Gelder en 2008

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

THE REANALYSIS OF

GRAMMATICALIZED PREPOSITIONS IN
MIDDLE ENGLISH
Elly van Gelderen

Abstract. During the Middle English period, a number of prepositions gram-


maticalize. I examine three of these (namelyfor(to), to and on) and argue that they
come to be associatmwith verbal features as well as future, tense and aspect. As
this happens, they change their structural positions and are reanalyzed as
C(omplementizer), I(inflection) and ASP(ect) respectively. The latter categories
are assumed by the language user as a response to the affiliation of the verbal
features of for(to),to and on.

0. Introduction
Reanalysis and grammaticalization are sometimes seen as two conflicting
theories of language change. Reanalysis (cf. Lightfoot 1979) is considered
essential in theories that focus on acquisition, for instance, in Government-
Binding (Chomsky 1981) and Minimalist (Chomsky 1995) frameworks.
Language acquisition is seen as the construction of a grammar by the child
with the help of (a) universals and of @) the languagethe child hears spoken.
As language changes, the grammar of the child will differ from that of its
parents. Grammaticalization, on the other hand, is a gradual process
through which words lose lexical meaning, morphological independence,
and obtain more grammatical function (e.g. Lehmann 1985, Heine et al.
1991,Traugott &Heine 1991, and,Bybeeetal. 1994).Heineetal.(1991)
argue, for instance, that through metaphorical extension grounded in
conceptual structures, a locational preposition acquires a temporal
meaning.Bybee et al. (1994) argues that a spatial notion implies a temporal
one and that in change,the former meaning is lost. The process is assumed to
be unidirectional. The relationship between reanalysis and grammaticali-
zation has recently been examined in, for instance, Haspelmath (1993) and
Abraham (1993) and this paper shows what happens syntactically to
grammaticalizingelements.
I examine changes regarding for(ro),to and on that can be seen as
responses (reanalyses) by the language learner to grammaticalization. My
aim is to trace what syntactic effects grammaticalization has. I will show

I An early version of this paper was presented at the Jahrestagung der DGR, Jena, 4
March 1993. I wish to thank the members of that audience for comments, in particular,
Ulrike Demske-Neuman and Olga Fischer. In that form, it appeared in Groninger Arbeiten
zur Germanistik 36. Section 2 was presented at the Germanic Linguistics Annual Con-
ference in Ann Arbor, MI, 22 April 1995, the members of whose audience I would also like
to thank.

Srudio Linguisricrr 5 0 ( 2 ) 1996, pp. 106-1 24. Q The Editorial Board of Studia Linguislica 1996
Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 lJF, UK, and
238 Main Street, Cambridge, MA 02142. USA
The reanalysis of grammaticalized prepositions 107

that, by the end of the fourteenth century, for and to are no longer
exclusively prepositions indicating location. They come to introduce
clauses and indicate future and non-realization, i.e. non-tense, respec-
tively. I argue they lexicalizecertain features (V-features in early Minimal-
ism; inherent categorial features in later Minimalism). The progressive
construction with on involves a similar reanalysis of on from preposition
to aspect indicator. In Old and Middle English, a progressive form is used
which is preceded by on/an. Initially, on is a preposition but by the time it
appears as a, it is an aspect marker. All three changes are metaphorical
extensions from the spatial to the temporal. My goal is to indicate what
happens to the syntactic representation.
The outline is as follows. In section 1, I outline the Modern English
structure forfor(to), to, and the progressive. In sections 2 and 3, I discuss
the changes in Old and Middle English concerningfor and to respectively.
Section 4 deals with on. In section 5, I conclude by arguing that
grammaticalization results in the acquisition of inherent categorial
features and of tense, future and aspect features. Thus,for to is listed in the
lexicon as inherently verbal as well as future: to as inherently tense and on
as inherently progressive. In an early Minimalist framework (Chomsky
19922),this means that these elements check the V-features. In a later
framework (19 9 9 , they check the categorial features.

1. Modern ‘standard’English
In this section, I indicate the Modern English structural position and
feature content offor(to) and to. I also provide an analysis of the progres-
sive because even though the progressive with on is no longer ‘standard’,
the modern construction bears some resemblance to the older one.
In Modern English, for is a complementizer in (1) and (2). In (l), the
for-clause is an adverbial clause and in (2), it is a complement clause. The
verbs that subcategorize for for- clauses are deontic ones, such as Eike and
want, indicating a desire on the part of the speaker:
(1) Sue didn’t walk for she felt tired.
(2) I would like for you to do some homework.
For in (2) introduces an actiodevent that must be future with respect to
the actiodevent in the main clause. I take this to mean that the tense of the
infinitiveis determined by for. There are verbs, such as believe and know,
that cannot be complemented by for and their tense properties are very
different. These are called epistemic verbs and indicate the belief or
knowledge of the speaker:

Chomsky (1 992) appears as a chapter in Chomsky’s The Mznimalist Program published


in 1995. I continue to use Chomsky (1992) since many changes occur between that paper
and the chapter in Chomsky (1995) that is called ‘Categories and Transformations’. The
latter I refer to as Chomsky (1995).

