-day England was not anglicised until the seventh century (Jackson Reference Jackson1953: 208–9). It may turn out, however, that there was indeed more native survival in many other areas than has previously been thought (see Filppula et al. Reference Filppula, Klemola and Paulasto2008 and Laker Reference Laker, Aertsen and Veldhoen2008 for surveys of recent cross-disciplinary research into Anglo-Saxon–British contacts, including lengthy discussion of various examples of possible Brittonic influence on English).

Now, based on the assumption that native speakers of Brittonic, more specifically Late British (i.e. the Brittonic language spoken around the fifth–sixth centuries AD) as well as later attested Brittonic languages, learned what must have been an early, prehistoric form of Old English as a second language, an interesting process of second language acquisition must have ensued. Celticists are in agreement that [f, θ, s, v, ð] were all phonemes in Late British, just as in Old Welsh, Middle Welsh, and today in Modern Welsh. Accordingly, Britons learning early Old English would interpret the allophonic voiced fricative variants of early Old English as phonemes, thus imposing a phonemic split on English, at least in their own varieties of the language.Footnote 14 The development suggested here is well known from situations of second language acquisition. Weinreich (Reference Weinreich1953: 18–19) in his pioneering study Languages in Contact refers to this kind of external phonemicisation as ‘over-differentiation of phonemes’ and lists it as one of four basic types of possible phonological/phonetic interference to be expected in situations of second language acquisition and language shift. In Weinreich's own words, the process ‘involves the imposition of phonemic distinctions from the primary system on the sounds of the secondary system, where they are not required’ (Reference Weinreich1953: 18). A well-known modern parallel is that of English speakers who, when learning Spanish, typically identify [ð] – an allophone of [d] in Spanish – with the English phoneme /ð/; see Whitley (Reference Whitley1986: 14):

Consider the English minimal pairs den/then, day/they, ride/writhe. These show the distinction between the two phonemes /d/ and /ð/ in English, and it can be predicted that English speakers will be alert to the difference between She's riding and She's writhing. Spanish speakers, however, will not readily sense the difference since their language organizes [d] and [ð] as allophones of the same phoneme. There are no minimal pairs such as /rajd/ (ride) vs. /rajð/ (writhe) in Spanish, and in fact, the same word can be pronounced with either phone, depending on its environment: dónde is [dónde] in ¿Dónde está? but [ðónde] in ¿De dónde es?, and native speakers are unaware of the change in pronunciation unless they have received phonetic training.

Thus, the same type of acquisition process which takes place when English native speakers learn Spanish could, indeed must, have applied when speakers of Late British (and later Old British dialects such as Old Welsh) learned early Old English as a foreign language: the voiced fricative allophones of early Old English had the status of phonemes in Late BritishFootnote 15 and so would have been understood as such during the acquisition process. The distribution of the fricatives /f, θ, s, v, ð, z/ in Late British (c. fifth–sixth century) and Middle Welsh (c. 1150–1450) is outlined in table 2 (note that in Late British, as well as in later Medieval Welsh, fricatives contrast for voice initially,Footnote 16 medially, and finally).

Table 2.

Voiceless and voiced fricative phonemes in Late British and Middle WelshFootnote 17

One argument which can be raised against the proposal suggested here is the fact that there are no grounds for assuming that Late British had a voiced sibilant [z] either phonetically or phonemically (see table 2), such that phonemicisation of the Old English variants [s(:)] vs. [z], which only occurred in medial positions, cannot be directly explained by this analysis. Rather, one would expect to find some evidence for substitution of [z] with [s]. There is indeed some evidence that the /s/ vs /z/ contrast is the least stable in Modern English. Especially varieties of northern English and Scots show a tendency for Standard English [z] to be realised as [s] (see Jones Reference Jones and Jones1997: 324);Footnote 18 but /s/ and /z/ are rarely differentiated in the orthography at earlier periods except initially, hence it is difficult to gain much historical insight into this matter.

