Variation and variety

5 Orthography

This section, from §5.2 on, is based on the parts of Aitken (1971) that are concerned with orthographic variation, shortened and revised for the purposes of this Preface, and with some additional material from Aitken (2002). Meurman-Solin (1993a) tested some of Aitken's observations using her Helsinki Corpus of Older Scots (HCOS): some of her findings are mentioned below.

5.1 The origins of OSc orthography

Since Aitken's (1971) account, Kniesza has added significantly to our knowledge of OSc orthography.

As we have seen, the i-digraph spellings <ai, ei, oi, ui>,[81] which originally belonged to diphthongs, are used in MSc also - and, in practice, chiefly - for long vowels. There is also a rarer <yi> spelling, applied to Vowel 1. The origins of these spellings has been much debated in the past, and the date of the merger of Vowel 8 with Vowel 4 has been particularly crucial to the question whether <ai > in Vowel 4 words is a reverse spelling reflecting this merger (see Aitken, 2002: §12.2). As Aitken points out, there were some conditioned mergers in ESc of Vowel 8 /ai/ with Vowel 4 /a:/ [82] and also of Vowel 9 /oi/ with Vowel 5 /o:/ (see §6.15), but Kniezsa has shown, after much detective work, that the origin of the i-digraphs lies outside Scots (see particularly 1989, 1997). The suggestion is that their origin lies in nME, and that they reached Scots by diffusion in the course of the OSc period:

• the <ei, ey> spellings for Vowel 2 are AN, and were originally more widespread in ME, before becoming associated with nME and Scots;

• the <ai> spellings for Vowel 4 words arose within the Great Scandinavian Belt (see §2.2.3) as a result of contact with ON, which has many cognate words in Vowel 8. This led to Vowel 4 and Vowel 8 doublets (e.g. the place-name Stainburn appears as both <Stanburn> and <Stainburne> in the 11^th century), persisting as an orthographic convention even after the native Vowel 4 cognates prevailed. The orthographic convention then diffused northwards;

• <oi> spellings for Vowel 5 appear to have arisen in the West Riding of Yorkshire and in Lancashire, where OSL of o produced a diphthong, e.g. the place-name Goyt (< OE gota 'water-channel'). The corresponding spelling spread beyond the area in which it was phonetically accurate, giving <oi> as an orthographic convention for Vowel 5 generally in 15^th century Yorkshire and subsequently beyond;

• it is possible that the <ui> spellings for Vowel 7 should also be traced to the West Riding of Yorkshire, where the modern dialect pronunciation of this vowel is /ʊi/,[83] and where <ui> spellings appear in place-names from the second half of the 15^thcentury. In this case, however, the orthographic convention did not become widespread in Yorkshire and it is therefore hard to see how it could have reached Scots. Meurman-Solin (1993a: 198) finds that this digraph is "conspicuously late compared with other digraphs". Possibly it was generated on the model of the other i-digraphs.

Kniezsa (1997) also traces the origins of the interchangeable set <u, v, w> to the influence of Irish spelling on ONhb. Already in late ONhb <u, wu, uu> were used for both consonant and vowel values. (Since <v> is in origin a form of its membership of this set is not problematic, and <u, v> also interchange in ME.) ONhb also already had <ch> for /x/ and for /ʧ/.

Using A Linguistic Atlas of Late Mediaeval English (LALME), which is based on the period 1350-1450, she finds that there is "one orthographic continuum" throughout England and Lowland Scotland (1997: 32). A number of features previously considered to be characteristic of Scots orthography extend as far south as Yorkshire, for instance <sch>; the interchange of <u, v, w>; and -ar in the nomen agentis, e.g. <millar>, and comparative of adjectives, e.g. <lattar>. This area also has spellings of the <qu, qw, qwh> type (which are likewise found in the earliest Scots texts), in contrast to <wh>. More detailed comparisons with nME remain to be made, but Kniesza mentions as particularly Scots:[84]

• <w> as a vowel spelling word-initially, e.g. <wp>

• -ill e.g. <litill>

• <lʒ, nʒ> for l-mouillé and n-mouillé respectively (see §6.31.1)

• <ae> for Vowel 4, e.g. frae (see §8.2.1).

