Orthographic regularization in Early Modern English printed books : Grapheme distribution and vowel length indication

The present study is an attempt at assessing the level of consistency in the orthographic systems of selected sixteenth and seventeenth-century printers and at tracing the influence that normative writings could have potentially exerted on them. The approach taken here draws upon the philological tradition of examining and comparing several texts written in the same language, but produced at different times. The study discusses the orthography of the editions of The Schoole of Vertue, a manual of good conduct for children, published between 1557 and 1687. The orthographic variables taken into account fall into two criteria: the distribution and functional load of the selected graphemes and the indication of vowel length.


Aims, corpus and methodology
Orthography understood as a codified and socially binding norm is a relatively recent notion.Although spelling started to raise interest among early linguists already in the sixteenth century (e.g., Smith, 1568 andHart, 1569), it had not achieved standardization before the second half of the seventeenth century.Nonetheless, by the end of that century, it had become largely regular in printed documents (Scragg, 1974;Salmon, 1999;Görlach, 2001).The question, which so far remains controversial and under-researched, is whether the regularization and standardization of English orthography should be attributed to the early modern phoneticians, orthoepists, spelling reformers and lexicographers (Brengelman, 1980;Carney, 1994;Milroy and Milroy, 1999) or to printers and publishers (Krapp, 1909;Howard-Hill, 2006;Rutkowska, 2013).To date, few authors have undertaken detailed analyses of spelling patterns in early printed documents (Blake, 1965;Osselton, 1984;Aronoff, 1989;Horobin, 2001;Rutkowska, 2005Rutkowska, , 2013;;and Howard-Hill, 2006), but even these case studies point to the potentially significant role of the printers in the establishment of the orthographic system of Modern English.The present study is an attempt at tracing the development of spelling practices in several editions of a popular manual of good manners entitled The Schoole of Vertue, published between 1557 and 1687, as well as identifying potential influences of normative writings on the sixteenth-and seventeenth-century printers.In fact, no formal distinction is made in this study into publishers, printers and compositors, because these occupations were frequently Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM overlapping in the earliest printing houses (Duff, 1906: 72;de Hamel, 1983: 29).Individual printing houses which issued the particular editions studied here are treated as entities whose orthographic "house styles" are investigated (if these can be identified).
The approach taken here draws upon the philological tradition of examining and comparing several texts written in the same language, but produced at different times.The study discusses the orthography of twelve sixteenth-century and seventeenthcentury editions of The Schoole of Vertue.The first known edition is a manual of good conduct for children written by Francis Seager (Segar), printed and published by William Seares (Seres) in 1557.62All the subsequent editions contain both the manual and a set of various prayers.The manual covers two thirds of the text and the prayers one third.This compilation was prepared by Robert Crowley, and first published by Henry Denham63 in 1582.The other editions analyzed in this study comprise those printed by John Charlewood for Richard Jones (1593), G. E. (possibly George Elde) for T. P. (full name not known) and I. W. (conceivably John Wright senior) (1621), 64 M. (Miles?)Flesher for Robert Bird (1626Bird ( , 1630Bird ( , 1635Bird ( , and 1640)), M. Flesher for John Wright (junior, 1660),65 E. Crowch for J. (John) Wright ( 1670), an anonymous printer for J. (John) Wright (1677), as well as an anonymous printer for M. W. (full name not known) and George Conyers (1687).The texts under analysis have the form an electronic database of transcriptions of the editions based on the facsimiles available at Early English Books Online (henceforth EEBO).66The corpus contains 87,737 running words.The exact word counts of the analyzed copies of particular editions as well as their catalogue numbers are provided in Table 16.According to the English Short Title Catalogue (ESTC), two other editions were produced, in 1642 and 1698, but these are not available at EEBO.
As for the other physical features of the editions, presumably all of them were issued in the small, octavo format, and printed mostly in blackletter, with the title 62 The title of this edition suggests that it is not the first one, because it was "[n]ewely perused, corrected, and augmented", but neither Pollard and Redgrave's Short Title Catalogue nor ESTC provide any record of an earlier edition.Some years of publication and names of the printers (or publishers) have been reconstructed.These years and names are provided in square brackets in Table 16 and in the References.In the body of the chapter, these square brackets are avoided for aesthetic reasons.63 According to the colophon, he was "the assigne of W. Seres" (1582, D7v).In this and the other citations from The Schoole, the capital letter refers to the quire, the number to the leaf and the small letter to the side of the leaf, i.e., r to the front side (recto) and v to the back side (verso).64 George Elde and John Wright cooperated at least in some occasions (see Rutkowska, 2013: 68-69 for details).65 In 1642 John Wright junior took over 62 copyrights which had belonged to Robert Bird and Edward Brewster (Plomer 1907: 198).66 The transcription of the first edition of The Schoole can be consulted at EEBO, but I decided to prepare my own transcriptions of all the editions available in the form of facsimiles.
Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM pages, running heads, titles of sections, chapter initials, and some important words (mainly names) highlighted in the Roman typeface in the majority of the editions.The main focus of this chapter is the realization and development of several orthographic variables in all the editions under consideration.The variables taken into account fall into two criteria: the distribution and functional load of selected graphemes,67 and the indication of vowel length.The discussion of the findings of the present study is divided into two sections, (2) and (3), one for the each of the criteria.The indication of vowel length is listed beside orthographic distinction between homophones as well as the establishment of morphological and etymological spelling in Salmon (1999: 21), all of which are considered crucial in the research on orthographic standardization in English.Wherever relevant, comparisons are made with the findings of the extensive study analyzing printers' orthographic systems in thirteen editions of the Kalender of Shepherdes (henceforth KS) published between 1506 and 1656, summarized in Rutkowska (2013).The KS corpus is approximately ten times larger than that comprising The Schoole editions.
All the tables and figures refer to The Schoole editions.Unless indicated otherwise, the tables and discussions contain absolute values.The years of publication (not 67 When discussing the correspondences between graphemes and phonemes, I also use the terms (nouns) consonantal and vocalic in this chapter, with reference to graphemes corresponding to consonants and vowels, respectively (following Wełna, 1982: 10).A consonantal may, occasionally, refer to a sequence of graphemes, if they correspond to a single consonant, as for instance, in speech.
Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM sigils) will be used to refer to particular editions in order to facilitate comparisons across the editions with regard to the chronology of changes.

