Albeit a conjunction, yet it is a clause: a counter-example to unidirectionality hypothesis? (1).

1. Introduction

In this paper I discuss the expression al be it in the context of grammaticalization, as presented most fully in the book by Hopper -- Traugott (1993). I will claim that this phrase, which arose in English in the 14th century, had its heyday in late Middle English and early Modem English when it was grammaticalized down the clause-to-conjunction dine, was later becoming more and more obsolescent to end up in the late 20th century as a more and more frequently used marker of concessivity, or better contradiction, albeit in a limited syntactic context. Modem data seem to suggest that the conventional spelling of albeit together only disguises its genuine character of a clausal phrase.

The origin of the idiom is usually attributed to the calque from the Old French expression tout-soft-il (tut seit-il) used in the same function (cf. Mustanoja 1968: 317, 468 or OED s.v.) and is first attested in the 14th century (Cursor Mundi, Chaucer, Gower):

(1) Al be it pai be theues all, Pat pai war breper elleuen Pat ham, pai neuend me pe yongeist nam

(Cursor Mundi 4978 a 1340) (2)

However, there are some earlier native instances of the adverb all, which appears to have been grammaticalized as a conjunction introducing clauses of concession with inverted word order and the subjunctive copula, e.g.

(2) al were he ifulled of de [holi] goste 7 al were he puruh miracle of barain iboren ... 3et ne dorste he wunien among men.

(Ancrene Riwle 70/10 1225)

(3) Ne telle pu nawt edelich, al beo pu meiden, to widewen ne to iweddede.

(Hali Meidenhad 39/653 c. 1230)

The close relationship between concessivity and universal quantification was noticed by Konig (1985: 10) to be present in many languages and "a component which is also used as universal quantifier" is considered to have been one of the major sources for the development of concessive connectives (cf. English for all, although, all the same, however, French toutefois, tout ... que, Russian vse taki, Polish mimo wszystko, wszelako, wszakze; Konig lists other examples, also from non-Indo-European languages such as Hungarian or Chinese). Also Mustanoja (1960: 316) and Visser (1963-1973: [section]883) point to the intensifying function of all in Middle English, which was easily combined with other connectives and yielded if all, though all, although cf. also also, already, always, algate(s) etc., where all was further delexicalized as a mere prefix; thus all went down the adverb > conjunction > prefix dine). Tracing the origin of concessive conjunctions in Romance languages, Harris (1988: 80-83) notes the use of Fr ench adverbs such as tout, bien in this function. In his opinion "a situation depicted as being entirely at one end is clearly made to be used concessively, provided that the end specified is that least readily compatible with the main clause, which is nevertheless represented as true. We find for instance tut seit-il mort (literally 'entirely be he dead') in the sense 'though he is dead' ... English 'albeit' clearly has a similar origin."

2. Al be it in Middle English

In Chaucer's English al-be-it becomes very common, yet its syntactic status is unclear. It is difficult to determine whether we should treat it as a petrified expression equivalent to a conjunction or whether it is still a clausal phrase. In all the corpus of Chaucer's prose (cf. Molencki in press), out of 54 instances of at be it only four are spelt together (all of them in Book V of Boece), thus supporting the conjunction analysis (though, as we know, medieval spelling cannot be fully diagnostic in such cases), e. g.

(4) Also ymaginacioun, albeit so that it takith the bygynnynges to seen and to formen the figures ...

(Boece Vp. 4, 205)

Yet the other manuscripts which I have consulted have the separate spelling:

(4) a. al be it so pat it taketh ...

It is not until the early Modem English period that the word/phrase is consistently spelt together, e.g. Marlowe's (late 16th century)

(5) Albeit the world thinke Machevill is dead,

Yet was his soule but flowne beyond the Alpes

(The Jew of Malta, Prologue 1)

Coming back to Chaucer, an argument favouring the conjunction interpretation is that out of 64 instances of the al + Verb + Pronoun sequence m the prose, in 49 we invariably find the expression al be it so that S (analogous to though so be that Slif so be that S commonly found in Chaucer) and in four al be it that S, without what Nagucka (1968: 79) calls the factive pronominal so, which cataphorically refers to the that-clause. In this case, however, that could also be interpreted as the optional pleonastic marker of subordination, frequently combined with other subordinators in late Middle English (if that, when that, which that, etc.). Except for one instance in The Legend of Good Women:

(6) Al be hit that he ne kan nat wel endite,

Yet hath he maked lewed folk delyte

it invariably appears in its weak form. But even in later texts one can find the strong form hit, even spelt together (h might have been mute anyway), e.g.

