Being on the edge: the Force of at and till


                                                                 Elly van Gelderen

                                                            Arizona State University

                                                           ellyvangelderen@asu.edu

 

                                                      DGfS, 2 March 2000, Marburg

 

 

In this paper, I have two goals, a descriptive and a theoretical one. First, I describe the extent to which at and till are used in northern texts and what their categorial status is. Second, I examine these changes from the point of view of a split CP, as in Rizzi (1997) and Cinque (1999). I argue that even though Modern English does not provide evidence for a split non-finite CP, there is some evidence in Middle English (perhaps related to the C-orientation). Finite CPs show a split in all stages. I also claim that there is a reorganization of e.g. finite that/for from Fin to Force. This is probably related to features such as future/purpose, but this remains for further research.

 

 

1          Rizzi and Cinque

 

Rizzi (1997: 283) argues that "the complementizer system [is] the interface between a propositional content (expressed by the IP) and the superordinate structure (a higher clause or, possibly, the articulation of discourse, if we consider a root clause)". Therefore, the CP contains elements that look outside which he calls Force, and those that look inside, which he refers to as Finite. In addition, the CP (optionally) accomodates Topic and Focus. The structure of such an articulated CP is given in (1):

 

1.         ... Force ... (Topic) ... (Focus) ... Fin IP

            Rizzi's (8), p. 288

 

In Italian, the finite complementizer che is in Force but the infinitival di occupies Fin. The evidence for this comes from topicalization and left dislocation. The finite complementizer precedes the Topic, as in (2), whereas the infinitival one follows, as in (3), both from Rizzi (1997: 288). In (3), di could be in I, but Rizzi does not discuss that possibility:

 

2.         Credo che il tuo libro, loro lo apprezzerebbero molto

            believe that the your book they it appreciate much

            `I believe that they would appreciate your book a lot'.

3.         Credo, il tuo libro, di apprezzarlo molto

            believe the your book for appreciate-it much

            `I believe to appreciate your book a lot'.

 

            The English finite complementizers that and for behave like che, as (4) (sentence is taken from McCloskey 1991) and (5) show:

 

4.         She maintains that Irish stew she sort of likes t.

5.         ... for Irish stew I sort of like.

 

This means that and finite for are in Force not in Fin[1]. Historically, this was probably always the case for for, which is first attested as a finite complementizer around 1200, according to the OED, as in (6) and (7), but not for that, which was Fin probably till about 1500, as we'll see:

 

6.         York, 11, 286

            For hardely I hym heete

            for indeed I threaten him

7.         10,12

            For frenshippe we haue foune.

 

            The infinitival complementizer for in (8) is similar to Italian di in (3) in that a topic cannot follow it. The reason is that for needs to be adjacent to the subject for Case reasons. However, it is unlike di in that a topic cannot precede it either, as the ungrammaticality of (9) shows. This is unexpected if for is in Fin. Rizzi does not discuss for:

 

8.         *I expect for her homework her to do.

9.         *I expected her homework for her to do

 

If we test the position with other non-finite complementizers, the result is the same, i.e. topics cannot precede them:

 

10.       *He swam the chanel i[to England] in order to get ti.

11.       *I wonder i[to Alaska] whether to go ti

 

Thus, non-finite complementizers in English do not fit readily in the expanded CP.

            Cinque (1999) argues that CP and IP do not suffice because adverbs need to be accomodated. The CP needs to accomodate speech act, evaluative, and evidential adverbials (honestly, unfortunately, evidently respectively), and modal affixes in certain languages. Cinque does not use Rizzi's categories, unfortunately, and does not test the compatibility of these adverbs with topics and focus. Even though (12) seems slightly odd, it is acceptable to native speakers with honestly in ForceP and those books in the topic. However, (13) is a problem since, in Cinque's approach, frankly is higher in the tree than surprisingly, but the two cannot occur together as in (13) (Cinque's examples have the second adverb inside the IP):

 

12.       ?Honestly, those books, he should have read before class.

13.       *Frankly, surprisingly, he read those books.

