History of Scots to 1700

8. Chronological, genre and regional variation (CM)

8.1 Corpus linguistics

The most important recent development in the study of Scots has undoubtedly been in the field of corpus linguistics. DOST was an early adopter of the technology of electronic text corpora: Aitken and Bratley (1966) describe the Older Scots Textual Archive that they set up, containing over one million words of OSc literature. (Unfortunately, these methods did not exist until excerption for DOST was already at an advanced stage.) Now that DOST is available electronically, the body of quotations will form a very valuable corpus of material, provided it is used sensibly, and with reference to the peculiarities of the individual source texts. The shortness of the quotations will be offset by the very large body of literature from which they are drawn, much larger than any available corpus of connected text.

Like lexicography, corpus linguistics extracts data from a large body of texts. The fact that these are captured electronically allows very powerful searches – both in terms of completeness and complexity. Corpus linguistics thus avoids the greatest weakness of lexicography: the fallibility of the human excerptor. On the other hand, the weakness of corpus linguistics is its tendency to produce false negatives (i.e. incorrect statements to the effect that an item does not exist), because of the necessarily limited quantity of text used. Negative findings may be of interest, but only if the corpus (or sub-corpus) is large enough and representative enough, and the item in question frequent enough, for its absence to be notable (and negative findings can never, of course, be conclusive). In many cases it should be a simple matter for researchers to check against false negatives by reference to DOST.

The strength of corpus linguistics is that its results are systematic and complete for a particular corpus or sub-corpus. Since the late 1980s, Meurman-Solin has created, augmented and utilised a highly structured corpus, the Helsinki Corpus of Older Scots (HCOS), parallel to the Helsinki Corpus of English Texts. HCOS is divided into four time periods: 1450-1500, 1500-1570, 1570-1640 and 1640-1700. As far as possible, each time period is represented by (mainly non-literary) prose texts in each of a number of text types (genres). Even considered merely as a selection of texts, therefore, the corpus offers the researcher a set of pre-determined sociolinguistic parameters, with the additional advantage of comparability amongst studies based on this corpus and between them and studies based on the Helsinki Corpus of English Texts.

The pre-determined dimensions can, however, be something of a strait-jacket, as Meurman-Solin acknowledges. In some ways, MacQueen (1957), with her responsiveness to emergent patterns in the data, is more revealing; for instance her distinction between formal and less formal types of document, her specific division of the material into secular and clerical, and her classification of letter writers in terms of their known time spent outside Scotland. However, she does not quantify individual linguistic variables, and her rough figures for the second half of the 17th century are intended only for comparison with her data for the first half of the 18th century (where the status of all putative scotticisms has been checked against the dictionary record).

It is possible for the user of HCOS to select and analyse differences between texts in terms of regional dialect, social characteristics of the author (age, sex and rank), degree of formality, the author-audience relationship, and the situation of communication. Grammatical as well as lexical items can be investigated, and this can be done quantitatively, using either simple counts, or (preferably) variables as in sociolinguistics.[157] As in sociolinguistics, both linguistic and extra-linguistic dimensions of variability can be explored.

The crudest way to make use of an electronic corpus is to use it to generate a word-list. A more sophisticated product is a concordance, i.e. a list of all the words (which in this case means strings of letters) that meet predefined criteria, with a specified context, for example the five preceding and five following words. The output is arranged alphabetically, which then provides the researcher with the equivalent of a very comprehensive set of raw dictionary citation slips, in which homographs are not distinguished from each other. This approach generates large quantities of data. For this reason, and because of the amount of preparation required for more sophisticated searches, corpus linguistic studies never approach the size of the reading lists of the large historical dictionaries.

