6. Article à paraître

The grammaticalization of the preterite-present verb AGAN

Cet article a fait l´objet d´une communication lors de la 2^e Conférence Internationale sur la Modalité à Pau en 2004 et sera publié dans un volume concernant le temps et la modalité.

Abstract: The aim of this paper is to try and explain, on syntactic and semantic grounds, the grammaticalization of the semi-modal ought to in Old and Middle English (hence OE and ME), i.e. how this lexical item has become a grammatical one(note: ➳).

6.1. Introduction

In OE, agan means “possess”, but in Contemporary English (CE) it now means “owe, should”. Agan belongs to the class of preterite-present verbs: broadly speaking, they have a past form, but a present meaning (see Campbell (1959) for further details). The ME form comes from the past form ahte in OE.

In the first section, we shall deal with OE data, mainly focusing on the syntax of agan. In the second section, ME data will be analyzed, making a parallel with the preceding section. Finally, in the third section, we shall focus on grammaticalization, going back and forth the OE and ME periods.

6.2. Agan in Old English

6.2.1. The distribution of agan

Agan(note: ➳) can be found in different structures:

It can function as a lexical verb and then it takes a DP object,

(643)

Dryhten	hælend,	þu	þe	ahst	doma	geweald.
Lord	Christ,	NOM-you	NOM-who	IND.PRES-owns	GEN-justice	ACC-power.

Lord Christ, who owns the power of justice. (cocynew(note: ➳),86.723. 623)

It can be passivized(note: ➳),

(644)

Þa	cwæð	se	cyningc	to	his	mannum	siððan	Apollonius	agan	wæs...
Then	PRET-said	the	NOM-king	to	DAT-his	DAT-men	when	NOM-Apollonius	P.PART-had	IND.PRES-was...

Then the king said to his men when he had had control over Apollonius... (ApolT,

ApT:14.1.250)

or it can be an infinitive and follow another preterite-present,

(645)

Ac	se	ðe	þa	ecan	agan	wille	soðan	gesælða.
But	the	NOM-one	ACC-the	ACC-eternal	obtain	PRES-will	ACC-true	ACC-blessing.

But who will obtain the eternal true blessing. (comeboe,161.7.23. 52)

It can have the same structure as the auxiliary verb BEON/WESAN “be”(note: ➳); it is then followed by a past participle,

(648)

...	se þe	ah	lifes	wyn	gebiden	in	burgum.
...	ACC-who	IND.PRES-ought	GEN-life	ACC-gain	P.PART.	in	DAT-town.

...who ought to ask in town the gain of life. (coexeter,144.27.133)

(649)

hi	agon	on	agenan	hwilan	mid	earfedan	gewunnen.
NOM-they	IND.PRES-ought	during	DAT-proper	DAT-time	by	DAT-suffering	P.PART-remained.

they ought to remain in suffering for their own time. (WHom, WHom_20.2:51.1425)

What is striking concerning this verb is that we have not found examples displaying the structure AGAN + infinitive (which can be found with all the other preterite-presents) in Pintzuk & al. (2000). Meanwhile we did find examples of the structure AGAN + to + infinitive,

(650)

&	betæhte	him	æl	þæt	he	ahte	TO	bewitenne.
&	PRET-delivered	DAT-him	all	that	NOM-he	PRET-ought	TO	DAT-keep.

& delivered to him all that he ought to keep. (Heptateuch,Gen:39.4. 1564)

6.3. Agen in Middle English

At the ME period, it is commonly assume that morphology gets poorer, i.e. a great number of flexions tend to be blurred, or dropped, leaving the words bare. Because of this morphological impoverishment, grammar (and syntax) has to readjust, entailing changes of parameters, hence status of some items.

6.3.1. The distribution of agen

We now find two forms for AGEN: the present form owe and the past form ahte, both being used either as a lexical or a modal verb.

As a lexical verb, agen can be found as a transitive verb,

(651)

...	þet	hit	er	ahte.
...	that	OBJ-it	already	PRET-possessed.

