From sicker to sure: the contact-induced lexical layering within the Medieval English adjectives of certainty

Abstract

The major Old English adjective of certainty was (ge)wiss, which in early Middle English came to be replaced with sicker derived from very weakly attested Old English sicor, a word of ultimate Romance origin (from Latin sēcūrus). The relative paucity of occurrences of both adjectives in the Dictionary of Old English corpus is attributed to their use in mostly spoken language. The rapid increase in the usage of sicker in the thirteenth century is a mystery with possible, yet difficult to prove, Norse and/or Anglo-Norman influence. The fourteenth century marks the appearance of sure and certain borrowed from Anglo-Norman first by bilingual speakers and writers, and the quick diffusion of the new lexemes to all dialects and genres. This article looks at the adoption of the different senses of these polysemous adjectives into Middle English in the context of subjectification, which appears to affect not only semantic developments within one language but also the process of borrowing. When sure and certain were used epistemically, they tended to occur in the predicative position, usually following the copula. It took several centuries of lexical layering (coexistence of synonyms) before sicker was lost from Standard English in the sixteenth century.

1 Introduction

A significant change affected adjectives of certainty in Middle English when the language lost the earlier Germanic adjective (i)wis<gewis(s) and the Romance sicker<sicor in favour of French loanwords sure and certain. The adjectival (i)wis became obsolete in early Middle English and so did sicker in the sixteenth century. As is usual in such cases, the changes were not abrupt and involved several centuries of lexical layering. Adopting an etymological and historical approach in this article, I will trace the replacement processes which occurred due to borrowing and the subsequent diffusion of new lexemes, viewing it in the context of typology of lexical borrowings (e.g. Fischer 2001; Timofeeva 2017), widespread bilingualism in medieval England after the Norman Conquest (e.g. Schendl & Wright 2011; Ingham 2012) and contact-induced semantic change (e.g. Miller 2012; Lutz 2013; Durkin 2014; Lim & Ansaldo 2016). Borrowing is understood here according to Van Coetsem's (1988: 3) definition as ‘the transfer of material . . . from the source language to the recipient language’. A similar description can be found in Thomason & Kaufman (1988: 37): ‘the incorporation of foreign features into a group's native language by speakers of that language’, which I accept with Winford's (2005: 382) reservation that ‘the term “dominant” or “primary” language seems more suitable than “native” language’, especially in the context of bilingual speakers in medieval England.

In the literature the term layering is usually discussed in the context of grammaticalization:

within a broad functional domain, new layers are continually emerging. As this happens, the older layers are not necessarily discarded, but may remain to coexist with and interact with the newer layers. (Hopper 1991: 22)

Following Traugott (2008), Arista (2011, 2014) and Brems (2012), who use the concept when discussing both syntactic and lexical phenomena, I believe that the idea of layering can be extended to the lexical component given that languages can have and develop diverse expressions for the same function. Such multiple synonymy can be tolerated for centuries.

I will also take into account the issue of borrowing different senses of polysemous words from Anglo-Norman into English at different times with reference to the lexicographical databases. I will show that concrete senses tend to be borrowed first, before the subjective ones that convey the speaker's attitude, which agrees with the process of subjectification. The idea of subjectification, also known as subjectivisation, was introduced into linguistic studies by Ronald Langacker and Elizabeth Traugott in the late twentieth century. While Langacker (1990, 1999) discussed it with reference to synchronic semantic extension, Traugott's (1989, 1995, 2010) studies concentrated on diachronic semantic change. She described subjectification as the direction of the semantic extension from a more concrete towards a more abstract sense, from a meaning ‘based in the sociophysical world’ to a meaning ‘based in the speaker's mental attitude’ (Traugott 1989: 46).

When discussing the diachrony of adjectives of certainty one has to be aware of the obvious limitations connected with the absence of spoken corpora of Medieval English and the availability of only some indirect data in the form of dialogues, first-person narratives and early letters. Epistemic adjectives are characteristic of spoken discourse and are often used as pragmatic markers, so the borrowing processes discussed here occurred mostly owing to oral contact and intense bilingualism after the Norman Conquest. The subsequent language shift from Middle English and Anglo-Norman bilingualism to monolingual use of English in the fourteenth century resulted in thousands of French loanwords being adopted into English, and among them were the new adjectives of certainty sure and certain.

The illustrative language material comes from the Dictionary of Old English Corpus (DOEC) and the Middle English Compendium (MEC). The short titles of the sources follow the conventions used by the compilers of the dictionaries related to these databases: the Dictionary of Old English (DOE) and the Middle English Dictionary (MED). Some additional data come from the Oxford English Dictionary (OED). The examples from the other early Germanic languages have been taken from Gordon (1957) and Mettke (1970).

