Factors triggering P-lability - Ghent University Librarylib.ugent.be/fulltxt/RUG01/002/162/803/RUG01-002162803_2014_0001... · A diachronic investigation of factors triggering P-lability:

GHENT UNIVERSITY – FACULTY OF ARTS AND PHILOSOPHY

Factors triggering P-lability

From Old English to present-day English

JOYCE STROOBANT

Masterproef voorgelegd tot het behalen van de graad van

Advanced Master in Linguistics: Linguistics in a Comparative Perspective

2013-2014

Supervisor: prof. dr. Miriam Taverniers Co-supervisor dr. Leonid Kulikov

1

A diachronic investigation of factors triggering P-lability:

from Old English to present-day English Joyce Stroobant

Abstract The aim of this article is twofold, I will not only try to enumerate which factors might have triggered P-lability in English but I will also try to evaluate the relative weight of each of these factors. This paper is written from a typological point of view: if one scrutinizes the PDE verbal system, it immediately becomes clear that English is the only present-day Germanic language that has expanded its category of verbs showing P-lability to contain over 800 verbs – and in fact this number is still increasing – whereas in the other Germanic languages only a handful of such verbs are available. I will cover both language internal factors, i.e. phonological attrition of alleged valency changing affixes, dative-accusative case syncretism and several syntactical processes of reanalysis, as well as language external factors, particularly influence of Old Norse will be considered. I will argue, based in the evidence found in older language stages of several closely related West Germanic dialects, that English owes its special status mainly to Old Norse influence, especially on the development of reflexives form Old English to Middle English.

1 Introduction The aim of this article is to investigate the cause or causes that underlie the expansion of P-

lability from just a few P-labile verbs in OE to over 800 in PDE (McMillion 2006). Traditionally, verbs are called labile if their form remains the same in both transitive and intransitive uses. Particularly, two types of lability exist: A-lability on the one hand and P-lability on the other. The distinction between these two types of lability crucially resides in the alignment between the transitive subject, the transitive object and the intransitive subject. A-lability or Agent-preserving lability is actually more commonly referred to as ‘absolute transitive’ or sometimes also ‘object deletion’. Moreover, it is called ‘Agent-preserving’ because the transitive subject, i.e. John in (1a) which prototypically is an Agent, is also the subject of the intransitive verb, i.e. John in (1b), in which case it is also an Agent. With P-lability or Patient-preserving lability, on the other hand, the transitive subject, i.e. John in (2a) is eliminated in the intransitive use, whereas the object or Patient becomes the intransitive subject (2b).

(1) a. John is eating pancakes.

b. John is eating. (2) a. John opened the window.

b. The window opened.

2

The alternation exemplified in (2) is referred to in the literature as the causative-inchoative alternation (Haspelmath 1993), although some authors, Ottosson (2009) amongst others, prefer to use the term anticausative instead of inchoative because the latter refers to aspectual nuances rather than to a difference in valency. Cross-linguistic studies have pointed out that anticausativity is not always encoded by P-labile ity, as is the case in English. Particularly, reflexive morphology is often and in several of the IE language branches used to derive the anticausative alternant from the causative one, e.g. in Russian (3)1, German (4) and Spanish (5) to name but a few. I briefly refer to reflexive morphology already in the introduction because it turns out that reflexivity, although reflexives are fairly scarce in PDE, is of interest if one wants to investigate the diachronic evolution of P-lability from OE to PDE.

(3) a. Džon otkryl okno.

‘John opened the window.’ b. Okno otkrylos’. ‘The window opened’

(4) a. Johann öffnet das Fenster. ‘John opens the window’

b. Das Fenster öffnet sich nicht mehr. ‘The window does not open anymore.’

(5) a. Juan abrió la ventana. ‘John opened the window’

b. La ventana se abrió. ‘The window opened.’

