Arizona State University

ellyvangelderen@asu.edu

GLAC, 21 April 2001

Draft 15 April 2001

Particles indicating the `mood' of the speaker, hence MPs (Modal Particles) are said to occur in all Germanic languages except English. Abraham (1991a: 7, 247; 1991b: 357) connects the possibility of having them to having a `Mittelfeld', i.e. the area between V-last and V-2nd. In English, full phrases often occur to express discourse information, rather than particles, and they occur at the beginning (or end) of the sentence, not in the middle. Thus, there are two questions: (a) why does English not have particles to the same extent as the other Germanic languages (German, Dutch, MSc), and (b) what do older varieties of English which have V-2 and V-last do? In this paper, I show that OE and ME indeed have MPs in the same way that other Germanic languages do. I also provide a very tentative suggestion for (a).

First, I will say something on terminology, structure, use, and meaning.

In the Germanic literature, the term for the particles under investigation is mood or modal particles, hence MPs (e.g. Abraham 1991a; Vismans 1994); in the English literature, terms such as discourse marker (e.g. Schiffrin 1987), cohesive, or pragmatic marker (Brinton 1996), hedge (Östman 1981) are used, to name but a few.

The definition for English MPs is broader as well. Thus, Brinton (1996: 6), in examining the six Middle English "pragmatic markers" gan, anon, gelamp, bifel, hwæt, and I gesse, distinguishes the following functions:

Lenker (2000) talks about the OE discourse markers soþlice and witodlice, which are not the typical Germanic MPs. The literature is vast and will not be reviewed here (e.g. Lakoff 1972, Brown & Levinson 1978, James 1983, Heritage 1984).

The Germanic literature on MPs is also very extensive (e.g. Altman 1976, Weydt 1979, Abraham 1990; 1991ab; 1995; Hentschel 1986, Helbig 1988, Jacobs 1983; Foolen 1993). However, in this literature, phrases such as I guess and you know aren't seen as MPs, and MPs are defined syntactically as well as pragmatically. For instance, MPs do not occur in initial position; they can't be stressed; they can't be the questioned or negated; and they can be deleted without affecting the meaning of the sentence (see e.g. Abraham 1991ab; Vismans 1994: 45; Aijmer 1996). Full phrases are not included. I'll expand on these differences in section 2. In the Germanic literature, a distinction is made in that scalar/degree particles and Focus Particles (hence FPs) are seen as separate from MPs. Abraham (1991a: 1) focuses on modal/pragmatic particles (but mentions scalar ones as well); König (1991) concentrates on FPs; and Klein (1998) on degree particles (refered to as degree adverbs). The distinction between MPs and scalar particles is justified psycholinguistically by Bayer (1991). MPs and the other particles are related since expressing degree implies focus and focussing something implies subjective judgement, i.e. mood.

In what follows, I initially examine as broad a spectrum as possible, but restrict it in the later sections to MPs in the narrow sense. These have not received as much attention in the English literature. For convenience, I use MP for both English and the other Germanic languages.

In this section, I first discuss some differences between Modern English and the other Germanic languages and then some similarities.

As mentioned, one of the differences between English and the other Germanic languages is the position of the MP in the sentence. In English, they either occur on the periphery, as in (1a), or are modifying adverbs, as in (1b) to (3), which would be considered scalar in Germanic, or focussing, as in (4), discussed in Underhill (1988). They also occur (less typically) an IP adverbial, such as perhaps in (5):

1 a. Surprisingly, he looked well.

b. He looked surprisingly well.

2. He knocked the man right out.

3. They had quite a car.

4. They went like to the store.

5. Unfortunately, this is perhaps not feasible.

Quirk et al (1985: 445 ff.) discuss modifying adverbs and here a variety of MPs can be found, but they are not discussed as such. The difficulty in distinguishing degree adverbs, scalar particles, or FPs from MPs is clear in e.g. (1b) to (3). In (2), right modifies out but also adds a sense of exclamation on the part of the speaker. In (1b), surprisingly technically modifies well but is added to provide speaker comments. This is true in (3) as well. The way to distinguish scalar from MPs is to stress or focus them, as I'll show below. So, the position of the MP in the English sentences is either modifying a clause or a phrase.

In Dutch, which I will take most examples from in this paper, MPs occur in the `middle', as in (6), a translation of (5), and are less clearly phrasal or clausal modifiers:

6. Dit is jammergenoeg waarschijnlijk niet doenbaar.

The `middle' can be made precise as the area between the definite NPs on the left and the indefinite NPs on the right, as in (7) and (9). Abraham (1991b: 244) calls this the area between thematic and rhematic. Sentences (8) and (10) show that the MPs are not in the definite domain since hem in (8) is definite and not in the VP-domain since een boek in (10) is assumed to stay in the VP (in these sentences, dan and toch are the clear MPs):

7. dat ik het hem dan toch weer eens een keer uitgelegd heb.

that I it him then yet again once a time explained have

`that I have once again explained it to him'.

8. *dat ik het dan toch weer eens hem een keer uitgelegd heb.

9. dat ik dan toch weer eens een boek heb gelezen.

that I again once a book read have

`that I've managed to read a book again'.

10. *dat ik een boek dan toch weer eens gelezen heb.