The degree of orthographic as well as morphological variation in OSc, and the high proportion of homographs, preclude fully automated ‘tagging’ (grammatical parsing) of the corpus. In linked projects, Meurman-Solin and Keith Williamson, of the Institute for Historical Dialectology, University of Edinburgh, have now begun the arduous process of tagging corpora of OSc texts, including the earliest texts and the ones closest to the spoken language, at the level of word and morpheme (Meurman-Solin, 1997c). This will allow the production of text profiles, i.e. descriptions of the linguistic features of individual texts, not limited (as in concordance studies) by pre-selection. The process of anglicisation introduced so much variability into MSc that this has tended to dominate quantitative studies, at the expense of the study of internal developments in the language, but the text profile approach, it is hoped, will encourage the latter. The regional, social and stylistic distribution of the many variants mentioned in §5 should become clearer as a result.

8.2 Chronological variation

8.2.1 Early Scots

The earliest (14th century) texts[158] still occasionally show some OE features: <sc> for /ʃ/, <wer + C> e.g. werld 'world', <h> for /x/.

In ESc, the spelling of /ʍ/ is variously <wh, qu, qw, qh, qwh, qhu, qhw>. The consonant /j/ is variously <yh, y, ih, i, ȝh, ȝ> (with <h> forms later dropping out of use, see Ȝ). Some spellings that occur variably in the 14th century texts are shared with English, but later dropped: <wh, gh, sh, oo, ee[159]>. The French spelling <-ez> is not infrequently found for the -is inflection. Vowel 7 is spelled with <o> e.g. god 'good'. Kniezsa points out that the Vowel 4 spelling <ae> already occurs, to disappear and resurface in the 16th century:

suggesting that the grapheme either was invented at an early period of Scots and went out of use for a while, or has been completely neglected by the scholarly descriptions. (1997: 43)

The early texts, and even some in the early 15th century, have spellings in <d> (-ed, -yd) for the verbal inflection. Like some other spellings above, -ed reappears later as an anglicised form. Long vowels often have no length marker in ESc.

8.2.2 Dateable features

Some miscellaneous features are listed below that, taken together and bearing in mind that many vary quantitatively over time, help to date texts. A number are taken from van Buuren ed. (1982). See also Girvan (ed.) (1939) for differences within ESc:

  • Barbour and the author of Ratis Raving are the only poets to keep Vowels 2 and 11 separate in rhyme (§§3.5, 6.1);
  • Barbour and Wyntoun (the latter not consistently) are the only poets to keep the various sources of vowel 14 separate in rhyme (§6.9.5);
  • in the first half of the 15th century, /d/ becomes /đ/ in father, etc. (§6.31.4);
  • ESc still has doublets with /v/ in hevid, laverd. New v-deletions are to be dated after about 1450 (most word-final examples are not attested until the latter half of the 16th century) (§6.31.5);
  • ESc still has doublets of reflexive pronouns in -selvin (§7.3.11);
  • negation with pre-verbal na gives way to post-verbal nocht from the mid-15th century (§7.10.1);
  • relative pronouns: at is replaced by that in formal styles in the late 15th century. The quh- relatives are characteristic of the 16th century, with quhilk at first used for both non-human and human antecedents, replaced in the latter function by quha in the late 16th century (§7.13.1);
  • after the mid-15th century, til (§7.8.9) and ane (§7.5.1) are found before consonants, and na (negative adjective) before vowels (§7.10.3);
  • after the mid-15th century, l-vocalised forms and reverse spellings are found (§§3.3.2, 5.2.6, 6.23);
  • individual words that differ between ESc and MSc include:
    ESc: ȝude, havis, ȝauld, alswa, swilk or swik, anerly, forout(in)
    MSc: ȝeid, has, ȝeldit, als (as well as alswa), sic, alanerly, without;
  • the <ui> digraph is uncommon before about 1500 (§5.1);
  • in the course of the 16th century, confusion of long and short vowel spellings becomes increasingly common, as do compensatory double long spellings (§3.4.3.3);
  • <þ> becomes rare after the mid-16th century;
  • the -our suffix of the agent noun is occasionally used with native stems after the mid-16th century (§4.2.3);
  • Arabic numerals replace Roman ones from the late 16th century;
  • in the late 16th century, -and and -ing become confused (§7.8.10).[160]