0 The Editorial Board of Studba Linguiatica 1996


108 Elly van Gelderen

(3) I believe (*for) her to be Zoya.


(4) I knew (*for) her to be Zoya.
In (3) and (4), the action of the infinitive must be at the same time as that of
the verb in the main clause. This is expected if the infinitives can only have
believe and knew as reference points (‘anchor’ in the sense of Enq 1987).
In cases such as (2) where for is present, to can take the future of for as a
reference point; when for is not, to takes the main verb as its reference
point. Both for and to play a role in tense marking because when neither
for nor to are present, the action expressed by the subordinate is realized
and cannot be untrue, as in (5) and (6):
(5) I hear Zoya leave the house.
(6) I made Zoya leave the house.
In (5) and (6), to is not present and the leaving cannot later be denied by
the speaker whereas in (2) the homework has not been done yet and in ( 3 )
the being nice can be denied at some later point. So, to in sentences such as
(3) and (4) contributes a sense of uncertainty because the action in the
main clause is separate from that of the subordinate clause.
As to the position of the two elements, for occupies the C position
because it precedes the subject and to is in a position separate from the VP
(cf. Akmajian, Steele & Wasow 1979). A structural representation is
provided in (7). In this tree, I(inflection) is used in accordance with
Chomsky (1 986, 1995) but one might call it the T(ense)/AGR(eement)
complex as in Chomsky (1992) or AUX as in Akmajian, Steele & Wasow
(1979). The reasons for assuming that to is separate are that the VP can be
deleted without to deleting, as in (8), and that the VP can be moved
without taking to along, as in (9):

C
I
for
/ \ IP
(\ , I*,

her .VP

to

v NP
I I
be =Ya
(8) Zoya tried to irritate Amir and Anji tried to also.
(9) and [read a book] Zelda tried to.

0 The Editorial Board of Studia Linguistica 1996.


The reanalysis of grummaticulized prepositions 109

In early Minimalism (Chomsky 1992), Functional Heads such as I


contain N and V-features and, depending on whether or not these features
are strong, trigger overt NP and V-movement. In non-finite clauses such as
(7), one might argue that I has strong V-features and that to checks these.3
In Chomsky (1995), the V- and N-features are replaced by categorial,
Case and phi-features. One could assume that to is in the lexicon with
inherent categorial (verbal) features, but without Case and phi-features. It
checks the categorial features of I. Chomsky is not very specific with
respect to the different verbal elements. I argue that each of these can have
other features accompanying them, e.g. tense. Regarding C, I argue that it
too has categorial verbal features that are checked by for. In addition to
being specified for verbal features, for is inherently specified for future
and hence can only appear as a C to 'compatible' verbs such as want and
like. Like to, but unlike 'real' verbs, for has no phi- or Case-features to
check.
The analysis of progressive in Modern English, in a Minimalist
framework (cf. van Gelderen 1996), is as in (lo), where is has moved to I
to check the categorial or V-features in I, and reading has moved to ASP
to check the V-features there, as well as its own (progressive) features:
IP

/'
I / "\
I
ttl / / ASP'
is ASP \ VP

t a book

Even though the construction with on that is discussed in this paper has
disappeared in Modern English, structure (10) is relevant for older stages
in that as on grammaticalizes, it is placed in the head of ASPP and reading
moves to ASP to check its features.

2. From Preposition to Comp1ementizer:for and for to4


In this section, I show that for( to) changes from a preposition to a Case
marker and to a complementizer and that future features come to be
In van Gelderen (1996), I argue that features are always strong even in tensed finite
sentences. 1 d o not go into this question here, however.
Fischer (1995) examines the difference between for to and to infinitivals on the one
hand and bare infinitives on the other.

0 The Editorial Board of Studia Linguistica 1996


110 Elly van Gelderen

associated with it. This is a process of grammaticalization which results in


certain features becoming linked (listed as inherent to them in the lexicon)
to a particular lexical element, e.g. for acquires future and categorial
verbal features.
Bybee et al. (1994:25ff.) would argue that the temporal meaning,
already implied, is emphasized. I m e i n e et al. (1 991), the meaning of for
is generalized to future by metaphoric extension of the locational meaning,
i.e. ‘in front of to the temporal one, i.e. future. My emphasis is on the
structural position of for.