5.2 Variation and variety (A. J. Aitken)

From the general viewpoint of the lexicographer surveying MSc over its whole chronological, regional and stylistic range, the spelling system was a perhaps extreme example of a common medieval European type, in which free variation was a prominent and important feature. For instance, among suffix syllables sets like the following occur: -is, -es and -s; -ill and -le; -ioun, -iown and -ion; -our and -or; -ure and -our, -or; -ar, -are, -air and -er; -y, -ie and -e.[85] Some of the variations MSc simply inherited from nME.[86] Others were, it seems, largely its own, for instance a largely interchangeable set consisting of postvocalic <th, ch, tht, cht> and superscript <t>;[87] and spellings reflecting l-vocalisation.

5.2.1 Proliferation of variants

Throughout the period, the general trend was towards a proliferation of variants, as further items were assigned to variant sets. This seldom resulted in the total obsolescence of the less favoured variants; the only notable loss was the disappearance of <yh> and <ȝh> after the early 16^th century, leaving only <y> and <ȝ>. In the second half of the 16^th century the already considerable body of sets of alternative spellings was further supplemented by the introduction of graphemes from English:

<sh> was now added to <sch>;

<tch> to <ch> /ʧ/;

<gh> to <ch> /x/;

<ea, ee> to <e(-e), ei, ey>;

<oa> to <o(-e), oi, oy>;

<oo> to <u(-e), ui, uy>;

-ed to -it, -yt.

And though, in the course of the 17^th century, the native spelling system largely gave way to a much less variable anglicised system, the old tolerance of spelling variation has continued in ModSc down to the present day.

5.2.2 Individual writers' systems

It would be surprising, since presumably all writers of MSc had access to this general varying system, if any one text or writer proved to possess a completely consistent practice with one and one only spelling for each word, a fixed spelling system. As a glance at any full glossary or description of a MSc text will show, few if any MSc scribes (or printers) are wholly consistent. Some degree of free variation was normal in the spelling practice of individuals as well as over the system as a whole. And some scribes, like the writer of the Laing MS of Pitscottie's Chronicle, are indeed highly inconsistent, varying freely between different spellings of single items, sometimes in quite short stretches of text, in what, to us, may seem a curious and striking manner. Without this habitual tolerance of spelling variation, the changes in popularity of particular variants and the introduction of new variants, including the anglicised spellings common in later 16^th century texts, could hardly have occurred.

It is nevertheless true that all writers of Middle Scots seem also to display some greater or lesser degree of consistency in some at least of their spelling habits. For instance, Gilbert Hay (1456) has, in one of his works, only two examples of <mair> to 221 of <mare> 'more'. In the choice between initial <v> and <w>, some scribes[88] have a strong preference for <v>, spelling e.g. both <veill> 'veal' and <veill> 'well'; others prefer <w>; some have a 'modern' distribution of the two letters; some seem merely inconsistent; and there is one instance of an invariable preference of to either <v> or <w>, in the person of King James VI.[89] No doubt all writers had access to the general variant system; but each made his own personal and idiosyncratic selection from the alternatives available to him.

Thus a writer's individual assemblage of spelling-choices could be just as distinctive of him as his handwriting, as indeed was recognised in one celebrated trial at law when five unsigned, treasonable letters were brought home to an alleged author on just such grounds:

ffirst, that he newir vseit to wrytt ane 'ȝ' in the begynning of ony word, sik as 'ȝow', 'ȝouris', 'ȝeild', 'ȝea', and siclyk; bot ewir wrait 'y' in steid of the said 'ȝ'; that he wrait all wordis begynning with 'w' with ane singill 'v'; and quhan that letter 'v' fell to be in the myddis or end of ane word, he wrait ane doubill 'w'; that quhen he wrait 'quhen', 'quhair', 'qlk', or ony sik word, quhilk vsis to be writtin and spellit be vtheris with 'quh', he wrait onlie 'qh', 'qhen', 'qhair' and siclyk; quhaneuir ane word began with 'con', he newir wrytt 'con' at lenth, bot with ane '9'. Quhan euir 't' fell to be in the end of ane word, he wraitt it without ane straik throw the 't'; and did the lyk quhan ewir 't' fell in ony pairt of ane word.[90]