8.2.3 Standard Scots

The dominant trend between 1450 and 1700, as Meurman-Solin has shown in various writings (see particularly 1993a, 1997a), is at first the spread of a standardised form of Scots (in effect, i-digraph spellings);[161] and later convergence on StE. What she interprets as 'divergence' in the first phase is actually, as we have seen (§5.1), in tandem with nME; but the eclipse of ME dialects by StE after about 1450 means that MSc stands more starkly in contrast to the southern Standard.

The spelling <tht>, with excrescent <t>, is a case of change from above in the standardisation of Scots. Prior to 1500, Meurman-Solin (1997b) finds that it is significantly more frequent in the Acts of Parliament, and then diffuses:

from more formal styles to more informal styles, so that the feature spreads from statutory texts to persuasive prose with a didactic aim and ultimately to letters, first to official letters, later to private letters. Geographically, the spelling practice is introduced into texts directly related to the central administration; it then spreads to local administration in the periphery. (p.121)

After 1500, it is rare in the Acts, and begins to fall out of use, with Aberdeen Burgh Records and letters written by women retaining it longer than most other text types. By 1570, before the main impact of anglicisation, both <tht> and <cht> have practically disappeared from use.

8.3 Anglicisation

8.3.1 Overall trends

There is a general pattern of anglicisation over time, which takes the form of an S-curve (Devitt, 1989), familiar in other contexts in sociolinguistic research. The incoming variant appears as a minor alternative, then goes through a period of rapid adoption between about 1580 and 1640, and finally the outgoing variant tails off towards zero. The curve is also discernible with individual features, but sometimes only as a very broad trend with many conservative texts lagging behind (see Meurman-Solin, 1997a: Figures 1.2-1.4). The greater consistency of the S-curve in Devitt's research is, Meurman-Solin (1993a: 51) suggests, attributable to differences in the selection and definition of genres between the two studies, and a consequent smoothing of variability in Devitt's findings.

Mixed Scots and English forms, e.g. <quhome> with English vowel and Scots <quh>, and also -et, are common, especially, Devitt finds, in the phase of rapid anglicisation (1989: 147).[162]

Individual writers can be quite idiosyncratic in their choice of which features to anglicise and which to retain in Scots form. For instance, in Meurman-Solin's samples of Patrick Waus' letters, he uses 100% anglicised <ea> spellings as a schoolboy in the 1540s, but in his mature letters 100% <ei> (1993a: 201-2, Table XXXVII). Overall, some features anglicise more rapidly or at an earlier date than others. The more common <ai, ei> digraphs are retained later than <ui, oi> (Meurman-Solin, 1993a: 241). MacQueen, Devitt and Meurman-Solin all agree that morphological features are the first to anglicise.[163]

A factor in the rate of anglicisation may be the structural relationship between the Scots linguistic system and the English target. In the case of the Northern Present Tense Rule (see §7.8.2), the relationship between the two systems is complex. The rival forms are embedded in different grammars.[164] This could be regarded as a case of what Wells (1982: 113) calls ‘over-differentiation’, where the native variety makes a distinction that the target does not. Such a distinction tends to be carried over.[165] Even in heavily anglicised texts in Meurman-Solin's findings, the inflectional ending continues to be employed (albeit at a low level of frequency) according to the constraints of the Rule. MacQueen (1957: 136) found that the Rule was still fully applied in a sample of Aberdeen and Rothesay Burgh Records for 1700, still quite common in the Burgh Records in general up to 1730 and in Aberdeen up to the end of the period studied, 1750. It survived into the 18th century also as the one grammatical scotticism in the letters in her corpus (p.184). There are even occasional examples of the application of the Rule with the ending -eth (MacQueen, 1957: 138; Meurman-Solin, 1993a: 206, 251 n.10; Hume Douglas ed. Reid: liii).