 

This suggests that the English CP may be restricted. Checking some literary works, no combinations such as (13) are found. Thus, matrix CPs such as (12) and (13) show that the CP contains a ForceP with the sentence adverb and a TopicP with a topic.

            Cinque says very little about subordinating conjunctions, which is surprising since these originally motivated the use of C and CP (Chomsky 1986). However, testing what can come after (or before) a subordinating conjunction provides the following surprising results:

 

14.       a.         *I know that frankly she left

            b.         *I know frankly that she left.

15.       *I know that surprisingly he left.

16.       ?I know that allegedly he left.

 

It seems that that is in complementary distribution with the adverb. If that is in the Head of ForceP in (14b), the adverb cannot be in the Specifier (perhaps in keeping with Koopman's 1996 generalized doubly filled COMP constraint). The data are confirmed by using electronic corpora, e.g. in Jane Austen's Emma, happily can only occur in the CP domain, as in (17), if that is not present. In cases when other CP material is present, as in (18) and (19), it does not occur at the beginning but after the subject in the IP:

 

17.       Emma, ch 5

            Happily he was not farther from approving matrimony than ...

18.       Emma, ch 24

            It is what we happily have never known any thing of.

19.       Emma, ch 4

            As I am happily quite an altered creature

 

            Epistemic adverbs (e.g. probably) are in the IP domain, as (20) and (21) show:

 

20.       *I know that probably IP[he left].

21.       I know that IP[he probably left].

 

The IP also needs to include three tenses, an irrealis mood, and countless aspects (see e.g. Cinque 1999: 77), but these are left out of this discussion.

            Accordingly, the complementizer that is in complementary distribution with speech act adverbs (frankly in (14)) and evaluative adverbs (surprisingly in (15)), expected if both are in ForceP. It is marginally in complementary distribution with the evidential ones (allegedly in (16)). That expresses Force but, against Cinque, a lower CP category as in (14a) cannot be filled by an adverb, or the specifier of Force as in (14b). Sentence (21) shows that the complementizer is not in complementary distribution with the epistemic adverb, expected if the latter is part of the IP domain. Putting these facts together with those in (4), the English finite CP complex includes a ForceP and a TopP.

            The same complementarity between sentence adverbs and the complementizer can be seen with non-finite complementizers, as shown in (22) and (23). These sentences are only comprehensible if the adverbs are seen as manner adverbs modifying the subordinate VP, not as speech act or evaluative ones:

 

22.       *He made preparations in order frankly to invade the country.

23.       *He made preparations in order surprisingly to invade the country.

 

            Because of the existence of this complementarity and the incompatibility of topics with for, as in (8) and (9) above, I assume Modern English non-finite CPs have only one CP. For older varieties, I will argue that Rizzi's structure provides some insight. English finite CPs are split into a ForceP and TopicP.

 

 

2          At and til: general background

Throughout the history of English, the meaning of prepositions changes from concrete to abstract, e.g. from spatial to temporal to future tense. As a result, they are reanalysed as complementizers and infinitive markers. This happens with at as well, as I describe below. The situation with till is more complex. The use of at and till as I and C has not been discussed much; their categories not either. Strang (1970) just cites Mustanoja; Mustanoja (1960: 348-9; 362-427) mainly gives prepositional uses and refers to the MED for more. Mitchell (1985, I: 497ff) provides lists of prepositions and the cases. Bosworth & Toller (1898), and Baugh & Cable (1993: $77 and Appendix A) indicate northern influence.

            Cognates of at can be found, for instance, in Latin ad, Sanskrit adhi, and Scandinavian att. In the Scandinavian languages, att/å is both an infinitive marker and complementizer (e.g. Platzack 1986); and cognates of to are no longer found (see Norde 2000 for the grammaticalization of at in Scandinavian). In Modern Dutch and German, the situation is the reverse: cognates of to occur (te and zu) but cognates of at do not. In Modern (standard) English, to and at occur both but in differing functions: to is used as a directional preposition, as well as a Case marker, and a functional category; the use of at is much more limited and very complex: "a classification of its uses is very difficult" (OED entry for at), and "[o]wing to the subtlety of its meaning, at is perhaps the most troublesome preposition for foreign learners" (Lindstromberg 1997: 165). The typical Old English usage of at is as a preposition expressing a spatial relation.