2 Native Germanic adjectives of certainty

2.1 Old English (ge)wis(s) > Middle English (i)wis

The prevailing Old English adjective of certainty was (ge)wiss, inherited from the Proto-Germanic *ga-wissaz, the past participle of the verb witan ‘to know’. Its cognates are found in all the (early) Germanic languages, e.g. Old High German in example (1) and Old Norse in (2). Modern German continues to use the adjective gewiss meaning ‘certain’. In Old English, also within the same texts, the adjective could occur with or without the prefix ge- (examples (3) and (4), with an interesting manuscript variation in (3)), though the prefixed form appears to have been more common especially in earlier Old English texts as the DOEC data indicate. The derived adverbs were (ge)wisse and (ge)wislice, e.g. in (5), only the former surviving into late Middle English, e.g. (10) and (11). The adjective (ge)wiss was invariably used in the predicative position. A Thesaurus of Old English (Roberts, Kay & Grundy 2000) also lists cuðlic and witod as synonyms of (ge)wiss, but the former has only a mere seven occurrences in the attributive premodifying position in the whole corpus (e.g. example (6)) in various senses, including ‘intimate, friendly’ and only this sense survived into Middle English (see the MED, s.v. couthlic). Witod is predominantly used in the context of certainty determined by fate and/or Providence, as in (7), and its adverbial derivative witodlice ‘certainly, truly, indeed’ is a sentential adverb commonly found in the biblical translations and other ecclesiastical texts. Neither of the words survived as adjectives of certainty beyond Old English.

1. (1) Uns allen thaz giwis ist thaz thu selbo Krist bist. (Old High German Otfrid von Weissenburg Evangelienbuch III 12,25)

2. ‘To us all that is known/certain that you yourself are Christ.’

3. (2) Því at víss er dauðinn ef þú bíðr Gláms. (Old Norse Grettis saga 35/28)

4. ‘For you are sure to be dead if you wait for Glam.’

5. (3) 7 þeh ðe he gewiss geworden wære [Corp. Christi Oxf. MS 279B: þeah he wis geworden wære] þurh þa ætywnesse þære gesyhðe, nohte þon læs he his fore gearwade mid ðam gemyngadam broþrum. (Bede 9.412.5)

6. ‘And though he became assured by the appearance of the vision, nevertheless he prepared for his journey with the above-mentioned brethren.’

7. (4) and him ða for an ðuhte ðæt he þæs gewiss wære, ðæt he ðæs on æfen ælcne man gecneowe and ælc gecnawe hine. (LS 34 (SevenSleepers) 626)

8. ‘The only thing he seemed certain of was that later in the evening he knew everyone and that everyone knew him.’

9. (5) se ceorl sæde þæt Swyðun hine gehælde, forðan þe he sylf wiste gewissost be ðam. (ÆLS (Swithun) 107)

10. ‘The man said that Swithun healed him because he himself most certainly knew about it.’

11. (6) þa wæs se bysceop mycle þig reðran on godum weorcum þe he ymbe þa cuðlican mede gehyrde. (Mart 2.1 Ju 29, B.24)

12. ‘Then the bishop became more zealous on good deeds when he heard about the certain reward.’

13. (7) þonne bið us seo med æt Drihtene witod. (ThCap 1 21.331.99)

14. ‘Then the reward from God will be certain for us.’

Old Norse typically used the cognate adjective of certainty víss (as in (2) above), which may have affected the usage of (i)wiss in the Ormulum. Nonetheless, the adjective (i)wis(s) became obsolete in English by the early thirteenth century. The MED (s.v. iwis) gives only one doubtful quotation of prefixed adjectival iwis from a mid-twelfth-century herbarium, which was a copy of an Old English text, and several examples of unprefixed wis from the Ormulum, as in e.g. (8), the only attestations of adjectival wis arguably meaning ‘certain’ in early Middle English. The spelling of wis with a single ‘s’ in the Ormulum suggests that the root vowel was long according to the spelling system devised by Orm (see Heselwood 2013: 102). This may have brought about some confusion with another common adjective expressing the capacity of mind wīse ‘wise’, which in turn may have led to the disappearance of the adjective (i)wis ‘sure, certain’ in early Middle English and its being replaced with sicker. The development was the result of the natural tendency to avoid homonymy in similar contexts so that different meanings might be signalled clearly and unambiguously (see Hock & Joseph 2009: 214).

The good evidence of such ambiguity of early Middle English wis is the fact that the MED lists example (8) both in the entry wis (adj.) and wīs(e) (adj.). I believe that the latter interpretation makes more sense. If it were the continuation of Old English gewiss, Orm would have doubled the final -s, as he did in the 7 instances of iwiss and 58 of wiss in the whole poem, which all appear to be adverbs (from Old English gewisse, as in (5) above). And i-wis(s) survived as a quite common adverb of certainty throughout Middle English (examples (10) and (11)). Since this usage very much resembles the occurrence of the adverb víst ‘for certain’ in Old Norse (still found in modern Swedish and Danish vis(s)t), as in (9), the Scandinavian influence is very likely here, so in this case the foreign factor may have contributed to the preservation of the native item in a bilingual environment. However, a significant difference between Norse and English is the unusually long preservation of the prefix i- until (very) late Middle English, which made the word distinct enough from the adjective wis ‘wise’. Probably the initial i- was no longer perceived as a prefix in this adverb at the time when the reflex of Old English ge- had already disappeared in other English words, in particular in past participles. The MED provides few examples of the unprefixed adverb wis/wisse in contrast to numerous instances of iwis although, as observed above, the tendency in the early Middle English Ormulum was the reverse.