Importantly, it is known from more recent contact situations that if a phonetic feature is already utilised in a language contrastively, then it can be more readily transferred to other speech segments too. Illustration of this point is provided by the language shift of Irish to (Irish) English, which took place mainly in the eighteenth and nineteenth centuries. In brief, speakers of Irish had a larger consonantal inventory than English and often found equivalents or near-equivalents for Modern English consonants from their native Irish language, yet Irish possessed neither the voiced alveolar /z/ nor the voiced postalveolar fricative /ʒ/. Nevertheless, these voiced sibilants were generally acquired by Irish speakers on account of the fact that a voicing contrast was already well established for other fricatives in Irish, such that speakers effectively learned to voice native /s, ʃ/ in line with other fricatives. As Bliss (Reference Bliss and Trudgill1984: 137) explains: ‘No sounds corresponding to English /z, ʒ/ existed in Irish, but since these are merely the voiced equivalents of /s, ʃ/ they were easily articulated.’ Thus, the lack of a voiced sibilant /z/ in Late British and Medieval Welsh need not constitute as significant a problem as might first be presumed, given that voice was well established as a distinctive feature within the Late British consonantal system.

4 Five points in support of the proposed explanation

It should now be apparent that the proposed thesis of early phonemicisation of a voice contrast in fricatives through early language shift offers several advantages over previous accounts. First, it is evident that early phonemicisation of a voice contrast in fricatives provides an explanation for the early indications of phonemicisation even in Old English (Fulk Reference Fulk2001, Reference Fulk2002). It can explain, for instance, why already in Old English some voiceless fricatives came to stand in voiced environments through metathesis and simplification of geminate fricatives without apparently reverting to voiced fricatives in line with the inherited rules of free variation, e.g. cærse (< *cræsse) ‘watercress’, hyrse (< *hruss(i)j-) ‘mare’, compare also blosm ‘blossom’ (< blostm).

Secondly, phonemicisation of allophonic contrasts – especially at so rapid a rate and across all dialects – is more likely to occur as a result of language shift than through lexical borrowing and diffusion. To assume, as Thomason & Kaufman (Reference Thomason and Kaufman1988: 308) do, that English has ‘phonologized certain allophonic pairs though French influence’, even though ‘not a single foreign sound (phone) has been introduced from French’ requires special pleading. Rather, early phonemicisation of a voicing contrast in fricatives through British language shift accounts for why the English were so adept at adopting French voiceless and voiced fricatives in loanwords without any large-scale sound substitution. The introduction of numerous French loanwords certainly enhanced the number of voice oppositions in English, notably in the highly salient initial position; but the phonemic contrast of voice in fricatives was already in place long before any large-scale French borrowing.

Thirdly, it is now clear that preservation of voice in final consonants after apocope of unstressed syllables has little to do with avoidance of homonyms, pace Trnka (Reference Trnka1936: 63), but is due to the fact that phonemicisation of voice was already instituted before apocope occurred, e.g. bathe /beið/, house /haʊz/ (< Old English baþian, hūsian) vs bath /bɑːθ/, house /haʊs/ (< Old English bæþ, hūs). Indeed, there has been a tendency for English speakers to extend this contrastive feature analogically by creating new verbal forms from nouns, e.g. ModE mouth and grease, i.e. pronounced /maʊθ/ and /griːs/ as nouns, but /maʊð/ and /griːz/ as verbs (see Oxford English Dictionary: s.vv. for the derivation of these forms). Such a productive extension of the voice contrast in fricatives to morphology illustrates how crucial the distinction in voice has become in English, unlike its closest relatives Frisian, Dutch, and German.

Fourthly, although a voice contrast in some fricatives arose to varying degrees in other Germanic languages and dialects, in none was it established to such a degree and so early as in English. In particular, a contrast already arose to some extent in varieties of Middle Dutch and now in Standard Dutch, but it has to be agreed, with Nielsen (Reference Nielsen, Laing and Williamson1994: 23), that ‘the opposition between voiceless and voiced fricatives did not become nearly as well integrated into the Dutch phonemic system as it did in English’. In fact, even southern Netherlandic and Belgian Dutch varieties, with their characteristic initial voiced fricatives (unlike varieties of northern Dutch), display a great deal of internal and dialectal allophonic variation owing to external sandhi and other phenomena (see Schönfeld Reference Schönfeld and van Loey1970: 56; Collins & Mees Reference Collins and Mees1981: 120ff., 159ff.; Nielsen Reference Nielsen, Laing and Williamson1994: 54). In English, however, the voicing contrast in fricatives (often referred to as a fortis vs lenis contrast; see note 8) is steadfast, as in Late British and today in Modern Welsh.