            Regarding till, the OED mentions that the origin is probably a noun, with the meaning of `fixed point', cognate with German Ziel. The use of till as preposition is rare in Old English, e.g. Beowulf has none, and in the OE part of the Helsinki corpus there may be 4 instances whereas there are over 6000 instances of to. Where we now use (un)til, OE uses oş şæt (see Mitchell 1985, II: 388ff). Most of the examples in the OED are from Cursor Mundi, Ormulum, Rolle, and Hampole, i.e. from Northern texts. Bosworth & Toller (1898) mention explicitly that it occurs only in northern texts. Mustanoja (1960: 408-9) says that it is "typically northern", but has been in general use since c1300. A famous instance is from the Northumbrian version of Caedmon's Hymn, listed as (24) below:

 

24.       heben til hrofe

            `heaven for a roof'

 

            Of the use of till as an infinitive marker, the OED says that this use was not present in Old Norse. This is surprising if it occurs in Northern texts, i.e. where Norse influence was most pronounced. The use could have been extended during Old English. The examples given in the OED start with a version of CM, namely (25), which is curiously one of the Midlands` versions, not a northern one:

 

25.       CM Fairfax Ms 12988-9 (other 3 have for to)

            me şink şe ne ha3t haue na doute

            for tille be myne vnderloute

            `I think you ought to have no doubt to be my subject'.

 

3          Northern: Lindisfarne, Cursor Mundi, St Benet, and the York Plays

In the oldest OE Northern text, the Lindisfarne version of Matthew (c950 from Northumbria), there are 4 instances of prepositional for, 1 of til as a preposition indicating direction, and 1 of til as an infinitive marker. There are no instances of for to/forto and of at/t. Thus, to is the regular preposition and infinitive marker; the use of for and til is exceptional, and at does not occur, indicating that at and till are not characteristic of this stage of northern Old English. In terms of a structural analysis, some questions arise. There is never an overt subject to the infinitive or an overt complementizer. One could argue that C already expresses purpose and that to occupies C. This is unlikely since to is never split from the verb. Lindisfarne therefore does not shed much light on a split non-finite CP.

            There are a number of versions of Cursor Mundi. Morris's (1874-1893) edition has 2 Northern and 2 Midlands versions and hence these can be compared as to the use of infinitival markers. The Cotton Vespas. A iii is a northern text from the early 14th century (e.g. retention of -ande and of -a-; no palatalization; the use of qu- for wh-) from the early part of the 14th century and is written in 3 different hands.

            In the pages selected, there are 5 instances of til/l as preposition and one, as in (26), which is a complementizer. It could be that in (26), till is in Force and all oure bale in Topic:

 

26.       CM, 105-6

            Til all oure bale ai for to bete,

            Oure lauerd has made şat 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'.

 

I come back to this in connection to forto. In the rest of the text, till also functions as infinitive marker, as in (27):

 

27.       CM 5330 Cotton

            To lith a quil his word til her

            to listen a while his word to hear

 

In the other versions, til is replaced by at (Fairfax) and by to in the other two.

            Forto functions as an infinitival marker 7 times, as in (26) above. Typical for for to is that an object precedes it, as in (26) and (28):

 

28.       CM, 1

            Man yhernes rimes for to here

            `People yearn to hear rhymes'.

 

These constructions perhaps provide evidence for a split infinitival CP, as in (29) and (30), and they continue to occur well into ME. (However, (29) and (30) may not be correct since a subject never occupies Spec IP, if there is a topic):

 

29.                   ForP

            .                       For'

                        For                   TopP

                        till         .                       Top'

                                                Top                  FinP

                                    all oure .                       Fin'

                                     bale                             Fin                   IP

                                                                        for        .                       I'

                                                                                                I                       VP

                                                                                                                        V

                                                                                                to                     bete[2]

30.                   TopP

            .                       Top'

            rimes    Top                  FinP

                                    .                       Fin'

                                                Fin                   IP

                                                for        .                       I'