A second characteristic is the number of such elements, i.e. the possibility of stacking. In Dutch many are possible, e.g. sequences of 7 MPs are possible, as (11) shows, even though, according to van der Wouden (1999: 294), 6 are quite `spectacular' already:

11. Doe dat dan nu toch maar weer eens even over nieuw.

do that then now yet but again once just again new

`Do that again'.

The order is roughly that of tense-mood-aspect (dan/nu are tense; toch/maar mood; weer/eens/even aspect), indicating they are possibly in functional categories. In English, whenever two or more particles co-occur, they tend to become one phrase, e.g. already, also, all right, well now, although, as well, even though. This is much less the case in German and Dutch (see also van der Wouden 2000).

As mentioned in section 1, a third characteristic is that typically MPs such as dan, toch, as in (12), cannot occur in initial position. If they do, it is as temporal adverbs. The phonologically heavier ones do occur initially, as in (13):

12. *Weer vroeg ik het hem.

Again asked I it him

`I again asked him to do it'.

13. Misschien zou je dat kunnen doen.

Perhaps could you that be able to do

`Could you perhaps do that?'

MPs in Modern English and Germanic have a number of characteristics in common. They cannot be focussed through being negated, or questioned, or stressed. If they are, they again are no longer modal but temporal or spatial. Examples of negated, questioned, and stressed MPs are provided in (14) to (16) from English. They are ungrammatical if derived from (17), i.e. if they contain an MP:

14. *It is not like you still owe me money.

15. *How do you still owe me money?

16. *You LIKE still owe me some money.

17. You like still owe me some money.

Sentence (14) is grammatical if not negating or questioning the MP. In (14), like would become the complementizer `as if'. The same is true if the MP is on the periphery, e.g. well in (18). This is shown by the ungrammaticality of (19) and (20):

18. Well, now, if I didn't think it ...

19. *Not well, now, ...

20. *How well, now, ...

In Dutch, the situation is the same. Even MPs that occur at the beginning of a sentence, such as misschien in (13), obey these constraints. Examples are given in (21) to (23), derived basically from (9):

21. *Het was niet dan toch dat ik een boek las.

22. *Hoe las ik een boek?

23. *Ik las DAN TOCH een boek

This focussing criterion provides a way to distinguish between MPs and FPs, as many people have indicated, e.g. Bayer (1991), König & Stark (1991). Thus, in (24), gerade can only be scalar since it is negated:

24. Er war nicht gerade erfreut

`He was not exactly happy'. (from König & Stark 1991:313)

Thus, syntactically, MPs in Germanic and English share a number of characteristics, namely they don't allow focus through negation, questioning, or stress, but they differ in position in the sentence, in morphological shape, and in stacking possibilities.

What could the structure be? Abraham (1995; p.c.) suggests that there are at least three layers of modality above the VP. MPs are phrases, according to him, in adjoined positions. Morphologically and syntactically, they definitely look like heads since they cannot be modified. This would also account for the absence English-like auxiliaries in Modern German and Dutch. However, if they were in the head positions, this would stop verb-movement to C, hence this is unlikely. They can't be in the Specifier position, if one assumes a Kaynian tree because the VP would have to move there to account for word order facts. Hence, without drawing a tree, I will assume they are connected to FCs but do not form barriers.

At a first look, there are quite a number of MPs in Modern English. Table 1 lists the `simple' ones. They are mainly of Anglo-Saxon origin, and are comparable to MPs in the other Germanic languages in that most do not occur sentence-initailly, and cannot (typically) be questioned, negated, or stressed. As mentioned in section 2, unlike Germanic MPs, English MPs are often connected to a phrase and aren't stacked to the same extent as they are in Modern English or Dutch:

Part Cat FP MP Du/Gm

about Adv/P 1055 --

again Adv -- ?

all Adj ? ? al

but P/Adv/C -- --

even Adj 1600 -- even/eben

just Adj 1551 -- < French?

like Adj ?20th 20th

once Adv -- -- eens/einmal

quite Adj ?17th -- < French? `free' cf. Du kwijt

now Adv -- OE nu/nun --

right Adj/v 1400 ?20th

so Adv/C OE OE zo

still Adj/v 1600 1699

then Adv/C -- OE dan/denn

though Adv/C -- OE toch/doch

too Adv OE ?20th te/zu

well Adv OE OE wel/wohl

yet Adv/C OE 20th

In table 1, I have used dates from the OED; FP refers to focus, degree, or intensifying use; MP if the particle is used for speaker inference, probablity, comment. Comparing this table of 18 MPs with e.g. Abraham's (1991b: 352-3) table of 16 MPs (aber, auch, bloB, denn, doch, eben, (so)gar, etwa, halt, ja, nur, schon, wohl and the complex eigentlich, einmal, vielleicht), shows that the inventory is similar. König (1991: 173) mentions "twenty ... or so" and adds einfach, erst, nun(mal), mal, ruhig. Hentschel (1986: 3) lists 18 particles, 3 of which (vielleicht, eigentlich, einfach) are complex and grammaticalized in the MA or C15 period. So, the inventory is similar, and most MPs originate as Adverbs; some first as intensifying.