5.2.3 Copyists

It is evident that many individual MSc scribes had at least some preferences and more or less consistent habits. That copyists could often be more loyal to their own habits than to their author's apparent intentions is shown by the fairly numerous cases when it is necessary to restore an alternative variant to make a verse line metrical or to provide a rhyme. The rhymes of the Maitland MS copyist's version of King Hart, for example, require a number of emendations of this kind:

for betrayid (line 382) read betraysit, for suppleit (390) suppleid, for dunt (537) dint, for glaid (554) glad, and for iustifeit (574) iustifyid.

When different recensions of the same original can be compared, they often show consistent differences in their choice of variants. Whereas the copyist of the Edinburgh MS of Barbour's Bruce regularly has gan (the auxiliary verb), gres 'grass', hundir 'hundred' and leawte 'loyalty', his contemporary of the Cambridge MS as regularly prefers can, gyrs, hundreth and lawte, laute. Similarly, Sheppard writes: “The complete orthographical independence of the sixteenth-century scribes is shown by a comparison of the Boece manuscripts. Each scribe has a characteristic and (more or less) consistent mode of spelling” (1936: 211).

5.2.4 Graphological and phonological variation

Much that we know of MSc spelling and phonology implies a reasonable fit between the orthographic and phonemic systems, with a few exceptions such as the 'artificial' spelling of the indefinite article as ane before consonants as well as before vowels.[91] Many of the misfits are due to a tendency to retain established spellings after the sounds that they had formerly reflected had undergone change. For instance, the early 15^th century coalescence of /al/ with /au/ (l-vocalisation, see §6.23) is slow to be reflected in word-final conditions, so that hall 'hall' and haw 'livid' continue in general to be so distinguished in spelling throughout the period. Similarly, the emergence of MSc doublets in Vowel 4 as well as in Vowel 2 in words like dede 'dead' and dethe 'death', which had Vowel 3 in ESc, is partly concealed by a persistent preference of most writers for the traditional <e-e> or <ei, ey> spellings (which were shared by Vowels 3 and 2), in contrast with the rare and tardy occurrence of the <a-e, ai, ay> spellings that imply Vowel 4.

Assuming the spelling system is fundamentally phonemic, and taking into account rhyme evidence and modern dialect reflexes, we are led to infer the existence of a large number of sets of synonymous doublets in MSc, e.g. the pronunciations implied by:

mak and maik,[92] blak and blaik 'black', lak and laik 'lack', tak and taik 'take', brek and breik 'break', spek and speik 'speak', glad and glaid; waik and tway beside wauk, walk and twaw, giff and liff beside geve and leve, mekill and sekir beside mickill and sicker, abuif and abuin beside abouf and aboun, chese beside chuse, lese beside los, warld and warldly beside wardill and wardly, seildin beside seindill 'seldom', brin and thrid beside birn and third, brander and hunder beside brandreth and hundreth (and houndreth), broder and fudder beside brother and father, bus and mers beside busk and mersk, nar beside nere, neist beside nixt, nerrest and narrest beside nerest, heich and hey 'high', laich, lauch and law 'low', multiple sets like the various forms of 'great' (grett, greit, gritt, gryte, gert, gart and girt) and a similar set for 'grass'.

One very numerous group exists among the principal parts of most 'strong' and some 'weak' verbs, where analogical processes supplemented the simple phonological and morphological causes which gave rise to the variants already instanced. This phenomenon is an important fact of life for the student of MSc.