(F 414-415)

(7) Albehit that our enemys ... assembled nigh the same river. (after MED)

However, the transparency of the phrase for Chaucer's contemporaries is evidenced by several instances of a pronoun different from it in the frame al + Copula + Pronoun, e.g.

(Letters from War with France 71 1418)

(8) He may nat fleen it, thogh he sholde be deed

Al be she mayde, or wydwe, or elles wyf

(9) al be thow fer froo thy cuntre, thou n 'art put out of it.

(CT Knight's 1170-1171)

Kerkhof (1982: 49) lists several other examples from Chaucer's poetry with "the emphatic adverb al preceding the verbal form, which does not affect the inverted word order". Poutsma (1929: 712) discusses early Modem English conjunction albe, which "now quite obsolete, appears at all times to have been rare". He attests no instances in Shakespeare or the Authorized Version (which is confirmed by my search), but quotes one from contemporary Ben Jonson:

(Boece I p. 5,9)

(10) Ay, but his fear Would he ne'er be mask'd,

allbe his vices were

(Ben Jonson Sejanus IV.5.224)

and another one from the romantic poet Southey (late 18th century):

(11) And in their hearts, albe the work was rude, It rais'd the thought of

all-commanding might, Combin'd with boundless love and mercy infinite.

(Tale of Paraguay IV.XIX)

There are several comparable examples in Chaucer's poetry, e.g.

(12) But, al be that he was a philosophre,

Yet hadde he but litel gold in cofre

(CT Prol A 297-298)

In these sentences al-be appears to be a fully grammaticalized conjunction (perhaps with pleonastic that in (12)), as in all the examples there are other copulas in the albe-clause, which is not the case in other Chaucer's examples above or below with only one copula being part of the al-be phrase. The copula has a full range of inflectional forms, e.g. the plural been:

(13) al been they grevouse synne, I gesse that they ne been nat deedly.

(CT Parson's 449)

On the other hand, the lack of plural agreement in examples like:

(14) This is to seyn, she may nat now han bothe

Al be ye never so jalouse ne so wrothe

(CT Knight's 1839-1840)

(15) and whom it wol do boote

Al be his woundes never so depe and wyde

(CT Squire's 154-155)

may be an indication of the fact that the grammaticalization of al be was well under way, though Chaucer is known to have used -n variably. The preterite subjunctive were is not uncommon, either:

(16) Al weere it so that a riche coveytous man hadde a ryver or

a gotter fletynge al of gold, yit sholde it nevere staunchen his covetise.

(Boece III m.3,1)

(17) with swich vigour and strengthe that ne myghte nat ben

emptid al were it so that sche was ful of so greet age.

(Boece I p.1,9, which renders Latin quamuis ita aeui plena foret)

For (17) in another manuscript we find

(17) a. alle were it so that ...

Other manuscripts also have occasional instances of alle for the reduced at, which would support the clausal interpretation, as one of the features of grammaticalized items is their phonetic reduction. This would, however, presuppose original alle, which is found in none of the early manuscripts. In the examples with al were it tense agreement appears to be obligatory - whereas any Middle English tense may occur after at be it (so) that, only the preterite is used in the clause that follows at were it so that, even with present reference. Thus were appears to be the main verb for the following clause and the expression at were it is a clause:

(18) al were it so that she right now were deed, ye ne oughte nat, as for hir deeth, yourself to distroye.

Interestingly, we even find one instance of the indicative past was in The House of Fame:

(CT Melibee 982)

(19) Al was the tymber of no strengthe,

Yet hit is founded to endure

Visser (1963-1973: [section]883) says that al is "almost exclusively combined with be and were". But in Gower we find a form with procliticized negation:

(House of Fame 1980-1981)

(20) Such a loss he cawhte, at nere it worth a stre.