Fifthly, it may be noteworthy that even the phonetics of Modern English and Modern Welsh fricatives in initial, medial, and final positions are largely analogous. In English and Welsh voiced fricatives are usually fully voiced between voiced sounds but only partially voiced or unvoiced in initial and final positions. Consequently, some phoneticians talk about English and Welsh employing a fortis vs lenis contrast, as this can be viewed as the primary distinguishing feature in fricatives as it also is for the plosives in these languages; see Cruttenden (Reference Cruttenden2001: 178 cited in note 8) and Ball & Williams (Reference Ball and Williams2001: 103–4).Footnote 19 Of course, it is difficult to ascertain whether such similarities in phonation result exclusively from early contacts, especially as it is likely that dialectal variation pertaining to the laryngeal features of the plosives in Brittonic and English did exist at earlier periods and still does exist to some extent today (see Schrijver Reference Schrijver, Filppula, Klemola and Pitkänen2002: 97–8 and Wells Reference Wells1982: 370, 409 respectively). Still, it is curious that now in most varieties of English and Welsh the laryngeal settings of fricatives are remarkably similar. In brief, the inventories of fricativesFootnote 20 in English and Welsh, their three positions of contrast, and their respective phonation types are strikingly similar from an areal linguistic perspective. If the tenor of this article is correct, this parallelism need not merely be viewed as coincidental.

5 Conclusion

Late British to early Old English language shift would lead directly to the phonemicisation of a previously allophonic alternation of early Old English voiceless and voiced fricatives. Such early phonemicisation can better account for early evidence for contrasting voiceless and voiced fricatives resulting especially from various reductive processes in Old English and Middle English. Furthermore, phonemicisation of an allophonic contrast, such as here between voiceless and voiced fricatives, is more likely to occur as a result of language shift than through superficial borrowing from French. Contrary to the generally held belief that French lexical influence introduced an awareness of voiced fricatives which promoted them to phonemic status in Middle English, the thesis outlined in this article explains why the English were so skilful at adopting French voiceless and voiced fricatives in loanwords without any large-scale sound substitution, namely by the very fact that an awareness of contrasting voice in fricatives had already been established in English well before the introduction of French loanwords. French lexical influence greatly enhanced the number of voice oppositions in English, notably in the highly salient initial position, but it was not as fundamental as Brittonic language contact and language shift were at setting in place a distinction of voice in English fricatives.

Footnotes

2 There is disagreement about the phonemic status of the contrast of voice in Old English fricatives; I therefore put the fricatives in italics or square brackets ‘[]’ in this introductory section, thus avoiding a phonemic analysis with slants ‘/ /’. Based on the explanation set out in section 3 of this article, the status of voiced fricatives need not have been the same in all varieties of Old English.

3 The rules in table 1 apply to all Old English dialects with one exception, namely that initial fricatives were voiced in some southerly dialects of Late Old English, and possibly earlier (Fisiak Reference Fisiak1984; Dietz Reference Dietz1990: 296–307; Goblirsch Reference Goblirsch2003 and Reference Goblirsch2005: 83–101). More copious Middle English evidence shows that initial voiced fricatives were found, very approximately, to the south of a line running from mid Shropshire in the west to the Essex–Suffolk border in the east. See further McIntosh et al. (Reference McIntosh, Samuels and Benskin1986: 549, dot maps 1180, 1181), Kristensson (Reference Kristensson1995: 199, map 16), and the discussion in Dietz (Reference Dietz and Fischer1989: 161–5).

4 The form scipfyrd ‘fleet’, in parentheses, is given only to complete table 1. The word is clearly a complex of two words scip ‘ship’ and fyrd ‘army’; no simplex word with the combination voiceless consonant + f exists in Old English. Note, too, that the status of the feminine abstract noun suffix -þu, e.g. in the form strencþu, is debated among scholars; see discussion in main text.

5 Liberman (Reference Liberman2008: 1) thinks that <s> was originally voiceless in these forms. His conclusion is mainly based on the spelling addice, which was common in the sixteenth and seventeenth centuries. However, he cannot explain why there should have been no voicing of the sibilant in Old English. With reference to Luick's rule (cited above), he goes on to comment: ‘One would rather expect voicing under no stress.’