Some other grammatical features also remain into the 18th century at a low level, often in set phrases, e.g. uninflected past participles, and adjectives inflected for the plural (only in saidis). The former has the reinforcement of Latin, and is the one grammatical scotticisim especially favoured by the writers - typically schoolmasters - of Kirk Session records.

Scots vocabulary diminishes less than other aspects of the language between the second half of the 17th century and the second half of the 18th, and therefore dominates the scotticism count in the 18th century, with some individual lexical items (c150), mainly part of a legal and administrative terminology, e.g. complainer, defender, infeft, tailȝie, vacance, resisting replacement by their anglicised alternatives, in some cases up to the present day.

8.3.2 Social factors

Meurman-Solin's (1993a) broad classification of social ranks into 'high', 'high professional' and 'professional' does not reveal significant differences amongst writers. MacQueen, however, suggests that anglicisation in letters, in the third quarter of the 17th century, was led by the nobility and clergy, with some military and businessmen, followed by lairds and lawyers (1957: 178). (On legal Scots, see further below.)

The sex of the writers of private letters appeared to be a significant factor in Meurman-Solin (1993a), but in a forthcoming paper she concludes that the very high degree of variability amongst women writers precludes any generalisations about language on the basis of sex. In the earlier study she found that women were slower to anglicise, and this was particularly noticeable between about 1615 and 1640 when the curve of change was levelling off in favour of English for many individuals (1993: 161, Table XVII). However, for the 17th century, MacQueen attaches more importance to the extent of an individual's time spent in England (1957: 178-9).[166] Depending on the degree of correlation (which might be quite high for the social level from which women's letters survive) between sex and amount of time spent furth of Scotland, it could be difficult to distinguish the influence of these two factors on the extent of anglicisation, but this remains to be addressed quantitatively. If indeed some women were slower to anglicise, other things being equal, then it was presumably because they were more inclined to spell by ear, as suggested by the presence of females from the nobility along with local burgh clerks amongst the ranks of substandard spellers (§5.2.8).

Letters written by husbands to wives, and by parents to children, were also more conservative than other family letters, even in the late 17th century, reflecting relations of familiarity and power - the Scots forms are used in the same way as thow as opposed to ȝe (see §7.3.6).

8.3.3 Genre

Meurman-Solin finds that genres and texts addressed to a wide audience - secular and religious instruction, pamphlets, travelogues, handbooks, scientific works, histories and biographies - are both more likely to be anglicised and more likely to be printed.

Some texts remain Scots because they remain closer to speech, but MacQueen and Meurman-Solin both find that texts governed by tradition and addressed to a domestic audience are also conservative.[167] MacQueen identifies:

an "officialese" in legal and other formal documents which, it is interesting to discover, is the style most completely in the Scottish tradition. (1957: 23)

This is characterised by a residue of Scots vocabulary (see above) and grammatical features, especially the Northern Present Tense Rule, plural adjectives and uninflected past participles, all mainly in set phrases. In a few cases, she is able to demonstrate that individual writers, for instance Lauder of Fountainhall in the late 17th century, had both a formal Scots and a less formal, more modishly anglicised, style at their command (p.179). The tendency to anglicise was, she concludes, somewhat restrained in formal documents by "a sense of appropriate style", thus, for instance, thir presents rather than this letter (p.162). However, Scotticisms of spelling are generally few in the national records, and not much more frequent in the Kirk Session minutes, but higher in Burgh Records.

The writers of national records, and others who read widely, were influenced by English, and were giving up the residual old conventions after about 1715, followed by the schoolmasters who generally acted as Kirk Session clerks (more slowly in the south, especially the SW, than elsewhere), and finally by the burgh notars (see notar n.1), who were trained on old deeds and other formal documents, and therefore remained more strongly Scots to a later date, particularly in the NE (p.194).