There are also instances in which it is more than likely that we are seeing the conservatism of the spelling system, rather than the deliberate representation of the older phonological variants, as with the unreduced members of the pairs ak and act, effek and effect, temp and tempt, count and compt, nummer and number.[93]

The general trend was plainly towards the accumulation of a larger and larger number of variants, only partly offset by the obsolescence of items like ȝude 'went' to leave only ȝeid (itself later superseded by the uncommon analogical gaid, or by went) or hevid beside heid. Again, the habit of tolerating this sort of variation perhaps provided a suitable condition for the ready acceptance, in the later 16^th century and 17^th century, of a massive number of additional variants of English derivation like oath, more, most, quhom, quhich, so, only, owe, kingdom, much, either, any, from, if, would as alternatives to the native aith, mare, maist, quham, quhilk, sa, anerly, aw or aucht, kinrik, mekill, owther, ony, fra, gif, wald.

Variants of this type are of course not unknown in other dialects and languages (including ME and EModE). MSc, however, seems to have been quite exceptional in possessing an extremely large number for which, at present, no regional or other specialisation of distribution is apparent - which co-existed as free variants over extensive regions and often in single, including some holograph, texts.

At least in holograph texts, written variants of this type may imply that the writer either used or was familiar with the spoken form or forms to which these written variants correspond; that, for example, a writer who wrote greit and grett 'great' in fact knew the spoken forms with Vowels 2 and 16 respectively in his own or his neighbours' speech. But we cannot of course guarantee, even for holograph texts, that this was the case once these spellings had come into established use. What were phonemic variants for one writer may have been merely orthographic for another. We are often not certain whose favoured pronunciation is being revealed to us - that of the writer himself, his writing teacher, those writers who set the national MSc standard, or some other. There are also occasional spellings like neixt ‘next’ that blend phonemic variants (in this case, usually spelled nixt and neist) and are quite indeterminate as to the writer's preference between spoken doublets.

Indeed, some quite clear instances exist of scribal defiance of phonemic principles of spelling. Most NE scribes prefer standard MSc spellings like quhare or quhair although it is likely that their own pronunciation would have been /f/, and indeed a few of them have occasional spellings in <f>, such as for and fair. We may imagine, too, that in writing English forms like oath, if, from, would, writers in the early stages of anglicisation were misrepresenting their speech-habits.

There are ambiguities inherent in the basic system of phoneme-grapheme equivalence itself. In MSc, as in other orthographic systems of a similar type, individual graphemes commonly participated in more than one orthographic set.[94] Thus the graph <a> participated in the Vowel 4 set <a, a-e, ai, ay> as well as being the usual spelling for Vowel 17, so that such a spelling as lard, out of context, is completely ambiguous as between 'lard' and 'laird'. Forms like gret may be ambiguous, though grett (Vowel 16) and grete, greit (both Vowel 2+3) are as a rule not. Equally, the commonest spellings for 'two' and 'who', namely twa and quha, overlap between the sets <a, ai, ay> (Vowel 4) and (in a labial context) <a, au, aw> (Vowel 12); the unambiguous spellings tway and quhay are rather less common; and the unambiguous spellings representing the other doublets, twaw and quhaw, are rare. By studying the specific habits of individual scribes some of these may be resolved: thus some scribes avoid <a> for Vowel 4. But this will be defeated if the scribe is at all inconsistent in his habits and no doubt a large residue of unsolvable ambiguities will remain.

5.2.5 Regional variation

In some cases, we can be confident that we are seeing genuine reflections of dialect speech. For instance, a noticeable and apparently regular habit of the clerk of Wigtown Burgh Court in the early years of the 16^th century was that of writing <t> or less frequently <d> where his contemporaries wrote <th> or postvocalically <tht> or superscript <^t>, e.g.:

tyrd (fol. 8a), clat (9a), a towsand v hundyrd (9b), tre hundyr bollis (ibid.), tolboud (10a), triys, triis (12a), hunderet (14b), etc.

Cf. this late 17^th century comment on the speech of Galloway:

Some of the countrey people, especially those of the elder sort, do very often omit the letter h after t, as ting for thing, tree for three, tacht for thatch, wit for with, fait for faith, mout for mouth. (Andrew Symson, Large Description of Galloway, 1684, repr.1823: 97)

Although the NE dialect was already distinctive, NE features are sometimes very rare indeed in their incidence, as very occasional aberrations from more or less standard spelling-practice: thus, at least on the evidence of the printed text, the for spelling in the 1539 quotation below is a unique exception to this scribe's otherwise regular quhar.[95] In the quotations below, the NE forms (see §§2.4, 6.10.1, 6.25.1) are italicised:

And the quyntray was dangerfull throw this plage of pestilence; 1500-1 Aberd. B. Rec. I 68 (see quintra(y n.)