1. (8) & forrþi wollde ȝho ben wis Off þatt þurrh Godess enngell. (?c.1200 Orm. (Jun 1) 2279)

2. ‘And therefore she would be made certain/informed/aware of that by God's angel.’

3. (9) Eigi veit ek þat víst. (Brennu-Njáls saga 94/177)

4. ‘I don't know it for certain.’

5. (10) Iwis ȝe beod Ænglisce englen ilicchest. (c.1275(?a.1200) Lay. Brut (Clg A.9) 29481)

6. ‘You English are certainly most like angels.’

7. (11) Cossyne, and ȝe had desyred me..I wold have sene yow offtener. I wesse, cossyne, het greveth me &c. (c.1465 Stonor 1.70)

8. ‘Cousin, if you had desired me, I would have seen you oftener. Certainly, cousin, it grieves me, etc.’

The antonym of (ge)wiss formed by adding the Germanic negative prefix un- is well attested throughout the Old English period, but only with the prefixed variant ungewiss. No examples of *unwiss are found in the DOE corpus. Together with (ge)wiss ungewiss was lost by the mid twelfth century:

1. (12) Ungewiss com se deofol to Criste, and ungewiss he eode aweig; forðan þe se Hælend ne geswutulode na him his mihte, ac oferdraf hine geðyldelice mid halgum gewritu. (ÆCHom I, 11 271.150)

2. ‘Uncertain came the devil to Christ and uncertain he went away because the Saviour manifested not his power to him, but overcame him patiently with the holy scriptures.’

2.2 Old English sicor > Middle English sicker

Middle English sicker was the continuation of another Old English adjective of certainty, sicor, which was ultimately of Latin origin. It had been borrowed into West Germanic from Latin sēcūrus (from sē- ‘without + -cūrus, from cūra ‘care’, thus ‘without care’, also ‘safe, secure’; cf. Partridge 1983: 135; Durkin 2014: 114) at an early preliterate stage, definitely before the eighth-century Old High German Consonantal Shift, as evidenced by German sicher (<OHG sihhur) – compare Old Saxon (13), which preserves the original plosive consonant, with Old High German (14), where the plosive was replaced with the homorganic fricative (see Schwerdt 2000). The word is not attested in either Gothic or Old Norse and its modern Danish and Swedish cognates are attributed to direct borrowing from German (OED, s.v. sicker). Classical Latin securus appears to have undergone subjectification in either Vulgar Latin or early Romance, as can be seen in the subjective senses of the reflexes of securus expressing the speaker's attitude of certainty in many Romance languages, e.g. Italian sono sicuro, Spanish estoy seguro, Romanian sunt sigur, which all mean ‘I am sure, certain, confident’. This development explains the presence of such epistemic senses in early Germanic.

1. (13) Ni uuas iro so sikur enig. that he bi themu uuorde. themu uuibe gedorsti. sten anuuerpen. (c.850 Heliand 3875)

2. ‘He was not so sure that they, by this word, would dare throw a stone at that woman.’

3. (14) sichor múgun sin wir thés. (Otfrids Evangelienbuch 74)

4. ‘we can be certain of this.’

5. (15) ðeah we næbre eft swa ne don, gif we ðæt gedone mid nanum ðingum ne betað ne ne hreowsiað, ne bio we no ðæs sicore gif us ðæt ne mislicað ðæt us ær licode, ðonne ne bið hit no us færgiefen. (CP 54.425.3)

6. ‘Even if we never do so again, unless we somehow atone for and repent of what we have done, we may not be sure of that, unless we are displeased with what pleased us before, that it will be forgiven us.’ (Latin: ita et cum Deo delinquimus, nequaquam satisfacimus, si ab iniquitate cessamus, nisi voluptates quoque, quas dileximus, e contrario appositis lamentis insequamur.)

In Old English, however, sicor was a very rare word (cf. OED, s.v. sicker), indeed a hapax legomenon, and it was not until early Middle English that siker became the most common adjective of certainty after the demise of the adjective iwis in the twelfth century. The example from King Alfred's translation of the Pastoral Care (in (15)) is the only attestation of the word in the predicative position (following the copula) in the sense ‘sure, certain’ in the whole Old English corpus and, interestingly, it was used independently of the original Latin Cura Pastoralis, which does not even have the word securi in the sentence or any other explicit marker of certainty. In the tenth-century translation of St Benedict's Rule attributed to Æthelwold of Winchester the word in the form sicar is used in a different sense of ‘spiritually safe, secure, free from sin’, rendering Latin liber (example (16); cf. MED, s.v. siker 1c). Another instance of sicor is to be found in the prose from the transitional period between Old and Middle English (example (17)) known as Alcuin's De Virtutibus et Vitiis, included in both the DOEC and the Vespasian Homilies, usually classified as a collection of very early Middle English texts believed to have been written in Old English (see Morris 1868; Warner 1917). Unlike in the examples (15) and (16) sicore is used here in the attributive prenominal position. In the same collection we also find the derived noun sicornyss ‘certainty’ (example (18)), which shows that the root sicor- underwent regular native word formation processes in Old English.

1. (16) Heo eft on þam dome sicar bið & sacles. (BenRW 2.15.18)

2. ‘She will again be safe and innocent in the judgment.’