2.1. For as a Preposition


In Old English, for is used as a preposition indicating location. This use
continues at least until the time of Shakespeare as (11) shows. In modern
English, infiont of would be used instead:
(1 1) All’s Well IV, iv, 3 .
For whose throne ’tis needful to kneele.
For is used in many different ways but one grammaticalized use is that of
causation. It is already present in Beowulf, as (12) shows. As is well-
known, for also comes to be used as a (benefactive) Case marker:
(12) BeowulfllO
ac he hine feor fonvrzc Metod for by mane mancynne fram,
and he him far banished. The lord for the crimes mankind from
’The lord banished him far from mankind because of his crimes’.
(from the Oxford English Dictionary entry for for)
The first uses of for and for to as complementizers that the Oxford
English Dictionary (hence OED) lists are from around 1200. Examples
are given in (13) and (14). This is no longer Old English and thus for Old
English, we can conclude that for is a preposition introducing NPs:
(13 ) Alcuin’s Virtuesand V. 115
for heo synd godes gescaefte,
‘because they are God‘s creation’.
(14) Trinity College Homilies 2 3
For to be time cam bat he . . .,
‘Until the time came that he. . .’
The use of for as in (13 ) is restricted to introducing finite clauses. This
restriction does not hold for for as I show in the next section, even
though the OED only lists instances of finite clauses.
Thus, in Old English, for is a preposition indicating location and
causation and for to does not yet occur.

’ 1 will not pay attention to the orthographic distinction between forfo and for to.
OThe Editorial Board of StudidLinguistica 1996
The reanalysis of gramrnaticalized prepositions 111

2.2. For as AGR@


In the early thirteenth century version of Layamon’s Brut, i.e. the Caligula
version, the temporal, or metaphorical extension, has started and for to
mrroduces purposive infinitives, as in (1 5) and (17) a few times. No
element intervenes between for to and the verb and for is never separated
from to by an object or adverbial:
Caligula, 929-30
Brutus hine Iette witen . . .for to leten fondien of his main stronge,
Brutus him let bind . . . in order to let find of his great strength
‘Brutus had him bound in order to discover his great strength’.
Cal. 15011
For to uzstnen pa luuen,
for to vest the love
‘in order to confirm the love’.
Cal. 7803
pat he ilad weore. limen for to leosen,
‘that he called (led to him) was, limbs for to suffer the loss of.
As to the structural position, for to is never separate from the VP. In van
Gelderen (1993), I have argued that this shows that for to is a prefix on the
infinitive. In the light of Kayne (1994), where the underlying structure of
language is VO and where the objects move to Spec AGRoP, these word
order facts could be captured if for to were a preposition responsible for
the Case of objects. The structure for (17) would be as in (18) and limen
moves to check its Case in Spec AGRoP.
(18)
/
AGRoP’


’AGRo
AGRo’

\“P

“fa V- NP
I i
[Case] leosen t

Since all the examples involve purposive infinitives, the meaning of for
to is to express that the action of the infinitive is in the future or that it is
unrealized. The event in the main clause is not (yet) taken as a reference

In this and the next sections, I use the TACT program to make concordance of the
Oxford Archive versions 3f Layamon, Piers Plowman and Chaucer.
’ In this paper, I assume an AGRoP as in Chornsky (1992). In accordance with Chomsky
(1 995), one could, however, argue that the features to be checked are in the light verb.

0 The Editorial Board of Studia Linguistica 1996.


112 Elly van Gelderen

point. I will take this to mean that the future features of for to in (15 ) to
(17) can be taken as a reference point for the tense of the subordinate
clause even though it is situated in AGRo. That for to is in AGRo is not
unreasonable since it is often assumed that infinitives are nominal in older
English and can therefore not assign Case. For this purpose,forto appears
in Spec AGRoP.
Thus, in early Middle English,for to is in AGRo and has (inherent) Case
marking features. It is also associated with future features and these are
emphasized once the infinitive ceases to be nominal.

2.3. For to andfor as Complernentizers


In this section, I show that for to in Otho is in C and that for is in C in Piers
Plowman and The Canterbury Tales. For to in the latter two texts is ‘still‘in
AGRo.
The number of infinitives introduced by for to increases in Otho, the
late thirteenth century version of Layamon’s Brut. Most of the infinitives
are purposive infinitives as in (19): there is one example of a complement
infinitive (i.e. (20)). Some of these for tos are separated from their
infinitive by an object (in (21) to (25)); others (e.g. (26)) are like those of
the earlier version, indicating the situation is changing gradually.There are
other uses as a regular C in (27):

(19) Otho, 4075


for to se pi(s) cnihtes game,
‘In order to see the knights’ game’.
(20) Otho, 5523
moche he lofde echn(e) cniht. pat lofde for to segg(e) riht,
‘Much he loved every knight who loved for to say the truth’.
(21) Otho, 5496
for him to reade,
‘in order to counsel him’.
(22) Otho, 6419
for to heom kepe,
‘in order to keep him’.
(23) Otho, 6915
fo[r] to londes seche,
‘for to countries seek’.
(24) Otho, 8488-90
wide his men sende. for to hine finde,
‘Widehis men (he) sent for to him find’.
(25) Otho, 8570
for to worch makie,
‘in order to make work’.
Q The Editorial Board ofStudia Linguistica 1996
The reanalysis of grammaticalized prepositions 113