Table 2 lists the more complex English ones. Some are loans from French into ME; some are phrases; and some are the result of grammaticalization (al + swa). In their use as MPs, they have no cognates in Germanic. The second column lists their first use in their modern form in the OED:

after all 1850

also 1200

all right 1837

alright 1175

already 1380

anyhow 1740

anyway 1570

as well 1366

at all c1350

certainly 1300

doubtless 1340

flat out ?

indeed 1575

in fact 1707

maybe 1848

no doubt 1380

of course 1541

only 1297

perhaps 1534

personally 1398

realistically 1868

really c1350

truly a1000

whatever 1900

Table 3 is a partial list of expressions that are originally clausal in nature or are expressions of emotion. These would not be counted as MPs in the Germanic literature, but as interjections, and surrounded by commas or comma intonation. They occur mostly on the periphery:

OED

as I said

Let alone

(I) for one

(Today) of all (days)

See

I mean (see Schiffrin)

I dunno

You know (see Schiffrin)

I think

Oh 1548 (see Schiffrin)

Oh well

on the whole

Table 3: Partial list of clausal expressions and interjections used as `MP'.

So, in Modern English, MPs are CP-oriented, i.e. occur at the edge of a sentence (as topics), and are full phrases, or they are ambiguous with focus markers connected to a phrase but are not phrases themselves. In Dutch, they typically occur in between the definite NPs and the indefinite material connected to the VP.

Vismans (1994: 102ff.) shows that in Middle Dutch (up to 1500), toch/doch occurs as MP in all texts examined. (This is not unexpected since Gothic already seems to have had thau, see Hentschel 1986). Dan occurs in 3 texts and eens in one. In the early C17, ook and nou are added; and the number of MPs gradually increases.

Vismans provides a sociological reason (based on Norbert Elias) and works within a functionalist framework, and he does not analyze the syntactic constraints. He (1994: 58ff.) divides MPs in reinforcers and mitigators. His Appendix (pp. 224-235) provides all the sentences with MPs from the corpora he uses. Van den Vos Reynaerde, a fourteenth century text, provides the following examples:

25. Die conic sprac: `Nu segghet dan'

that king spoke: now tell then

26. Laet mi doch lesen twee paternoster

let me please read two prayers

Examining the other Middle Dutch sentences from Vismans' Appendix (as well as going through the 15th century Esmoreit), it turns out that nu is used in the beginning, as in (25), and doch (and eens) in the `middle'. As I show in section 5 below, this use is not that different from that in ME.

Doch in OHG is used in a similar way to Middle Dutch, according to Hentschel's (1986: 87 ff.) data. No data are provided on nu. She shows that in Gothic, in addition to thau, jah was used "metakommunikativ" (p. 74). The two other particles Hentschel examines are eben and halt, which are outside the scope of this paper. Wauchope (1991) also concludes that thoh, ia, thanne are MPs in OHG. Many examples are given, e.g. (27) is from Isidor, (28) and (29) from Otfrid:

The well is much deep; where take you then over the water flowing

Since the cognate of though appears to have been the earliest MP in Gothic as well as Middle Dutch and OHG/MHG, I'll first examine that MP. Looking through some texts, a possible MP is eke, a cognate of Dutch ook and German auch. Lastly, I look at now and then, which have remained MPs in ModE (be it in a different position), and at even which is a degree adverb. This review shows that by Late ME, the MP use of these particles was lost, and their position is on the periphery or modifying a phrase. This loss coincides with the loss of V-2 (see e.g. van Kemenade 1987), in accordance with Abraham's observations for German. It also coincides with the complementizer becoming optional in the subordinate clause (see van Gelderen 1993). Thus, as the verb ceases to move to C and the I position becomes more important, MPs become restricted to the CP periphery. In section 6, I speculate on what causes this relation between the loss of MPs and the loss of V-2nd.

Though is a cognate of Dutch/German toch/doch. The OED lists some instances of `adversative particle' though from OE on, and these are definitely comparable to the Middle Dutch examples above. For instance, (30) and (31). These, however, are the only examples provided by the OED:

30. Blickl. Hom. 37, from 971

Ne ma3on þis þeah ealle men don

`Not might this though all men do'.

31. Lamb. Hom. 119, from 1175

ac hi þah ledað to deðe on ende

`but they though lead to death in the end'.

The position of the MP is very much like that in Modern Dutch and German, after the definite pronouns.

Examining the Helsinki Corpus confirms that the use of though as MP might be rare (but that may also be due to the genre of the sources). In the period before 950, Alfred uses it, as in (32) to (35); between 950 and 1150, there are a few examples as in (36) to (38):

32. Orosius, 64.17

hie þeh Philippus besirede mid his lotwrencum

they though Philippus deceived through cunning

33. Idem, 66.6

Philippuses yfel mehte þeh þagiet be sumum dæle gemetlic þyncan

Philippus' evil could though then-yet to some extent sufficient seem

34. Idem, 98.6

& he þeh sige hæfde

and he though victory had

`and he was victorious'.

35. Idem, 123.2

hie þeh þurhtugon þæt hie ofslogon Lucius

they though achieved that they killed Lucius

36. Rushworth Gospels, Matthew 15.20

37. Wulfstan

þonne gebyrhst ðu þeh þinre agenre sawle

38. Chronicle

& he wæs þæh full cyng ofer eall

`and he was though fully king over all'.

In (36), though is a C (which occurs quite frequently in that period), and in (35), the position of the MP is strange but may still be the Mittelfeld (if the CP extraposes).