Distribution of graphemes
One of the elements in the regularization of spelling in the books printed in English in the early modern period was the introduction and spread of the modern rules of grapheme distribution and graphophonemic correspondences.The graphemes affected by these changes include, most characteristically, <u> and <v>, as well as <i>, <j> and <y>.The findings of the present study concerning their distribution show the interrelations between particular graphemes (and phonemes) as well as between spelling and typography.
The results regarding the distribution of graphemes in The Schoole editions are compared with the findings of the study on the KS (see particularly Rutkowska, 2013: 117-130), and, where relevant, with the practice and views of early modern theoreticians and lexicographers.

<v> and <u>
English medieval manuscripts generally observed the principle of the complementary distribution of <u> and <v>, established earlier on the continent, according to which these graphemes could correspond to both the vowel /u/ and the consonant /v/ but <v> was used word-initially and <u> word-medially (see e.g., Scragg, 1974: 14;Salmon, 1999: 39).This old rule is followed exceptionlessly in the editions of The Schoole published between 1557 and 1621.In the editions printed between 1626 and 1635, the distribution of the graphemes <u> and <v> depends on the typeface.In 1626, the passages in the Roman typeface are set according to the modern principle, where <u> functions as a vocalic and <v> as a consonantal, with only one exception (euening).In the same edition, the compositor(s) setting the text in blackletter usually observed the medieval complementary distribution rule, but not consistently, allowing for several instances of spellings such as unto, up, utterance, us, upon, heavenly, and deliver.It also contains the form uertues, which is certainly a typo.In 1630, the confusion between the distribution rules is avoided: the old rule is observed in blackletter and the modern one in the Roman typeface, without a single exception.The 1635 edition displays the same arrangement, but this time a few exceptions are allowed, namely ungentlenesse, us, unfainedly, and up, where the new principle sneaks into some passages in blackletter.Eventually, in 1640 and all the subsequent editions, <u> and <v> are distributed in the modern, contrastive way.This seems rather late, as according to OED (s.v. "U, n.1"), the modern distribution had become established by 1630.
An analogical change in the distribution of the graphemes can be noticed in the KS editions, with those published between 1506 and 1631 following the old principle, and the last edition, of 1656, the new one.However, single exceptions to the old rule are recorded already in the earliest editions (see Rutkowska, 2013: 118-119 for details).
The remarks on the employment of <u> and <v> made above concern lower-case letters.In both the KS and The Schoole, no functional difference is made between the capital versions of these graphemes.In blackletter, the same <U>-like shape corresponds to the vowel /u/ and the consonant /v/, for example in Unto, Upon and Up as well as Uice, Uertue and Uicars.In the relatively infrequent passages in the Roman typeface, the only words recorded are those starting with the consonantal, a <V>-like character (e.g., Vertue and Verdict), so no conclusions can be drawn on the potential functional differentiation.

<i> and <j>
In England, the first attempts to replace the grapheme <i> with <j> where it corresponded to the consonant /dʒ/ go back to the late sixteenth century, but it became generally established after 1630 (OED, s. v. "J, n.").This replacement can also be traced in The Schoole editions.No occurrence of the grapheme <j> can be found in those published between 1557 and 1621.In 1626, word-initially, eighteen tokens of lower-case <j> are recorded in jangle, jewell, joy, joyfull, judge, judgements, judging, just, justice, and justly, printed in blackletter, as well as one upper-case token in the Roman type in Jacobs.However, not all the tokens of the relevant lexemes are spelt in the modern way, as this edition contains also the variant forms ioyfull, iudge, iudgement(s) and iust in blackletter, lower-case (five tokens), Iudges and Iust in blackletter, upper-case, as well as one token of Iacobs in upper-case, in the Roman typeface.Word-medially, <i>, in all the words where it corresponds to /dʒ/ in the previous editions, is replaced by <j> in 1626 (i.e., Majesty, reject, rejoyce, subjects, and vnjust, all in blackletter).Surprisingly, in the following edition (1630) one can see a switch to the old system, without <j>, apart from one instance of Jacobs in the Roman typeface.By contrast, in 1635 the modern distribution of <j> is in place, except when a word begins with a capital letter.In the latter case, the old variants are used (i.e., Iacobs, Iudge(s), and Iust).In the 1640 edition, <j> is fully established word-initially when a word starts with a lower-case letter, and for the capitals, the use depends on the typeface, with Iudge(s), Iust and Iustice in blackletter, and Jacobs in the Roman typeface.The only difference between 1640 and 1660 is that in the latter Iacobs (in the Roman typeface) is also spelt in the old way.In the last three editions (1670-1687), the modern spelling is the standard, except when the capital occurs in blackletter (Iust, Iudgements, Iudg, Iudges, Iustice), which simply does not seem to have a separate character for <J>.
By comparison, in the KS, <j> is recorded solely in the last edition (1656), in journey, joynt and judg(e)ment, printed in the Roman typeface.This grapheme is not employed at all to refer to /dʒ/ in blackletter, even in lower-case.Nevertheless, in the editions issued in 1506, 1528, 1580-1585, and 1604-1618, it is occasionally used as a final allograph of <i> in Roman numerals (e.g., iiij and xij).The absence of <j> in 1656 may have resulted from the lack of a separate character for it in blackletter in the printing house which issued it.

<i> and <y>
It is usually claimed that the graphemes <i> and <y>, acting as vocalics in wordmedial position, were used interchangeably and haphazardly by the early printers of books in English (e.g., Blake, 1965: 65, 71;Brengelman, 1980: 349;Salmon, 1999: 28;Nevalainen, 2012: 144).Nevertheless, some studies show that already in early sixteenth-century books, the distribution of <i> and <y> is not as random as has been suggested.More precisely, <i> is regularly used in function words (e.g., this, which, his, with)68 and <y> in content words (Aronoff, 1989: 68-69, 93-94;Rutkowska, 2013: 121-130).69 Such division can be noticed in the earliest KS editions (1506-1528), but in the second half of the sixteenth century and the first half of the seventeenth century <y> is gradually replaced by <i> also in content words.Interestingly, him(self), their, and will behave differently from the remaining function words, displaying the spelling patterns characteristic of content words (see Rutkowska, 2013: 123-124 for potential reasons of this differentiation).Consequently, it does not seem surprising that the only function words which still contain <y> in The Schoole edition published in 1557 are will, their, and him(self).In 1593, theyr is the last remnant spelling variant of a function word containing <y>.By contrast, in 1557, <y> is routinely employed word-medially in content words.In 1593 and later editions, however, its use steadily decreases to the benefit of word-medial <i>.By 1626, the sole old-system spellings containing <y>, which cannot be found in PDE, are voyder, lyar, lye(s), obeysance, and tryall.Of these, tryall is replaced by triall in 1630, and voyder is replaced by voider in 1640.In light of these developments, it is quite remarkable that the edition of The Schoole printed in 1582 by Henry Denham, does not contain any words with medial <y> apart from Physician, Pythagoras, Egypt, and tyrant, all of Greek origin.Unexpectedly, even in those forms where it is standard in PDE (across the morphological boundary), Denham avoids <y>, using, for instance, saieng, plaieng, enuieng, ioifull, and praier (cf. PDE saying, playing, envying, joyful, and prayer, respectively).