(26) Otho, 3123


And lawes he sette stronge. his folk for to stewe,
And laws he made strong his people for to restrain
‘And he made strong laws in order to restrain his people’.
(27) Dtho, 3770
Alle dai was pat fiht; forte hit were dorcke nipt,
‘All day lasted that fight until it was dark night’.
For to in (19) and (20) may still be a preposition in AGRo as in the
Caligula version; in (26), it clearly is. However, in (21) to (25), it is a
complementizer preceding the object in Spec AGRoP. Forte in (27) is
mainly a preposition because it is followed by pat as in (29) and mostly
introduces finite clauses, as shown in (27). It is, however, not a transparent
preposition since it introduces an infinitival once (in line 8286) and is
never complemented by an NP (cf. Modern English while) in Caligula and
Otho:
(28) Otho, 1147
for 3if pou were Brutus his sone,
‘because if you were Brutus’ son’.
(29) Otho 7700
forte pat ich mihte bet borewes a-winne,
until that I might better boroughs win
‘Until I might better win the boroughs’.
As to the features associated with for to, these are future in (20) because
segge ‘say’is future from the point of view of the event in the main clause,
i.e. lofde ‘loved’. In Modern English, love has a complement where the
action of the infinitive seems future with respect to the form of love, as
(30) indicates:
(30) I would love for Zoya to get into a good school.
This means that the tense of the infinitive is anchored in C and not in the
main verb (cf. EnG 1987 and also Stowell 1982). However, in (20), the
action of the infinitive may be simultaneous with that of the action in the
main clause. There are too few instances in Layamon to decide whether
this is the normal case and hence differs from Modern English use.
Thus, there are several differences between Caligula and Otho
indicative of further change of for and to. As to the position, for to in Otho
is no longer exclusively a preposition in AGRo but is also a complemen-
tizer. I suggest that these are present in purposive adverbials and in com-
plements to deontic verbs introduced later than Layamon. I now turn to
two texts written about a century after the Otho version: Piers the Plow-
man and The Canterbury Tales.
The B-version of Piers the Plowman is assumed to be from around
1377 and, unlike in Otho and Caligula, for to is often used to introduce
D The Editorial Board of Studia Linguistica 1996.
114 Elly van Gelderen

complement infinitives. Examples are (31) and (32). Unlike Otho, for to is
never separated from the infinitive, indicating it is exclusively positioned
in AGRo. For, on the other hand, has become a clear Complementizer as
(33) and (34) show. Forto also occurs as one word but its distribution is
exactly as that of for to, i.e. it always immediately precedes the infinitival
verb:
(31) Piers Plowman, P 172
And 3if him list for to laike,
and if him please for to play
‘and if he desires to play’.
(32) Idem, 5 304
Now bigynneth glotoun for to go to schrifte,
‘Now began Glutton to go to confession’.
(33) Idem, 12 15
It is but murth as for me to amende my soule,
‘It is only joy as for me to amend my soule’.
(34) Idem, 20 25 1
for loue to ben holy,
‘for love to be holy’.
In Chaucer’s late fourteenth century The Canterbury Tales, for to
introduces subordinate verbs several hundred times as in (35) to (39). As
in Piers Plowman, it is no longer restricted to occurring with purposive
infinitives as some of these sentences show. Again as in Piers Plowman, for
to can still immediately precede the infinitive and is not a complementizer
because a second fur (i.e. a real C) is possible, as in (39), and because
extraction of an object is possible as in (38). Constructions such as (40)
and (42) indicate that for is separate from to. Thus the situation in
Chaucer is quite similar to that in Piers Plowman. The only difference is
that for to is not written as one word?
(35) Canterbury Tales, Prologue, I, 12-3
Thanne longen. . . palmeres for to seken straunge strondes,
‘Then long pilgrims for to seek strange shores
‘Then pilgrims long to find foreign shores’.
(36) The Knight’s Tale, I, 1095
This prison caused me nat for to crye.
(37) The Wife ofBath’s Tale, 111,920
as he hopeth for to finde grace.
(38) The Clerk’s Tale, IV, 533
This child I am comanded for to take.
(39) Prologue, I, 225-6
For unto a povre ordre for to yive Is signe that a man we1 yshryve.

This is true for the Canterbury Tules. Searching the entire Chaucer corpus, I found 3
instances of forto in The Romaunt of the Rose. These do not behave differently from for to.

0 The Editorial Board of Studia Linguistlca 1996


The reanalysis ofgrammaticalized prepositions 115

(40) The Wife of Bath’s Tale, III,96 1


for all this world to wynne,
‘in order to win the entire world.
(41) The Miller’s Tale, I, 3270
for any good yeman to wedde.
(42) The Parson’s Tale, X, 785
Whan man or womman preyen for folk to auancen
hem, oonly for wikked flesshly affeccioun,
‘When man or woman prays for people to help
them only out of wicked devotion’.