MacQueen concludes that even in the early years of the 18th century Scots remained a national language, with hardly any local usages appearing in the records (p.149), and only a tardiness in anglicisation characterising some dialects.

8.3.4 Printing

Printing as a factor favouring anglicisation has already been mentioned (§2.5.2). Bald (1926) compares letters, manuscripts, academic works and published sermons, and concludes that "some printers - or it may be their authors - evinced a quite definite theory as to the sort of books which should be anglicised. ... English - or an anglicised Scots - was regarded quite early as the correct diction for academic works" (p. 109). By the 1620s, Scottish printers like Andro Hart could produce works more or less in Standard English. Thereafter, only antiquities, such as the poetry of the makars, were printed in Scots,[168] until the 'vernacular revival' of the 18th century.

8.3.5 Religion

8.3.5.1 Religious versus secular

MacQueen compared the records of Parliament (1540-1707) and the General Assembly of the Church of Scotland (1560-1740) and found the latter considerably more anglicised over the period: see Figure 15. She also compared the burgh and Kirk Session records of Rothesay and Aberdeen for the period 1650-1750, and found the Kirk Session records considerably more anglicised than those of the burgh in each place: see Figure 16. (In keeping with the regional findings mentioned below, Aberdeen burgh records were intermediate between Rothesay burgh records and both sets of Kirk Session records.)[169]

8.3.5.2 Religious faction

As yet, there has been no objective testing of Law's observation, in his Introduction to the Catholic Tractates, that:

When the vernacular literature, in the hands of the dominant ecclesiastical party after 1560, was becoming rapidly Anglicised, mainly under the influence of the English Bible, a remnant of the old Church maintained or affected a certain linguistic conservatism, and made it a point of honour to adhere to their mother tongue. (1901: vii)

Apart from the jibes of Winȝet and Hamilton against anglicisation (see Scottis B 1 (1) and knap v.2), the anti-Reformation faction may really have been more concerned to make capital out of the difficult Latinate style of supposedly accessible vernacular writing (see §9.5.2.2). A contrasting 'plain' style of Protestant prose is remarked on by Lyall (1988).

8.3.5.3 Popular preaching

The influence of the Kirk on the history of the Scots has come to be seen, because of the introduction of Biblical English, as tending wholly towards anglicisation. But the popular sermons of the presbyterians were remarkable for earthy and colloquial passages (which the printers chose to print in Scots). Indeed, they may have worked against Scots by bringing it into disrepute as a medium for public speaking. The Scotch Presbyterian Eloquence was assembled as anti-presbyterian propaganda, and concentrates on extreme instances, but there is no reason to doubt the genuineness of the material, which can be compared with colloquial passages in published sermons like the so-called Pockmanty Preaching of James Row. This preaching was passionate, abusive, and, above all, extempore, so that it was on occasion wildly unscriptural. Decorum was breached by speaking of the sacred in terms of low life imagery, e.g. “I fear we'l find you like Ephraim, a Cake unturn'd, that is, it's stonehard on one side, and skitter-raw on the other” (p. 80). In contrast to this there was the predictable, carefully prepared Biblical English (probably scotticised in pronunciation) of the episcopalians with their liturgies and set prayers, and gentlemanie (q.v.) preaching in general. This may well be the context in which Scots and English first became polarised, not just as old-fashioned versus modern, but as vulgar versus polite, though it is prefigured by Knox (himself following Calvin):

Each of the reformers tends to use an elevated diction, when putting his ideas forward, but resorts to indigenous words, when stating the opposing case. (Jack, 1971: 21)

8.4 Regional variation

The North-Eastern and Southern dialects must already have been distinctive in some ways by ESc (cf. §§6.9.2.1, 6.9.2.2, 6.9.5.2, 6.10.4, 6.15.2.1, 6.25.1, 6.27.2, 6.29.1), but regional features tend not to be manifested in spelling or rhyme: hence the impression that Smith (1902) and others had of the uniformity of MSc as a literary language, despite its rampant variability in orthography (Aitken, 1971). Even in late MSc, when some revealing sub-standard spellings appear (§5.2.8), the peculiarities of Southern Scots, being in vowels rather than consonants, are not so obvious within the elastic MSc orthography.[170]