Theoreticians' opinions on the distribution of graphemes
The graphemic conventions found in the early printed books replicate, at least to some extent, the patterns common in medieval manuscripts, which are due to continental influences.This certainly applies to the distribution of the graphemes <u> and <v>, as well as to the frequency of <y> word-medially.The latter was originally motivated by graphetic considerations, as <y> (compared to <i>) increased the legibility of those words where it immediately preceded graphemes such as <m> and <n> (e.g., in him(self), thing, bring, crime, time, and plain).These "received" rules were contested by early modern spelling reformers.For example, John Hart urged his readers to distinguish between <u> and <v> as well as between <i> and <j>, already in his The opening of the unreasonable writing of our inglish toung..., an unpublished pamphlet written in 1551 (Dobson, 1957, I: 69).70 Hart's endorsement does not seem to have influenced the contemporary printers, as even his own works, Orthographie and Methode, published in 1569 and 1570, respectively, were printed in accordance with the old principles as regards the distribution of <u> and <v> and the absence of <j>.Nonetheless, it is noteworthy that both Hart's books contain very few instances of word-medial <y>, although they were issued by different printers, Orthographie by William Seres and Methode by Henry Denham.Seres printed the first edition of The Schoole before his cooperation with Hart, whereas Denham issued its second edition over a decade after such cooperation.Perhaps the latter's views concerning the distribution of the vocalics <i> and <y> affected Denham's spelling practices, which are characterized by the extremely low number of tokens with word-medial <y>.
Later on, Butler (1633: "To the Reader"), explicitly stated the distribution principle according to which <v> and <j> should correspond to consonants and <u> and <i> to vowels.The new rule is observed already in the book in which he expressed this opinion, printed by William Turner.Butler's recommendation seems to have become authoritative because a general change in spelling practices soon occurred in printing houses.It is also recorded in an edition of Coote's handbook as well as an edition of Cockeram's dictionary, both published in 1637.

Indication of vowel length
This section examines the three main methods which early modern printers employed in the documents under consideration for indicating long vowels.These include doubling the vocalic corresponding to the relevant vowel, adding the final <e>, and the combination of both solutions.Attention is also paid to the use of the digraph <ea>, an aspect relevant both to vowel length and to qualitative differentiation between vowels.The functional significance of <ea> is complex, due to both qualitative considerations and later quantitative modifications, that is, mainly the shortening of the vowel (e.g., bread, head, wealth, health etc.); therefore, for clarity of the findings, the lexemes with the shortened vowel are not taken into account here.Also other lexemes with shortened vowels are not examined here (e.g., book, good and love).
Nearly all the words analyzed in this section are monosyllabic, with few exceptions.Singular forms are taken into account with regard to nouns, and unmarked forms with regard to verbs.Also, only native words are considered; loanwords are excluded because the spelling patterns displayed by them might be at least partly obscured by foreign orthographic influences.Some other groups of words excluded from the analysis are those ending in /r/ and in consonant clusters, due to the complex histories of the vowels in such phonotactic contexts (often associated with shortening), as well as words ending in /v/, /s/ and /z/ (e.g., leave, prove, house, and choose) because they do not show variation and are consistently spelt with the final <e> (with few exceptions).
The analysis focuses on the spellings of the lexemes containing the reflexes of the Middle English vowels /eː/, /ɛː/, /oː/ and /ɔː/, that is, those which are generally considered to have undergone the so called Great Vowel Shift (henceforth GVS, see for instance Lass, 1999: 72-73).Those words in which the vowel derives from ME /aː/ (e.g., make, name, same, and wipe) are spelt consistently with the final <e> in all The Schoole editions, so they do not provide evidence relevant to a comparative study.Also the lexemes containing the reflexes of ME /iː/ (e.g., life, time, and pride), as well as the reflexes of ME /uː/ (e.g., about, out, and without) do not show sufficient spelling variation and modification with regard to vowel length indication (though they often display the variation between <i> and <y>), to allow a quantitative comparative study.
Similarly to the previous section, the findings concerning vowel length indication obtained on the basis of The Schoole editions are confronted with, on the one hand, the results of the study on the KS (see Rutkowska, 2013: 141-161) and, on the other hand, with the opinions of early modern language authorities.16 present the ways of indicating vowel length in words ending in a closed syllable,71 with the stem vowel originating in Middle English /eː/, which developed into /iː/ before 1500 (Dobson, 1957, II: 651).The words whose spellings are considered here include deed (n), deem (v), feed (v), feel (v), heed (n), indeed (adv), keep (v), meek (adj), meet (v), meet (adj