The Helsinki Corpus points towards Alfred as using though as MP, especially in (33) with multiple MPs in medial position. Alfred's Pastoral Care has 426 instances of (swa)though and 46 of these occur medially, as in (39) to (44), with the MP typically in the mittelfeld. These `feel' very much like the Dutch or German MPs. However, note that the position of though in (40) is similar to that in (35). This position would be different in Modern Dutch since the V would have to be in second position. All other instances conform though:

39. PC 51.7 Hatton (similar in Cotton)

Ond næs swaðeah to anwillice ne forbuge he

`and yet he must not decline it too obstinately'

40. PC 99.21-2

& sua suiðe sua he ... he ðeah gehwyrfde his heortan eage ... to

`and although he ... he still directed his mind's eye to ...'.

41. PC 199.10-1

gedo he ðeah ðæt his hieremenn ongieten ðæt he ...

do he still that his underlings understand that he

`let him make his subjects understand that he ...'

42. PC 147.12

Ac hit is ðeah suiðe earfeðdæde ðæt mon ...

but it is though very difficult that one ...

43. PC 263.16

Nalles ðeah sua egeful ðæt hie ealneg ðurhwunigen on ðæm ege

not-want though so fearful that they always continue on that fear

`yet they shouldn't fear as much as to always continue fearing'.

44. PC 265.6

hit mon sceal ðeah geendigean for sumes godes lufum.

`yet he ought to finish from the love of something good'.

Most instances of though are conjunctions, as in (45), and they are mainly clause-initial (and occur mainly after and), with a few clause-final:

45. PC 31.6

ðeah ðæt folc ðyrste ðære lare

though the people thirst for instruction

In Middle English, the MP use of though dies out. Layamon's Caligula version from the 13th century has about 35 instances (that I could find) of variants of though (þah 10 times; þeh 22 times, and the rest þeah, þeih, þaih, not searched for þæh). These all introduce clauses and are less often combined with and, al, nu than they are later. In late ME, e.g. in Chaucer, though is never an MP. It occurs 378 times and only functions as a clause introducer, often followed by that or preceded by and, but, for, as, all, and a few times by ne or eke. There are 90 instances of a variant of although, used the same as though.

The first use in the OED of although is 1325, as in (46), (but note that though al occurs in Layamon) and of even though in 1697 (entry 9c even adv)

46. Song Mercy 168

Al þau3 i kouþe, yf þat i wolde.

In Modern English though is still used on the periphery of the clause, as in Middle English. Some of the 19 instances from Tom Sawyer are:

47. TS l. 1938

Well, I'll just bet I will, though.

Although is limited to complementizer in the text, and occurs only 6 times.

Concluding the discussion of though, its MP use in OE is rare, except perhaps in certain texts such as Alfred, and its use dies out by Early ME.

The OED lists ac as a conjunction cognate with OHG oh. It lists eke `also, in addition' as an adverb, < OE eac (Beo 3131), but the origin is uncertain. The two are not said to be related. I will consider them together, however, since in ME they are hard to distinguish.

In Old English ac is quite frequent, especially introducing a clause, as in (48):

48. Chron. (Plummer 1965/1992P, p. 136)

Ac for eallum þissum se here ferde swa he sylf wolde.

In the OE4 part (=1050-1150) of the Helsinki Corpus, there are 265 acs, mainly initial, as in (48), and sometimes before swa þeh. There are no forms of eke, but eac occurs 203 times, as in (49), (and in 10 of these, swa follows, already pointing to a grammaticalization of also):

49. Wulfstan Homilies, p. 221

Hi namon eac him ða þat to wisdome þurh deofles lare

They took also him then that to wisdom through devil's lore

However, unlike ac, which always occurs clause-initially, eac frequently occurs in the `middle' of a sentence. In fact, only 45 (=22%) occur after &/and, and 28 (=14%) occur at the beginning of a sentence. That leaves 130 (=64%) sentence-medial, as in (49), i.e. possibly as MP. Checking Alfred's PC shows that eac is quite frequent (i.e. 342 times), that many occur medially, namely 112, not counting when they occur after initial adverbs and conjunctions, e.g. ac, ne, for ðæm, oft). Many of the medial ones, however, still have the adverbial meaning, as in (50):

50. PC 163 1-3

Eac often occurs together with other adverbs, which would mean it might be a MP. Thus, oft eac occurs 20 times; sua eac 19 times; ge eac 10 times; ne eac 9 times; hwilum eac and ac eac 5 times each. It also occurs in possible MP environments such as wh-interrogatives, three times, as in (51) and (52), 10 times after gehi(e)ren, as in (53), and 5 times after pre-modals, as in (54):

51. PC 165.3

Hwa bið medtrum, ðæt ic ne sie eac for his ðingum seoc?

who is sick that I not am also on his account sick

`Often also we can better teach the proud as we wish'.

The meaning is predominantly adverbial though. Ac occurs 496 times and is always clause-initial.