For to as in (35) ‘still’precedes the VP in the sentences that occur with


forto but it introduces a complement much more often than in Otho and is
associated with sentential subjects as in (39). The tense in those com-
plementing verbs is also straightforwardly future since the verbs used are
deontic ones such as hopeth in (37) and longen in (35). This indicates that
the tense of the infinitive is anchored through the features associated with
for to.Yet, the position offor to is in AGRo since objects always precede it
(cf. (42)). It is interesting to compare afor to complement as in (37) with
one without for as in (43):
(43) Chaucer, Boece V, pr. 3,117
as science comprehendeth it to be. (Visser 1963-73:2309)
The tense of the infinitive in (43) must be simultaneous with the tense of
the main verb unlike the case of for(to),e.g. as in (37), (38) and (42).
Again, adapting EnG (1987), if for is a future marker, the tense in the
infinitive (which cannot be independent) is linked to that of the future
marker. Future Cs are selected by deontic verbs such as expect, want and
love.
As in Piers Plowman, a reanalysis of for as a C is taking place in
Chaucer, as (40), (41) and (42) show. I have found 11 of these in The
Canterbury Tales but most of these are not complements to verbs, which
Visser (1963-73:224-5) confirms (perhaps too categorically): ‘the idiom
is hardly ever met with before the beginning of the twentieth century’. For
in (40), (41) and (42) is analyzed as other fors, introducing clauses. The
features associated with for are the same as those associated with for to,
namely unrealized tense/future. This latter use of for starts for adjuncts in
the latter half of the fourteenth century as in (44) and (33) and (34) above;
for complements as in (42) above, it starts gradually (verb by verb) after
1380; for subjects it starts after 1567 (Visser 1963-73:957):

(44) Wyclif, Acts 23, 24


make 3e redi a hors for poul to ride on,
‘make a horse ready for Paul to ride on’.
(Visser 1963-73:988)
0 T h e Editonal Hoard of Studia Linguihtica 1996
116 Elly van Gelderen

In conclusion to 2.3, I have described the situation in Otho’s version of


Brut, Piers Plowman and The Canterbury Tales. In Otho, for to can be
either in C or in AGRo; in the other two texts, it can only be in AGRo but
for by itself is situated in C. Infinitives introduced by for to become more
frequent and shift from being used as adverbials to complementing
deontic verbs. The increasing future sense of the preposition makes this
possible. It also causes the language user to reanalyze for(to) as being
situated in C .

3. From Preposition to Inflection


In van Gelderen (1993), I show that in Middle English to is first a preposi-
tion of location, then a Case marker, then a tense marker. This indicates
grammaticalization. As in the case of for(to),the position in which to is
situated changes from AGRo to I. I first briefly indicate the situation in
Old English before outlining the changes in Middle English.

3.1. To as a Preposition
In Old English, to is a preposition used to indicate location as in (45) and
(46). As is clear from (46), the meaning of to is one of motion towards but
‘without any implication of reaching’ (OED entry for to). The latter
meaning, I will suggest, is the source of the Modern English meaning of to,
namely, connected with unrealized tense features as in (47). In the course
of Old English, fo also gets to be used as an indicator of dative Case as in
(48):’
(45) Dream of the Rood, 2
hwaet me gemztte to midre nihte,
‘what I dreamt at mid night’.
(46) OE Chronicle, an. 1132
Dis ?ear com Henri king to pis land,
‘In this year came King Henry to this country’.
(47) Laws 42, Elf., Intr. c. 49a
God self sprecende waes to Moyse
3him bebead to healdanne,
‘God himself was talking to Moses
and ordered him to be loyal‘
(Callaway 1918:45).
(48) O.E. Chronicle, an. 1123
se biscop . . . side to bam kyng,
‘the bishop. . . said to the king’.
(Visser 1963-73:624)

According to Mustanoja (1960:96), this is found in late Middle English prose.

0 The Editorial Board of Studia Linguistica 1996.


The reanalysis of grammaticalized prepositions 117

Mustanoja (1 960:95) says that the replacement of the dative case by to as


in (48) only occurs in late Old English. Visser (1963-73:624) claims that
it occurs at the beginning of Middle English. Thus, it seems that to starts to
change in late Old English.

3.2. To as AGRo
With respect to the position of to, the to which grammaticalizes into a
tense marker immediately precedes the VP in early Middle English texts
such as Caligula’s version of Brut, Katherine and Wohunge of Ure Louerd,
which are from the first part of the thirteenth century. This is similar to the
position offor to in section 2. In (49) and (50),the object precedes to and
in ( 5 l),an adverb does. The structure for these is as in (18) above:
(49) Caligula, 8874
heo wenden hine to finden,
‘they went to find him’.
(50) Wohunge, 37-8
a1 engles lif is ti neb to bihalden,
all angels’ life is thy face to behold
‘the angels’ wish is to behold your face’.
( 5 1) Katherine, 3 12
‘t feng on bus to speokene,
‘and started to speak thus’.
In these constructions, the action of the main clause is separate from that
of the subordinate clause. As mentioned in section 1, in Modern English,
to is responsible for this and I assume that to is gradually assuming those
features here even though it is ‘still’involved in Case marking and situated
in AGRo.
As mentioned in connection with (23) and (25) above, the sitution in the
Otho version of Layamon’s Brut is that for together with to is in C because
objects follow it:
(52) Layamon Otho 8489-90
wide his men sende for to hine finde,
‘he sent his men wide for to find him’.
At the same time that this shift occurs, forte is also being used as a
complementizer with finite clauses in Otho as (53) shows but not in the
earlier version (i.e. Caligula) as (54) shows:
(53) Layamon, Otho 2834
forte he com to Rom,
‘until he came to Rome’.
(54) Idem, Cal.
bat heo come to Rome buri.
0 The Editorial Board of Studia Linguistica 1996.
118 Elly van Gelderen