That na burges . . . sal haue nay forstaller vnder him to pas in quintray; 1507 ibid. 435

In calling of hir commond vyld freris hvyr that scho wes that hes ane pek of lyis betuix thi shoulderis. I sell leid the to the place for the freir swewyt the quhar thou tynt the pendace of thi belt in the hie publict gett; 1539 ibid. 159 (see quhare adv. and conj. C 1 a (d))

Ane phingar; ibid. 161 (see phingar var. of quhingar n.)

To heid the blokhouse with faill and put ane fulse rief thairon, thykit with faill; 1542 ibid. 184 (see ref(e n.³)

Dauid Anderson, maister of wark to the stein wark of the sayme [gable of the parish kirk]; Cullen Chron. Aberd. 33 (see stane wark n.)

And thair eftir I paist to Dunnotter, fair I beheld his grace at his supar; ibid. 53 (see fair var. of quhare adv. (Additions and Corrections, vol. II))

Swa that neyn belewit his lyif; 1596-7 Misc. Spald. C. I 85 (see nan(e A 3 (1) (c))

The bairne suld be ane las that was in hir wymb; ibid. 98 (see wam(e n. 3 (e))

Wytit on be the cummer, ibid. 92 (see wait v.¹ III 7 (e))

The counsel … does … grant hir liberatione from the inner hous to the hous abow the samyn how sein the dore lock and chambers of the said roume be helpit; 1674 Elgin Rec. I 319 (see son(e adv. 4 d (1))

In the forme of ane four futit beist, and speciallie lyk ane futret, and sum tyme lyk ane catt; 1597 Misc. Spald. C. I 148 (see quhitret n. (b))

Seing he wes ane man so guidlyk and ritche . . . and scho ane vgle harlot quyne; ibid. 178 (see quine n.¹2).

5.2.6 Register variation

Still another class of occasional phonemic variant spellings exists in which the distribution of the implied spoken forms was apparently by register rather than by region. This is most obvious with certain items of that group of words which underwent phonetic reduction by the loss or assimilation of intervocalic or final consonants, resulting in the emergence of reduced and unreduced doublets, such as aw and all, deil and devil, mow and mouth, and so on. In every case both doublets persisted at least as orthographic variants. Since many of them are still represented in the modern Scots dialects by spoken doublets, it seems that for MSc most of them can also be classed as phonemic variants. The reduced members of these variant sets are evidenced also in reverse spellings like ewine for ein 'eyes' (after evin, ein, 'evening') (Leg. S. ii 557) and send for sen 'since' (after send, sen, 'send') (Cullen Chron. Aberd. 50), and in innumerable instances in rhyme, like evin (evening) with wene (Kynd Kittok 12) or sen ‘send’ with den (Henr. Fab. 556).

Some examples of this sort of variant show no obvious tendency towards specialisation of distribution. It appears, for example, that hauch, bow, trow, know, row and gouff were as common in all classes of text as halch, boll, troll, knoll, roll and golf, and fouth and stouth seem to have completely superseded fulth and stulth. Clais seems to have been as widely and freely used as clathis; ill and evill, though unrelated, were apparently regarded as variants of this kind and interchange freely (for example, in different recensions of the same text); and the specialisation of fow as opposed to full is apparently by semantics rather than by dialect or register. With sets like aboun, abone and abovin, abuvin and lesum and leve-, leifsum, the reduced variant is much the more common, without any evident specialisation of distribution of either alternative.[96]

But, when all these have been excluded, there remains a large residue of reduced forms that seem to have been restricted (see §9.3.7). It is possible also that some, though clearly not all, of the aphetic forms that are such an obvious characteristic of the MSc morpheme system may have a similar distribution: possibly for instance chete, cheit (beside escheit) and tach and tachment (beside attach and attachment); though apparently not fect, feck (beside effect), levint (beside elevint, alevint eleventh) or mendis (beside amendis).