... that (he) already owned it. (CMLAMBX1,31.377)

(652)

...	the	love	that	men	to	hym	owen.
...	the	love	that	SUBJ-men	to	OBJ-him	PRES-owe.

... the love that men owe him. (CMCTPARS,313.C2.1087)

As a modal verb, agen is found in two types of structures: modal + infinitive (unlike the OE agan) and modal + to + infinitive(note: ➳)

(653)

...	þt	euch	mon	ahte	hersumin	&	herien	in	eorðe.
...	that	each	SUBJ-one	ought	obey	&	praise	on	earth.

... that each one ought to obey & praise on earth. (CMKATH,23. 75)

(654)

þus	ahte	ech	of	us	him	seluen	TO	cnowen.
thus	ought	each	of	SUBJ-us	him	OBJ-self	TO	know.

thus, each of us ought to know himself. (CMTRINIT,123.1649)

(655)

[Arthure]	owen	al	þe	worlde	TO	deme.
[SUBJ-Arthur]	PRES-ought	all	the	OBJ-world	TO	deem.

Arthur ought to deem all the world. (CMBRUT3,81.2468)

(656)

first,	we	owen	vndirstonde	it	bi	the	lettre.
first,	SUBJ-we	ought	understand	OBJ-it	by	the	letter.

we first ought to understand it word by word. (CMPURVEY,I,52. 2123)

6.4. Syntax and grammaticalization

Let us now turn to the syntax of agan/agen in OE and ME. In all the examples we parsed, we found two types of uses : a lexical and a modal use.

6.4.1. Old English syntax of agan

In examples (643) to (648), agan is a lexical verb. It is then a V within a vP(note: ➳); in (650), it is a control verb. In Roméro (2005), we have shown that the syntax of the preterite-presents and the syntax of TO in infinitivals were quite similar.

Let us have a look at the following examples, where ∅ indicates an ellipsis of the verb:

(657)

Ure

æghwylc

sceal

ende

gebidan

worolde

lifes;

se

þe

mote

domes

ær

deaþe

∅.

GEN-Us

NOM-every

PRES-must

ACC-end

wait

GEN-world

GEN-life;

NOM-who

that

PRES-has to

GEN-judgment

before

DAT-death

∅.

Everyone of us must wait for the end of life of the world, and before death, one has to wait for the judgment. (cobeowul,43. 1386.1148)

(658)

willaþ	þæt	gewrecan	gif	we	magon,	þeah	we	beotiaþ	TO	∅.
FUT-will	ACC-that	avenge	if	NOM-we	PRES-can,	yet	NOM-we	PRES-pride ourselves	to	∅.

(we) will avenge it if we can, yet we pride ourselves on doing it. (BlHom, HomS_10_[BlHom_3]:127.446)

According to those examples (two among many others), the preterite-present verb mote (“must, have to”) and TO seem to have the same behaviour concerning ellipsis. Mote (i.e. a special kind of v) heads a TP, and to (i.e. a C) heads a TP).

But, in that respect, agan does not behave like the other preterite-presents: it is the only one to be structured with TO.

This leads us to question the syntactical status of agan in OE. From examples (643) to (650), agan is a lexical verb. But the structure of (650) is interesting since agan is followed by TO: here we assume it is a control verb, hence the following structure,

(659)

We also assume that when agan is followed by a bare infinitive, it has a different grammatical status: it has become a semi-lexical verb (what we called v_Modal in Roméro (2005)), what example (657) seems to indicate for preterite-present verbs.

So, in OE, we can find agan either as V followed by a TP, or as a control verb when used with TO, i.e. followed by CP. Let us now have a look at ME syntax of agen.

6.4.2. Middle English syntax of agen

Unlike OE, ME agen can be found both in AGEN + bare infinitive or AGEN + to + infinitive structures, and strikingly enough, we find more examples of OWE(N) + to + infinitive than AHTE + to + infinitive. If we just follow our analysis, agen is a control verb in the structure AGEN + to + infinitive, and semi-lexical in AGEN + bare infinitive(note: ➳).

