CAPITULO II MARCO TEORICO
COMO SE REALIZA?
Through a series of changes which can be viewed collectively as a form of palatalisation, /sk/ generally became /ʃ/ in Old English (except medially before back vowels and, in most instances, finally after back vowels). Representative examples are sċip ‘ship’, wasċe ‘I wash’, fisċ ‘fish’, with palatalisation (palatalised /sk/ is indicated in this section as 〈sċ〉); no palatalisation is found in ascaþ ‘ he asks’, tusc ‘tusk’ which retain /sk/, as confirmed by Modern English pronunciations. There is a general consensus that the change took place via a number of stages, probably [sk] > [sc] > [sç] > [ʃ] (see Hogg 1992a: 272, Minkova 2003: 193). The problem is that it is difficult to put a date on each stage of the proposed development and regional and sociolectal differences must also have existed. Still, the frequent appearance of a diacritic 〈e〉 after the cluster 〈sc〉 in Old English sources – e.g. sċeacan ‘shake’, sċeamu ‘shame’ – is generally regarded as confirmation that a palatalised stage was reached, though it is of course unclear from such spellings precisely which supposed stage of palatalisation was reached, i.e. [sc], [sç] or [ʃ] (see Minkova 2003: 133). One further way of gaining information about the palatalisation of /sk/ has been to investigate the effects of Scandinavian influence on English. It is generally held that several Old English place-names had a palatalised initial [ʃ] (< /sk/) but that this was later replaced with /sk/ by Norse speakers who had difficulty pronouncing /ʃ/. Representative examples are, for instance, Skipton (North and West Riding of Yorkshire) and Skirlaugh (East Riding of Yorkshire), instead of more frequently attested Shipton (Hampshire, Oxfordshire etc.) and Shirley (Derbyshire, Hampshire, Warwickshire), from Old English sċīptūn ‘sheep farm’ and sċīrlēah ‘wood belonging to the shire’ respectively. Yet, as Ekwall (1963: 89) argued years ago, it is unlikely that /sk/ had developed as far as /ʃ/ during the time of Scandinavian contacts (ca. 900) ‘for it is not quite easy to understand the common substitution of sk, if the stage [š] had already been completely reached’. According to Ekwall, if the stage /ʃ/164
had indeed already been reached, a more natural substitution for this would have been Scandinavian /s/.165 Consequently, Ekwall thought an intermediate stage of palatalisation, e.g. [sç], would have been more prone to Scandinavian replacement with /sk/. However, as an alternative scenario we may wonder
164 I use the International Phonetic Association symbol /ʃ/ instead of Ekwall’s [š]. 165
Some such substitutions are posited by Ekwall (1963: 89), e.g. Silpho (North Riding of Yorkshire) < OE scylfhōh ‘ridge with a peak’.
153
whether the same blocking effect on the development of /sk/ to /ʃ/ could at least in some instances have resulted from earlier British influence, for, just like Norse, [sc], [sç] and [ʃ] were equally foreign to Late British speakers, who only had the cluster /sk/. The problem here is, naturally, that both Scandinavian and British influence can be posited throughout most of northern England, and so it is not easy to decide whether a British or Norse pronunciation of a place-name ultimately gave rise to a modern /sk/ pronunciation.
A more convincing argument for Brittonic influence can be made in areas where Norse influence was not particularly strong. In a recent article, Kristensson (2005) has focused on the frequency of place-names with initial /sk/ in Devon, where in fact many place-names (approximately one sixth of those listed in Gover et al. 1931–32) show unexpected initial /sk/, e.g. Landskerry, Scarhill, Skillaton, Score, Scruell. After surveying the evidence, Gover et al. 1931–32: 100) conclude that ‘the distribution of the names showing this development is such as to make it impossible to suggest that it is due to Scandinavian influence, which might perhaps be expected in some of the coastal districts’. Instead, Kristensson argues that Cornish speakers in Anglo-Saxon times (as well as subsequent arrivals of Cornish speakers after the Old English period) could instead have forestalled or reversed the development of /sk/ to /ʃ/.