The copula is also found as an auxiliary in passive clauses as in (2) or:

(Confessio Amantis 5,997)

(21) al be he sodeynly caught with drynke, it is no deedly synne but venyal.

(22) I wol have moneie, wolle, chese, and whete,

(CT Parson's 822)

Al were it yeven of the povereste page

Or of the povereste wydwe in a village,

Al sholde hir children sterve for famyne

In Chaucer's prose corpus there is only one instance of a verb different from the copula, viz, do, which, however is a special verb even as early as Middle English:

(CT Pardoner's Prologue 448-451)

(23) Thanne cometh scornynge of hys neighebor, al do he never so weel

In the poetry have appears in this structure both as a lexical verb:

(CT Parson's 510)

(24) I holde hym riche, at hadde he nat a sherte

and as an auxiliary:

(CT Wife of Bath's 1186)

(25) Al hadde man seyn a thyng with bathe his yen,

Yit shul we wommen visage it hardily

There are occasional instances of modal verbs used in this syntactic frame, as can be seen in (22) and

(CT Merchant's 2272-2273)

(26) For thefte and riot, they been convertible,

Al konne he pleye on gyterne or ribible

(27) They wolde hym folowe, al wolde he fle.

(CT Cook's 4395-4396)

OED and Visser (1963-1973: [section]883) attest the fully lexical verb speke in the General Prologue to the Canterbury Tales:

(Romaunt of Rose 6268)

(28) He moot reherce as ny as evere he kan

Everich a word, if it be in his charge,

Al speke he never so rudeliche and large

MED adds to this list two examples with al telle I noght (Knight's Tale 2264) and al bledde I not a drape of blod (Romaunt of Rose 1754). In The Legend of Good Women I also found verbs like founde, made and swere, which shows that the pattern was quite productive, e.g.

(CT GP 732-734)

(29) And evere shal, til that myn herte dye

Al swere I nat, of this I wol not lye

(30) This was his wit, al made he to Jasoun

(LGW 57-58)

Gret chere of love and of affeccioun

As can be expected in concessive clauses, mostly personal pronouns are found in the frame al + Verb + NP, with few exceptions, as in (15), (19), (22), (25) or

(LGW 1420-1421)

(31) If in the hondes of som wrecche I falle,

I nam but lost, al be myn herte trewe

Discussing the origin of albeit, OED (in the entry for all) states that "all be it (that), in full all though it be that" is "only a particular instance of all with a verb in the subjunctive, in which the conjunctive phrase becomes a quasi-word" and that with the subjunctive mood "though or if being expressed by the reversed position of vb. and subject (as in be they = if they be) were omitted, leaving all apparently = although. Thus al be I = although I be". Such an interpretation is supported by the presence in Chaucer's corpus of the examples such as

(Troilus & Criseyde V, 705-706)

(32) yit hath the moment som porcioun of it, although it litel be.

Thus, in the 14th century al was losing its original meaning of an intensifying adverb and started sliding down the grammaticalization cline to become a prefix attached to the following copula. Its use, however, appears to have been optional, because the concessive relation could also be indicated by the inversion itself, as in earlier English (cf. Kerkhof 1982: 48-49):

(Boece II p. 7, 100) (3)

(33) Thow most tellen it platly, be it never so foul ne so horrible.

Compare (32) with the following two sentences, where there appears to be a free variation:

(CT Parson's 1023)

(34) It is ful perilous, al be it never so lite.

(35) thynkynge that oure Lord Jhesu Crist quiteth every good dede, be it never so lite.

(CT Parson's 294)

Nevertheless, it appears that Chaucer's English was still the intermediate stage of the process by virtue of which the clausal expression al be it became a conjunction. In later English, however, its use was becoming more and more limited. In all Shakespeare's corpus the concordance program search has provided me with merely 14 occurrences of albeit, invariably spelt together and replaceable by (al)though, which proves that it had become a fully grammaticalized conjunction, e.g.