Before I proceed to the actual overview of possible causes for the rise of P-lability in section 3, I

will describe the situation for OE (in section 2.1) and compare it to the situation in the other Old West Germanic dialects, e.g. Old Saxon, Old Frisian, Middle Dutch and Old High German (in section 2.2) for which I rely heavily on Hermodsson’s Transitive und Intransitive Verben im ältern Westgermanischen (1952). Subsequently, in section 3 I will try to give a systematic overview of possible factors that might have caused English to develop an extraordinary number of P-labile verbs. However, I will not merely enumerate, but I will also try to evaluate their overall significance. Since English is the only Germanic language in which P-lability has become so prominent, the obvious question is indeed to ask why this happened in English but not in the other Germanic languages. Thus in section 3.1 I will look at some phonological developments, in section 3.2 I will investigate dative-accusative case syncretism, in section 3.3 syntactic factors are scrutinized and in section 3.4 I will argue, following McWorther (2002) that Old Norse has crucially influenced the development of P-lability in OE. Finally, in section 4, I will put the changes that were under investigation in section 3 in a broader Indo-European perspective.

2 P-lability in old West Germanic

2.1 The situation in Old English In this section I will provide some background on the number and semantic nature of labile verbs

in Old English. Subsequently, in section 2.2, I will compare the Old English data with that for the other old West Germanic dialects, namely Old Saxon (OS), Old Frisian (OFris), Middle Dutch (MD), and Old High German (OHG), with special attention for OHG and MD. We will also briefly turn our attention to Gothic which, although it does not pertain to the group of West Germanic, is interesting to us for its antiquity. In my view, such a comparison is useful because it places the OE situation into perspective. Many authors, when discussing the extraordinary number of PDE labile verbs, refer to the fact that such verbs were already present in OE. However, as we shall see, P-lability is not

1 Russian example taken from Paducheva (2003: 174). In (3b) the Russian verb otkryl ‘open’ is marked by the reflexive particle –sja (with allomorph –s’) when it is used anticausatively.

3

exceptional for the old West Germanic dialects. For most of the data presented in this section I will rely on an extensive description of reflexive and intransitive verbs in old West Germanic by Lars Hermodsson, published in 1952. Although this work is in some respects outdated, it has provided me with a vast quantity of examples for the historical periods of the languages mentioned above. We will begin our discussion with a description of OE labile verbs.

I will try to present the OE labile verbs (and also those of the other old West Germanic dialects) in a systematic manner. Following Hermodsson, I will make a first broad distinction between denominative and non-denominative verbs, secondly the non-denominative verbs will be subcategorized into a number of semantic classes, while the denominative verbs will be classified according to the derivational suffix that was used, i.e. –jan or -ôn. This will prove to be particularly interesting for our comparison in the next section. For now we can just observe that OE labile verbs included verbs of motion, ‘spread’ verbs, ‘show’ verbs, ‘break’ verbs, ‘change’ verbs, ‘open/close’ verbs and approximately 40 denominative verbs.

Table 1 ‘Labile verbs in Old English’

Denominative verbs (1952: 196-201) with -jan- suffix (1st class weak verbs)

Old English transitive meaning intransitive meaning 1 brædan to broaden (tr.) to broaden (intr.) 2 gecelan to cool (tr.) to cool(intr.) 3 dælan to divide (tr.) to divide (intr.) 4 gedieglan to hide (tr.) to hide (intr.) 5 drygan to dry (tr.) to dry (intr.) 6 geeaðmedan to humble to humble oneself 7 afierran to remove, to lead away to go away 8 hliewan to warm (tr.), to care for/

cherish to warm (intr.)

9 lengan to lengthen (tr.) to lengthen (intr.) 10 stillan to calm, to silence to be calm, be silent 11 twæman to divide (tr.) to divide (intr.) 12 ðiedan / þeódan to join (tr.) to join (intr.)

with -ōn-(-ian-) suffix Old English transitive meaning intransitive meaning 13 abiterian to make bitter, exacerbate to become bitter /worse 14 blissian to make rejoice to rejoice 15 cwiccian to make alive to come to life 16 adrugian dry (tr.) dry (intr.) 17 afeorsian to remove, to lead away to go away 18 gaderian to gather (tr.) , to assemble

(tr.) to gather (intr.) , to assemble (intr.)

19 gladian to make rejoice to rejoice 20 godian to do good, make good,

improve (tr.) to become good, improve (intr.), get better

21 heardian to harden (tr.) to harden (intr.) 22 hefegian to make heavy to become heavy 23 hlænian to make lean to become lean 24 hluttrian to make clear, purify to become clear 25 ahnescian to weaken (tr.), soften (tr.) to weaken (intr.), soften (intr.) 26 liðigian to soften (tr.), assuage, calm to be soft or yielding 27 lytlian to make little, lessen (tr.),

diminish (tr.) to become little, lessen (intr.), diminish (intr.)