In the Helsinki Corpus ME1 (from 1150-1250) section, there are 154 acs (many ac gef `but if'), 62 ecs, 35 eacs, and 2 sentence-final ekes (Caligula has the first). The distribution of ac is mainly sentence or phrase initial (=91%); that of ec as well (=71%), except for 9 sentence-final (!6 from Ormulum) and 9 medial, and ec may be `picking up' some of the `eacs'. Eac occurs less frequently (!19 initial and 16 medial) but `feels' definitely modal, as in (55) and (56) from C12:

55. Bodley Hom, p. 98

Swa eac nu mæg ealc mon deofel ofercumen

so also now may every man devil overcome

56. Idem, p. 106

hit is nu eac on þare niwæ læge æfter bocæ tæcinge rihtlic.

it is now also on the .....

In (56), the 2 MPs, if that is what they are, occur in the same place as in Modern Dutch.

In ME2 (from 1250-1350), there is a total of 171 acs; they are all sentence-initial, and function as complementizers. There are 34 eks, of which 16 initial, 5 final, and 13 medial, as in (57). The latter are not stacked, as in (56), and don't function as MPs:

57. Robert of Gloucester 2925

& ibured was ek þer

`and buried was also there'.

58. Idem 2941

Elianore of brutaine. deide ek þulke 3er.

`E died also that year'.

There are 10 ekes, 9 of which occur after `and' and function as coordinators; 1 is clause-final.

In ME3 (from 1350-1420), 37 eek, 13 ek, 15 eke, almost all after `and' (but see table 5 for details). However, in Chaucer's works (parts are included in ME3), variants of `eke' occur frequently (namely 500 eek of which 286 after and, 266 ek, 236 eke, of which 113 after and, and 1 eeke) and occasionally, as in (59) to (62), the form is possibly an MP:

59. Clerke's Tale, 1205

In jalousie I rede eek thou hym bynde

`I suggest you bind him in jalousy'.

60. p. 679

And tak eke anon the altitude of any sterre fix

61. WoB 14

Herkne eek, lo, ...

62. Thou shalt eek considere alle thise causes

In ME4 (1420-1500), there are 23 ekes, but they have become real adverbs with the meaning `also' and only 4 follow and. In EMODE1 (1500-1570), there are only 2 ekes left after `and' (in Roister Doister by Thomas Udall), probably also has `taken over'. The numbers are summarized in Table 4:

ac ec ek eac eke eek

OE4 265 - - 203 - -

ME1 154 62 - 35 2 -

ME2 171 - 34 - 10 -

ME3 - - 13 - 13 37

ME4 - - - - 23 -

EMOD1 - - - - 2 -

Table 4: The use of `ac', `eac' and its variants in the Helsinki Corpus

Thus, in OE and ME, ac and ec are complementizers and eac is a manner adverb and possibly an MP. Eek/eke are `successors' to eac, but even so many occur after and, suggesting the MP is located in the periphery. In table 5, I have shown the variants of eke in sentence-medial position. By the end of ME, the use is very rare:

eac ec ek eke eek

OE4 130/203 (=64%) - - - -

ME1 16/35 (=46%) 9/62 - 0/2 -

ME2 - - 13/34 0/10 -

ME3 - - 2/13 0/13 10/37

ME4 - - - 14/23 -

EMOD1 - - - 0/2 -

Eac/eke/eek also has a competitor in also. The latter grammaticalizes during C12 from the adverb swa, reinforced by the degree particle al, and may have contributed to the demise of eac/eke. For instance, in Chaucer, there are 424 instances of also, 8 of al so, and one of all so. These often occur after and in the beginning (=80 times), or final, as in (63) and (64), which are unusual in that both eek and also are present:

63. 3631

and eek his wenche also.

64. 894

And eek the false empoysonere also.

Also is somewhat frequent after modals (=9 times), as in (65), and 10 times in existentials, as in (66), mainly to indicate change-of-topic:

65. Thou shalt also have in suspect...

66. Ther was also a Nonne, a Prioresse

So unlike though, eac/eke continues into the ME period, but dies out by lME, certainly as MP.

The Modern English MP now has been discussed in Aijmer (1988) and Schiffrin (1986). Kryk-Kastovsky (1996), in examining the historical developent of Germanic and Slavic `now', shows that in IE, `now' was both a temporal adverb and "a particle emphasizing requests or questions, expressing doubts or second thoughts" (p. 323). If this is true, it is not surprising to see `now' as MP in Middle Dutch, as in section 4 above. In this section, I provide some examples of MP use through the history of English.

Now is used in commands in OE, as in (67) and (68), from the OED. Looking through the 171 instances of nu in OE1-2 from Helsinki Corpus confirms the use of nu as MP. It often occurs in the `middle' in (69) to (71), unlike in Middle Dutch above:

67. Vesp Ps ii

And nu, cyningas, on3eotað ...

`And now, kings, learn/see ...'

68. Bl Hom

Cleopian we nu in e3lum mode ...

`Speak we now in loathsome manner'.

69. Alfred, Boethius XVIII, $ 3

þeah he nu maran wilni3e, he ne mæ3 furðun þæt forðbringan ...

`though he now more wishes for, ....'

70. Kath 977

þis is nu þe derfschipe of þi dusi onsware

this is now the strength of thy foolish answer

71. Hali M, Bodley l. 12

Ant hwet is nu þis lare, þet tu nimest se deopliche

`And what is now this teaching that you take so seriously'.