                                                                        I                       VP

                                                                        to                     V

                                                                                                here

 

A similar construction occurs in the two northern versions, namely (31ac), replaced by (31bd) in the other two:

 

31.       a.         CM 18441-4 Cotton

                        fader ... grantes til his, Sinful men şaim for to ledd In paradise

            b.         Idem, Fairfax

                        ffader ... sent to his Synfull men for to thus lede In paradice

            c.         Idem, Göttingen

                        Fadir ... sent till hijs, Sinful men şaim forto lede In paradis

            d.         Idem, Trinity

                        Fadir ... sent to his Synful men for to şus lede In paradis

 

            For is either a preposition (8 times), or a complementizer (11 times). There is, however, an interesting construction where for is a non-finite complementizer, and at an infinitival marker, as in (32):

 

32.       CM, 232; 236

            şis ilk bok is es translate ...

            For the commun at understand

 

The structure would be as in (33), very similar to (29) and (30):

 

33.                   FinP

            .                       Fin'

                        Fin                               IP

                        for        .                                   I'

                                    the commun      I                       VP

                                                            at                     V

                                                                                    understand

 

Visser (pp. 1039-40) lists a number of instances, especially in the Northern Cotton and Midlands Fairfax where to and at are separated from the infinitive by a pronoun or nominal object, as in (34ab), (35ab), (36a), and (37ab). The Göttingen and Trinity versions have to immediately before the infinitive in (34) to (36):

 

34.       a.         CM, 7746 Cotton

                        Yee send a man at it receiue

            b.         Idem, Fairfax

                        3e sende a mon atte hit resceyue

            c.         Idem, Göttingen

                        3e send a man it to rescayue

            d.         Idem, Trinity

                        sendeş a mon hit to resceyne

35.       a.         CM, 8318 Cotton Fairfax

                        To temple make he sal be best

            b.         Idem, Fairfax

                        to temple make he sal be best

            c.         Idem, Göttingen

                        şe temple sal he make wid ese

            d.         Idem, Trinity

                        şe temple bi him made shal be

36.       a.         CM, 23784 Cotton

                        We fine noght to it al fulfill

                        `we end not to fulfill it all'.

            b.         Idem, Fairfax

                        we fine neuer hit to fulfille

            c.         Fine we neuer to ful-fill

            d.         Hit to fulfille we fuyne nou3t

37.       a.         CM, 27363 Cotton

                        şar nan has might to oşer boru

                        `that noone has a right to rescue the other'.

            b.         Idem Fairfax

                        şat nane has mi3t at oşer borou

            c.         missing

            d.         missing

 

Since most of these occur in Cotton and Fairfax, the use is not arbitrary, but indicative of a structural difference. In van Gelderen (1993), I argue that split infinitives are indicative of having to and at in I. There, I argue that I is only activated later that century. The evidence in (34) to (37) shows that to is in I at least half a century earlier. These data fit well with earlier trees, with the addition that the object is in Spec AGRoP.

            In summary, in CM, there is some evidence for a split infinitival CP. Till may be in Force and for in Fin, e.g as in (26). At and to are infinitival markers, most likely in I.

            In The Rule of St Benet, a Northern ME text (e.g. third person s, lack of palatalization in sal `shall' and mikil `much') from the early fifteenth century, att is only used as a preposition, as in (38). At is used as a preposition, as in (39); an infinitival marker, as in (40) to (44); a finite complementizer, as in (45); and a relative marker, as in (46). This difference in function between at and att is an indication that phonological reduction goes hand in hand with grammaticalization, as mentioned above. As to the kind of infinitive at introduces, it can be purposive in (40) but is most often used after semi-auxiliaries, as in (42) to (44):

 

38.       Benet, 145/6

            Att şe bygynnyng of şe mese şe madyn şat salbe mayde nun sal sit in ...

            `At the beginning of the mass the maiden that is to become nun shall sit'

39.       Benet, 1/13

            it is to obey hym at al tyme

            `It is to obey him all the time'.