3. (17) Swyðe sicore forgyfonysse se mæig him biddan æt Gode, se þu nu wyle forgyfen þan þe wið hine agylteð. (Alc(Warn35) 146 a1150(OE) Vsp.D.Hom. (Vsp D.14) 95/31)

4. ‘Such a certain forgiveness one may pray from God, he who will forgive those who trespass against him.’

5. (18) Gode mænn..habbeð mycele mede, for heo habbeð blisse for þære sicornysse Godes rice. (a.1150 (c.1125) Vsp.D.Hom.Elucid. (Vsp D.14) 143/17)

6. ‘Good people have a great reward, as they have the bliss of certainty of God's kingdom.’

On the subject of the life cycle of words, Fischer (2001) remarks that what matters in vocabulary diffusion is not only attestations of words, but also their intensity. In the case of sicor the intensity of use of the borrowed item is very low according to the data from the DOEC, which comprises all the Old English texts that survived. On the other hand, the fact that somebody used the word, albeit once, with proper Old English phonology and morphology suggests that sicor was a part of the English lexicon at the time. Otherwise the text would simply not have been understood. I suspect that one of the reasons why sicor is so poorly attested is that the word might have been used mostly in (colloquial) spoken Old English, as is usually the case with adverbs of certainty – cf. Kärkkäinen (2003) and especially Biber (1994: 186), who finds that in Present-day English the ratio of the occurrences of sure per one million words oscillates between 234 in the written Longman–Lancaster corpus and 427 in the spoken London–Lund corpus.

An additional argument for the presence of sicor in (spoken) Old English is the fact that in Middle English (including early texts) we find numerous instances of si(c)ker, e.g. 443 matches of the spelling siker in the Corpus of Middle English Prose and Verse (CMEPV). Although no cognate word is recorded in the preserved Old Norse texts, this does not exclude the possibility of a cognate of siker occurring in spoken Old Norse¹ and then promoting its use in English in bilingual environment, as in example (20) from the Ormulum, whose author has a Scandinavian name. Besides, we should bear in mind that most of the Old Norse data are several centuries later than Old English and that we have rather scarce evidence for northern dialects of early Middle English where the Scandinavian influence was most pervasive. In the preserved Middle English texts sicker is found in several senses, from objective ‘secure, safe, free from danger’ through ‘strong’ to subjective ‘sure, certain’ (examples (19)–(22)), with the majority of the earliest thirteenth-century examples of the concrete, objective use of the word. This semantic development provides the corroboration of Traugott's (1989, 1995, 2010) idea of subjectification defined in the Introduction.

The word was sometimes spelt secure in the fourteenth century (especially in the Fairfax MS of the Cursor Mundi, e.g. in (23) below, where another manuscript has the form siker), although the new Latinate secure is not attested in English until the 1530s according to the OED (example (27)). Sicker also developed an antonym by adding the negative prefix un-, as can be seen in (21). Alongside sikerli and sikerlice, siker itself could also be used in the function of an adverb of certainty, as in (24). In the fourteenth century sicker was still quite common, but in late Middle English its frequency decreased considerably in favour of new Romance words sure and certain (see sections 3.1 and 3.2 below). This long co-occurrence of older and newer adjectives of certainty supports the idea of lexical layering advocated by Hopper (1991), Traugott (2008), Arista (2011, 2014) and Brems (2012), which is discussed in the Introduction.

1. (19) Men weneð bon siker þurh walle and þurh diche. (a.1225 (?c.1175) PMor.(Lamb 487) 41)

2. ‘People think that they are safe thanks to the wall and to the ditch.’

3. (20) Beo þu sikerr þat he shall Þe ȝifenn eche blisse. (?c.1200 Orm. (Jun 1) 4844)

4. ‘Be certain that he shall give you all bliss.’

5. (21) Eahte þinges nomeliche leaðieð us to wakien eauer. . .Vre sunnen þe beoð se monie, deað þet we beoð siker of & unsiker hwenne. (c.1230(?a.1200) Ancr.(Corp-C 402) 76/6)

6. ‘Namely eight things that lead us to keep a constant vigil. . .our sins which are so numerous, death that we are certain of but uncertain when (it will come).’

7. (22) Zykere hy byeþ of zuyche lyue. (c.1350 Ayenb.App.(Arun 57) 269/9)

8. ‘They are certain of such life.’

9. (23) bot quam so god helpis wiþ-alle. ful secure may he wende oueralle. (a.1400 Cursor (Frf 14) 4008) [Cotton Vesp. MS: Ful siker mai he wend. . . and Trinity MS: May sauely go.–.–.]

10. ‘But the one whom God helps fully can go very safely everywhere.’

11. (24) Siker þu ert myd him a galilewis mon. (a.1300 I-hereþ nv one (Jes-O 29) 286)

12. ‘Certainly you [Peter] are with the man of Galilee.’