(CT Parson's 688)

(36) Who are you? tell me for certainty,

Albeit Ile sweare that I doe know your tongue

(37) I haue as much of my father in mee, as you, albeit I confesse your comming before me is neerer to his reuerence

(Merchant of Venice II,vi.27)

Visser quotes some instances of albeit it be-clauses, which are evidence of full grammaticalization:

(As You Like It I,i,41)

(38) albeit it be dayly vsit ... it sufferis na iniures.

If the late Middle English or early Modern English pronunciation of albeit were available to us, we could be even more certain that the phrase was grammaticalized, as often the phonetic reduction may be one of the manifestations of the process. But this is an obvious limitation of any historical study of the language. However, some instances of the idiom from the late 15th century (alle bette, albut, all bote) quoted by MED show that the be-it part had merged and the vowel had an undetermined value, most likely /[LANGUAGE NOT REPRODUCIBLE IN ASCII]/. There might also have occurred contamination with but, e.g.

(Kennedy Litel Tracteit 129,2 (ed. Kuipers) c. 1560)

(39) He had gret fere, Albut paw hit nere no nede.

(40) Bod alle bette paw he had ben quene ... mekeliche he wolde abyde in pe quere at euery tyde.

(St. Editha 2124 1460)

Very interesting here is the combination of the two concessives albut and paw(= though).

(St. Etherlreda 285 a. 1450)

To sum up, in early Middle English concessive al appeared in various constructions:

Pattern One al + V + NP

with various verbs, including be. The most common noun phrase is the pronoum (h)it. In the mid-14th century one can find the first instances of

Pattern Two al be it (so) (that) + S

i.e. the clause al be it is followed by another full clause. Towards the end of the fourteenth century a complex reanalysis appears to have occurred. The clause at be it is reinterpreted as a conjunction introducing the concessive clause

Pattern Three albeit S

All the three patterns coexist in Chaucer's English, but Pattern One is recessive and becomes obsolete by the end of the Middle English period.

3. Howbeit 

Dictionaries also attest a synonymous form how-be-it such as in Shakespeare's

(41) So doe the Kings of France vnto this day,

Howbeit they would hold vp this Salique Law 

To barre your Highnesse clayming from the Female

(Henry V I,ii.91)

(42) How be it that this dyuyne essence ... maye not be perfyghtly knowen ... yet there is not any mortall persone but that he woll confesse there is a god.

How in many languages appears to have been another source for concessive connectives (English however, Dutch hoewel, French combien que, Polish jakkolwiek, etc.). OED (s.v.) and Mustanoja (1960: 429) quote some instances of how-be-it from the Middle English period:

(Trevisa Barth. De P. R. 1398) (4)

In Chaucer's prose there are no instances of how-be-it, but there are a few examples of how (so) that. Kerkhof (1982: 458) quotes an instance of how so it be that (= although it be, al be it) from Chaucer's Troilus and Criseyde:

(43) How so it be that som men hem delite

With sublyl art hire tales for to endite,

Yet for al that, in hire entencioun,

Hire tale is for som conclusioun

Some dictionaries list the form how were it, but no clear examples of the idiom equivalent to al be/were it are given. I have not come across it used as conjunction either in the Helsinki corpus or OED. Thus, parallel with al-be-it, how-be-it seems to have undergone the same process from the clause (with the meanings of all its elements fully preserved) down to the conjunction status. In Shakespeare's Othello, however, there is an example of howbeit followed by that, which is either a late reflex of the pleonastic subordinating that, characteristic of late Middle English or, as the original punctuation seems to suggest, a complementizer following the clause how-be-it, similarly to Chaucer's al be it that, as in (4a). (5) On the other hand, the elision be it > be't is the evidence to the contrary (cf. (39) and (40)).

(T&C II, 256-259)

(44) The Moore, howbe't that I indure him not,

Is of a constant, louing, noble nature.