Aside from place-names, hundreds of English words with initial /sk/ have also been viewed as being the result of Scandinavian influence. Numerous words with initial /sk/ are explained as being either Scandinavian loanwords with /sk/ (e.g. sky, skirt, skull) or as being English native words that were subject to Scandinavian phonological influence (i.e. Norse speakers reversed an earlier unfamiliar palatalised variant such as [sc], [sç] or [ʃ]). Yet in many instances it is not easy to decide whether we are dealing with a Norse loan or the effects of putative phonological influence, e.g. skin could simply be a Norse loanword (cf. ON skinn) or it could be a native English word (cf. OE sċinn) that was subject to the same substitution /sç/ > /sk/ as posited for some place-names (see Hogg 1992a: 275–6). Wright (1905: 247) noted that even when non-dialect words such as scab and scaffold are taken out of consideration, the English Dialect Dictionary contains 1,154 words with initial /sk/, leading him to conclude that ‘either the dialects contain a far larger number of Norse words than is generally supposed, or else it is not certain that initial sc has under all circumstances become ʃ in native words in the dialects’. If the
latter possibility is indeed true, one wonders whether Late British influence may also have blocked the development of /sk/ to /ʃ/ in some English dialects. On the whole, some degree of British influence (in addition to just Norse influence) does appear to be a realistic possibility and could just as easily help to explain the presence of doublet forms such as skift vs. shift and skelf vs. shelf in dialects of Cumberland, Westmoreland, Yorkshire, Lancashire, Scotland and Ireland (see Wright 1905: 247). To sum up, many place-names and dialect words that have /sk/ instead of expected /ʃ/ could result not only (or exclusively) from the influence of Norse speaking settlers but also from earlier Brittonic influence on the English language.
154 7.2.3 /hC/ clusters
According to most handbooks and grammars, (Pre-)Old English started out with four preaspirated clusters /hn, hl, hr, hw/ (spelled 〈hn, hl, hr, hw〉 in Old English), these later became voiceless sonorants /n̥, l̥, r̥, w̥/, before ultimately merging with the already existing voiced sonorants /n, l, r, w/. In a detailed discussion of the orthographic evidence, Dietz (2006: 245–65) has reasoned that the (usually) sporadic appearance of 〈nh, lh, rh, wh〉 spellings (i.e. 〈h〉 follows the sonorant) around the turn of the first millennium could indicate that the preaspirated sonorants had become voiceless monophonematic sequences at around this time.166 Probably soon after, in the eleventh and twelfth centuries (though later in Kentish and some other southerly dialects), /n̥, l̥, r̥/ merged with existing /n, l, r/. In Southern and South Midlands dialects, voiceless /w̥/ also seems to have become /w/ by about 1200, though only in unaccented function words, while the merger of /w̥/ at /w/ in accented words began around 1300 and continued for many centuries (Dietz 2006: 257–9). In northern dialects the development of OE /hw/ was very different and will be dealt with separately below.
Before investigating what effect, if any, contact with Late British may have had on the development of OE /hn, hl, hr, hw/, it will be useful firstly to survey what motivations for the loss of these initial clusters have been suggested in the literature. In the main, the development of the preaspirated consonants has been viewed as a natural process of phonological change, as such /hn, hl, hr/ and somewhat later /hw/ simply dropped out of usage over time. Since the same merger of /hn, hl, hr, w/ with /n, l, r, w/ took place in other closely related Germanic languages – Frisian, Dutch and German – at roughly the same period,167 their development in English does not arouse particular suspicion. Only the fact that several varieties of English have held on to voiceless /w̥/ until the present day is noteworthy. Vachek (1976: 205–06) thought that the loss of the voiceless sonorants /n̥, l̥, r̥/ was unsurprising, due to their rarity and lack of functional load, while the delayed loss of the voiceless approximant /w̥/ could be put down to its important signalling function in a small but important set words, namely interrogative pronouns and adverbs who, what, which, when, where, why, etc. Alternatively, Lutz (1991) viewed the loss of the Old English preaspirated sonorants as part of an ordered process of /h/ loss in the language, arguing that /hw/ was retained longer than /hn, hl, hr/ because the sequencing of /h/ + /w/ is universally more preferred, and consequently more frequently attested, than /h/ + /n, l, r/.
Another impetus for the loss of /hn, hl, hr/ clusters has been suggested by Schreier (2005), who argued that while there is some evidence for early loss of preaspiration in Old English, the rapid increase in 〈l, r, n〉 spellings between 1080 and 1200 implies that Norman French contact provided an extra catalyst for the loss of the preaspirated sonorants and/or voiceless sonorants. A weakness of Schreier’s account is that it is too Anglocentric. A similar sudden elimination of 〈hn,
166 Dietz (2006) also makes a convincing case here and elsewhere in his monograph that the 〈wh〉 spelling (and often other similar spellings where a consonant is followed by 〈h〉) was influenced by Anglo-Latin scribal practices rather than those of Anglo-Norman.