5.2.7 Standard Scots

Perhaps the narrowest limits of variation in this way will prove to be those adhered to in some of the printed prose and verse of the later 16^th century, as if the Scottish printers were moving, like their English opposite numbers, towards a fixed spelling; this movement, if it existed, came to an end when, early in the 17^th century, the Scottish printers abandoned the native spelling tradition altogether for an imported English one. Among the 16^th century manuscript sources, however, the great national registers (of Parliament, Privy Council, and the Great and Privy Seals) and almost all the existing copies of the major literary texts in prose and verse, also conform to relatively narrow limits of variation. And much the same range of variants is adhered to by many other writers of that century, including local clerks, minor officials and the writers of private records. If we assume that it was writers like these royal and literary clerks who set the standards of spelling and of other literary usages, then we may regard this limited, majority practice as the 'standard' form of written MSc. Meurman-Solin's data confirm "that there was a fairly uniform … standard not only up to the Scottish Reformation but quite clearly - in the majority of texts - up to the first half of the seventeenth century" (1993a: 239). That there was a contemporary belief in the possibility of some such standard is implied in the frequent complaints of or apologies for “wrang ortographie and fals spelling” (see orthographie n. (a)).

5.2.8 Substandard spellings

But the boundaries between this standard system of spelling and less conformist systems was neither sharply defined nor immutably fixed. Occasional irregular or substandard spellings do turn up in otherwise 'correctly' spelled texts, like thua 'two' and bethuix 'betwixt' in one early MS of Bellenden's Boece, and staw (p.t. of stele 'to steal') in another. Even so, the very rarity of these is strong proof of the general rule of the standard spelling: staw, for example, appears to be recorded only once in a MS of over 300 folios, as against the commoner stall. The available glossaries to other literary prose are equally unproductive of these irregular spellings. The highest single yield, that from the Laing Pitscottie, includes schoissin 'chosen', schyre 'chair', staw (p.t. of stele), stowin (p.p. of stele) and schowne, shone and sowun among the variants of 'soon', in each case only as a much less common (sometimes unique) variant of the regular spelling(s). But as more data become available it may turn out that there were others among the late 16^th- and early 17^th century literary and official scribes and men of letters who, as well as enlarging their scope of spelling variation with the new anglicised spellings, were also relaxing some of the traditional Scottish prohibitions. With spellings like ga 'gall' and stowin 'stolen', blaithe 'blithe' and whait 'white', swyt 'sweat', sweik (= swilk or sic 'such'), schosin(e) 'chosen', lainthe 'length' and strainthe 'strength',[97] Rev. James Melvill is at least one instance of a highly literate writer of late MSc who was no longer conforming strictly to the earlier rules of correct orthography. Most of these and others like them, however, appear in Melvill's personal narrative; in copying official documents and formal declarations, including those originally composed by himself, his spelling standards were perhaps less relaxed.

Writers of texts of the strikingly substandard category include a number of local clerks (for example, certain of the burgh clerks of Aberdeen, Ayr, Dunfermline, Elgin, Inverness, Peebles, Selkirk, Stirling and Wigtown) and some writers of private documents, such as letters and accounts (see also §8.3.2).

Some of the spellings employed by writers such as these are extremely idiosyncratic: John Wallwod's habit of doubling his initial <ll > (on analogy with <f>) in e.g. lledein 'leading' or 'loading' and llent (1600 Misc. Hist. Soc. X 67), or his unique spelling of 'silver' as selder (1602 ibid. 73) (a reverse spelling on the analogy of the loss of <d> after <l> in some other words, and with <e> for ), or the practice of the Stirling clerk of writing <sy-> for normal <sw->, e.g. in syourd 'sword', syene 'swine', syoerne 'sworn' (1525 Stirling B. Rec. I 24, 25) or the same clerk's <kt> for normal <cht> in slaktir (1526 ibid. 26), or his quite logical (but unique) lujene (for luging 'lodging') (1525 ibid. 23), or Skipper Morton's unique phonemic spelling of 'narrow' as nawreye (a1600 Skipper's Acc. (Morton) 48a).