Johnston's (1997b) map of the ModSc dialect areas, based on the data of LAS3, shows that since the late 19th century (when the dialect areas shown in Map 10 were originally delineated), Central Scots has been expanding, and this is probably a continuation of a process already happening in OSc, of which we have a few hints. Features whose geographical range appears to have been greater than now include:

  • Vowel 14 in dauch 'dough', etc. (§6.9.2.1);
  • ? merger of the reflex of OF ǖ with Vowel 6 (§6.9.5.1);
  • Vowel 14b(ii) (§6.9.5.2);
  • ? unmerged Vowels 14a and 14b (with modern merger as 14b in Orkney) (§6.9.5.3);
  • ? the Vowel 13 reflex of ald, etc. (§6.13);
  • a monophthong distinct from Vowel 4 in day, etc. (§6.29.1);
  • that and this with plural nouns (§7.5.4);
  • ?? negation with pre-verbal na (§7.10.1).

As Aitken points out (§§5.2.8, 6.2 here), the regional distribution, if any, of many OSc variants is unknown. However, the following are clearly regional:

  • Vowel 13 in douchter 'daughter', etc. (§6.9.2.2);
  • the development of vowel 4 after /w, ʍ/ (§6.26);
  • Southern merger of Vowels 12 and 17 (§6.28.3);
  • the development of Vowel 7 before the voiceless velars (§6.10.2);
  • SW failure of smoothing of /ai/ before front fricatives (§6.15.2.1);
  • Northern conditioned early merger of Vowel 4 with Vowel 3, thus also steen ‘stone’, etc. (§6.25.1);
  • Insular Scots /ð/ > /d/ (§6.28.2) (and likewise /θ/ > /t/);
  • /θ/ > /t/ in Galloway and perhaps more widely (§5.2.5).

A number of changes were of course spreading in the MSc period, without reaching completion in ModSc:

  • vocalisation of al (§6.23);
  • merger of Vowels 3 and 2, at the expense of merger between Vowels 3 and 4, with a residue of unmerged Vowel 3 (§6.25.2);
  • the SVLR (§6.28);
  • merger of non-final Vowel 8 with 4 (§6.29.2.2);
  • preiotation of initial Vowel 4 (§6.27.2). (Preiotation is unusual in that it spreads northwards into the Central area);
  • unrounding of Vowel 7; merger of the unrounded vowel in SVLR-short environments with Vowel 15 (§6.10.2);
  • merger of Vowels 5 and 18 (§6.28.3);
  • reduction of /nd/ to /n/ (§6.31.3).

There is also a regional time-lag in the spread of standardised Scots - in Meurman-Solin's (1997a: 7) data for texts pre-1500, the proportion of 'Scottish' features[171] is lowest in local records from Aberdeen (54%): contrast Acts of Parliament (71%) and local records from Peebles (78%).

There is a similar time-lag when Scots later begins to converge on StE, with local records in general, but especially those from the NE, lagging behind the Acts. Even in the period 1640-1700, Aberdeen records retained ane + C and -it at the level of 75%, and NE texts stand out in this period for their Scots lexis (Meurman-Solin, 1997a, 1997c). Indeed, most of the 'dialect' features that Meurman-Solin is able to identify in HCOS are time-lag features (1999: 318).