Yet, when to occurs by itself in Otho (as in Caligula but that is expected),
objects precede it still. Thus, in (55), his lond precedes to, indicating that
the latter is in AGRo:
( 5 5 ) Layamon, Otho 2432
healp his lond to winne,
‘help his land to get’.
In 3.2, the to used to indicate non-finiteness is argued to be in AGRo. So
far, this development runs parallel to that of for(to)when it is situated in
AGRo. I will now show that to comes to be situated in I and is associated
with non-tense-features. This change solidifies the independence of to.

3.3. T o i n I
As I argue in van Gelderen (1993), around 1380, pro-infinitives as in (56),
do in its modern English use as in (57) and split infinitives as in (58) start
to occur. To becomes an independent auxiliary‘” in this way. Handlyng
Synne is a text from 1300 and Chaucer and Wyclif are from around 1380.
In a Minimalist framework, this means that the (inherent) features of to
change from Case-marking ones to categorial verbal features:
( 5 6 ) Handlyng Synne, 80 2 3-4
But wyle 3e alle foure do
A Pyng pat y preye 3ow to,
‘but will all four of you do
a thing that I ask you to’.
(57) Chaucer, The Monk’s Tale, VII 2432
Fader, why do ye wepe?
(Visser 1963-73:1552)
(58) Wyclif, Matthew, 5,34
Y say to ~ O U to
, nat swere on a1 manere,
‘I say to you, to not swear completely’
(Visser 1963-73: 1040)
In ( 5 8 ) , to can be separated from the infinitive; in (56), it can be left when
the VP deletes. The ‘introduction’ of modals, not shown here, and the
appearance of do as in (57), in complementary distribution with modals
and to, also indicate there is now a special position, i.e. I as in Modern
English (7) above. Accusatives with Infinitives also start to occur around
1380 as (59) and (60), from a text from that period, show. If one analyzes
these as in Massam (1985), they involve IPS and would therefore be

‘ I ’ As against e.g. Haspelmath (1989:296) who argues there is grammaticalization of zu in


German but asserts that to in OE is independent and a prefix in Modern English.

0 The Editorial Board of Studia Linguistica 1996


The reanalysis of grammaticalized prepositions 119

expected to occur once I becomes available. Thus, this analysis involves


treating to as occupying I as in (7) above:
(59) Wyclif, Luke 8,46,
I have knowe vertu to haue gon out of me,
‘I have known virtue to have gone out of me’.
(60) Idem Acts 27,27,
the schippe men supposiden
summe cuntre to apere to hem,
‘the sailors supposed some country to appear to them’.
(Visser 1963-73:23 13; 2315 respectively)
In section 3, I show that to changes its position from AGRo to I. This
can be described as to acquiring inherent verbal features and non-tense.

4. From Preposition to Aspect Marker: on


In 4.1, I provide examples of the prepositional use of on. Subsequently (in
4.2. and 4.3.), I show that on grammaticalizes and is reanalyzed as having
verbal and aspect features. It also changes its structural position.

4.1. On as a Preposition
In‘OldEnglish, on is used to indicate a position ‘outside of, but closer to or
near, any surface’ (OED entry for on).It is used in an extended sense as in
(61) and (62):
(61) O.E. Chron. an. 1122
feole scip-men on s z and on wzter,
‘many sailors on sea and on the water’.
(62) Klfric, Deut x, 1
On p z r e tide Drihten cwzb to me,
‘In that moment, God spoke to me’.
In (62), the temporal sense of on is already evident. Bybee & Dahl(l989)
provide examples of other languages where a locative preposition evolves
in a similar fashion. The question is at what point the category of on is
reanalyzed from preposition to aspect marker.
On as a progressive marker can be found in Old English, as (63) and
(64) show, according to Visser (1963-73:1993) but its category is a
preposition as I will show. This use is continued later, as in (65) and (66):
(63) De Eccles. Gradibus
Exorciste beod on getacnunge Cristes gespellan,
‘E. is on teaching Christ’s story’.
(Visser 1963-73: 1998)
OThe Editorial Board of Studia Linguistica 1996
120 Elly van Gelderen