167
155
hl, hr, hw〉 spellings also occurs in Old High German in the course of the ninth century and in Old Dutch in the eighth and ninth centuries (see Braune/Reiffenstein 2004: §153, Quak 2000: 715), without the effects of Norman Conquests into these territories. Importantly, Schreier does not explain why French speakers apparently did not also accelerate the loss of /hw ~ w̥/ too, which, according to his reasoning, would be expected since (Norman) French did not posses /hw/ either.168
Let us now address what effect contact with Brittonic languages might or might not have had on the development of the Old English preaspirated sonorants. First the possible effects of Late British on /hn, hl, hr/ will be considered, since aspiration in these clusters was lost earliest. In a situation of language shift, two replacement strategies for /hn, hl, hr/ may be envisaged. Firstly, Pre-Old English /hn, hl, hr/ could have been replaced by Late British /xn, xl, xr/. It is agreed that Late British /k/ became a velar fricative /x/ after certain words (e.g. a(c) ‘and’, ni(d) ‘not’, see 4.2.4). Thus, spirantisation of /k/ in the consonant clusters /kn, kl, kr/ yielded /xn, xl, xr/, and clusters of the type /xC/ would certainly have constituted close counterparts for Pre-Old English /hC/-type clusters.
As an alternative scenario, we may consider the existence of preaspirated or voiceless rhotic and liquid /hr ~ r̥/, /hl ~ l̥/ in Late British (Jackson 1953: 480); however, there is neither orthographic evidence from Medieval Brittonic languages nor phonetic evidence from modern Brittonic dialects to suggest there was ever a preaspirated or voiceless nasal /hn, n̥/ in Late British. Thus, Pre-Old English /hl, hr/ could have been replaced by the Late British voiceless liquids /l̥, r̥/ and a new voiceless nasal /n̥/ could have been created on their analogy. Let us now consider in greater detail the theoretical and evidential basis for this scenario. The origin of the preaspirated or voiceless liquids in Late British is complex and based mainly on the evidence of Welsh. The best hypothesis seems to be that Welsh /r̥h
/ results from external sandhi (*/-s/ + /r-/ > */-h/ + /r-/ > */hr̥-/ > Modern Welsh /r̥h
-/) but became the default word-initial rhotic (see 4.2.2.3). A voiceless allophone /hr̥, r̥/ is also retained in some Breton dialects and appears to have survived in Cornish until the tenth century. The same sandhi phenomenon may also have applied to other consonants, giving rise to */l̥/ (< /hl/ < */s/ + /l/) and, in theory, possibly */n̥/ (and even /m̥/). Furthermore, due to the fact that /s/ generally changed to /h/ in Brittonic in sonorous surroundings (see 4.2.2.2), the prehistory of /s/ + consonant clusters would seem to provide a good breeding ground for preaspirated clusters too. It is agreed that PCl. */sl/ yielded a voiceless lateral in Welsh /l̥/, e.g. W llu ‘troop’ (< Late British *Lʉ̄ɣ < PCl. *slouga).169 On the basis of
168
Other issues that beset Schreier’s proposal concern the numbers of Anglo-Norman settlers and their interaction (or lack of it) with the native English masses and the fact that French influence on English is not typically found in written English of the period 1080–1200.
169
By a similar token, one may consider the developments of the Proto-Celtic initial clusters */sr, sn, sm/. In fact, there is no confirmation that these yielded Late British */hr, hn, hm/. Rather, */sn/ and */sm/ appear to have merged with /n/ and /m/ respectively, while */sr/ appears to have produced either /str/ or /fr/. See Schrijver (1995: 440–4) for detailed discussion on this point.
156
medieval Welsh, as well as possibly evidence from Cumbric, the voicelessness of both of the lateral and rhotic was sharpened and increased to the extent that they are now in present-day Welsh (Jackson 1953: 480). In brief, it is likely that Late British had equivalents to /hl, hr/, though these were by no means as prevalent as they came to be later in Welsh. There is no scribal evidence to suggest that /hn/ or /n̥/ existed in Late British or in later attested Brittonic languages, though it seems impossible to rule out that it once existed as a result of the sandhi phenomenon described above. Nevertheless, even if /hn/ or /n̥/ had not existed in Late British, the fact that Late British had other preaspirated or voiceless resonants would suggest that additional preaspirated or voiceless sonorants could have been acquired on their analogy. On the whole, then, the presence to some extent of voiceless sonorants and/or */xn, xl, xr/ in Late British could have facilitated the acquisition of Pre-Old English */hn, hl, hr/. But Late British contact need not necessarily have supported their long-term development, given that voiceless or aspirated /l/ and /r/ were lost in Cornish and in most dialects of Breton. In short, the development of /hn, hl, hr/ in Old English does not show any unexpected traits that would either confirm or dismiss any claim of Late British contact.