The idiom seems to have been favoured by the translators of the King James Bible, where we find as many as 93 instances of howbeit and only nine of albeit. The last occurrence of the conjunction howbeit is dated by the OED for 1634 (the adverb is recorded last in Ruskin's writings from 1887):

(Othello II, i. 297)

(45) I ... would fain have access and presence to the King ... euen howbeit I should break up iron doors. (Rutherford Letters I,110)

but Poutsma (1929: 712) records a much later example from Walter Scott:

(46) Our good father Eustace says that, howbeit we may not do well to receive all idle tales of goblins and spectres, yet there is warrant from holy Scripture to believe, that the fiends haunt wastes and solitary places.

which, however, may have been a deliberate archaism.

(Monastery XX/231)

4. Albeit in Modern English

As for albeit, its usage seems to have declined in the 17th, 1.8th and 19th centuries. Noah Webster (1828 [1970]: s.v.) qualifies it as "now antiquated", but, what is interesting, in his definition makes a reference to its discourse function by saying that albeit is "equivalent to admit or grant it all". The searcher program has provided me with no examples of the item in the selected corpora of such authors as Milton, Dryden, Congreve, Defoe, Austen, Burke, Mary Shelley, Darwin, Melville, Woolf, Scott Fitzgerald and Faulkner. Nevertheless, we find occasional instances of this obsolescent word in Victorian writers, e.g.

(47) Albeit she was angry with Pen, against her mother she had no such feeling.

(Thackeray Pendennis I, Ch. XXI,215)

(48) Maisey would have wept at the least encouragement, but Dick's indifference, albeit his hand was shaking as he picked up the pistol, restrained her.

(Kipling The Light that Failed Ch. 1,9)

(both quoted after Poutsma 1929)

In all the Victorian English examples available to me albeit is used as a conjunction (in both the examples above it is replaceable by (al)though, I believe), except for a single occurrence of albeit that from Stretton's Chequered Life published in 1862:

(49) From that day to this we have never met -- albeit that he has had my best wishes.

(Stretton Ch .L.I, 125 after OED)

This is a rather too late occurrence to be interpreted as a combination of a conjunction and a pleonastic subordinator, so the clausal analysis seems more plausible in this case: al be it that S, which might have been a reflex of Middle English Pattern Two.

In the 20th century albeit is becoming rare. Some short or concise English dictionaries (especially those in the range of 40,000-50,000 entries) do not have this entry at all, e.g. 1994 editions of Chamber's Minidictionary, Compact Dictionary, School Dictionary, Essential Dictionary. Nor is it included in Mini-Oxford Dictionary or Little Oxford Dictionary. The dictionaries that have albeit usually give the following definition:

(conj.)unction (al)though

and sometimes qualify it with such epithets as "not colloquial", "formal" (e.g. Hornby's Advanced Learner's Dictionary of Current English or dictionaries published by Collins, Cassell, Harrap, Longman or Webster). Quirk et al. (1985: [section]14.12) mention albeit among the "archaic concessive subordinators" that "still have a limited currency". However, as we will see below this is not the case at all. According to the modern data the role of albeit as a subordinating conjunction is marginal, whereas it is becoming a more and more common discourse marker of contrast functioning like a sentential adverb. It is only in the BBC English Dictionary (1993) and Collins/Cobuild (1994) (both are based on the Birmingham Concordance) that we find a more satisfactory definition:

CONJ: You can use albeit to introduce a fact or comment which contrasts in some way with what you have just said (a formal word)

where it is still classified as a conjunction, yet its discourse function is given most prominence.

5. Albeit towards the end of the 20th century

Since my intuition was telling me that there was something wrong with the modern definitions of albeit, in order to determine its present status I applied a concordance program first to a relatively small corpus of modern English texts of c. 8,000,000 words and then to a much larger one of more than 155,000,000 words. The data that I found were most revealing. The smaller corpus included some issues of Time Magazine, extracts from the British Times of 1992, lectures about road-building, a leaflet on AIDS and a short story Small is Beautiful by E.F. Schumacher. The data from Time are particularly interesting, as they show that contrary to expectations the usage of albeit has been gradually increasing in recent times. In the five decades between 1920-1970 there were barely 10 occurrences of the word. In 1989 there are as many as 12, in 1990-17 and in the first six months of 1991 the writers used it 16 times. Curiously enough, albeit does not appear even once in any of the 88 interviews scanned (spoken register? ), starting with the 1933 interview with Joseph Stalin. Unlike the 19th century examples such as Thackeray or Kipling above, it is hardly ever used to introduce a full concessive clause that would contain a verb, as does (al)though, with which albeit is supposed to be synonymous according to most dictionaries of Present-day English. Out of 69 occurrences (which gives the average of one albeit per c. 115,000 words) 17 introduce an adjective, as in

(50) His initial speech was an eloquent, albeit quixotic, lament over the racist lyrics in the official state anthem.