6 Fulk concedes that the modern pronunciation of Horsa (i.e. with a voiceless sibilant) could be bookish. He notes, however, that voiceless sibilants are also found in place names containing the Old English elements Horsa, cærs, and gærs, and these, in his view, demonstrate retention of [s], e.g. Horsington (Lincolnshire) < OE Horsing + tūn ‘farmstead associated with a man named Horsa’; Cassington (Oxfordshire) and Carsington (Derbyshire) < OE *cærsen + tūn ‘farmstead where cress grows’; Garsden or Garsdon, (Wiltshire) and Garsington (Oxfordshire) < OE *gærsen + dūn ‘grassy hill’.

7 The second referee points out that it is difficult to determine when phonemicisation takes place, especially given the fact that very often the authorities’ criteria differ. Useful theoretical discussion is given to these issues in Scobbie & Stuart-Smith (Reference Scobbie, Stuart-Smith, Avery, Dresher and Rice2008). The authors discuss the complexities of assigning phonemic status to a number of vowels and consonants in present-day Standard Scottish English. Due to the fuzzy status of some segments, some phonologists prefer to view contrastiveness as a scalar property. The fact that voicing of Old English fricatives can be determined in part from the morphological structure of words may support the idea that voiced fricatives had quasi-phonemic status. However, presumed lack of voicing in words such as blosm ‘blossom’ (< blostm), i.e. where voicing is not determined by morphological structure, would seem to suggest phonemic status, though, admittedly, examples of this type are few in number.

8 Unlike French, where voiced fricatives are voiced throughout (Price Reference Price2005: 31–2), in Modern English voiced fricatives are usually fully voiced between voiced sounds only, as Cruttenden (Reference Cruttenden2001: 178) notes: ‘Like the voiced plosives and affricates, /v, ð, z,ʒ/ tend to be fully voiced only when they occur between voiced sounds, e.g. in cover, other, easy, leisure, a van, all that, by the zoo. In initial and (especially) in final positions, the voiced fricatives may be partially or almost completely devoiced, e.g. initially in van, that, zoo (i.e. with silence preceding) only the latter part of the friction is likely to be voiced, and finally in leave, breathe, peas, rouge /ru:ʒ/ (i.e. with silence following) the friction is typically voiceless, though the consonant remains lenis—-[, , , ].’ Since English fricatives do not simply contrast for voice, some phoneticians refer to fortis and lenis fricatives instead of ‘voiceless’ and ‘voiced’ fricatives, implying that the former type are produced with more force than the latter (see Cruttenden Reference Cruttenden2001: 178; Roach Reference Roach2000: 35, 49–50); however, these terms are not unproblematic (see Giegerich Reference Giegerich1992: 121–4). For convenience's sake, I shall continue to refer to voiceless and voiced fricatives in this article, because most of the literature on the historical development of English fricatives works with the notion of voicing.

9 Gąsiorowski (Reference Gąsiorowski2005) is, I think, right to assume that rare interdialectal loans, such as vat and vixen, showing initial fricative voicing of southern dialects (cf. note 3), do not explain the phonemicisation of /v/, but rather presuppose phonemicisation of /v/, pace Jordan/Crook (Reference Jordan and Crook1974: 192).

10 See also Wakelin (Reference Wakelin and Fisiak1988: 636) for more examples of voicing of initial voiceless fricatives from French loans.

11 This is not to deny that there were local tendencies to devoice fricatives in line with the existing phonotactic system, and a few words preserve such devoicing in Modern English, e.g. sheriff (< Old English scīrġerēfa). See Dietz (Reference Dietz1997: 482–4) for further examples and discussion.

12 Actually the idea was already suggested by Trnka (1934/5 [Reference Trnka and Fried1982]: 212), though in less detail.

13 The idea has been mooted before, but it has never been investigated in any detail. In his paper on parallels between English and Celtic, Raymond Hickey mentions that voice contrast is a general underlying trend of English and Celtic languages (Hickey Reference Hickey1995: 114), Schrijver (Reference Schrijver, Filppula, Klemola and Pitkänen2002: 105) makes a note that ‘the consonant system of Brittonic included close counterparts of all consonant phonemes that were present in early West Germanic, including the fricatives f, θ, χ, v, δ, γ’; and David White (Reference White and Tristram2006: 314) has referred to the ‘distinction of voice in [English, SL] fricatives’ as a Brittonicism.