40.       3/17

            şus kennis ure lauerd us his werkis, at folu his cumandementis

            `In this way, our lor teaches us his works, (in order) to follow his commandments'

41.       2/31

            şe riht şidir at cume şai sal haue

42.       5/4

            wat man az at be abot

            `what kind of man ought to be abbott

43.       5/11

            He ah na şing at cumande

44.       11/10

            yef I leuid at se to mekenes

            If I cease to see to mekeness

45.       4/22

            şanane byhouis şam fezte şam ane, at god es tar best help

            `for this reason, it is necessary to fight them alone, that God is their best help'.

46.       şe preste at makis hir sal sai or sing

            `the priest that makes her shall say or sing'.

 

            Till, tyl, tyll in St Benet are used as prepositions and postpositions. Til is not only a preposition and postposition but also a complementizer, as in (47), but this use is rare:

 

47.       21/12-3

            Wylys sho is in sentence, sho ne sal noht be in cuuent, til it be amended.

 

Thus, as expected, the phonologically reduced form is more grammaticalized. This is also true for att and at, shown above. So, in St Benet, at is in I and til is a finite complementizer.

            The York Plays are written down in a mid C15 ms, and, as the place name implies, quite northern (e.g. sal). What is interesting is that at has `lost the battle' to for to and to as complementizer and infinitival marker. There are 72 instances of till and 77 of tille, and one marks an infinitive, others are adverbs, prepositions, and finite complementizers. The majority of at is preposition or infinitival marker. In York, the infinitival marker at is (once) separated from the verb, as in (48), indicating its categorial status is unclear:

 

48.       York 1.20

            Baynely in my blyssing at here be

            Willingly in my invocation to be here

 

            What is the status of for to? Objects optionally precede it, as in (49) and (50), indicating that for may be in the expanded CP of Rizzi. Occasionally, a construction as in (52) to (54) is encountered, but without an object preceding it:

 

49.       York 5.20-1

            Myght Y hym so betraye

            His likyng for to lette

            `Might I betray him so to put an end to his liking'.

50.       York 7.24

            That place of price for to restore.

51.       York 7.38

            Me for to were fro warkes were.

52.       York, 10. 139

            I wolde be glade for hym to dye

53.       York 19.236

            I schall nott lette for nothyng

            To saie as we haue sene

            `I shall no allow for ...

54.       York 21, 144

            For the to touche haue I grete drede

 

            Till is used in a variety of ways. It is most often an adverb, preposition, and finite complementizer, as in (55); only once an infinitival marker, as in (56):

 

55.       York 16G.267

            Go we nowe till şei come agayne.

56.       York 9.132

            For till accounte it cost şe noght.

 

            In summary to St. Benet, as in CM, there is evidence that non-finite for is in the head of the FinP. At and to are infinitival markers, and till is used as finite C and adverb mainly. The latter already shows the modern situation.

 

 

4          Non-Northern

In the early 13th century Caligula version of Layamon's Brut, i.e. Early Middle English from the Midlands, till does not occur, and the use of at is restricted to indicating places, as in (57), and activities, as in (58). There are 106 such instances (in Otho, there are 116):

 

57.       Caligula, 3

            He wonede at Ernle3e

58.       4591

            biræiuie hine at liue

            bereaved him of living

 

            For precedes the subject in Caligula, as in (59), but there is no topic in the same sentence to check if it is in Force or Finite:

 

59.       Caligula 1554

            for hire fader heo scunede

            `because her father shunned her'.

 

That is true with both finite and non-finite clauses. In Otho, to is sometimes in C together with for, sometimes it immediately precedes the infinitive. In the earlier Caligula, to is never separate from the infinitive. Whether with to or without, there is no evidence in Otho for for being in For or Fin. An interesting construction in both Caligula and Otho in finite clauses is (60), which occurs over 20 times, with either ne followed by the finite verb or contracted with it. This indicates for is perhaps higher than C:

 

60.       Otho 15772

            for ne may ich hit ileue

 

Infinitives follow for in Otho, as in (61), but it is not clear if that means the infinitve has moved to C or not:

 

61.       Otho 713

            for hunti deor wilde

 

            Thus, the position of non-finite complementizers is no different from Modern English. The finite complementizer for is quite different, i.e. it may be in ForceP, in (60). At and till are no longer used as complementizers or infinitive markers, the latter is in fact not used at all.