Early Modern English /iː/ reflecting Middle English /eː/
, sweet (adj), and weed (v), as well as been, feet, and teeth, which comprise the past participle of the verb be and the plural forms of the nouns foot and tooth, respectively.Most words listed above yield between one and three tokens in each edition.Some lexemes, namely deed (n), feed (v), heed (n), keep (v), and need (n), have 5-10 tokens in most editions.In the first edition of The Schoole (1557), similarly to the four earliest editions of the KS, the printers indicate the length of the vowel in the stem almost exclusively by adding the final <e> (90% of the tokens).72By comparison, the <eCe> pattern is recorded in 99.5-100% of the tokens in the KS in the years 1506-1528, and 93% in 1556.In 1582 and 1593, this method is abandoned in The Schoole; instead, the doubling of the vocalic and the combination of the doubling with the final <e> compete with each other, but the latter slightly prevails, accounting for approximately 55% of the tokens.In contrast, although the share of the <eCe> spellings decreases sharply between 1570 and 1585 (from 78.5% to 14%), in the KS it remains in use throughout the sixteenth century.During the same period, the number of the <eeCe> spellings in the KS increases from 22% to 73%, and that of the <eeC> spellings moves from 0% to 13%.In The Schoole, the rivalry between <eeCe> and <eeC> continues until 1670, when the modern spelling becomes the standard one.Yet, <eeC> becomes dominant already in the 1620s and the 1630s, approaching 60%, and exceeds 87% and 90% in 1640 and 1660, respectively.In the KS, the <eeCe> pattern is the major method of indicating vowel length between 1600 and 1611 (68-71%), and then drops to 22% in 1618, giving ground to the modern spelling (78%); it regains some importance in 1631 (49%), but is totally ousted by <eeC> in 1656 (i.e., even earlier than in The Schoole).
Apart from the main three spelling patterns, in the 1557, 1582 and 1621 editions of The Schoole, single tokens of speach(e) (in all the three editions) and streate (only in the first edition) are also recorded.In the case of the noun speech, the uncertainty concerning spelling may be due to the influence of the graphemic shape of the semantically related verb speak.
In open syllables, the graphemes corresponding to the vowel /iː/, derived from ME /eː/, were subject to different tendencies in content words and in function words.The vowel length in content words, including agree (v), degree (n), flee (v), free (adj), knee (n), see (v), and three (num), is not marked in The Schoole in 17 out of 28 tokens (61%) in 1557 (mostly in see), but in all the remaining editions these lexemes are spelt in the modern way.A lower level of unmarked vowel length, that is, 40% is recorded in the 1556 edition of the KS.73 Nevertheless, in the three earlier KS editions this value is much higher, amounting to 89% in 1506, 71% in 1518, and 73% in 1528; then the popularity of the <Cee> pattern increases sharply from 60% in 1556 to 80% in 1570, over 90% in 1580-1585, and 100% in 1600.72 With regard to the KS, the following lexemes are taken into consideration: green (adj), keep (v), meek (adj), meet (v), need (n), need (v), sleep (n), sleep (v), speech (n), heed (n), and sweet (adj) (for more details see Rutkowska, 2013: 143-145).73 The list of lexemes examined in the KS covers degree (n), flee (v.), free (adj), knee (n), see (v), three (num), and tree (n).In function words full regularization takes more time in The Schoole, but the tendency is clear already in the first edition.Table 18 makes it clear that function words including be (v), he (pro), we (pro), and ye (pro), are almost invariably spelt with a single <e>.The spellings with the <Cee> pattern account for 0.5-7.6% of the tokens, depending on the edition.The object form of the pronoun thou (pro) (i.e., the(e)), has been excluded from the count because of its unusual frequency in this corpus.In 1557, the accounts for 14.6% of the tokens of thee (7 out of 48), but single instances of such spelling can be found only in 1593, 1621, 1640, and 1660.Apparently, this pronoun follows the pattern of content words, presumably due to avoidance of homography with the definite article the.The spelling of the relevant function words follows an analogical trend in the KS (see Rutkowska, 2013: 146 for more information).

Early
Modern English /ɛː/ and /eː/ reflecting Middle English /ɛː/ This subsection examines the graphemic representations of the lexemes which in PDE are spelt with <eaC>, and in which the vowel goes back to Middle English /ɛː/.They include beat (v), clean (adj), deal (v), each (a, pro), eat (v), heat (v), lead (v), lean (v), mean (adj), mean (n), meat (n), reach (v), read (v), seam (n), speak (v), and teach (v), as well as break (v) and great (adj).During the Early Modern English period, /ɛː/ was raised to /eː/ in these words, but the new pronunciation was accepted in careful speech in the later part of the period (Dobson, 1957, II: 607, 619).In PDE the majority of the words listed above have the vowel /iː/, due to a post-GVS merger with words such as meet and seem (see for instance Lass, 1999: 73), with the exception of break and great, which contain /eɪ/.All the lexemes considered here are represented by close syllables.The only relevant lexeme which ends in an open syllable and contains a vowel originating in ME /ɛː/ is sea, but it does not show spelling variation in The Schoole editions.
Table 19 and Figure 17 show the spelling patterns of the relevant lexemes, apart from break and great.In 1557 and 1582, <eaCe> predominates, accounting for 73% and 65% of the tokens, respectively.In the first edition, there are also remnants of the old <eCe> pattern, but it is recorded only in each (24% of all the tokens), and so is the <eC> pattern in 1582 (17.5%).The fact that the latter does not contain any indication of vowel length implies that the vowel in each could have been perceived as short, perhaps because it was followed by a voiceless affricate.As regards single instances of <eeCe>, these concern solely the lexeme clean (adj).The modern pattern, <eaC>, yields merely one instance in 1557, but it quickly rises in importance, accounting for 16% of the tokens in 1582, and for 65.5% in 1593.There is some variation between <eaCe> and <eaC> in the years 1621-1660, with the modern spelling prevailing in 1593, losing leadership in 1621, regaining it in 1626 and 1635, showing equal importance as <eaCe> in 1630, and eventually becoming the standard in 1670.<eaC> remain equal and the old spelling pattern is gone (though single instances of <eC> occur until 1611), and between 1611 and 1656 the number of the modern spellings increases from 55% to 98%.In fact, the <eaC> spellings reach nearly 79% already in the 1618 edition of the KS, which seems early, compared to 70% still in the 1660 edition of The Schoole.
The lexeme great acquires the modern graphemic shape already in the second edition of The Schoole (1582), whereas in the 1585 edition of the KS, the form great accounts for almost 94% of the tokens of great and in 1600 it nears 100%, with a single token of greate found.By comparison, break is spelt predominantly as breke until 1580, and then breake until 1656.In The Schoole, in turn, breake is used between 1557 and 1635, and break in the following editions.