14 I am aware that Anglo-Saxonists assume that early Old English also had a close counterpart of the labiodental fricative [v] in the form of a bilabial voiced fricative [β], an allophone of /b/. Early eighth-century Old English glossaries sometimes distinguish [v] as <f> (e.g. cefr ‘beetle’, wulfes ‘wolf's’, geroefa ‘reeve’) and [β] as <b> (e.g. gaebuli ‘tribute’, halbae ‘halves’, sceabas ‘sheaves’). But the large amount of inconsistency in this distinction suggests they had merged in the dialects of the scribes, though possibly not in the language of the original exemplar from which they were copying (Hogg Reference Hogg1992: 284). I assume that in a situation of language contact with Late British or even Old Welsh both [v] and [β] would have been replaced, either by a labiodental fricative phoneme /v/ or a bilabial fricative phoneme /β/. Note that it is agreed that both voiceless and voiced labial fricatives were originally bilabial in articulation (i.e. /ɸ, β/) in Brittonic rather than labiodental as in Modern Welsh (i.e. /f, v/). Indeed, David Willis argues that labial articulation of these consonants was still found in varieties of Old and Middle Welsh (see Willis forthcoming). I will use the more familiar symbols /f/ and /v/ in this discussion, however, because the place of articulation is not at issue.

15 Late British */v, ð, ɣ/ originate from lenition of Proto-Celtic */b, d, g/ in sonorous environments. Phonemicisation of */v, ð, ɣ/ occurred at latest when apocope and syncope destroyed the original conditions under which lenition occurred. Jackson dates apocope to the end of the fifth century and syncope to the mid sixth century (Reference Jackson1953: 695–6); however, Sims-Williams puts syncope in the fifth century (Reference Sims-Williams, Bammesberger and Wollmann1990: 245–7). British */s/ always existed as a reflex of unlenited Proto-Celtic */s/. */f, θ, x/ arose variously (see Schrijver Reference Schrijver1995: 460–1), though earliest in medial positions from various ‘s + consonant’ clusters; see table 2 and Schrijver Reference Schrijver1995: 374–458. By apocope these medial fricatives could also come into final position. Further, voiceless fricatives derive from spirantisation, dated mid or later sixth century (Jackson Reference Jackson1953: 696) or earlier (Sims-Williams Reference Sims-Williams, Bammesberger and Wollmann1990: 250). In particular, initial /θ/ arose via spirantisation, i.e. aspirate mutation (see Jackson Reference Jackson1953: 634–8). Further, */f, s/ also entered Late British in numerous Latin loans during the Roman period.

16 Initial voiceless fricatives (excluding */s/) cannot have existed even as allophones during the early Roman period (unlike voiced fricatives). It is thus possible that some early varieties of British Latin were possibly influenced by this constraint. Whether, then, southern varieties of Old English which show voicing of initial fricatives (also varieties of southern Dutch) could have been influenced by a process of Early British > Vulgar Latin substratal influence must be left for future research to determine.

17 The forms and reconstructions given here are largely taken from Schrijver (Reference Schrijver1995). Middle Welsh forms are cited, though many forms are attested in Old Welsh too, and examples from Medieval Breton or Cornish would have served the same purpose. The orthography of Middle Welsh (in contrast to Modern Welsh) is ambiguous; importantly, the graphemes represent the phonemic value of fricatives as given in the Late British reconstructions. Initial /θ/ is formed by aspirate mutation (AM), which is posited for Late British, as it is attested in Cornish, Breton, and Welsh. Initial /v/ and /ð/ are derived from soft mutation (SM). Although aspirate mutation and soft mutation are triggered by preceding words, e.g. a(k) ‘and’ or y ‘his’ (see table 2), phonemic status of /θ/, /v/ and /ð/ is assumed due to the fact that these are independent phonemes in medial and final positions.

18 Christian Kay, drawing particular attention to Scots and northern English dialects, also notes in the appended discussion to Nielsen's article (Reference Nielsen, Laing and Williamson1994: 57): ‘[s z] particularly is unstable.’

19 Since English plosives and affricates contrast in terms of them being fortis or lenis rather than voiceless or voiced, it could be presumed that the same opposition was simply transferred to English fricatives too. The assumption need not necessarily follow: German plosives contrast in terms of fortis vs lenis but German fricatives contrast for voice, i.e. the latter are regularly voiced not only medially but also word initially (see Beckman et al. Reference Beckman, Jessen, Ringen, Baumer, Montero and Scanlon2006: 76 and passim).