(64) AeTGen 14.22


Ic ahebbe mine hand to bam healican gode,
se Tie ys agnigend eorban ond heofonan,
‘I raise my hand to the holy God who is possessing earth and heaven’.
(Mosse, I, 82)
(65) Layamon, Brut Caligula 6139
he was an slaeting,
‘he was hunting’
(66) Pepys’Diary 31 Dec
I am upon writing a little treatise.
These sentences must have been construed by the language learners as
involving PPs because different prepositions occur: an in (65), upon in
(66) but in other cases on, at, a and in also occur. There is, however, never
a determiner, as there is, for instance, in Dutch, shown in (67). The English
structure is as in (68); the Dutch as in (69):
(67) Ik ben een boek aan het lezen,]
I am a book on the reading
‘I am reading a book’.

on V
I
writing
I
a book

eenboek AGR; PP
P /‘ DP

lezen t’2

‘ I Sentences such as (i) to (iii) might seem similar to the Dutch ones but they start occurr-
ing relatively late:
i. He is on the run.
ii. He is on the make.
iii. She is on the take.
Checking the OED, sentences such as (i) turn out to be rather recent innovations. Partridge’s
Slang Dictionary lists on the run as introducedin the latter part of the nineteenth century and
on the make as introduced in the 1890s from US slang.
I * In English, there is an AGRoP as well responsible for Case to the object but
NP-movement into the Specifier of AGRoP, in e.g. Chomsky (1992) is not overt. In Dutch,
movement is overt and I have therefore added AGRo.

0 The Editorial Board of Studia Linguistica 1996.


The reanalysis of gramrnaticalized prepositions 121

4.2. From Preposition to ASPect


In both the Caligula and Otho versions of the thirteenth century Brut, the
progressive is preceded by a preposition. Both the verbal endings and the
prepositional forms vary widely. In both manuscripts, an, a , at as well as
-yng(e), -ing(e),-enge occur. It seems that the situation is very unstable
around 1250 and this makes it possible for the status of on to be re-
considered.13The confusion may have been ‘helped’by the (phonological)
change of on to a. Thus, sentences such as (70) and (72) change to
sentences such as (71) and (73), i.e. ones where on becomes a prefix (and
as a result the PP becomes an Adverb Phrase). This indicates gram-
maticalization since the latter involves loss of morphological independ-
ence:
(70) Cal. 161
Wa wes him on hue,
Woe was him in life
‘He was anguished in life’.
(71) Otho, Idem
Wo was him a-liue.
(72) Cal. 1494
mare ban is on hue,
‘more than (one who) is a-live’.
(73) Otho, Idem
more pan alle bat his a-liue.
The phonological change occurs first with ons preceding NPs as in (70)
to (73). At the end of the fifteenth century (OED entry for a ; Visser 1963-
73:1993 ff.), a starts to precede progressives and may form one word with
this verbal element as in (74), from 1530, and (75), from 1526, before it
disappears as in (76), from 1879:
(74) J. Redford, Wyt &Science 695
Art a-answering, to?

Progressive constructions with a Verb ending in -ende, or -o&e) have occurred all
through the history of English (Visser IY63-73:1993). Old English examples are (i) and (ii):
i. Aelfred, Boethius 18,
mid p a n beowum ic eom ealne bone hefon ymbhweorfende,
‘with these servants, I am all the heaven encompassing’.
(MossC 1938,1:79)
ii. Larnbeth Homilies 41
be per were wuniende,
‘who there were living’.
(MossC 1938,1231)
The form in -ing is a Middle English innovation, but whether it is a direct continuation of the
-ende form is contested (see Mosse 1938, II:36; and Jespersen 1 940:415).

0 The Editorial Board of Studia Linguistics I996


122 Elly van Gelderen

(75) Tindale, Luke 19,33


as they were aloosinge the coolte.
(Visser 1963-73:1998)
(76) Meredith, The Egoist 249
if he is calculating. . .
(Visser 1963-73:1997)
Checking Visser (1963-73), (66) is the last instance with a preposition
other than a h . The variation dies out in the sixteenth century and this
indicates a loss of preposition status. As mentioned, on changes from an
independent preposition indicating location as in (6 1) to a prefix
indicating aspect as in (75). Subsequently, the overt aspect marker
disappears in (76). Syntactically, a structure such as (68) is reanalyzed as
(10). In (lo), the verb moves to ASP to check its features, whereas in ( 6 9
the verb moves to P (possibly covertly, i.e. at LF). The change involves the
present participle overtly moving (incorporating) to check the aspect
features that on has acquired. Thus in (63), it is not clear that the verb has
moved overtly to P, but in (74) and (75) it has moved to ASP. Again, the
grammaticalization of on can be seen as a category change from preposi-
tion to aspect marker. I assume this change takes place in the sixteenth
century when a becomes common.