15 an adverb:

(51) Saddam is hailed --albeit posthumously -- as a hero of the Arab masses.

6 a noun phrase:

(52) But mostly it is because they couldn't accept that fellow Germans, albeit Communists, could create such a frightful economic mess.

21 a prepositional phrase:

(53) Job shedding is likely to continue, albeit at a slower rate.

There are 8 instances of the pronominal one, either modified by an adjective:

(54) for the year 2000 onwards, contemporary art will still be "of this century", albeit a brand new one.

or a prepositional phrase:

(55) She was as ferocious with her employees as a bulldog, albeit one with a face-lift

or a relative clause:

(56) She is also an anomaly: an influential woman in a macho society, albeit one that claims to have eradicated sexism.

The only example in the whole corpus that does introduce a clause has an intervening that. Admittedly, it represents so-called legalese, thus rather unusual English, as it comes from a judgement passed by the Queens Bench Divisional Court on June 30, 1992:

(57) Mr Lyons submitted that the condition precedent to the justices having jurisdiction to order destruction of the dog under section 5(4) was not satisfied because the applicant had been prosecuted, albeit that it had been discontinued.

Syntactically this sentence resembles the latest occurrence in OED for albeit that given under (49).

The observations from the smaller corpus are confirmed by the huge corpus of the CD-ROM editions of all the four 1993 British quality daily newspapers. Table One below presents the statistical occurrence of albeit:

Table One

Statistical occurrence of albeit in the corpus

Corpus               Number of words  Number of albeit
                                        occurrences


The Times              38,755,029          681
The Guardian           28,728,001          428
The Daily Telegraph    43,372,564          573
The Independent        44,557,571          633

 TOTAL                 155,413,165         2,315

Corpus                 Occurrence of
                     albeit per number
                         of words

The Times                 55,603
The Guardian              67,121
The Daily Telegraph       75,693
The Independent           70,391

 TOTAL                    66,672
Table Two shows the syntactic distribution of albeit in the corpus:

Table Two

The syntactic distribution of albeit in the corpus

Structure                                Number of    Percentage
                                         occurrences

albeit + PP                              949          40.92
albeit + Adj                             489          21.09
albeit + Adv                             465          20.05
albeit + NP (including the pronoun one)  366           5.78
albeit + that -S                         30            1.29
albeit + S                               11            0.47
albeit + it                              8             0.35
albeit, S                                1             0.04

It looks that in Present-day English albeit is only marginally a conjunction introducing concessive clauses and its main role is that of a marker of contrast. This function has its roots in Middle English, as occasional instances of usage similar to the modem examples above are found in Chaucer, e.g.

(58) and by a maner thought, al be it nat clearly ne parfitely,

ye loken from afer to thilke verray fyn of blisfulnesse.

(Boece III, p.3,4)

Yet in Chaucer's English such usage was marginal, the vast majority of the examples being found in the frame al be it (so) (that) -S.

The 1993 British dailies corpus also shows that albeit that is used not only in legalese, but is also found in the sports column:

(59) A poorish performance by Poland in midfield, albeit that they had a 5-4 advantage was contributory, yet England's authority could not be denied.

It is also there (and in economy reports) that we find the item used as a conjunction equivalent to although:

(60) Importantly for Scotland, their seven-match development tour of the South Sea islands began with a victory, albeit they almost let the game slip away in the dying minutes as they wilted in the heat.

(61) Analysts are predicting the company will return to profit this year, albeit their tax projections are for a figure of only [pounds sterling]500,000.