The final elimination of the word sicker in the fifteenth–sixteenth century might have been brought about by the reborrowing of Latinate secure at the time, especially since one of the senses of sicker ‘free from danger, (objectively) safe’ was exactly the same as the primary meaning of secure. The adjectives had similar sound and similar meaning, which ran counter to the tendency to avoid homonymy in similar contexts (Hock & Joseph 2009: 214), so one of the words had to disappear. Example (27) is the earliest occurrence of secure in the OED in the sense of ‘without care’, but, as observed above, spellings <secure> for sicker are found in English two centuries before. At the same time secure is used in the sense of ‘sure, certain’ until the eighteenth century, which in fact might be late occurrences of sicker in disguise, e.g. in (28). Interestingly, the noun security is attested earlier than the adjective secure in both Anglo-Norman and Middle English ((25) and (26)), rendering Latin securitas and synonymous with sickernyss used in the variant manuscripts. Both the Anglo-Norman Dictionary and the Middle English Dictionary have an entry securite/security, but not secure, though the conservative spelling segur (for seur) is occasionally found in some Anglo-Norman texts, e.g. in (30).

1. (25) refere la dite toure e mur pur greienur securité de la dite cité. (Rot Parl 1 i 275) ‘refer the said tower and wall for the greater security of the said city.’

2. (26) As hit is..seide, Paradise..hathe securite [Trev.: sikernesse and suerte; L securitatem] to the whiche seyenge the altitude of the place berrethe testimonye. (?a.1475(?a.1425) Higd.(2) (Hrl 2261) 1.77)

3. ‘As it is said, Paradise is a safe place and its altitude bears testimony to the saying.’

4. (27) But we be secure [= without care] and vncarefull, as though false Prophets could not meddle with vs. (?1533 Latimer Let. to Morice in Foxe A. & M. (1583) 1742/2)

5. (28) I am secure that no man will so readily take them into Protection. (1713 Johnson Guardian No. 4 _6)

As can be seen in the OED entry sicker, the word was lost in English after the sixteenth century except for in the northern British dialects (see also the entry in dictionaries of Scots: The Dictionary of the Older Scottish Tongue (DOST) and The Dictionary of the Scots Language (DSL)).

3 New Romance words

3.1 Anglo-Norman and Middle English seur/sure

In the late thirteenth century new adjectives of certainty borrowed directly from Anglo-Norman appeared in Middle English: sure and certain. The former, first recorded in French in Chanson de Roland (c.1080), is the regular phonetic development of Latin sēcūrus (*sekur>segur>seur>sûr) and is lavishly attested in many orthographic variants in both continental Old and Middle French and Anglo-Norman:

1. (29) Aucun espiritel homme dont vus (var. vus estes) seures e certeines. (Ancren2 180.22)

2. ‘no spiritual human of whom you are sure and certain.’

3. (30) Molt soi tient segur e sein Qui met la serpent en son sein. (Dial Greg 70rb)

4. ‘He feels very confident and healthy who puts a snake on his breast.’

5. (31) Et les aultres xx. li. serront paiez a quell heoure qe nous sumus sure dez ditiz terrez. (c.1380 John Stoke to Edmund de Stonor)

6. ‘And the remaining £20 will be paid at the hour when we are certain of the said land.’

As was the case for many other Medieval French loanwords, the word seur entered Middle English through language contact in a bilingual environment, most likely first among educated speakers and professionals in London where in the early fourteenth century French still had a wide currency and was associated with higher prestige (see Wright 1996; Britnell 2009; Schendl & Wright 2011; Lutz 2013). In this volume both Ingham and Timofeeva convincingly show that the bilingual clergy (priests, friars, preachers) may have been the most important milieu responsible for the diffusion of French loanwords, many of them belonging to the general word-stock of the language, to the monolingual Middle English-speaking population. This is confirmed by our data, as most of the early occurrences of sure and certain cited in the Middle English Dictionary appear in ecclesiastical texts.

The inclusion of the Romance adjectives of certainty into the word-stock of English was obviously nothing unusual at the time when thousands of Anglo-Norman and Old French words were being borrowed. It also supports the view that native-like knowledge of French was a common phenomenon in the higher and professional classes of the fourteenth-century English society (Rothwell 1991, 2001; Trotter 2003; Ingham 2012, 2015; Durkin 2014: 232; Ingham & Marcus 2016). However, with each generation English was becoming a more common and natural tool of communication among original bilinguals, whose French was waning, though French was still the preferred medium in some restricted sociolinguistic domains of the language, e.g. those of law, the Church and the Chancery. In the situation when two languages were spoken side by side for such a long time, a considerable transfer of words from one language to the other was inevitable, especially in the case of formal and/or abstract vocabulary, for which bilinguals often found it difficult to find suitable English equivalents. Instead they preferred the French words that they had been used to, including common phrases such as je sui seur/certein, which became I am sure/certayn in their English. The use of such words and phrases appears to have been quickly adopted by the monolingual English-speaking population. Durkin (this volume) discusses the inclusion of French loanwords into the basic areas of the lexicon in late Middle English.