Early Modern English /uː/ reflecting Middle English /oː/
In this subsection, lexemes with the <ooC> pattern in PDE, containing the long vowel /uː/, a reflex of ME /oː/, are considered.The group of (close-syllable) words analyzed here comprises brood (n), doom (n), food (n), mood (n), noon (n), school (n),74 smooth (adj), spoon (n), and soon (adv).Apart from school (13-37 tokens, depending on the edition) and food (9 tokens, except in 1557, where one token is found), these lexemes occur one to two times per edition, so school is definitely overrepresented here.In The Schoole, in 1593, the number of tokens decreases considerably (see Table 20), which is due to the split of the running head into two parts (The Schoole on the verso pages and of Vertue on the recto ones), and the consequent drop in the instances of schoole in this edition.The running head is not split in any other edition.Table 20 and Figure 18 indicate that the old <oCe> pattern is used only in the first edition of The Schoole, in 13.6% of the tokens.In the KS,75 expectedly, the <oCe> 74 The lexeme school is a loanword from Latin, but since it was present already in early Old English, and seems to have been fully naturalized, I have decided to include it into the count in this study.75 In Rutkowska (2013: 151-152) the tokens of choose (v), move (v), and prove (v) have also been taken into account, but these have been omitted in the present study.The <ooCe> prevails between 1557 and 1660 in The Schoole (i.e., for over one hundred years) accounting for 86.4% in 1557, and between 67.6% and 76% in the editions published in the years 1582-1660.In the KS, this pattern dominates between 1556 and 1631, rising from 55% in 1557 to 85% 1585, but it starts to decrease in 1600, and it is virtually eliminated in 1656 (i.e., earlier than in The Schoole).
The cumulative method of vowel length indication is fully replaced by the modern <ooC> spelling in 1670 in The Schoole, but already between 1582 and 1660 it accounts for 24%-27% of the tokens.By comparison, in the KS, single instances of the modern pattern are recorded already in 1506 and 1518, then it is dropped for eighty years, to reappear in 1600; between 1600 and 1631, it covers 20-32% tokens, with the exception of 1618, where it reaches as much as 48%; finally, it replaces <ooCe> almost entirely, with 98.5%, in the final edition.There are few words in The Schoole containing /uː/, the reflex of ME /oː/, in an open syllable.They include do (v), too (adv, always spelt too), and two (num, spelt twoo and two in 1557 and two elsewhere).Among these, only the lexeme do is sufficiently common in the corpus to merit a separate comment.As can be understood from Table 21 and Figure 19, do is the dominant form in the first edition, covering 87% of the tokens.In 1582 and 1593 it is replaced with doo, which, in the former, is used by Henry Denham with surprising consistency (100% of the tokens).In 1621 and 1626 doe takes over the leadership among the do forms, accounting for approximately 91% of the tokens.In 1630 do becomes current again, with 47.8%, but between 1635, 1640 and 1660 doe retakes the lead, with 94%, 77%, and 81%, respectively.In 1670 and the remaining two editions do has the modern spelling.
In the KS, the modern variant prevails in most editions and, surprisingly, it is used most consistently in the earliest versions of the book, especially in de Worde's edition of 1528, where it is the only spelling of do, but also in the other editions published in the years 1506-1570, never dropping below 97% of the tokens.In 1580 and 1585, do is overtaken by doo, which yields 48% and 59% of the tokens, respectively.Interestingly, the printer of these two editions was probably John Charlewood (who also issued the 1593 version of The Schoole); this can be responsible for the popularity of doo in all the three editions.The modern variant is, again, used nearly exclusively between 1600 and 1611 (91-94%) in the KS, but its share drops in 1618, to the benefit of doe, which accounts for 55% of the tokens in this edition and 95% in 1631.In the final edition of the KS, do covers over 79%, and doe nearly 21%.In view of the high status of the modern variant in the 1656 KS edition, one may wonder why there are still so many instances of doe (over 80%) in 1660 in The Schoole.Perhaps the choice of the prevailing variant was determined by the fact that the latter edition was issued by Flesher, the printer of all The Schoole editions in the years 1626-1660, who could have been simply accustomed to this spelling.The lexemes containing the reflexes of ME /ɔː/ recorded in The Schoole include abroad (adv), bone (n), cloth (n), hope (n), go (v), nose (n), oath (n), so (adv and conj), sloth (n), stone (n), stroke (n), and yoke (n).In the majority of these words, the vowel was raised to /oː/ in the sixteenth century, or at the latest, before 1650 (Lass, 1999: 85, 94-95), and started to be represented by the spelling <oa>, modeled on the digraph <ea> (Wełna, 1978: 219).The change did not affect abroad (adv) and cloth (n), where the vowel remained mid-open (Wełna, 1978: 219).In bone (n), hope (n), nose (n), and stone (n), the spelling pattern <oCe> is the only one employed in The Schoole editions and is the standard one in PDE.The first edition does not contain a single token of <oa>.The patterns <oaC> and <oaCe> are first recorded in the 1582 edition, in the forms abroad and abroade.The former remains the main variant of the lexeme abroad (adv) throughout The Schoole editions, but the spelling with the final <e> is also found in 1640 and 1660.No other lexemes are spelt with the digraph <oa> in the sixteenth-century editions.The situation changes in 1621, where the forms sloath and stroake, replace the previously used sloth (slouthe in 1557) and stroke, representing sloth (n) and stroke (n), respectively, but the new forms continue to alternate with the old ones in the following editions.76In 1621 and later editions, <oa> can be found also in cloath, sloathfull, and oathes, alternating with <o>; however, these forms cannot be treated as evidence of vowel length indication because the vowel in cloth underwent shortening, slothful (adj) ends in a derivational suffix containing a short vowel, and oathes, representing the plural form of oath (n), ends in an inflectional suffix.The digraph <oa> is also recorded in board (for the noun board) in 1626 and the following editions, alternating with boord (up to 1660), and in yoak (for the noun yoke) in 1670 and in 1687, as well as in the gerund form cloathing (of the verb clothe) in the years 1670-1687.
In the KS editions, the digraph <oa> appears in 1585 for the first time in the context which can be interpreted as likely to indicate vowel length (cf.1582 among The Schoole);77 it is recorded in broade (beside the earlier form brode), representing broad (adj), spelt exclusively as broad in the seventeenth-century editions.Between 1600 and 1656, it can also be found in throat(e) and smoak(e), the representations of the lexemes throat (n) and smoke (n), respectively.These three lexemes are represented by altogether 20-22 tokens, depending on the edition.In 1600 and 1604, the old <oCe> pattern still accounts for some 5% of the tokens, <oaCe> for 52-55%, and <oaC> for 40-43%.In 1611, the old pattern is not recorded any more, and <oaC> starts 76 They yield only single tokens, so no percentages are provided here.77 See Rutkowska (2013: 160) for the information concerning earlier (and later) occurrences in polysyllabic words.
In The Schoole, two lexemes with the reflexes of ME /ɔː/ are represented by open syllables, i.e., so (always with the modern spelling) and go.The lexeme go (yielding six tokens) is spelt in the modern way already in the first and second editions of The Schoole.However, in 1593, goe takes over and remains the main spelling of go between 1621 and 1660, with only single instances of go in 1621 and 1640.In 1670 and the remaining editions, go regains its position as the only exponent of go (the forms with inflectional endings are not counted here).By comparison, in the first edition of the KS (1506), the spelling go accounts for 88% of the tokens, compared to 12% reserved for goo, a form which does not emerge in The Schoole editions at all.In the years 1518-1556, go is the only spelling, but between 1570 and 1585, the share of go falls from 93% to 60%, with goe rising in importance.Then, go regains the main position in the years 1600-1611, with 95-98% of the tokens, the goe tokens increase from 47% to 79% from 1618 to 1631, and finally go covers 57% of the tokens in 1656 (see Rutkowska, 2013: 158-159 for the relevant tables and diagrams).
Although the corpus of The Schoole editions is too small to offer sufficient information on the tendencies concerning the spelling patterns in the word groups examined in this subsection, some findings of Rutkowska (2013: 158-160) seem confirmed here; that is, the dominance of go in the sixteenth century, and the temporary rise in importance of the variant goe in the 1630s.