Examples like (59)-(61), however, are exceptional and, I believe, are fossilized instances of the earlier usage. As is well-known, language change does not come about abruptly and there always is a transitional period when one can find instances of earlier forms and functions which are used parallel with the innovations. It appears that in Present-day English the predominant function of albeit is not that of a conjunction, but it is first of all found as a marker of contrast. It functions as a comment referring to a preceding sentence element and expresses surprise, regret, reservation, downplaying and the Like. In not a single example in the huge corpus of the British papers is albeit used at the beginning of a sentence, as still was the case in the 19th century (cf. (46) here). The fact that the syntactic status of albeit is unclear for Modem English speakers is borne out by the following examples with the apparently redundant pronoun it:

(62) But the real reason Buckingham Palace has thrown its gilded gates open (albeit it for a hefty [pounds sterling]8 per head) is that the Prince of Wales has been lobbying hard for such a move.

(63) Yet she had lived long enough, albeit it in her mother's womb.

Finally, I have been able to find an instance of albeit without any accompanying words, used as a discourse marker equivalent to nevertheless. The sentence from the Daily Telegraph is merely a single example in the whole huge corpus, but shows another interesting development in the career of albeit in English. It also supports the diachronic path adverbial phrase > sentence adverbial > discourse marker proposed by Traugott (1985). Albeit, which may be regarded as a sentence adverb in sentences (50)-(5 6), has acquired a new pragmatic function of a comment to the prior discourse:

(64) Admirably, Her Majesty has resisted the pressure on the Foreign Office, to preclude her meeting with Turkish Cypriots. Albeit, it will take place on British territory on the island.

The vast majority of the late 20th century examples, thus, do not exclude the clausal analysis of al-be-it, and what is more, less than one per cent of the instances support the conjunction interpretation. As for spoken English, I also think that modern native speakers tend to treat albeit more like a sort of a clausal phrase than a conjunction proper. The expression is stylistically marked, perhaps deliberately used to achieve an effect of sounding more sophisticated, somewhat archaic. Having heard several instances of al-be-ft in both the British and American mass media, as well as from numerous native informants, I noticed that all the three elements of the phrase are distinctly pronounced as if they were three separate words, with the long vowels and pauses: [o:1 bi: 'It] (6) so the phonetic reduction characteristic of delexicalized elements does not seem to have taken place and one might claim that we are back in Chaucer's time, where al is the concessive conjunction, be the main verb in the subjunctive form and it the pronominal subject. Middle English Pattern Three appears to be recessive now and earlier Pattern Two is gaining more and more ground, though contrary to Middle English al-be-it is rarely followed by a full clause. Instead, we typically find NP, AdjP or AdvP.

Thus the history of al-be-it in English might provide a counterexample to the strong hypothesis of unidirectionality (Hopper - Traugott 1993: 126), which claims "that all grammaticalization involves shifts in specific linguistic contexts from lexical item to grammatical item, or from less to more grammatical item, and that grammaticalization chines are irreversible. Change proceeds from higher to lower, never from lower to higher on the chine". Hopper - Traugott (1993: 126), however, believe that "Extensive though the evidence of unidirectionality is, it cannot be regarded as an absolute principle. Some counterexamples do exist". Albeit apparently is one.

(1.) This article is a modified version of the paper I delivered at the 8th International Conference of English Historical Linguistics in Edinburgh in September 1994. I am most grateful to Elizabeth C. Traugott, Stanford University, for her useful comments and suggestions. I would also like to thank Jacek Lipinski, University of Silesia, for his help with the concordance programs. Obviously, I myself am responsible for all the errors and inconsistencies that remain.

(2.) Interestingly in Morris's parallel edition of four manuscripts, for Al be it in Cotton and Fairfax we find All-pou and Alpouze in Gottingen and Trinity, respectively and there is a similar alternation with another instance of al be it in Cursor Mundi attested by Visser (1963-1973: [section]905).