Looking in Chaucer, i.e. late ME, now occurs very frequently (i.e. 1409 times), and is often used as MP, as in (72) to (74). Most of the time, it occurs in the beginning of a clause, as in the first instance in (72) and in (73), but sometimes in the middle, as in the second instance in (72). At most they ccur with one other possible MP, e.g. but now is frequent (=72 times):

73. 3136

Now herkneth, quod the Millere, ...

74. 831

Lat se now who shal telle the firste tale

75. 2082

From a structural point of view, now triggers V-2 frequently when it is in initial (or semi-initial position), as in (76); subject first, as in (77), is less frequent (e.g. with first person singular, there are 23 instances of Now I Vfin... but 122 of Now Vfin I ... in Chaucer). This means now is already in Spec CP, and show that the loss of V-2nd was not directly responsible:

76. 1234

But now wolde I fayn that ye wolde condescende ...

77. 2343

In the remainder of this section, I show some of the characteristics of now in Modern English by looking at Mark Twain's Tom Sawyer. This text was randomly selected and the use seems quite `modern'. There are 284 instances of now. Very few of these, 10 to be precise, occur in the `middle', and the ones that do are temporal, as in (78) and (79). A few occur in expressions such as now and then and they are not counted as `middle':

78. He was now become to them ...

79. the doctor, who was now standing.

The vast majority have a comma on either side and occur at the beginning, or after another MP, as in (80) and (81):

80. Well, now, if I didn't think ...

81. Oh come, now, you don't mean to ..

Most of these are MPs, as in (82) and (83):

82. Now, children, I want ...

83. There, now, that's a good boy.

The ones at the end are often temporal:

84. it's not much of a secret now.

So now, unlike though, survives as an MP but already in ME, it is related to the CP layer.

Then is closely related to now (e.g. Schiffrin 1986: chap 10). In their adverbial uses, they are opposite temporal deictic markers; in their MP use, now indicates "speakers' attention to upcoming talk" (p. 317), and then "makes warranted requests". Then has received some attention in a historical context, e.g. see the review in Brinton (1996: 9-12).

According to the OED, then has been a `particle of inference' from OE on. However, there is only one OE example, as in (85). ME has more, as in (86), and from 1600 on, examples such as well then, and now then become frequent, often surrounded by commas:

85. Blickl. Hom 39

Us is þonne mycel nedþearf þæt we 3ebu3on to him.

us is then much need that we obey to him

`We need to obey him'.

86. Hali M, 5

Nis þenne sariliche ... akast

It is difficult to examine then/than in a corpus since the forms vary so and are used as adverbs as well as demonstratives. In the Helsinki Corpus, OE4 has 2 þanne, 3 þane, 272 þone, 249 þonne (and several dental fricatives followed by ænne/æne), and 1294 þa. ME1 has 231 þanne; 46 ðanne; 161 þenne; 3 ðenne (231 þanne but no þænne) and over 1100 þa. I will examine Chaucer's use of then, since the use of eke as MP occurs in Chaucer.

Chaucer's entire works have 41 then and 979 than; these are most often comparative. There are also 10 thenne and 19 thennes, mainly temporal. Thanne occurs 871 times, mainly clause-initial, but some are MPs, as in (87):

87. l. 2965 (p. 64)

But shortly to the point thanne wol I wende.

88. For thanne agayn ...

89. by this manere thanne, althoughe ..

So as with eke/eek, then occurs in Chaucer as MP. The word order facts are similar too, at least checking first person singular: 6 times thanne is followed directly by the subject I and then the finite verb, whereas it is followed by a finite verb and then I in 97 instances. This suggests then is connected to CP.

By Early Modern English, in the Helsinki Corpus MODE1), the use is much reduced. There are 573 instances of then and 5 of thenne (many comparative, and initial, no MPs), 288 than and 27 thanne, 1 thone. So, the use of then as MP is comparable to the use of eke/eek as MP: still used in Chaucer, but not much after 1500.

In the Modern English period, then occupies the periphery of the clause. For instance in Twain's Tom Sawyer, most of the 306 instances are sentence initial, as in (90), almost half occur after and, as in (91), and very few occur at the end, as in (92) to (94), and two in the middle, (95) and (96):

90. Then she had a new inspiration ...

91. and then you'll look back ...

92. What will you come for, then?

93. Ah, how would she feel then?

94. Why, don't you DO it then?

95. Is this, then, Life?

96. He then held a position at the gate for some time...

Apart from the sentence-final ones and one of the medial ones, these instances are not very MP-like and, in keeping with that, do not occur regularly with other particles, e.g. twice with well, once with well, and three times with but.

Then functions as MP in OE and ME but comes to be sentence-initial or final by Early Modern English.

So far I have looked at MPs that originate in temporal and aspectual adverbs and complementizers. It may be the case (Abraham, p.c.) that the development of MPs that originate as Focus Particles is different. In this section, I therefore look at degree/focus adverbs briefly.

Based on the Helsinki Corpus, Traugott (1998) shows that in the period up to 1500, there are three clusters of meaning for even: `smoothly, in equal degree, and exactly'. Even though scalarity is involved, these meanings are not scalar/focus, according to Traugott, and focus even is said to occur only from 1570 on. Traugott does not mention the MP use and checking even in the 19th century, e.g. Emma, only provide instances of focus, with the Dutch meaning zelfs:

97. Emma, ch 8

"Even your satisfaction I made sure of".