The major senses of seur listed in the AND are 1. ‘certain, sure’; 2. ‘resolute’; 3. ‘reliable’; 4. ‘safe, secure’; 5. ‘confident, sure’. All of these are also reflected in the more descriptive MED definitions: 1. ‘free from danger, sickness, etc. safe; also, fully recovered’; 2. ‘safe against attack, secure’, as in (32) below; 3. ‘trustworthy, dependable; reliable’, as in (33); 4. ‘assured, confident, convinced [as in (34)–(36)]; having a confident opinion; certain (to do sth.)’; 5. ‘strong, firm, resolute’. Middle English appears to have copied all of the Anglo-Norman senses of seur, as the data and definitions from the AND and MED indicate; cf. Haspelmath (2009: 39), who talks about loan meaning extension as ‘an extremely common (and often unnoticed) process whereby a polysemy pattern of a donor language is copied into the recipient language’. But this is not always the case – cf. Kay & Allan (2016), who argue that meanings are usually borrowed selectively and what is core in a donor language might be only a minor sense in the borrowing language. And indeed, the Anglo-Norman senses appear not to have been taken over wholesale into Middle English at exactly the same time. In the case of sure the earliest English attestations have the concrete objective sense of ‘safe, secure’, and the abstract meaning of ‘certain, confident’ is found at least one generation later, which is in line with the idea of subjectification discussed above. This principle appears to be valid not only in semantic change within one language but also in the case of borrowing processes: the concrete senses of the polysemous items discussed here are borrowed before the abstract ones.

The earliest attestation of sure in English listed in the MED is in the Cotton Caligula manuscript of the Owl and Nightingale copied c.1275 (example (32)), and the same word is used in the later Jesus College manuscript. The composition of the original version of the poem is usually dated c.1200 and the dialect was that of Kent or Surrey, i.e. near London, where the number of bilingual Anglo-Norman and Middle English speakers was the highest (see Dunn & Byrnes 1973: 54). The dating of evidence is always a problem in historical linguistics, as it can be either the date of the copying of the earliest manuscript or the presumed date of the composition of the original text (see Durkin 2014: 228). The occurrence of sure in the Owl and Nightingale, however, appears to be an isolated example before 1300, which is used most likely to provide a rhyme for bure and the sense is ‘secure, safe against attack’ rather than abstract ‘certain’. It is not until the mid fourteenth century that sure becomes common in Middle English texts (examples (33)–(36)) and the first uses in the sense of ‘(subjectively) certain, confident’ are found in texts written in London, e.g. in the Auchinleck manuscript of Degare c.1330 (example (34)). The epistemic uses are particularly common with the first-person subject (examples (34)–(36)), further proving that they were characteristic of spoken discourse. But there are also some relatively early examples listed in the MED of epistemic sure from Leicestershire in the East Midlands and as far north as Cheshire. Although the introduction to the MED does not say explicitly that the illustrative quotations start from the first attested examples, the order of their presentation is always chronological, so it is reasonable to accept the first quotations as the first occurrences, as is also the case in the OED.

1. (32) He hire bileck in one bure Þat hire was boþe stronge & sure. (c.1275 (?a.1216) Owl & N. (Clg A.9) 1082)

2. ‘He locked her up in a castle which was strong and safe for her.’

3. (33) Þer sche him puruaid..an hundred kniȝtes..And fond hem armour and stede, Boþe soure and gode at nede. (c.1330 (?a.1300) Arth.& M. (Auch) 7676)

4. ‘There she provided for him a hundred knights and found armour and war-horses for them, both reliable and good in need.’

5. (34) Ich am al sure, He þat bette þat fure Wil comen hom ȝit to-niȝt. (c.1330 Degare (Auch) 761)

6. ‘I am sure that he who kindled that fire will come home yet tonight.’

7. (35) fforþermore I knowe wel and I am right seur þat he schal noþing do in þis neede withouten my counseil. (c.1390 Chaucer CT Mel. 2953)

8. ‘Furthermore I know well and am very certain that he shall do nothing in this matter without my advice.’

9. (36) and I am sure she wilbe glad iff I tell here of ȝour prosperous welfare, whan I come home. (1481 Stonor Letters 292)

10. ‘And I am sure she will be glad if I tell her about your prosperous welfare when I come home.’

In a trilingual treatise Femina (c.1400), described as an advanced manual for teaching French (Rothwell 2005: ii), there are several instances of sure and certein in the Anglo-Norman version rendered by identical items in Middle English (e.g. (37) and (39)), which is good evidence of ‘bilingual thinking’ on the part of the compiler. I believe that if a person is bilingual and often switches codes, there is a tendency to preserve similar, if not identical, meanings of the items that are borrowed from one language into another (see e.g. Bhatia & Ritchie 2013). However, it may not always be the case, as in the text we also find one instance of sicker for French sure (example (38)).

1. (37) ffaitez un gars luy garder sure Si q'il ne ceppe ne chie ceo Mez gardez en suerté vs Make a boy hym kepe sure So þat he stomble nat ne fall But kepet in suerte. (c.1400 Femina (Trin-C B.14.40) 12.6)

2. ‘Make a boy keep stable so that he may not stumble or fall but retains stability.’

3. (38) Et donque est le measoun tout sure vs And þanne ys þe hous al seker. (75.3) ‘And then the house is all safe.’

4. (39) Mez pur estre en tout certeigne, Vous die de la cowe du mayne vs But for to be in al certayn, ȝow y say of þe wrest of þe hand. (20.4)

5. ‘But in order to be fully certain I will tell you from the wrist.’