28 gemetgian to moderate, temper to become moderate, to moderate

4

oneself 29 micelian to make great, increase the

size or quantity of a thing to become great, increase in size or quantity

30 minsian to lessen (tr.), diminish (tr.), make small

to lessen (intr.), diminish (intr.), become small

31 nearwian to make narrow to become narrow/ contracted 32 openian open (tr.) open (intr.) 33 samnian to collect (tr.), assemble (tr.),

gather (tr.) to collect (intr.), assemble (intr.), come together

34 strangian to make strong, confirm, comfort

to be(come) strong, prevail

35 swiðian to make firm, fix to be(come) strong, prevail 36 swutulian to manifest, make clear to become manifest 37 ðiccian to thicken (tr.) to thicken (intr.) 38 aðiestrian to darken (tr.) to darken (intr.) 39 ðynnian to make thin to become thin 40 wlitigian to make beautiful to grow beautiful 41 yfelian to do evil, maltreat, injure,

afflict, wrong to get bad

Non-denominative verbs (1952: 201-210) A. Motionverbs – movement from point A to B

42 brecan to break (tr.) to break forth 43 cierran, cyrran to turn to be turned, to turn himself 44 dragan tod rag, draw to draw oneself, draw, go 45 sceadan to separate (tr.), divide (tr.) to separate (intr.), to go away 46 strican to stroke, strike , smooth,

rub, wipe to go, move, run

47 teon to draw along (horizontal movement), pull, bring, lead

to draw (nigh), go, proceed

48 ðringan to press (tr.), crowd, throng to move with violence, press forward, force a way (through)

49 ðryccan to press, crush, oppress, repress, trample (tr.)

to press (intr.), force a way

50 wendan to cause to move, alter the direction or position of sth.

to go, proceed, turn around

51 windan to twist, roll (of motion that results from a blow, swing, or other impetus) to fly, leap, start

B. Motionverbs – movement on the spot 52 biegan, bygan to bend (tr.), curve (tr.) to bend (intr.), curve (intr.) 53 cierran, cyrran to turn to return, go 54 hebben to heave, lift up, raise to rise, mount 55 hieldan to bend (tr.), incline (tr.),

heel (tr.), tilt (tr.) to bend (intr.), incline (intr.), heel (intr.), tilt (intr.)

56 oðhydan to hide (tr.) to hide (intr.) 57 miðan to hide (tr.) to hide (intr.), be hidden 58 settan to set, place, put, cause to

take a certain position to settle, abate, subside

59 styrian to stir, move (tr.) to be in motion 60 ðrawan to twist (tr.), rack, torture to twist (intr.), turn round 61 wendan to turn (tr.) to turn (intr.) 62 windan to turn (tr.) to turn (intr.)

C. ‘spread’ verbs

5

63 mengan to mix, mingle, combine to mingle together, stir up 64 sceadan to separate (tr.), divide (tr.) to separate (intr.), to divide (intr.) 65 stregdan to strew, spread (tr.), scatter

(tr.), sprinkle (tr.) to scatter (intr.), disperse

D. ‘show’ verbs 66 ætiewan to show (tr.) to show (intr.), appear

E. ‘break’ verbs 67 brecan break (tr.) break (intr.) 68 slitan to slit, tear (tr.) to tear (intr.)

F. ‘change’ verbs 69 bregdan to move to and fro (tr.), to

change (tr.) to move (intr.), be pulled, (of lightning) flash (intr.)

70 ci(e)rran, cyrran to turn (tr.) to change (intr.) 71 lædan to lead, take, carry (Lat. ferre,

ducere) to take a certain direction, to emerge

72 wendan to change (tr.) to change (intr.) G. ‘open/close’ verbs

73 ondon open (tr.) open (intr.) 74 behlidan close (tr.) close (intr.) 75 lucan close (tr.) close (intr.) 76 ontynan

< tynan (tr.) ‘(en)close’

open (tr.) open (intr.)

Table 1 clearly indicates that in OE labile verbs, both denominative as non-denominative were

quite common. We should note, however, that it is possible that the transitive use of some of these verbs might only marginally occur, or vice versa. Furthermore we must keep in mind that for some of these verbs, e.g. wendan which occurs more than once in the table because it is polysemous, also reflexive uses are attested. What should be retained from the table above, are not absolute numbers, but the number of different semantic groups labile verbs occur in.