20 Part of the reason for the similar inventories arises through English preserving the interdental fricatives /θ/ and /ð/, which some scholars think could have resulted from Brittonic phonological support (see Filppula et al. Reference Filppula, Klemola and Paulasto2008: 119–20). Further, English borrowing has led many speakers of Welsh to integrate /z/ into their inherited fricative system (cf. table 2), thus: /f, v, θ, ð, s, z/. And many speakers of Welsh also now contrast /ʃ/ and /ʒ/ as a result of borrowing too (though this is more typical of southern Welsh since /ʃ/ had already developed there as a result of a native development). Thus, phonological influence has ultimately gone both ways in Britain, especially as a result of bilingualism. English, however, has not introduced any new fricatives, e.g. the Welsh lateral and uvular voiceless fricatives /ɬ, χ/, following modest Welsh lexical borrowing.

References

Ball, Martin J. & Williams, Briony. 2001. Welsh phonetics. Lampeter: Edwin Mellen Press.Google Scholar
Bammesberger, Alfred. 1988. Voice opposition in the system of Old English spirants. In Kastovsky, Dieter & Bauer, Gero (eds.), Luick revisited, 119–26. Tübingen: Gunter Narr.Google Scholar
Baugh, A. C. & Cable, Thomas. 2002. A history of the English language, 5th edn.Upper Saddle River, NJ: Prentice Hall.Google Scholar
Beckman, Jill, Jessen, Michael & Ringen, Catherine. 2006. Phonetic variation and phonological theory: German fricative voicing. In Baumer, Donald, Montero, David & Scanlon, Michael (eds.), Proceedings of the 25th conference on formal linguistics, 76–86. Somerville, MA: Cascadilla Proceedings Project. www.lingref.com, doc. #1435 (4 August 2008).Google Scholar
Bliss, Alan. 1984. English in the south of Ireland. In Trudgill, Peter (ed.), Language in the British Isles, 135–51. Cambridge: Cambridge University Press.Google Scholar
Campbell, Alistair. 1959. Old English grammar. Oxford: Clarendon Press.Google Scholar
Collins, Beverley & Mees, Inger. 1981. The sounds of English and Dutch. The Hague: Leiden University Press.Google Scholar
Cruttenden, Alan. 2001. Gimson's pronunciation of English, 6th edn.London: Arnold.Google Scholar
Crystal, David. 2003. The Cambridge encyclopedia of the English language, 2nd edn.Cambridge: Cambridge University Press.Google Scholar
Dietz, Klaus. 1989. Die historische Schichtung phonologischer Isoglossen in den englischen Dialekten: II. Mittelenglische Isoglossen. In Fischer, Andreas (ed.), The history and the dialects of English: Festschrift for Eduard Kolb (Anglistische Forschungen 203), 133–75. Heidelberg: Carl Winter.Google Scholar
Dietz, Klaus. 1990. Die Südaltenglische Sonorisierung anlautender Spiranten. Anglia 108, 292–313.CrossRefGoogle Scholar
Dietz, Klaus. 1997. Die stimmhaften Spiranten des Mittelenglischen und ihre phonologische Entwicklung. Anglia 115, 165–92, 469–86.Google Scholar
Filppula, Markku, Klemola, Juhani & Paulasto, Heli (eds.). 2008. English and Celtic in contact. London and New York: Routledge.CrossRefGoogle Scholar
Fisiak, J. 1984. The voicing of initial fricatives in Middle English. Studia Anglica Posnaniensia 27, 3–16.Google Scholar
Fulk, Robert D. 1997. Ambisyllabicity in Old English: a contrary view. In Rauch, Irmengard & Carr, Gerald F. (eds.), Insights in Germanic linguistics, vol. II: Classic and contemporary (Trends in Linguistics, Studies and Monographs 94), 29–45. Berlin and New York: Mouton de Gruyter.Google Scholar
Fulk, Robert D. 2001. Conditions for the voicing of Old English fricatives, I: Phonology. Interdisciplinary Journal for Germanic Linguistics and Semiotic Analysis 6 (1), 55–77.Google Scholar
Fulk, Robert D. 2002. Conditions for the voicing of Old English fricatives, II: Morphology and syllable structure. English Language and Linguistics 6 (1), 81–104.CrossRefGoogle Scholar