So (654) and (655) still display control verbs, but not (653) and (656). The changing point lies in (656): in OE agan was not found with bare infinitives, it was only a control verb. We assume this is due to the grammaticalization of agen (and generally speaking of all the preterite-presents class), as well as the reanalysis of TO(note: ➳) (and what Dekeyser calls the "HABERE/DEBERE interface).

Syntactically, we thus have in ME, two competing structures for agen: the “control” structure, which is identical with (659), and a “grammaticalized” one, which is as follows,

(660)

6.4.3. What about grammaticalization?

Grammaticalization is the process in which a lexical item becomes a grammatical one. With respect to agen, this is, as Dekeyser (1998) says, a “braintwister”: in CE, ought does not have a past meaning anymore, but more an irrealis one (a conditional), and it is considered as a “semi-modal”: grammaticalization seems to have not been completed. And we must add that it always functions with TO, which was reanalyzed because of the loss of the infinitive morphology(note: ➳). Strikingly enough, at the same period, we have not found examples of dare + to + infinitive, nor þurven “need” + to + infinitive(note: ➳)

Moreover, we have two forms of agen: present owe(n) and past ahte(note: ➳). In our corpus (Kroch & Taylor (2000)), we have found 20 examples of OWE(N) + to + infinitive, but only 7 with AHTE. Yet, only the latter became a grammatical item. We shall see in the next section that the choice of ahte over owe seem to be due to the meaning of the morphological form(note: ➳). On semantic ground, by the end of the ME period, owe “own, possess” is to be used more as a lexical verb, and ahte “ought, obligation” as a modal one (see Dekeyser (1998) for more details).

So we have two processes at work for what is to become the modal periphasis ought to: the reanalysis of TO into a TO_(Irrealis) (now followed by a bare infinitive) and of AHTE as an irrealis item, with an obligation meaning.

6.4.4. Grammaticalization and tense

In the following examples, we give illustrations of ´past´ forms of modals which have conditional meanings. Some are already grammaticalized (examples (661) and (662)), some on their way to grammaticalization (examples (663) and (664)).

(661)

´Though

I

wiste

that

neither

God

ne

man

ne

sholde

nevere

knowe

it,

yet

wolde

I

have

desdayn

for to

do

synne.

“Though

SUBJ-I

PRES-know

that

neither

SUBJ-God

nor

SUBJ-man

not

should

never

know

OBJ-it,

yet

would

SUBJ-I

have

disdain

to

do

OBJ-sin”.

´Though I am aware that neither man or God should never know it, yet I would scorn sin.´ (CMCTPARS,290.C2.94)

(662)

The

yeer

of

oure

Lord

1391,

the

12

day

of

March

at

midday,

I

wolde

knowe

the

degre

of

the

sonne.

The

SUBJ-year

of

our

Lord

1391,

the

12

day

of

March

at

midday,

SUBJ-I

would

know

the

OBJ-degree

of

the

sun.

On March 12th in the year of our Lord 1391, I would know the degree of the sun. (CMASTRO,669.C1.189)

(663)

but,

certes,

he

sholde

suffren

it

in

pacience

as

wel

as

he

abideth

the

deeth

of

his

owene

propre

persone.

but,

certainly,

SUBJ-he

should

suffer

OBJ-it

in

patience

as

well

as

SUBJ-he

PRES-awaits

the

OBJ-death

of

his

own

proper

person.