            Looking at Chaucer, who wrote in the Late Middle English period (lC14), at is not a complementizer or infinitival marker either. The only remnants of grammatical use are two instances of what look like an infinitive marker, as in (62), and a number, as in (63), with at preceding a verbal noun, expressing action in progress:

 

62.       Troilus & Cressida 4.1659 (also 3.1003)

            I kan na more; it shal be founde at preve

63.       at reste/at mete

 

Kenyon (1909), in his study of infinitives in Chaucer, does not mention this infinitival marker. The total number of instances of at is 921, as opposed to 9922 instances of prepositional and infinitival to. Thus, in non-northern texts even as early as 1250, the role of at is reduced to preposition and possibly aspect marker, not a C or I.

            The use of til/l is different. In Layamon, as mentioned, variants of (un)til/l do not occur. In Chaucer, even though there are fewer instances of till than of at (341 til and 16 till), they are often finite complementizers, as in (64) and (65). It is interesting that many tils are followed by that, as in (66), indicating till is in Force and that in Fin. The latter fits with having many wh that constructions as in (67):

 

64.       til we be roten l. 3875

65.       til she the cradel fond l. 4221

66.       til that he cam to Thebes l. 983

67.       She loved Arite so

            That when that he was absent any throwe,

            Anon her thoghte her herte brast a-two ll. 93-5

 

There are also 4 until/ls in Chaucer, both as C and P:

 

68.       until a smyth men cleped daun, l. 3761

69.       until to-morwe, l. 1070).

 

            Till has an initial purpose meaning even though that meaning is lost. This may indicate it is in the ForceP. Evidence for that is its use with that at least up until Shakespeare's time, as in (70) and (71):

 

70.       Macbeth 79

            Till that Bellona's husband, lapt in proofe

71.       Hamlet 3173-5

            Till that her garments, ..., Pul'd the poor wretch from her melodious buy, To muddy death.

 

This shows purposive till is in Force and that in Fin. In (72) and (73), till would also be in ForceP and from my bones and by some elder masters in Focus positions:

 

72.       Macbeth

            Ile fight Till from my bones my flesh be hackt.

73.       Hamlet 3700-2

            I stand aloofe ... Till by some elder Masters of knowne Honor, I haue a voyce

 

However, there is also evidence from the middle of the 15th century that that (by itself) is already in Force since it precedes a Topic, as in (74) and (75), but also from Shakespeare, as in (76):

 

74.       PL 4022

            And I told him that, as for such mony that shuld ..., I wold ...

75.       PL 7181/11230

76.       Hamlet 1009

            I am sorrie that with better speed and iudgement I had not quoted him

 

The number of wh that constructions is very limited, one in Hamlet, none in Macbeth.

            One scenario for the loss of the expanded CP with till and non-finite for is that till is reanalyzed from a purpose marker to a time marker, i.e. in the head of FinP) and that as a result till that is generally lost by 1500 (occurs in Shakespeare, ME4 in Helsinki but not in EMOD1). New purpose markers are slow: in order that/to only appear in 1711.

 

 

5          Conclusion

Historically, til/l starts out as a preposition (derived from a noun) and a purposive infinitive marker, but ends up as a finite temporal complementizer. Structurally, it shifts from Force to Finite, and its grammaticalization involves a shift of purpose features (in Force) to tense features (in Fin). That shifts from Fin to Force, which indicates degrammaticalization, but the motivation is not clear. At starts as preposition, becomes an infinitive marker and then disappears as such, perhaps because it is an aspect marker as well. For emerges as both a non-finite and finite complementizer, i.e. a clear C; and to an I. For some time, there is evidence that non-finite for is in the head of FinP, but then the non-finite CP is no longer split. In order to/that appears in 1711, according to the OED.

            As to a split CP, Modern English provides evidence that a finite CP is split, but not a non-finite one. In Middle English, the non-finite CP shows evidence of some split too. This development runs counter to other developments that show an increase in the number of functional categories in the history of English, see e.g. van Gelderen (1993; 1997).