Furthermore, it is interesting to note that some of the verbs above, especially denominatives, were ‘labile’ from the beginning, whereas others, mainly non-denominatives, have undergone a process of intransitivation or transitivation. Contrary to Visser (1963: 127), who suggests that a “wholesale process of transitivation” has taken place in the history of English, Hermodsson claims that transitivation of originally intransitive verbs occurs only marginally. In section 4, I will attempt to explain why these author’s opinions on one and the same phenomenon are so divergent; most importantly it concerns their respective attitudes towards what transitivity is. In the following paragraph I will continue to explain why Hermodsson thinks intransitivation is more appropriate.

A first factor in favour of intransitivation is the higher frequency of transitive uses of the verb, as opposed to the relatively low frequency of intr. uses; for hebban, settan and slitan for example the intr. use occurs only seldom, hence Hermodsson’s assumption that their basic valency must have been transitive. Secondly, based on his comparison with OHG, Hermodsson concludes that, since OHG reflexive verbs are in OE frequently represented by intransitive forms, the intransitive use is the younger one because the reflexive forms in OHG are also younger, i.e. derived from the transitive form, thus the transitive use must be older. I should, however, stress that Hermodsson does not assume that the OE intr. forms were originally reflexive. Many authors have claimed that the ellipsis of the refl. pron. can cause P-lability, but, according to Hermodsson, this scenario is highly dubious for OE because reflexive forms have never been very productive in this language in the first place. In other words, if there is nothing to ‘elide’, “the refl. pron. ellipsis-hypothesis” cannot apply. We will come back to this in section 3.3.2.

2.2 The situation in other old West Germanic dialects In this section I will compare the OE material, discussed in the previous section, with Gothic and

the other Old West Germanic dialects, with special attention for MD and OHG. It will become clear

6

that OE conforms to the West Germanic standard of the time, contrary to what one might expect. In fact, it is OHG which deviates from the norm in having an elevated number of inherently reflexive verbs and a relatively low number of labile verbs. In accordance with, e.g. Abraham (1997) and Ottosson (2009), we will conclude that in German the opposition causative-anticausative is to a large degree paradigmatically determined. As was the case in OHG, in present-day German anticausativity is clearly marked, either by means of umlaut, derivational affixes, reflexives or auxiliaries; formal identity between the causative and anticausative member of the alternation seldom occurs. The only context under which the paradigmatic opposition between causative and anticausative in German is neutralized is perfectivity (Abraham 1997: 19), but we will come back to this issue in section 3.3.3 when I discuss syntactic factors triggering P-lability.

A first glance at the Gothic material does not seem to reveal an elevated number of labile verbs, however, the Gothic data may be misleading. It is well known that the available corpus material for this language is quite scarce2 due to the antiquity of the language, therefore, although absolute numbers are not very high (i.e. approx. 20 labile verbs are attested) we may conclude that Gothic does show a considerable amount of labile verbs already. Furthermore, according to Hermodsson (1952: 108), bimaitan constitutes further proof for this claim. The fact that a verb as rare as bimaitan ‘circumcise’ exhibits P-lability must be a clue to the overall frequency of this phenomenon in Gothic. Moreover, the fact that P-lability occurs in a language attested as early as Gothic and, as we shall see, also in all other Germanic dialects discussed here indicates that P-lability is not unique to OE. Table 2 ‘Labile verbs in Gothic’

Denominative verbs (1952: 94-97) with -jan- suffix

Gothic adjective/ subst. tr. meaning intr. meaning 1 gabairhtjan bairhts (<bright3) to make clear, declare appear 2 ufarfulljan fulls (<full) to overfill (tr.) to become full to

overflowing, to be abundant

3 bleiþjan bleiþs (merciful) to have mercy for to be merciful 4 balþjan balþs (<bold) to dare (tr.)(confront

boldly, have sufficient courage to contend against)

to dare (intr.) (have sufficient courage)

5 ananiujan niujis ( < new) to renew (tr.) to renew (intr) 6 sildaleikjan sildaleiks (wonderful)

silda > seldom to marvel at something to be surprised,

amazed 7 ufarassjan subst. ufarassus

(excess) ufar (<over)

increase (tr.) increase (intr.)

with -ôn- suffix 8 galeikon leiks (alike) to compare, treat as equal to be equal, to act in

the same manner Non-denominative verbs (1952: 97-103)

Gothic etym. info tr. meaning intr. meaning 9 ataugjan (cf. OE: ætiewan, MD.

togen & tonen) Goth. augjan (tr. show) maybe desubst.

to show (tr.) to show (intr.)