The new adjective of certainty sure became a high-frequency word from the mid fourteenth century onward. The search in the Corpus of Middle English Prose and Verse has provided us with more than a thousand matches with various spellings, the most common being sure (888 matches). The diffusion of the new item was very quick, as the word is recorded in 75 out of 139 Middle English corpus texts. All the dialects are represented and the text types range from scientific and theological treatises (e.g. Lanfranck's Science of Cirurgie), legal documents (of the Chancery), historical chronicles (e.g. Higden's Polychronicon) through poetry and drama (e.g. La Belle Dame sans Merci, the York Plays) to family letters (e.g. the Stonors’ correspondence in both French and English) and dialogues in narrative texts (e.g. The Right Plesaunt and Goodly Historie of the Foure Sonnes of Aymon). The greatest number of matches (154) is found in The English Register of Godstow Nunnery, near Oxford written c.1450, which again supports the view that the French loanwords were particularly popular among the clergy.

3.2 Anglo-Norman and Middle English certain

Old and Middle French certein(e) was derived from Vulgar Latin *certanus < Classical Latin certus (originally the metathesized past participle of cernere ‘sift, discern, decide’, thus cretus/certus meant ‘decided, certain’; see Onions 1966: 159) and according to Greimas (2007: 91), it is first attested in Continental French in Le Roman d'Enéas written c.1160. Witness the following Anglo-Norman and Middle French examples:

1. (40) com il fu en certeyn leu a tel certeyn jour. (BRITT i 100)

2. ‘how he was in a certain place on that certain day.’

3. (41) Certain est que quant ilz veullent entrer en Angleterre ilz sont tous a cheval les uns et les autres excepté la ribaudaille qui les suit a pié. (Jean Froissart Chroniques 34.10)

4. ‘Certain is that when they wanted to come to England, they all – the first ones and the others – went on horseback except for the ribaldry who followed them on foot.’

5. (42) De la victoire sunt certain. (Brut3 2773)

6. ‘They are certain of their victory.’

The AND enumerates the following major senses of the adjective certein: 1. certain, sure, assured; certain, acknowledged as true (as in (41)); 2. convinced (as in (42)); 3. steadfast; reliable, trustworthy; genuine; 4. prescribed, specified; authorized, authenticated; 5. fixed, regular; 6. certain, particular (as in (40)); individual, personal. As in the case of seur/sure, almost identical meanings are listed in a more detailed way in the MED entry for certain in a slightly different order: 1. specified, fixed, prescribed (time, place, amount, quantity); determined; a definite but unspecified, as in (43); 2. real, genuine; true, trustworthy; 3. sure to occur, inevitable, predetermined; 4. indubitable, reliable; trustworthy; 5. ben certein, to be sure, convinced, or confident, as in (44), (45); 6. ‘safe, secure’, of which senses 3 and 5 are clearly epistemic. The first attestations of Middle English predicative certain in the above epistemic sense 5 ‘sure, confident’ are more or less half a century later than those of sure, most of the examples quoted in the MED being from the latter half of the fourteenth century onwards (examples (44) and (45) below). The adjective expressed certainty mostly in the predicative position.² When it was used as a premodifier (as in (43)), it usually referred to one particular thing or person, without specifying exactly which one. The attributive uses of certain in the sense ‘specified, particular’ and ‘definite but unspecified’ are found as early as c.1300 and they correspond to similar uses in Anglo-Norman (cf. AND, s.v. certein), while the earliest epistemic senses are recorded in the late fourteenth century according to the MED data.

Being a longer word than monosyllabic sure, certain was probably perceived as the more formal of the two, similar to pairs such as deep vs profound, let vs permit, wish vs desire, etc. Most of the examples of certain quoted in the MED are taken from religious, philosophical and scientific texts. The CMEPV corpus search has shown that certain was more common than sure in written Middle English: there are more than 3,000 matches in several orthographic variants, but the search obviously does not distinguish between attributive and predicative uses. The word is not attested in the York Mystery Plays and the morality play Everyman, although there are several occurrences of predicative sure in the dialogues of these dramas. We also do not find any instances of sicker in these texts.

1. (43) Certeyn rente schal beo itake ech ȝer. (c.1300 SLeg.Becket (Hrl 2277) p. 17)

2. ‘A certain rent shall be taken every year.’

3. (44) I my silf am certeyn [Latin: certus] of ȝou, for and ȝe ȝou silf ben ful of loue. (c.1384 WBible(1) (Dc 369(2)) Rom.15.14)

4. ‘I am certain myself of you, because you yourself are full of love.’

5. (45) Also certeyn as we knowe echon That we shul dye, as vncerteyn we alle Been of that day whan deth shal on vs falle. (c.1395 Chaucer CT.Cl. (Manly-Rickert) E.125)

6. ‘Also as certain as we all know that we shall die, as uncertain we all are of the day when death shall come to us.’

7. (46) Wommen kan holde a man ful narwe‥Day be day‥To stonde vnsur³ betwixe hope & drede. (c.1425(a.1420) Lydg. TB (Aug A.4) 3.4853)

8. ‘Women can hold a man very strictly, day by day, to stand uncertain between hope and fear.’