5 . Conclusion
I have given three instances where prepositions grammaticalize and
change position in the sentence. I have argued that for to, to and on come
to have categorial features (for verbal category as well as for future, tense
and aspect respectively). Grammaticalization can thus be seen as an
association of a lexical element with a set of features. Once the verbal
features are associated with a particular element, an appropriate Func-
tional Head is assumed by the language learner (C, I and ASP respect-
ively). The latter accounts for the changes in word order.
This paper extends what is meant by categorial features in a Minimalist
framework. I assume that the verbal categorial features of lexical items are
specified further, namely for future, tense and progressive features.

Texts Used
Brook, G. & Leslie, R. (eds). Layarnon: Brut. Oxford: Oxford U. Press,
1963 edition of EETS 250.
Einenkel, E. (ed). The Life of Saint Katherine. London: EETS 80,1884.
Klaeber, F. (ed). Beowulf. Boston: Heath & Co. 1950, edition of 1922.
Oxford Text Archive. Chaucer’s Canterbury Tales, computer readable
version. Oxford.
0 The Editorial Board of Studia Linguistica 1996
The reanalysis of grammaticalized prepositions 123

Oxford Text Archive. Layamon’s Brut, computer readable version,


Oxford.
Oxford Text Archive. Langland’s Piers Plowman, computer readable
version, Oxford.
Robinson, F. Chaucer’s Complete Works. Oxford:Oxford U. Press, 1974.
Skeat, W. The Vision of Piers The Plowman, Vol 1. Oxford: Oxford U.
Press.
Sullens, I. (ed.), Robert Mannyng’s Handlying Synne. Binghampton, 1983.
Thompson, W. (ed.) b e Wohunge of Ure Lauerd. EETS, Vol. 241, Oxford
U. Press, 1958.

References
ABRAHAM, W. 1993. Grammatikalisierung und Reanalyse: Einander ausschlie-
Bende oder erganzende Begriffe? Folia Linguistica Historica 13, 1/2,7-26.
AKMAJIAN, A,, STEEL, S. ~r WASOW, T. 1979. The category of AUX in universal
grammar. Linguistic Inquiry 10, 1-64.
BYBEE, J. DAHL,0. 1989. The creation of tense and aspect systems. Studies in
Language 13,51-103.
BYBEE, J., PERKINS, R. & PAGLIUCA, W. 1994. The evolution ofgrammar. Chicago:
Chicago U. Press.
CALLAWAY, M. 1918. The infinitive in Anglo-Saxon. Washington: The Carnegie
Institute.
CHOMSKY, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.
CHOMSKY, N. 1992. A minimalist program for linguistic theory. MIT Occasional
Papers in Linguistics I.
CHOMSKY, N . 1995. Categories and transformations. The Minimalist Program.
Cambridge: MIT Press.
EN^, M. 1987. Anchoring conditions for tense. Linguistic Inquiry 18,633-57.
FISCHER, 0. 1995. To and bare infinitival complements in late Middle English.
Diachronica 12, 1-30.
V A N GELDEREN, E. 1993. The rise of functional categories. Amsterdam: John
Benjwns.
V A N GELDEREN, E. 1996. Restraining functional projections. WCCFL 1995
Proceedings, ed. J. Camacho et al., 111-122. Stanford: CSLI.
HASPELMATH, M. 1989. From purposive to infinitive. Folia Linguistica Historica,
287-310.
HASPELMATH, M. 1993. Functional categories, X-bar theory, and grammaticaliza-
tion theory. Berlin, ms.
HEINE, B. CLAUDI, U. 8r HUNNEMEYER, F. 1991. Crammaticalization. Chicago: U.
of Chicago Press.
JESPERSEN, 0. 1940. Modern English grammar V. London: Allen & Unwin.
KAYNE, R. 1994. The asymmetry of syntax. Cambridge: MIT Press.
LEHMANN, C. 1985. Grammaticalization: synchronic variation and diachronic
change. Lingua e Stile 20,303-31 8.
LIGHTFOOT, D. 1979. Principles of diachronic syntax. Cambridge: Cambridge U.
Press.
MASSAM, D. 1985. Case theory and the projection principle. Ph.D. diss., MIT.
+
Moss<, F. 1938. Histoire de la Forme Pkriphrastique dtre participe prksent en
germanique, Vol. I et II. Paris.
0 The Editorial Board of Studia Linguistica 1996.
124 Elly van Gelderen

MUSTANOIA, T. 1960.A Middle English syntax. Helsinki: Societe Neophilologique.


Oxford English Dictionary 1993. Oxford: Oxford U. Press.
PARTRIDGE, E. 1979. A dictionary of slang and unconventional English. London:
Routledge.
STOWELL, T. 1992.The tense of infinitives. Linguistic Inquiry 13,560-70.
TRAUGOTT, E. HEINE, B. 1991. Grammaticalization Z and ZZ.Amsterdam: John
Benjamins.
VISSER, F. 1963-73. A n historical syntax of the English language: 3 Volumes.
Leiden: Brill.
Elly van Gelderen
Arizona State University, English Department
Tempe, A Z 85287-0302
USA
(elIyvangelderentGasu.edu)

OThr Edilorial Board o f Studia Linguisrica I W b

You might also like