2 Hermodsson has used the Gothic Bible, translated by Wulfila in the 4th century AD from Ancient Greek. 3 Gothic dictionary with etymologies by András Rajki, 2004, URL: http://web.archive.org/web/20100416081032/http://etymological.fw.hu/Gothic.htm

7

>augo 10 biugan * bheug- to bend (tr.), curve (tr.) to bend (intr.), curve

(intr.) 11 daupjan *dheub- (< diups)

‘deep’ to baptize to wash oneself,

to be baptized 12 haban

[labile under Greek influence]

appears as intr. + adv., always as translation of Gr. ἔχειν 4

to have, possess to find oneself + adv.

13 bimaitan uncertain; related verbs: afmaitan, gamaitan, usmaitan = only tr. subst. bimait (circumcision)

to hack, cut, circumcise to have oneself circumcised

14 skaidan derived verbs: afskaidan (tr./ refl.) – disskaidan (tr) – gaskaidan (refl.)

to separate (tr.), divide (tr.) to separate (intr.), to divide (intr.)

15 anastodjan, dustodjan

<PIE *stō (cf. standan, intr.) subst.: anastodeins ‘begin’

begin (tr.) begin (intr.)

16 ustiuhan *deuk- (vgl. Goth. tiuhan ‘pull’, PDG. ‘ziehen’)

to lead out (of), to lead away, to execute sth., to accomplish

to come to an end

17 aftiuhan tiuhan in Goth. is only tr., although maybe rarely also already intr. with meaning: ‘to stroll, wander’

to pull away to move away

18 þwahan (Eng. > towel) subst. þwahl ‘bath’

to wash sth. to wash oneself

19 gawandjan the simplex wandjan is attested once as tr. and once as refl.

to turn (tr.), to steer, to convert (tr.)

to turn (intr.), turn back (intr.), convert (intr.)

20 wasjan, gawasjan

PIE *ues- ‘to clothe’ gawaseins ‘clothing’ wasti ‘dress’

to dress sb. to dress oneself

As mentioned above, Hermodsson takes Middle Dutch as a measure for comparison and not Old

Dutch which, time wise, would have been a more obvious choice5. Therefore, since Old Dutch material is scarce and fragmentary, Hermodsson uses Middle Dutch data. He remarks that MD exhibits more P-lability than the other old West Germanic dialects he investigated, including OE. This surely does not come as a surprise, since data for MD is more abundantly available for MD than for OE and is considerably younger. Hermodsson’s findings are, nevertheless, interesting. In fact we must, again, not look at sheer numbers, i.e. Hermodsson provides a list of 210 P-labile verbs for MD

4 ἔχειν: 1. (tr.) to have, hold, occupy; 2. (intr. + inf.) to be able to, know how to; 3. (intr. + adv.) to be like ‘meaning of the adverbs’, e.g. εὖ ἔχειν, ‘to be alright’; (middle/passive) 1. to find oneself in a place/situation, to be held in captivity; 2. to follow immediately after. 5 An interesting article about the periodization of the Germanic languages is: Versloot, Arjen. 2005. Why Old Frisian is still quite old. Folia Linguistica Historica (25):257-302.

8

(6), but at the distribution of the labile verbs over different semantic verb classes. Interestingly, MD compared with OE shows three additional semantic groups of non-denominative labile verbs, e.g. ‘begin/end’ verbs, ‘mental state’ verbs and so-called ‘privative’ verbs which are exclusively found in MD and express removal or withdrawal.