The situation with /hw/ is different. It is agreed that /sw/ yielded /xw/ in Late British and that it survives in medieval and modern dialects, e.g. OCo. huir, OB guoer /xwoer/, MW chwaer ‘sister’ (< PCl. *swesūr). Due to the fact that /hw/ was, with certainty, a stable cluster in Late British and has subsequently survived in all modern branches of Brittonic, it is reasonable to think that Old English /hw/ would have been acquired by Brittonic speakers who may indeed have helped prolong its survival in English, as opposed to Dutch, German and not much later in Frisian. However, more than this, there was also a development of Old English /hw/ to a uvular or velar fricative /χw ~ xw/ in northern and some Midlands English dialects. This development, as we shall see in 7.2.4.1, may add weight to the hypothesis of Brittonic phonological contact on English, since this English dialectal divide finds parallels in Brittonic dialects, with southern Welsh and Cornish displaying a preaspirated reflex /hw/ and Northern Welsh, and evidently other northern British dialects, attesting a uvular fricativised variant /χw/.
7.2.4 /Cw/ clusters
Six initial /Cw/ clusters were present in (Pre-)Old English: /tw, kw, dw, θw, sw, hw/. The situation in Late British was different: only the cluster /xw/ existed.170 In general, there do not appear to have been any significant changes that affected these clusters, at least not in varieties that were influential to the formation of English standard pronunciations. Yet it is uncertain whether any British influence would be expected anyway. After all, Late British did have as part
170
As indicated in 7.2.3 above, a different realisation of /xw/, namely [hw], seems to have existed in southern dialects of Late British. In Proto-Celtic the labial-velar stop */kw/ existed too, but this famously developed to /p/ in Brittonic, e.g. MW pymp, MCo. pemp, B pemp ‘five’ (< LBr. *pemp < PCl. *kwenkwe).
157
of its consonantal inventory the very consonant segments which make up such clusters – /t, k, d, θ, s, h/ and /w/ – and it may not have constituted much of a challenge to string two such consonants together. There is, however, one development which took place in many dialects, and with probable beginnings in the Old English period, which involved the loss of /w/ in several such /Cw/ clusters. There was a tendency for /w/ to be lost in the cluster /sw/ when a back rounded vowel followed. Representative examples, which first appear in early Middle English manuscripts, include such, so, also, sword (< OE swylc, swā, ealswā, sweord). Other words that are attested in Middle English or appear in modern dialects include suster ‘sister’, sūp (dialectal) ‘sweep’, sūmd (dialectal) ‘swum (past tense)’ (< OE swuster, swāpan, swum).171 As interesting as this development appears, similar developments can be found in other closely related Germanic languages, such as Old Frisian sā ‘so’, salik ‘such’, suster ‘sister’, similarly Old High German sō, sulik, soster and Middle Dutch so, sulk, suster (< PGmc *swǣ, *swa-leika-, *swester-). However, more examples seem to occur before other vowels in English than in other related West Germanic languages, which could suggest that loss of /w/ before labial vowels was more widespread. In particular, /kw/ clusters were sometimes simplified, e.g. Old English variant forms cuc, cucian, cudu ‘quick, quicken, cud’ (< PGmc *kwik-, kwid-), and there are a few examples of /w/ loss after other consonants, e.g. tū (< *twō) ‘two’, hū ‘how’ (< *hwō), hō ‘who’ (< *hwā). Yet in these examples there seems to be a reasonable phonetic explanation for /w/ loss without the need for positing British phonological influence. The combinations of labial consonant /w/ plus labialised vowel /u, o, ɔ/ appear to have been prone to assimilation. Why this assimilation typically occurred when other consonants, and specially /s/, preceded /w/ is not entirely clear, but since such parallel developments are found in other older Germanic languages, these changes fall short of being credible candidates for Brittonic contact influence. A more likely candidate for Brittonic influence is the dialectal change involving the clusters /kw/ and /hw/ to be discussed in the next section. It is a change, or rather a merger, that does not rely on any phonetic conditioning and is not found in any other old or modern Germanic language.