There are of course many individual linguistic items that have a restricted geographical distribution. Some examples (drawn partly from Aitken, 1971: n.8, n.9) are:

  • Orkney and Shetland: *austercope, *ayr, *callow, cloggand or clowgang, hafe-wrak, *haifreis-dyk, *hailye, hallow n.3, handband, *heavie, *helm(e)in, quy n.2. For Richard Holland's Orkneyisms see s.vv. lang reid and *idy;
  • Caithness: *hasty;
  • Easter Ross: schafe;
  • North-East: codrach(t, daker v.1, half-disch or had(d)isch, kard n.2;
  • East Central and Southern: clour, herring-drave, hirst n.3, hope n.2, husband-land, kimpill, kirk-master n. 2, lokman n.1. The author of King Hart employs two Southern Scots items: waile (if this is to be read as wyle 'ropetwister' – for the <ai> spelling of Vowel 1, see §§5.2.8, 6.25.1) and rewle = (1) 'ravelling', (2) 'revelling';
  • West Central and South-West: clat v., *halfe-gavill, *half-manure, *half-net n.2, *heft n.2, lime-craig, lime-holl; "Father James Dalrymple's possible Ayrshire origins [are confirmed] by his use of clachan and inspreth" (Aitken, 1971: 178);
  • East coast: heave-eel, *hettell, likarstane, linkis, lippie.

More narrowly localised items include: Aberdeen: inland n. 3 'inner portion of a tenement'; Perth: cumling 'incomer';[172] Moray Firth burghs: greveschip 'precinct'; Kirkcudbright: kirk-master n. 1 c 'church-officer'; Inverness: kist n. 5 'fish-cruive'; Sheriffhall colliery: murement 'spoil'; Haddington: lowand-ill 'only in Knox and in the Haddington records'; Elgin: marschal(l n. 7 b 'an executioner'.

Note that those items with asterisks are from the Additions and Corrections to Volumes I and III.

8.5 Highland Scots

(Broken) Gaelic speech is occasionally represented in literature, from Holland onwards. The word that seems to have been universally known is the greeting bannachadee (q.v.). It was evidently regarded as insolent to use this to a non-Gaelic speaker, rather as it is nowadays potentially insolent to speak Scots to an English monolingual.[173] In the incident related in Wallace, the English soldiers are trying to make it the ground of a quarrel:

“Sen ȝe ar Scottis ȝeit salust sall ȝe be -
Gud deyn, dawch Lard, bach lowch, banȝoch a de.”
Ma Sotheroune men to thaim assemblit ner.
(from McDiarmid ed., 1968: VI: 139-41)

The other fragments of broken Gaelic here and in Holland (ll.795ff.) are not regarded as having been borrowed into Scots, and are not entered in DOST (see glosses in editions of the poems).

There are also representations of Highland Scots, characterised by a number of recurrent, stereotyped features. One of these is the use of the third person instead of the first person singular pronoun (see hir pron. 1 d, s(c)ho pron. 1 c), which has no warrant in any modern speech form. Despite Lauder's purportedly objective observation (s.v. hir), SND (s.v. she B 4) is no doubt right to regard it as a purely literary stereotype.

Hir is frequently used in combination with nain-sell (q.v.). [174] Nain-sell is only recorded in DOST following his/hir (meaning 'his'), almost entirely as a stereotype of Highland Scots. It occurs in the ModSc period both in the Highland stereotype his/her nainsel and as General Scots (SND s.vv. ainsel(l, nain). If the combination of (n)ain and self bears any relationship to Gaelic fhèin and Scots usages based on it (see §4.2.2.2), then it is an exaggerated one (Macafee and Ó Baoill, 1997).

Other features are credible representations of a strongly Gaelic-accented Scots, with <p> for /b/, <t> for /d/, <sh> for /ʃ/ and <t> for /θ/ (see quotations under the words cited above).

[157] A simple count (occurrences per 1000 words) is used, for instance, by Dons and Moessner (1999) in their study of Scots -and and English -ing, but the usefulness of their figures is limited. Without knowing the possible number of occurrences, we cannot tell whether low figures for -and mean a high degree of anglicisation or a low usage of present participles, or a combination of both.