Gąsiorowski, Piotr. 2005. Zounds! – The rise of voiced fricative phonemes in Middle English. Paper presented at the 4th Medieval English Studies Symposium, Adam Mickiewicz University, Poznań, 27–28 November 2005.Google Scholar
Giegerich, Heinz J. 1992. English phonology: An introduction. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Goblirsch, Kurt G. 2003. The voicing of fricatives in West Germanic and the partial consonant shift. Folia Linguistica Historica 24 (1–2), 111–52.Google Scholar
Goblirsch, Kurt G. 2005. Lautverschiebungen in den germanischen Sprachen. Heidelberg: Winter.Google Scholar
Hickey, Raymond. 1995. Early contact and parallels between English and Celtic. Vienna English Working Papers 4 (2), 87–119.Google Scholar
Hogg, Richard M. 1992. A grammar of Old English, vol. 1: Phonology. Oxford: Blackwell.Google Scholar
Jackson, Kenneth H. 1953. Language and history in early Britain. Edinburgh: Edinburgh University Press.Google Scholar
Jespersen, Otto. 1928. Monosyllabism in English. Proceedings of the British Academy 14, 341–68. London: Humphrey Milford. [Reprinted in Jespersen 1933b: 384–408.]Google Scholar
Jespersen, Otto. 1933a. Voiced and voiceless fricatives in English. In Jespersen 1933b: 346–83.Google Scholar
Jespersen, Otto. 1933b. Linguistica. Selected papers in English, French and German. Copenhagen: Levin & Munksgaard.Google Scholar
Jespersen, Otto. 1937. Voicing of spirants in English. English Studies 19, 69–71.Google Scholar
Jones, Charles. 1997. Phonology. In Jones, Charles (ed.), The Edinburgh history of the Scots language, 267–334. Edinburgh: Edinburgh University Press.CrossRefGoogle Scholar
Jordan, Richard. 1974. Handbook of Middle English grammar, vol. 1: Phonology, revised and translated by Crook, E. J.. The Hague: Mouton.Google Scholar
Kemenade, Ans van. 1994. Old and Middle English. In König, Ekkehard & van der Auwera, Johan (eds.), The Germanic languages, 110–41. London: Routledge.Google Scholar
Kristensson, Gillis. 1995. Survey of Middle English dialects 1290–1350: The East Midland counties. Lund: Lund University Press.Google Scholar
Kurath, Hans. 1956. The loss of long consonants and the rise of voiced fricatives in Middle English. Language 32 (3), 435–45.CrossRefGoogle Scholar
Laker, Stephen. 2008. Changing views about Anglo-Saxons and Britons. In Aertsen, Henk & Veldhoen, Bart (eds.), Six papers from the 28th symposium on Medieval Studies held at the Vrije Universiteit Amsterdam on 15 December 2006, 1–38. Leiden: Leiden University Department of English.Google Scholar
Lass, Roger. 1992. Phonology and morphology. In Blake, Norman (ed.), The Cambridge history of the English language, vol. II: 1066–1476, 23–155. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Lass, Roger. 2006. Phonology and morphology. In Hogg, Richard & Denison, David (eds.), A history of the English language, 43–108. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
Liberman, Anatoly. 2008. An analytic dictionary of English etymology: An introduction. Minneapolis, MN: University of Minnesota Press.Google Scholar
Luick, Karl. 1914–40. Historische Grammatik der englischen Sprache, 2 vols. Stuttgart: Bernhard Tauchnitz.Google Scholar
McIntosh, Angus, Samuels, M. L. & Benskin, Michael (eds.). 1986. A linguistic atlas of late mediaeval English, vol. 1: General introduction, index of sources, dot maps. Aberdeen: Aberdeen University Press.Google Scholar
Nielsen, Hans F. 1994. On the origin and spread of initial voiced fricatives and the phonemic split of fricatives in English and Dutch. In Laing, Margaret & Williamson, Keith (eds.), Speaking our tongues: Medieval dialectology and related disciplines, 19–30, discussion 51–66. Cambridge: Brewer.Google Scholar
Oxford English Dictionary, online edition, Simpson, J. A. et al. (eds.), Oxford University Press. www.oed.com (15 July 2008).Google Scholar
Price, Glanville. 2005. An introduction to French pronunciation. Oxford: Blackwell.CrossRefGoogle Scholar