The search in the full Shakespeare corpus of thirty-seven plays has provided us with 292 matches of mostly predicative uses of sure vs less than a hundred of predicative certain in the dialogues, clearly showing that sure was more frequent in speech. One may presume that a similar situation occurred in late Middle English.

4 Concluding remarks

Old English used the common Germanic word gewis(s) as the usual predicative adjective meaning ‘sure, confident, certain’ alongside far rarer attributive cuðlic and witod. Additionally, in the Alfredian prose of the late ninth century there is a singular occurrence of predicative sicor conveying the speaker's attitude of certainty, with the other two instances of the word in the DOEC displaying different semantic and syntactic properties. I believe that the predicative sicor must have been much more common in spoken Old English, as epistemic adjectives are predominantly characteristic of speech (Biber 1994; Kärkkäinen 2003). Otherwise, it would be difficult to explain the frequent occurrence of si(c)ker in early Middle English, as we do not find direct evidence of its being borrowed from either Old Norse or Anglo-Norman. The adjective (ge)wiss became obsolete in the twelfth century, but the word survived into late Middle English as an epistemic sentential adverb, usually preserving the reflex of the original prefix in the form iwis, ywis, etc.

At the turn of the fourteenth century English acquired new adjectives of certainty through language contact with Anglo-Norman: sure and certain, which were polysemous with both (earlier) concrete, objective senses of ‘safe, secure’ and (later) abstract, subjective senses of ‘sure, certain, confident’. This evidence supports Traugott's (1989, 1995, 2010) idea of subjectification whereby abstract senses usually develop from earlier concrete ones. I believe that the process had taken place in French (if not in Proto-Romance) before the words were borrowed into English. The comparison of the AND and the MED material and definitions does not bring out any striking differences between their usage in Anglo-Norman and Middle English. On the other hand, the fact that the earliest attestation of sure in Middle English is a concrete rather than abstract use of the word may point to subjectification also governing borrowing processes. In the subjective sense sure and certain had syntactic restrictions: they mostly occurred as predicators, usually following the copula.

Another factor that may have contributed to the gradual obsolescence and final loss of the adjectives (i)wis and sicker is the natural tendency to avoid homonymy, especially in the context of words referring to similar concepts (Hock & Joseph 2009). It appears that the adjective (i)wis ‘certain’ disappeared because it was too similar to another important adjective expressing the capacity of mind, namely wis ‘wise’. Likewise, sicker became obsolete when English borrowed its Latin etymon secure in the sixteenth century. Its meaning ‘safe’ overlapped with one of the senses of sicker, and the OED material shows that there was some confusion in the use of doublets sicker and secure in all senses in late Middle and early Modern English. Besides, sicker was also phonetically similar to secure unlike its French-derived cognate sure, which was a more distinct and shorter word that specialized as an adjective of certainty in English, with certain becoming its more formal synonym.

The widespread societal bilingualism and common code-switching among the better-educated population of medieval England (see Rothwell 2001; Schendl & Wright 2011; Ingham 2012) was the key factor responsible for the process of replacement and the rapid diffusion of Romance lexemes in English in the fourteenth century. This study also confirms the view proposed by Ingham and Timofeeva (this volume) that the role of the clergy and preachers was instrumental in spreading French loanwords. By the end of Middle English sure and certain had become the usual adjectives expressing subjective certainty on the part of the speaker/writer after the demise of the original Old English words. We obviously do not and cannot know everything about the replacement process given the limited data of Medieval English available to us in terms of genre, style and register. Epistemic adjectives and adverbs are more common in spoken discourse and we do not find many of them in religious texts (predicative sure and certain are conspicuously rare in Wycliffe's Bible), chronicles and poetry, which make up the majority of Medieval English texts that have been preserved. Nevertheless, the relatively scarce data from the dialogues in medieval plays and first-person narration in, e.g., Chaucer and early correspondence are indicative of the tendency for such uses of the words to occur mostly in speech.

As observed above, the borrowing of sure must have first happened among bilingual Anglo-Norman and Middle English speakers. Except for the singular occurrence of sure meaning ‘safe’ in the early thirteenth-century poem The Owl and the Nightingale, the first attestations of sure and certain are recorded c.1300, i.e. at the time when more and more bilinguals were switching to English as their first language, and transferring numerous French lexemes to their newly adopted mother tongue. There was a rapid diffusion of the new Anglo-Norman adjectives of certainty across the late Middle English dialects, social groups and text types, as the rich lexicographical and corpus data from the MED and the CMEPV confirm. From the mid fourteenth century onward both sure and certain became high-frequency adjectives of certainty as a result of this contact-induced lexical diffusion.

However, the obsolescence of the Germanic adjectival (i)wis in the twelfth century and then of sicker in late Middle English and their replacement with new Romance words sure and certain acquired through (oral) language contact was gradual. Sure was used alongside its earlier English doublet sicker for at least two centuries. Thus, the history of the successive English adjectives of certainty, which coexisted and competed with one another for several centuries in Middle English, strongly confirms the view of layering, described by Hopper (1991) as the tendency to preserve multiple synonymy in languages, not only in the context of grammaticalization, but also with reference to lexis. In early Modern English predicative sure and its more formal synonym certain became the dominant adjectives of certainty and their status has not changed for the last five centuries.