[158] Collected by Slater (1952), MacRae (1975), and ongoing computerisation at the Institute for Historical Dialectology, University of Edinburgh. Some early texts are reproduced, with transliterations, in Facs. Nat. MSS, and see also Simpson (1973: Text 8). Early place-name forms are often cited in the place-name literature, and see also Barrow (1980: Appendix C).

[159] Except for final <ee> in French loans such as citee 'city', which continues. <ea> is not an anglicism in (originally) hiatus position, e.g. really, realm. <burgh> is a frequent variant spelling of burch.

[160] They are also confused in Troy-bk. I am grateful to Catherine van Buuren for this point.

[161] Her figures have to be used with some caution. She regards the i-digraph spellings as particularly salient for Scottishness, but some of the variants with which they contrast are equally Scots. She likewise regards the l-mouillé and n-mouillé forms as salient, and treats forms without the palatal consonants as English, although many are also Scots (see §6.11). In the 1993 figures, the proportional figures for -ed are inflated by the omission of -yt forms (p.126 n.2), though this is corrected in the 1997a figures; and the omission of <a> may have similarly affected the proportional figures for <aw> versus <ow> (1993: 157-60).

[162] Meurman-Solin (1997b), however, regards -et as reflecting a loss of distinction between the unstressed vowels of -ate and -it in the 16th and early 17th centuries. -id may also be a blend of -it and -ed, but since the ending had been reduced to /d/ in a number of phonetic environments (see §9.1.2.5), it is safest to follow Meurman-Solin (1993: 51, 56) and reserve judgment.

[163] In Devitt's findings, the earliest features to anglicise are the ending of the present participle and the form of the negative particle (anglicised not - but see §9.1.2.5). <wh> in relative clause markers enters the phase of rapid adoption later (c.1600) than the inflectional ending -ed, and likewise in Meurman-Solin's (1993, 1997a) findings, <wh> spellings in general become predominant later than -ed, with the personal pronoun they (as against Scots thai) intermediate (see Meurman-Solin, 1997a: Figures 1.2-1.4).

[164] For discussion of a similar case in Hiberno-English, see Harris (1984).

[165] ScStE similarly retains the two Vowel 1 diphthongs, where Scots over-differentiates in comparison with English (see Macafee, 2004).

[166] In the 18th century part of her sample, it no longer seems to influence the extent of anglicisation.

[167] These are Burgh Records and Acts of Parliament in HCOS (Meurman-Solin, 1997a). A few writers of official letters still retain a Scots residue, particularly of i-digraph spellings, in the second half of the 17th century (Meurman-Solin, 1993: 156, Table XV).

[168] This Scots was increasingly anglicised in superficial and not so superficial ways. See Geddie (1912).

[169] Devitt found that religious treatises were the most highly anglicised genre, despite her inclusion of sermons in that category. Although Meurman-Solin (1993: 244) questions the validity of this result - her own findings for the Religious Instruction and Sermon categories are more heterogeneous - it confirms MacQueen's findings.

[170] Meurman-Solin (1999: 312, 315) has examples of <a> for Vowel 16, e.g. sand 'send', which is reminiscent of the modern Southern Scots pronunciation [æ], but the regional affiliations of the writers are not sufficiently clear to identify the significance of these spellings.

[171] See note 161 above.

[172] See 1555 Perth Hammermen pp. 79, 84.

[173] In terms of Brown and Levinson's (1989) theory of politeness, this is a case of unwanted familiarity, with the additional implication, since the person addressed is not actually expected to understand, that he or she is not part of an in-group.

[174] Ainsel, on which it must be formed by false division of mine/thine ainsel, is not recorded until the 19th century.

Macafee, Caroline and †Aitken, A. J. (2002) ‘A history of Scots to 1700’ in A Dictionary of the Older Scottish Tongue vol. XII, xxix-clvii. Online http://www.dsl.ac.uk/about-scots/history-of-scots/chronology/