(3.) In different editions of Shakespeare we find variant versions, e.g. for

Albeit I make a hazard for my head

(Henry IV. Part One I,iii,128)

in Wells and Taylor's (1986) original spelling edition we find

Although it be with hazard far my head

(4.) OED also records the adverbial function of howbeit equivalent to however it may be as in

How be hyt I wyl not fayle you

(Malory Morte d'Arthur xi 1470-1485)

(5.) Another argument for treating how-be-it as a clause is an instance of the idiom that I came across in the Paston Letters, where the pronoun is used in its strong form hit without the phonetic reduction:

Syr, hit nedith not, I trow, to send yow the tidyngys of these parteyes, how be hit

I have thryes send yow such as here were in entent that ye shuld send us yowrys

(Letters 775/7 1477)

but the spelling may be misleading - h may well have been mute, as in (6) and (7).

(6.) The 1992 edition of the American Heritage Dictionary of the English Language (Third edition) gives two variant phonetic transcriptions - for al-, apart from more common [0:l], we also find [ael].

REFERENCES

Benson, Larry Dean

1993 A glossarial concordance to the Riverside Chaucer. New York, London: Garland Publishing Inc.

Bowers, Fredson (ed.)

1973 The complete works of Christopher Marlowe. Cambridge: CUP.

Fisiak, Jacek (ed.)

1997 Studies in Middle English. Berlin, New York: Mouton de Gruyter.

Haiman, John - Sandra A. Thompson (eds.)

1988 Clause combining in grammar and discourse. Amsterdam: Benjamins.

Harris, Martin

1988 "Concessive clauses in English and Romance", in: John Haiman - Sandra A. Thompson (eds.), 71-99.

Hopper, Paul J. - Elizabeth Closs Traugott

1993 Grammaticalization. Cambridge: CUP.

Kerkhof, Jelle

1982 Studies in the language of Geoffrey Chaucer. (2nd edition.) Leiden: E.J. Brill/Leiden University Press.

Konig, Ekkehard

1985 "On the history of concessive connectives in English. Diachronic and synchronic evidence", Lingua 66,1: 1-20.

Kurath, Hans - Sherman Kuhn (eds.)

1954- Middle English dictionary. Ann Arbor, Mi.: University of Michigan Press.

Molencki, Rafal

1997 "Concessive clauses in Chaucer's prose", in: Jacek Fisiak (ed.), 351-372.

Morris, Richard (ed.)

1874-1893 Cursor mundi. A Northumbrian poem of the XIVth century in four versions. London: Published for the EETS by K. Paul, Trench, Trubner & Co.

Mustanoja, Tauno

1960 Middle English syntax. (Memoires de la Societe Neophilologique de Helsinki 23.) Helsinki: Societe Neophilologique.

Nagucka, Ruta

1968 The syntactic component of Chaucer's Astrolabe. (Zeszyty naukowe Uniwersytetu Jagiellonskiego CXCIX. Prace jezykoznawcze. Zeszyt 23.) Krakow: Wydawnictwo Uniwersytetu Jagiellonskiego.

Oizumi, Akio (ed.)

1991 A complete concordance to the works of Geoffrey Chaucer. Hildesheim, Zurich, New York: Olms & Weidmann.

Poutsma, Hendrik

1929 A grammar of late Modern English, Vol. I: The sentence, Part II: The composite sentence. (2nd edition.) Groningen: P. Noordhoff.

Quirk, Randolph et al.

1985 A comprehensive grammar of the English language. London: Longman.

Robinson, Fred N. (ed.)

1957 The complete works of Geoffrey Chaucer. (2nd edition.) Oxford: OUP.

Simpson, John Andrew - Edmund Simon C. Weiner (eds.)

1989 Oxford English dictionary. (2nd edition.) Oxford: Clarendon Press.

Traugott, Elizabeth Closs

1995 "The role of the development of discourse markers in a theory of grammaticalization", (Paper delivered at the 12th International Conference on English Historical Linguistics, University of Manchester, England, August 1995.)

Visser, Frans Theodor

1963-1973 An historical syntax of the English language. 3 vols. Leiden: E. J. Brill.

Webster, Noah

1828 An American dictionary of the English language. New York, London: Johnson [1970] Reprint Corporation.

Wells, Stanley - Gary Taylor (eds.)

1986 William Shakespeare. The complete works. (Original-spelling edition.) Oxford: Clarendon Press.