According to the OED, the use as FP did not arise before C16. One of the early examples is from Shakespeare's Timon. This is certainly in accordance with Traugott's observations:

98. Timon, I, i, 82

Make sacred euen his styrrop.

The OED is not clear about the origin of even, whether it comes from `level', or has some unknown Teutonic origin.

Looking at table 1, it seems that in cases where MPs derive from words that also become FPs, the latter is always first. If even is not an FP in ME, one wouldn't expect it to become an MP either, certainly not in Modern English, where heads are not often used.

In summary to section 5, I have shown that MPs occur in OE and ME in a limited way. Most previous research has focussed on the pragmatic function of discourse markers in OE and ME, or on complex markers such as in fact, indeed, certainly. I have focussed on their position in the sentence and their status as heads, since that makes MPs in German and Dutch different from those in Modern English.

Abraham, as mentioned above, has related having a Mittelfeld to having MPs. Above, I show that certain MPs disappear by ME, certainly those in medial position. The loss of Verb-2nd, i.e. the loss of a possible Mittelfeld, is usally seen as a gradual process, starting in ME. The VO word order is introduced by 1250. So, the relationship is there but is difficult to pinpoint precisely. For instance, Chaucer is still quite V-2nd (as indicated a few times above), but not OV, and has some MP use. In the Early Modern English period, both overt V-movement and medial MPs die out.

Showing there is a possible correlation between the loss of MPs and the loss of a Mittelfeld does not account for this relation. It is possible to argue that the loss of Verb-second freed up the CP, and made it possible for MPs to be in CP. This cannot be correct since the structure of the CP was quite elaborate in ME, more so than in later stages of the language (when the language became more I-oriented). Thus, sentences such as (99) occur, with a structure as in (100), adapted from Rizzi (1997):

99. CM, Cotton, 105-6

Til all oure bale ai for to bete,

Oure lauerd has made þat maiden.

so all our sorrow always for to heal our lord has made that maiden

`Our lord has created that woman in order to heal our sorrow forever'.

100. ForP

. For'

For TopP

till . Top'

Top FocP

all oure . Foc'

bale Focus FinP

ai . Fin'

Fin IP

for . I'

I VP

V t

to bete

Sentences such as (99) show that around 1300 the CP was more elaborate (or certainly as elaborate) than it is at present. That leaves us with the question why sentence-medial particles disappear.

Another explanation is to build on the insights that stacking gives us, namely that they show a TMA sequence, as in (11) above, and that their loss coincides with the grammaticalization of auxiliaries. However, as we'll see this runs into theoretical problems. Abraham (1995; p.c) argues there are at least 3 functional MoodPs above the VP as well as TP and ASPP to accomodate MPs in German. If this is true in Middle English, positions such as TP and ASPP will be taken over by grammaticalizing auxiliaries (see Lightfoot 1979, Traugott 1972 for changes in auxiliaries). So, the development of auxiliaries banishes MPs to CP. This would also explain that FPs lag behind.

The problems with this account are theoretical. If MPs are heads (and they look like heads), V-to-C movement should not be able to occur in Modern German and Dutch, since it would skip the MP heads. If MPs are heads that do not form barriers for head movement, the difference between English and Dutch would be that the former uses CP more for expressions of mood since the head positions of the TMA categories can be filled by auxiliaries. Some evidence for this account can be found in Chaucer. Chaucer starts combining modals and perfect and passive auxiliaries, but their combination with MPs is interesting, in that the original manner adverb ek/eke always follows the perfect and modal auxiliary. It never precedes, as expected if MPs occupy specific FCs, and manner is aspectual an close to the V:

101. I have ek understonde ...

102. myghte eke fallen ...

103 shal eke ...

Though never occurs after an auxiliary since it is originally a sentence adverb or complementizer, i.e. higher. The temporal adverb now mainly precedes the auxiliary in Chaucer. For instance, there are 40 instances of `now wol I/ye' but only 3 of `I wol now'. Looking through the 1411 instances of now, this ratio seems representative. This indicates that unlike ek/eke which is a manner adverb, now may be being pushed out since tense has to be filled.

If there is a problem having MPs as heads, why not have them as specifiers? This would be a possibility but then the CD between emerging auxiliaries and disappearing MPs would be lost. It would also be problematic for asymmetric approaches.

In section 5, I show that there are quite some MPs in OE and ME. A lot of variation exists, e.g. though in OE; eac in OE and eME; now and then throughout the history of English. The position of the elements is interesting, e.g. in Chaucer many are already clause-initial, but by the modern period, they are dominant in peripheral position. This is quite unlike the other Germanic languages. So, the answer to the question posed in the introduction is that older varieties of English do have MPs in a way similar to other Germanic languages (namely clause-medially). MPs are not stacked as much as in Modern Dutch (but Middle Dutch didn't either and that may be the problem with lacking spoken data). They also occur quite a bit on the periphery already (but then Middle Dutch did too).

As to being able to correlate the loss of Verb-2nd, i.e. the loss of a possible Mittelfeld, this is difficult to say with certainty. Chaucer is still quite V-2nd (as indicated a few times above), but not OV, and has some MP use. By eModE, both overt V-movement and medial MPs die out. An explanation I have suggested but which faces numerous problems is to relate the stacking of the auxiliaries in Modern English with the loss of medial MPs.