            In terms of grammaticalization, the results are somewhat unexpected too: that switches from Fin in ME to Force in ModE, but finite for was always in Force. Non-finite for and finite till lose the ability to head a split CP. The latter may be related to the switch from C-oriented to I-oriented.

 

 

References

Abraham, Werner 1997. "The base structure of the German Clause", in German: Syntactic Problems. Werner Abraham and Elly van Gelderen (eds.): 11-42. Tübingen: Niemeyer.

Baugh, A. & Cable, Thomas 1993. A History of the English Language. ...

Beadle, Richard 1982. The York Plays. London: Arnold.

Belden, Henry 1897. The Prepositions in, on, to, for, fore, and æt in Anglo-Saxon Prose. Baltimore.

Bosworth, Joseph & T. Northcote Toller 1898. An Anglo-Saxon Dictionary. Oxford: Oxford University Press.

Carpenter, H. 1905. Die Deklination in der Northumbrischen Evangelienübersetzung der Lindisfarner Handschrift. Bonn.

Chomsky, Noam 1995. The Minimalist Program. Cambridge: MIT Press.

Cinque, Guglielmo 1999. Adverbs and Functional Heads. Oxford: OUP.

Demske-Neumann, Ulrike 1994. Modales Passiv und Tough Movement - Zur strukturellen Kausalität eines syntaktischen Wandels im Deutschen und Englischen. Tübingen: Niemeyer.

Dusenschön, Friedrich 1907. Die Präposition æfter, æt und be in der altenglischen Poesie. Kiel.

Gelderen, Elly van 1993. The Rise of Functional Categories. Amsterdam: John Benjamins.

- 1996. "The Reanalysis of Grammaticalized Prepositions in Middle English", Studia Linguistica 50.2: 106-124.

- 1997. Verbal Agreement and the Grammar of its `Breakdown'. Tübingen: Niemeyer.

- 1998. "For to in the History of English", in the American Journal of Germanic Language and Literature 10.1: 45-72.

Gottweiss, Reinhard 1905. "Die Syntax der Präpositionen æt, be, ymb in den Aelfric-Homilien und andern Homiliensammlungen". Anglia 28: 305-93.

Haumann, Dagmar 1997. Syntax of Subordination. Tübingen: Niemeyer.

Heine, Berndt, Ulrike Claudi & F. Hunnemeyer 1991. Grammaticalization. Chicago: University of Chicago Press.

Kenyon, John 1909. The Syntax of the Infinitive in Chaucer. London.

Koopman, Hilda 1996. UCLA ms.

Lindstromberg, Seth 1997. English Prepositions Explained. Amsterdam: John Benjamins.

McCloskey, Jim 1991. "Verb Fronting, Verb Second, and the Left Edge of the IP in Irish", ms.

Mitchell, Bruce 1985. Old English Syntax I and II. Oxford.

Morris, Richard (ed.) 1874-1893. Cursor Mundi, 7 Parts. Trübner & Co.

Mustanoja, Tauno 1960. A Middle English Syntax. Helsinki.

Norde, Muriel 2000. "The Grammaticalization of Adpositions in the Hsitory of Swedish", UvA ms.

Platzack, Christer 1986. "COMP, INFL, and Germanic word order". Topics in Scandinavian Syntax ed. by Hellan, Lars & Kristi Koch Christensen, 185-234. Dordrecht: Reidel.

Rizzi, Luigi 1997. "The Fine Structure of the Left Periphery", in Elements of Grammar. Liliane Haegeman (ed): 281-337. Dordrecht: Kluwer.

Strang, Barbara 1970. A History of English.

Visser, F. 1963-1973. An Historical Syntax of the English Grammar, Vol I-IIIb. Leiden: Brill.

 

Notes



[1].         Notice that, unlike topicalization in (4), left dislocation is only possible in a main clause as in (i) and not in a subordinate as in (ii):

 

i.          Those books, I read them.

ii.          *I know that those books I read them.

 

[2].         I have left ai out but it would be in Spec FinP.