More often, however, these irregular or substandard spellings conform to recurrent patterns common to this group of writers as a class.

• It was apparently chiefly writers of this kind who were given to writing <sch-> where standard spelling prescribed <ch->,[98] e.g.:

schallans (= chalange), schairge (= charge), schertee (= cherite 'charity'), scheis, schosing (= chese, chosin, 'choose, chosen')

and, though apparently less frequently, the converse of this in e.g.:

chep, cheip 'sheep' (1468 Peebles B. Rec. I 157, 1503 Dunferm. B. Rec. I 132, 1557 Inverness Rec. I 10), flecht and flechour (as well as flescheir) (1522 Stirling B. Rec. I 15), feych 'fish' (1532 Selkirk B. Ct. (ed.) 121).

• A similarly widespread tendency in many of these texts is to interchange <th> and <t> spellings in e.g.:

tha tyngis (14. . Liber Calchou 449), efter the theching of this trethis (ibid. 451), at thwa termis (1456 Peebles B. Rec. I 113), ragrathing (1520 Stirling B. Rec. I 7), scheit ('sheath') makar (1521-2 ibid. 14), thwa siluer ringis (1540 Elgin Rec. I 49), outout (= outouth 'outwith') (1541 ibid. 65).

Other recurrent 'substandard' variants are:

• <i, y> for Vowel 2 (standard <e, ei, ey>) for instance by our Stirling clerk and by Mary Queen of Scots), reflecting the raising of the vowel by the GVS;

• <ou, ow> for standard , both Vowel 19, e.g. hourt, soun[99] and Vowel 7, e.g. goud, schoun 'shoes';[100]

• <e> for Vowel 15 (standard ), e.g. meln, mestour, begit (= biggit 'built'), kel (= kill 'kiln') (1456 Peebles B. Rec. I 115ff. ); well (= will), hem (= him), mecht (= micht) (Lady Home); selder and wettell (= victual) by John Wallwod. This reflects the lowered phonetic realisation characteristic of many modern dialects;

• Meurman-Solin (1999: 317, 320 n.3) adds <wo> for <wi>, i.e. Vowel 19 (see §3.5), e.g. wosdum 'wisdom' (cf. 15^th century <o> forms of wympil(l n. and v., windo(w v., wisp n.)

In addition, some rather more widely distributed variants are especially common in recognisably substandard contexts, for instance:

• <ai, ay> for Vowel 1 (standard <i, y, yi>), reflecting diphthongisation by the GVS;

• -ene, -ein, -eyn as an addition to -ing, -yng, -in, -yn for the verbal noun-ending (cf. e.g. lledein above);

• ra- beside re- in e.g. ragratour, ralef, raward. Meurman-Solin (1993a: 243), however, finds this in all text types in the period before 1570.

Some of these variations are no doubt phonological in origin. Evidently, like some illiterate spellers today, these substandard spellers were given not only to inexplicable idiosyncratic modifications of the spelling system, like the Stirling scribe's <sy->; but also to improvements in the direction of greater phonological accuracy, like -ene, -ein for the verbal noun ending, confirmed in modern speech and early rhyme; it may well have been, in spoken use, more common than its Vowel 15 doublet which the standard spelling -ing, -yng on the face of it represents.

But of course in the absence of external confirmation of this kind we must be much more cautious. It may be that the interchange of <sch> and <ch> noted above originally arose from an assimilation in some or all positions of /ʧ/ to /ʃ/ in some dialects (as in modern Caithness) or idiolects, and that the spellings noted above either directly reflect this or follow as reverse spellings. Similarly with the interchange of <th> and <t>, reflecting a merger of /θ/ with /t/ (as in 17^th century Galloway and modern Orkney and Shetland, as well as many varieties of English). This explanation would have the advantage of accounting for the tendency of these spelling habits to recur in widely dispersed texts. At the same time, we cannot exclude the possibility that we are dealing with purely orthographic phenomena.