- (ed.) 1991a. Discourse Particles. Benjamins.

- 1991b, in Traugott & Heine.

Altmann, Hans 1976. Die Gradpartikeln des Deutschen. Niemeyer.

Bolinger, Dwight. 1972. Degree Words. Mouton.

Brinton, Laurel 1996. Pragmatic Markers in English. Mouton.

Conradie, Jac. 1993. "On Subjectification of Modal Adverbs", ICHL talk.

Foolen, Ad 1993. De Betekenis van Partikels. Nijmegen diss.

Gazdar 1979 see Bayer's biblio

Helbig, G. 1988. Lexikon deutscher Partikeln. Leipzig: VEB Verlag.

Jacobs, J. 1983. Fokus und Skalen. Niemeyer.

Kerkhof, J. 1966. Sts in the Lg of Geoffrey Chaucer. UP.

König, Ekkehard 1991. The Meaning of Focus Particles. London: Croom Helm.

König, Ekkehard & Detlef Strang 1991. in Abraham

Luke, Kan Kwong 1990. Utterance Particles in Cantonese Conversation. Benjamins.

OED

Östman, Jan-Ola. 1981. You know: a discourse functional view. Benjamins.

Schiffrin, Deborah 1987. Discourse Markers. CUP.

Swan ** 1988 in Neval.

Sweet, Henry 1871. Gregory's Pastoral Care. London: OUP [EETS]

Traugott & Heine 1991. Grammaticalization. Amsterdam: JOhn Benjamins.

Underhill Robert 1988. "Like is, like, focus". American Speech 63: 234-246.

Weijnen, A. nd. Zeventiende-eeuwse Taal. Zutphen: Thieme.

Weydt, Harald (ed) 1979. Die Partikeln der deutschen Sprache. Berlin.

- 2000. "Collocational behavior in the modal domain", talk DGfS.

In the 15th century Paston Letters, for instance, about is a temporal and spatial preposition, as well as a preposition and particle with a verb, but is not really a MP:

104.

The OED shows it is aspectual from the 16th C on, e.g. in (100):

105. 1535

They were aboute to go for to descrybe the londe

Four hundred years later, in Austen's Emma, the use is exactly the same. The reason it did not grammaticalize into an MP is because of its aspect function (He about went there):

106.

Sooth as adverb from Beo on; as interjection in the OED CM 7739, and Malory, Shakespeare.

True starts out in OE as being reliable of a person (1303 Handl S þat loueþ trew), then a statement , not till 1594 does `it is true ...' occur.

Already, yet and still

All three refer to actions that go on: typically, already to the past, yet to the unrealized, and still to the present. Of the three, ...

Already is in group 2 since it is a compound of all and ready and likely to be grammaticalized. In the 1380s, it means `complete' and is usually written separately, as in (85) and (86), or `by this time':

107. Chaucer (Harl MS) Wife's Tale 169

Al redy was his answer

108. Chaucer, Astrolabe ii, 11

... ben departid by 15 degrees al-redy.

It is used in non-standard, according to the OED, "in Yiddish influenced speech", as in `Enough, already', and in Afrikaans-English. In the PL, there are 2 occurrences, as in (87):

109. all redy to haue goon to the barge

In Emma, the use is as in Modern (standard) English:

110. Frank had already written.

[ADD: Traugott & Waterhouse (1969), already and yet, but not historically. Underhill like: focus, softener, C: CP is relevant]

The use of yet as an MP is quite recent. According to the OED, it means `in addition' or relating to time in OE but is of obscure origin, expressed as *noh in some of the other languages. An early example is (41):

111. Beo 47

Early was the use to strengthen a comparative, now taken over by even and still, and the use with a negative. There are interesting eME examples of as yet and of the use as a conjunction. Ironic use is from 1936 on "imitating the Yiddish noch". Looking at the PL, yet is used primarily after a negative, having temporal meanings, as in (42), and as a C, as in (43). The use of (44) is rare:

112. PL but sche is not yet come to Norwich

113. PL yet I made an appell

114. PL there thei kepe hem yet wyth-ought any warant.

Four centuries later, in Emma, the use seems no more MP than in the PL. It seems used less after negatives, still as a C, and sometimes as an intensifier as in (85):

115. was yet a dearer, tenderer recollection.

Still has a temporal meaning in Modern English, as in (51), but is used as a discourse marker in (52):

116. Mr Elton was still talking.

Still drives from the Adjective stil, meaning `quiet'. The adverb starts out as a manner adverb refering to noise (first use Layamon), then extends to motion and to action/condition around 1300, to time around 1500, and is used as a FP in Spenser's FQ. The first use of it as a possible MP is a1699, as in (53):

118. ... and yet be good still?

The PL shows an interesting collocation of after all. It is used only 4 times but always before `due' and of those 3 are `due recommendation' and one is `due reverence and recommendation'. So the use is quite grammaticalized.

Grammaticalization from tense/aspect to speaker oriented:

In Brinton's intro, she discusses:

her, nu: text deictic

þa

this

OED VI 23a has OE speaker acceptance of the situation, i.e. MP, but always at the beginning.

Verdam 1911 al < al `langs' (see Barbiers 57)

nog < nu (tot nog toe) (Franck 1936, see Barbiers 57)