Indication of vowel length in The Schoole editions: summary of changes
A comparison of the changes affecting the sets of spelling patterns within the variables considered in this section shows that the old spelling pattern <VCe>78 is still strong in the first edition of The Schoole in the group of closed-syllable lexemes which contain the vowel /iː/, originating in ME /eː/ (see Figure 16), and less so in the lexemes containing the reflexes of ME /ɛː/ and /oː/ (Figures 17 and 18).The cumulative <VVCe> patterns, including both vocalic doubling and the final <e>, and the <VVC> ones coincide for a few decades in lexemes with high vowels (originating in ME mid-close ones), but in 1670 the modern spellings are already regular in three out of four groups of words considered here in detail.
The digraph <ea> is already well established in the first edition of The Schoole, but the corresponding <oa> digraph starts to be used systematically as late as 1621, and is not yet stabilized in 1687, affecting the level of spelling regularization of the group of lexemes containing the reflexes of ME /ɔː/.
In open-syllable words, containing /iː/, a systematic orthographic differentiation between function words (spelt with a simple vocalic) and content words (with a double vocalic) is visible already in the second edition.As regards the lexemes containing back vowels in open syllables, both the types and tokens are few, and no orthographic differentiation of this kind can be noticed.In the case of do, vowel length is indicated by doubling in the 1580s-1590s, and by the addition of the final <e> in the seventeenth century, before 1670.The final <e> is also employed to emphasize the length of the vowel in go in the years 1593-1660.In 1670 and subsequently, however, both lexemes are spelt in the modern way, unmarked in respect of vowel length.
In 1670 and in the later editions (as in PDE), the patterns <VCe> and <VVCe> are mostly reserved for the lexemes in which <V> corresponds to a diphthong, e.g.knife, rise, thine, make, name, same, as well as those ending in /z/ or /v/ (e.g., leave, cheese, these).
The spelling of particular long vowels as well as particular lexemes (and groups of lexemes) seems to have become regularized by different dates.Sometimes, the house styles of some printers (e.g., Denham's consistent spelling of do as doo) and the avoidance of homography (e.g., the spelling of the function words the and thee) may have played a role.
Despite a few minor differences concerning the exact time of the adoption of particular spelling patterns, the findings reported in this section generally confirm those formulated in Rutkowska (2013: 141-161), but thanks to the time span extending beyond the 1650s, the present corpus provides also information on further developments in the English orthographic system.