Roach, Peter. 2000. English phonetics and phonology: A practical course, 3rd edn.Cambridge: Cambridge University Press.Google Scholar
Schönfeld, M. 1970. Historische Grammatica van het Nederlands. 8th edn, ed. van Loey, A.. Zutphen: Thieme.Google Scholar
Schrijver, Peter. 1995. Studies in British Celtic historical phonology. Amsterdam and Atlanta, GA: Rodopi.CrossRefGoogle Scholar
Schrijver, Peter. 2002. The rise and fall of British Latin: Evidence from English and Brittonic. In Filppula, Markku, Klemola, Juhani & Pitkänen, Heli (eds.), The Celtic roots of English, 87–110. Joensuu: Joensuu University Press.Google Scholar
Scobbie, James M. & Stuart-Smith, Jane. 2008. Quasi-phonemic contrast and the fuzzy inventory: Examples from Scottish English. In Avery, Peter, Dresher, Elan & Rice, Keren (eds.), Contrast in phonology: Theory, perception, acquisition, 87–113. Berlin and New York: Mouton de Gruyter.CrossRefGoogle Scholar
Sims-Williams, Patrick. 1990. Dating the transition to Neo-Brittonic: Phonology and history, 400–600. In Bammesberger, Alfred & Wollmann, Alfred (eds.), Britain 400–600: Language and history, 217–61. Heidelberg: Winter.Google Scholar
Sledd, James. 1958. Some questions of English phonology. Language 34 (2), 252–8.CrossRefGoogle Scholar
Suzuki, Seiichi. 1994. Breaking, ambisyllabicity, and the sonority hierarchy in Old English. Diachronica 11 (1), 65–93.CrossRefGoogle Scholar
Thomason, Sarah G. & Kaufman, Terrence. 1988. Language contact, creolization and genetic linguistics. Berkeley and Los Angeles: University of California Press.CrossRefGoogle Scholar
Thomason, Sarah G. 2001. Language contact. Edinburgh: Edinburgh University Press.Google Scholar
Trnka, Bohumil. 1934–5. Je prokázán Verner ův zákon pro novou angličtinu? Časopis pro moderní filologii 21. 154–61. [Translated into English by Vilém Fried as ‘Can Verner's Law be applied to Modern English?’. In Trnka 1982: 210–18.]Google Scholar
Trnka, Bohumil. 1936. On the phonological development of spirants in English. In Jones, Daniel & Fry, D. B. (eds.), Proceedings of the Second International Congress of Phonetic Sciences, 60–4. Cambridge: Cambridge University Press.Google Scholar
Trnka, Bohumil. 1938. The phonemic development of spirants in English. English Studies 20, 26–31. [Reprinted in Trnka 1982: 224–31.]Google Scholar
Trnka, Bohumil. 1982. Selected papers in structural linguistics: Contributions to English and General Linguistics written in the years 1928–1978, ed. Fried, Vilém. Berlin and New York: Mouton.CrossRefGoogle Scholar
Visser, Willem. 1997. The syllable in Frisian. The Hague: Holland Academic Graphics.Google Scholar
Wakelin, Martyn F. 1988. The phonology of South-Western English: 1500–1700. In Fisiak, Jacek (ed.), Historical dialectology, regional and social, 609–44. Berlin and New York: Mouton de Gruyter.CrossRefGoogle Scholar
Weinreich, Uriel. 1953. Languages in contact: Findings and problems. New York: Linguistic Circle of New York.Google Scholar
Wells, John C. 1982. Accents of English, vol. 2: The British Isles. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
White, David L. 2006. On the areal pattern of ‘Brittonicity’ in English and its implications. In Tristram, Hildegard L. C. (eds.), The Celtic Englishes IV, 306–35. Potsdam: Potsdam University Press.Google Scholar
Whitley, M. Stanley. 1986. Spanish/English contrasts: A course in Spanish linguistics. Washington, DC: Georgetown University Press.Google Scholar
Willis, David. Forthcoming. Old and Middle Welsh. In Ball, Martin J. & Müller, Nicole (eds.), The Celtic languages, 2nd edn.London and New York: Routledge.Google Scholar
Winford, Donald. 2003. An introduction to contact linguistics. Oxford: Blackwell.Google Scholar
Figure 0

Table 1. Distribution of voiceless and voiced fricatives in Old English4

Figure 1

Table 2. Voiceless and voiced fricative phonemes in Late British and Middle Welsh17