But he should certainly endure it with patience, as well as he awaits his very own death. (CMCTMELI,217.C1b.20)

(664)

for

ase

ofte

ase

ʒe

þrefter

breken

eni

of

ham,

hit

walde

to

swiðe

hurten

ower

heorte

ant

maken

ou

swa

offered

þet

ʒe

muhten

sone,

as

God

forbeode,

fallen

i

desesperance...

for

as

often

as

SUBJ-you

thereafter

PRES-break

any

of

OBJ-them,

SUBJ-it

would

too

greatly

hurt

your

OBJ-heart

and

make

SUBJ-you

so

P.PART-offered

that

SUBJ-you

might

soon,

as

SUBJ-God

PRET-forbade,

fall

in

despair...

for, as often as you break any of them this way, it would too greatly hurt your heart and make you so offered that you might soon be desperate, which God forbade... (CMANCRIW,I.46.65)

Let us now turn to agen. The situation is somewhat different: there are more examples with OWE (FOR)TO (morphological present form) than with AHTE TO (morphological past subjunctive form).

(665)

þogh

ʒe

mow

wyth

glosyng

wordys

desayve

me

and

say

ʒe

ben

yn

full

charyte

as

ʒe

owen

FORTO

be,

...

although

SUBJ-you

may

with

glosing

words

deceive

OBJ-me

and

say

SUBJ-you

PRES-are

in

full

charity

as

SUBJ-you

ought

to

be,

...

although you may deceive me with glosing words and say you are indulgent as you ought to/should be, ... (CMMIRK,130.3483)

(666)

and	heruth	God	seruyce	as	cristen	men	owen	FORTO	do;
and	IMP-hear	God	OBJ-service	as	christian	SUBJ-men	ought	to	do;

and hear God service as Christian men ought to; (CMMIRK,138. 3674)

(667)

&	[þe	eorl]	benam	him	al	ðat	he	ahte	TO	hauen.
&	[the	SUBJ-earl]	PRET-took	OBJ-him	all	that	SUBJ-he	ought	to	have.

The earl took from him all that he should have had. (CMPETERB, 58.529)

(668)

When

ye

wyl

aske

any

þing

at

a

ryche

man

in

erthtt,

ye

aske

ful

mekelike

yure

erande

wyd

mekil

mare

deuociun

ahte

ye

þanne

AT

pray

to

god,

of

whaim

þat

al

ʒude

cumis.

When

SUBJ-you

will

aske

any

OBJ-thing

to

a

rich

man

on

earth,

SUBJ-you

PRES-ask

full

much

your

OBJ-request

with

lot

more

devotion

ought

SUBJ-you

then

to

pray

to

God,

of

whom

that

all

OBJ-good

PRES-comes.

When you ask anything to a rich man, you will fully ask your request with much more devotion, should you then pray to God, from whom all good comes. (CMBENRUL,19.667)

When owe is used, TO has been reanalyzed ((665), (666) and (668)), whereas when it is ahte, TO has not yet but the modal has (667).

What does this syntactically mean? The modal is no more generated in V, but in Mod/T: it no more has a past meaning, nor is it a lexical item since it functions as a [+irrealis] head, i.e. a subjunctive, and it weakens so that it cannot move to T anymore(note: ➳). Hence it is no longer a control verb, but a raising one. By the end of the ME period, the syntax of agen would thus be,

(669)

[_TP [_Spec Subject_i [_T ahte [(Neg) [_TP [_T to [_VP [_Spec t_i [_V verb]]]]]]]]], that is,

(670)

which is different from (659).

6.5. Conclusion

With the loss of the infinitive morphology, we showed that the reanalysis of TO went along the one of agen and specially ahte. We also underlined that the meaning shift of ahte, from “past” to “conditional”, or broadly speaking to irrealis, was partly responsible for its grammaticalization. Semantically, agan/agen no longer means “possess” but “obligation”. As for syntax, ought, a control verb in OE, turned into a raising verb generated under T and taking a TO-complement (like “seem” or “appear”) by the end of the ME period.

Yet, dealing with the grammaticalization of such a verb is not that easy since the very process is still obscure, and Contemporary English displays the two “layers” of agan: the lexical verb ´own/owe´ and the semi-modal “ought to”(note: ➳). For reasons of space, we did not investigate the root/epistemic interface.