Theoreticians' opinions on the vowel length indication
The convention of doubling the vocalic to mark a long vowel goes back to the fourteenth century (Mossé, 1952: 12;Fisiak, 2004: 19).Also, the final <e> was a usual sight in medieval manuscripts.However, although the latter method is very common in early printed books, the former was apparently neglected by the early printers since it is rare in books published in the first half of the sixteenth century.The first surviving explicit recommendations of the use of these two methods for marking vowel length seem to be John Rastell's remarks published in c. 1530 (see Salmon, 1989: 297).79Rastell, a lawyer and printer, recommends the spellings flee, knee, see, bleed, breed, seek, speed, spleen, week and ween,80 in the words where the vowel 79 Brengelman (1980: 347) credits Levins, Mulcaster, and Coote with the first prescriptions of the use of final <e> to mark vowel length, and Scragg (1974: 61, 79) attributes it to Mulcaster.80 All these and the following examples occur in the facsimile of two pages from Rastell's book made available in Salmon (1989: 300-301).
Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM originates in ME /eː/ (both open and closed syllables), but has also bee beside be 'be', wee beside we 'we', so he does not consistently distinguish orthographically between content and function words.81The revival of the <eeC> pattern and the support for the employment <ee> in open syllables (which is already a variant covering one fourth of tokens in the KS editions of 1518 and 1528) could have had some impact on printers, but this is difficult to assess on the basis of the corpora taken into consideration in this study because the first editions examined here, following Rastell's publication, come from as late as 1556 and 1557.In fact, Huloet (1552) still uses predominantly the old <eCe> and <oCe> patterns, which means that the revived rule of vocalic doubling still might not have been very widespread in the early 1550s.Yet, the second edition of this dictionary, Huloet and Higgins (1572), has <eeCe> in green, keep, meek, meet, and heed, and <ooCe> in book, cook, look, and soon, but <ooCe> is recorded in both editions of this dictionary in moon, noon, and root, which suggests that the cumulative spelling method started to be used earlier in the group of words containing the reflexes of ME /oː/ than in those with ME /eː/.This correlates with the findings in The Schoole and the KS, but the process seems to be more advanced in both Huloet (1552) and Huloet and Higgins (1572).So does the spread of <ee> in open-syllable content words, for instance in degree, knee, three, and tree (free and see start to follow this pattern in the second edition).Such spellings are common in normative writings from the second half of the sixteenth century.The doubling of the vocalic is also occasionally recorded in function words (bee, hee, wee) in Hart (1569Hart ( , 1570)), Laneham (1575), Bullokar (1580b), Coote (1596), Evans (1621), and regular in Hodges (1644Hodges ( , 1649Hodges ( , and 1653) ) and Wharton (1654).Huloet and Higgins (1572) has wee and yee as headwords (also numerous instances of hee, mee, shee, and thee in definitions and examples).The form thee, which follows the spelling tendencies of content words, is recorded already in Huloet (1552).Later, Coles (1674: 15) lists both single and double <e> variants of be and me, but provides only the modern spelling for we.Nonetheless, although the spelling of <ee> in function words is quite widespread in normative writings, apart from thee (temporarily also hee, wee, and yee), these do not catch much ground in the editions of The Schoole and of the KS.Dobson (1957, I: 44) reports that, in the mid-sixteenth century, John Cheke prescribes the use of double vocalics to indicate vowel length but employs both this and the cumulative <VVCe> method in his writings.In contrast, his contemporary Thomas Smith proscribes the use of silent letter and recommends diacritics (Dobson, 1957, I: 51).Surprisingly, although in his earlier unpublished manuscript (1551) he claims that the final <e> should be abolished (Dobson, 1957, I: 63, 69), a few years later Hart (1569: 33r) assures his readers that it should be preserved, and seems to favour the <VVCe> in his own books (e.g., keepe, neede, schoole, soone).In the group of words with the reflexes of ME /eː/ and /oː/, the cumulative patterns, <eeCe> and 81 Admittedly, he employs mostly be and we, and only single instances of bee and wee.
As regards the words containing the reflexes of ME /ɛː/, Rastell (c.1530) prescribes the spellings mede 'meadow', rede 'read', fle 'flay, skin', se 'sea' and never employs <ea>, so he cannot be held responsible for the high increase of spellings containing this digraph in the mid-sixteenth century.According to Scragg (1974: 67), the rise in the popularity of <ea>, a convention inherited from Anglo-Norman scribes, was due to a transfer of scriveners' orthographic practices to printing houses during the sixteenth century.However, it is worth emphasizing that great and greate are the variants found already in the early sixteenth-century editions of the KS; for instance, great covers 41%, and greate 11%, of the tokens in 1506, and great covers 96% in 1518.The extension of these patterns to other lexemes could be then interpreted as part of lexical diffusion (see also Taavitsainen, 2000: 144;Nevalainen, 2012: 146;Rutkowska, 2013: 167).This diffusion is likely to have been enhanced by the presence of relevant spellings in works of reference, such as dictionaries.For example, in Huloet (1552) as well as in Huloet and Higgins (1572), <eaCe>, the dominant spelling of words containing ME /ɛː/ from the mid-sixteenth century until the early seventeenth century in the KS and The Schoole, is already regular in eat, heat, meat, speak, and teach, which could have supported the spread of the pattern in the books printed in the 1580s.This cumulative pattern is also used by Hart (1569) in mean, read, speak, and weak (but not in each, spelt eche), by Puttenham (1589Puttenham ( /1968) in break, eat, heal, heat, and meat, but not in great (spelt great), teach (teach) and each (each, eche, and ech), and by Cawdrey (1604, 1609, 1613, and 1617) in heat, meat, and break.
Although Mulcaster (1582) appreciates the old <eCe> pattern for marking the vowel originating in ME /eː/, he is simultaneously the first authority to systematically promote the modern spelling patterns <eaC> and <ooC> for lexemes such as, for example, break, deal, eat, mean, meat, read, sea, speak, teach (but not each, spelt mostly ech), book, cook, doom, moon, smooth and soon.Nonetheless, in later normative writings, the variants with <eaCe> and <eaC> still often concide, and <ooCe> predominates in, for instance, Clement (1587), Puttenham (1589/1968), Coote (1596), and Gil (1619).Eventually, by the 1640s, the consistent modern distribution 82 The lexemes blood and flood have more spelling variants but they are irrelevant here, because they were not recorded in The Schoole editions (for more information see Rutkowska, 2013: 154-156, 161, 165-166).
Among the lexemes subject to orthographic experiments aiming at vowel length indication, do merits a separate comment due to its frequency and the range of variants.In the sixteenth-century editions of The Schoole and the KS, until 1580, it is almost uniformly spelt in the modern way.Then, in the 1580s and 1590s, the variant doo is the major or even the only spelling of do (Denham's 1582 edition of The Schoole).As regards normative writings, doo is recorded (in variation with do and doe) a few years earlier, in Huloet and Higgins (1572), and it is the only spelling of do in Laneham (1575).Puttenham (1589Puttenham ( /1968) favors doo, but employs also doe, and occasionally do.On the whole, doo is not a widespread variant.Though he also publishes his work in in the 1580s, Mulcaster (1582) uses do consistently (without a single instance of doo or doe).Additionally, Baret (1574) employs a rather exotic variant dooe with vocalic doubling combined with the final <e>, but his usage does not seem to have had any impact on other writers.In the early seventeenth century, do regains ground in the editions examined here, only to cede it to doe in the 1620s and 1630s.This variant was already regular in Hart (1569 and1570) and Bullokar (1580b), though they occasionally resort to do.Clement (1587) uses both do and doe, without clear preference.In Coote (1596), there are slightly more tokens of do than of doe.Yet, Coote's original orthography is modified by the publisher of later editions, and in the 1655 edition of this handbook only do is used.As regards lexicographers, Cawdrey (1604) has doe, doo, and do but retains just doe in the remaining editions of his dictionary (1609, 1613, and 1617); Cockeram uses doe between 1623 and 1639 but switches to do in 1655.Eventually, between 1656 and 1670, the modern spelling starts to dominate again in the KS and The Schoole, respectively.
The spelling of the words containing the reflexes of ME /ɔː/ is not uniform across lexemes (compare, for instance, coat and bone), so it is not surprising that even in the last editions of the books examined in this study, there is some diversity.As for the views of theoreticians and lexicographers, Huloet (1552) as well as Huloet and Higgins (1572) have othe, smoke, stroke, yoke.They have spelling variation in some other lexemes (e.g., abrode ~ abroade, bote ~ boate, brode ~ broade, throate ~ throate).Mulcaster (1582: 118) treats the digraph <oa> as superfluous, because to him "the qualifying e" sufficiently indicates the length of the preceding vowel (e.g., in bone, bote, cote, and mone).Nonetheless, it is recorded in Clement (1587, e.g., in abroade, throate) and Puttenham (1589/1968, e.g., in abroad, broad, boast, andsmoake).Coote (1596) recommends such spellings (e.g., abroad, broad, boat, cloathes, coate, goat and oath).Also the seventeenth-century orthoepists, lexicographers and grammarians use <oa> regularly (e.g., Cawdrey, 1604Cawdrey, -1617;;Gil, 1619;Evans, 1621;Cockeram, 1623Cockeram, , 1626Cockeram, , 1639Cockeram, , 1655;;Hodges, 1644;and Wharton, 1654), mainly in the lexemes abroad, boat, broad, coat, goat, load, and throat.Cockeram (1623) has <oa> also in smoke, yoke (used as a verb).His spellings smoake and yoake, recorded in the editions between 1623 and 1639, are changed to smoak ~ smoke and yoke in 1655, and smoke and yoke in 1670.Variation continues also in later authors; for example, Coles (1674) lists the spellings smoak and stroak, but yoke, whereas Ellis (1680) has smoak, but stroke, yoke.Some of the above mentioned authorities could have influenced the printers of the editions under consideration as regards the increase of spellings with <oa>, but the level of variation in this set of lexemes among the theoreticians precludes any evidence of strong and concerted influence on the preferences for particular forms among contemporary printers.

Conclusions
Several conclusions can be drawn on the basis of the results of the present study.First of all, a few stages can be identified in the process of orthographic regularization in the corpus under analysis.The most spectacular changes are visible between the 1557 edition and the 1582 one, both in the distribution of the vocalics <y> and <i> and in the indication of vowel length, consisting in the replacement of (the remnants of) the old system <VCe> pattern with the new, <VVCe> one.Two other important dates are 1640 and 1670.The former corresponds to the establishment of the graphemic distributional principles without exception, and the latter to the full installation of most modern vowel length indication patterns, at least in The Schoole editions.Salmon (1999: 32) refers to the year 1660 as "marked by the establishment of an orthography which was, in most respects, that of the twentieth century".The findings of the present study suggest that this boundary should perhaps be moved to the year 1670.One can argue, obviously, that the relatively small size of the corpus limits the possibility of generalizing the results, but in fact the findings confirm the main trends identified by Rutkowska (2013) for the KS, a much larger corpus, despite some time differences concerning the settlement of particular features.
As concerns the analysis of particular graphemes, the difference in typeface as well as the difference between lower-case and upper-case letters should be taken into consideration in assessing the findings concerning graphemic usage in early modern books printed in English.Also, the distribution of the graphemes <v> and <u> as well as <i> and <j> in the corpus under examination casts some doubt on the dating of The Schoole editions allegedly published in 1626, 1630, and 1635.My findings suggest that the 1626 edition could have been printed after 1630 (and possibly also after 1635).Thus, the analysis of orthographic features can affect the dating of particular documents in cases where the date has been reconstructed, and not explicitly stated in the colophon.
With regard to vowel length indication, the methods (adding the final <e>, vocalic doubling, and both) and speed of the regularization process depend on individual variables (i.e., the groups of lexemes to which a given word belongs) and the division into these groups is connected with the ME vowel in the stem of a word.With hindsight, it seems that at least some spelling patterns established in the early modern period are used to avoid homography, so they observe the heterographic, rather than phonemic (or phonological), principle.83For example, meet, containing a reflex of ME /eː/ is spelt differently from meat, containing a vowel derived from ME /ɛː/.However, the fact that printers spell meat, sea and meet, see, differently, and do not use any orthographic distinction between the words such as break, great and meat, sea, on the one hand, and theoreticians do not (yet) suggest the this should be changed, implies that neither the diphthongization (of the reflex of ME /ɛː/) nor the merger (of ME /eː/ and ME /iː/) had been completed by that time.84Consequently, these spellings are likely to have coincided with the phonemic principle at the time when they were established -they ensured the orthographic differentiation between mid-close and mid-open vowels.
The "house styles" of particular printing houses are difficult to identify on the basis of The Schoole corpus but, at least in one edition, this turns out to be possible.Henry Denham, the printer of the 1582 version of The Schoole, is surprisingly consistent with regard to the spelling of do (always doo) and the use of word-medial vocalic <i>.The combination of these two features cannot be easily correlated with the recommendations or usage of any particular language authority.In turn, the spread of the digraph <oa> from abroad to other lexemes in the 1621 and subsequent editions of The Schoole could have been triggered or at least supported by the lexicographers and other authors of publications on language usage.
On the whole, some more or less exact correlations between the recommendations of theoreticians and the practice of the printers can be made, but it is difficult to prove a cause-and-effect relationship between the two.For example, a few years after Butler (1633) prescribed the modern distribution of <v> and <u>, this principle started to be generally observed, but the change was not sudden, and rather clear signs of it can be identified already in the 1626 edition of The Schoole.
It is also debatable to what extent the new spelling patterns and usages were the ideas of theoreticians and to what extent they were a product of printers' house styles, or else still, of the negotiations between theoreticians and printers during the publication process.Hence, the question of the extent to which theoreticians, on the one hand, and printers, on the other, have impacted on the regularization of Early Modern English spelling is likely to remain a chicken-and-egg problem for quite a long time.Definitely, more studies based on other sets of editions of early printed books correlated with investigations into language authorities' views are needed to refine our overall knowledge concerning the stages and speed of the orthographic regularization and standardization in English.

Table 16 .
The analyzed editions of The Schoole of Vertue

Table 17 and
Figure

1557 1582 1593 1621 1626 1630 1635 1640 1660 1670 1677 1687
Words with /ɛː/ or /eː/ < ME /ɛː/ in closed syllables: tendencies of vowel length indicationIn the KS, the relevant lexemes taken into consideration include each (adj and pron), eat (v), heal (v), heat (n), meat (n), speak (v), and teach (v).The first three editions contain nearly exclusively the forms with the <eCe> pattern, with <eeC> as a minor pattern in 1518 (4%) and 1528 (10.5%) in heat and meat.In 1556, <eCe> and <eaCe> are equally frequent; thus, compared to 1557 in The Schoole, the change to the new pattern seems less advanced in this corpus.The <eaCe> spellings dominate in the editions published in the years 1570-1585, staying at the level of 57-59%, which is roughly comparable to 65% found in 1582 in The Schoole.In the same period, approximately 5% of the forms in the KS are spelt <eCe>.In 1600 and 1604, the shares of <eaCe> and Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM

Table 20 .
Words with /uː/ < ME /oː/ in closed syllables: distribution of vowel length indication Brought to you by | Uniwersytet im.Adama Mickiewicza Poznan -Adam Mickiewicz University Authenticated Download Date | 10/11/16 11:30 AM pattern prevails in the first three editions, but it gradually decreases in importance throughout the century, from 79% in 1557 to 1.4% in 1600; in 1580 it still accounts for 23.5% of the tokens and in 1585 for 15%, so the lack of this pattern already in the second edition of The Schoole is surprising.

Table 21 .
Spelling patterns in do: distribution