(6) denominative; achteren, veranderen, argen, armen, baren, beteren, verbitteren, bleken, verbliden,

verblinden, verbloden, bosen, verbouden, breden, verdeluwen, diepen, verdieren, verdikken, donkeren, verdorren, bedroeven, drogen, verdullen, dunnen, verduusteren, verdwasen, veredelen, vergramen, vergrimmen, vergroenen, vergroven, heelen, heeten, hogen, verhovaerden, verjogeden, claren, coelen, corten, vercouden, cranken, crenken, crommen, lagen, verlamen, langen, leeliken, lengen, lichten, liken, luteren, mageren, matten, meerderen, meersen, menichvouden, vermenigen, menken, meren, merenen, minderen, moeden, morwen, nauwen, nederen, vernieuwen, veroevelen, openen, ouden, plompen, quaden, verriesen, riken, roden, ronden, sachten, sachtmoedigen, sadden, saluwen, scherpen, schonen, verseren, ontsienen,slaken, slappen, slechten, smalen, gesmieden, soeten, soren, verspaden, verspaken, sterken, stiden, stillen, stiven, verstouten, stricken, suren, swacken, swaren, swarten, tragen, tregen, tweevoudigen, ververschen, vesten, vetten, vulen, vullen, verwackeren, warmen, weken, werden, widen, witten, verwoesten, verwreden, baren, openbaren, beteren, vereffenen, verenen, verenigen, gaderen, vergaderen, verledigen, verootmoedigen, rechten, reden, reiden, gereiden, toereiden, samenen, versamenen, stricken, verren, vervreemden; non-denominative, motion verbs: inbreken, dragen, dringen, keren, cruden, dorecruden, porren, roeren, scheiden, setten, striken, tien, trecken, wenden, boren, opboren, draeyen, heffen, keren, neigen, porren, roeren, vouden, wenden, winden; ‘begin/end’ verbs: hemmen, houden, letten, merren, stoppen, versumen, toeven; ‘spread’ verbs: mengen, mengelen, scheifelen, sparsen, spreiden, strooyen, werren; ‘show’ verbs: togen, tonen; ‘mental state’ verbs: verbinden, dragen, opgeven, hoeden, laten, maten, schicken, setten, stellen; ‘break’ verbs: breken, cleven, clieven, cloven, riten, schoren, liten, spliten, spouden; ‘change’ verbs: verdraeyen, verheffen, keren, wenden; ‘open/close’ verbs: ontdoen, luken, ontpluken, sluten; ‘privative’ verbs: ontbinden, ontdecken, laten, ontnaeyen, ontspannen, vlaen

Hermodsson concludes his discussion of MD labile verbs by suggesting that MD was well on the

way to expand its number of labile verbs to a degree reminiscent of PDE. However, in Late Middle Dutch the expansion was hindered by the intrusion of the OHG reflexive pronoun sich in the Dutch verbal system. Before, Middle Dutch, like Old Saxon, Old English and Old Frisian, i.e. all North Sea Germanic dialects, used the non-reflexive personal pronoun in the accusative case – and sometimes also in the dative case (later these two cases syncretized in Nord-West Germanic, hence accusative forms became identical to dative forms) - to express ‘reflexive meaning’. Consider the following passage:

“Es sei ferner bemerkt, dass im Mnl. auch eine beträchtliche Zahl von ursprünglich intr. Verben in doppelter Funktion erscheint. Aus alledem ergibt sich, dass die Doppelfunktion der Verba im Mnl. sehr häufig war. Die Entwicklung in Mnl. Zeit führte zu einer immer stärkeren Verbreitung dieser Erscheinung. Das Mnl. befand sich nach meiner Auffassung auf dem Wege, einen ähnlichen zustand zu erreichen, wie er im heutigen Englisch vorhanden ist. Diese Entwicklung ist jedoch im Spätmnl. Zeit aufgehalten worden, und das Nnl. weicht in dieser Hinsicht von der früheren Entwicklung stark ab. Die Aufnahme des hd. refl. Pronomen ins Nl. bedeutete eine wichtige Veränderung. Der Gebrauch von intr. Verben im Nl. hat sich nach dieser Aufnahme wesentlich anders gestaltet als früher (1952:262)”. [emphasis mine, JS]

Despite the fact that sich has not completely replaced the older forms in all dialects of Dutch, i.e.

especially in Flanders and Brabant6, it has had far-reaching consequences in the verbal system. According to Hermodsson this was possible because it did not just concern the replacement of one morphological form by another, i.e. hem (m. and n.)/ hare (f.) by sich, but because the whole

6 In non-standard Dutch of Flanders and Brabant – it occurs less in Limburg – the older forms are still regularly encountered, e.g. Hij beeldt hem in dat… ‘he imagines that’ or Zij wast haar ‘She washes (herself)’.

9

syntactical context in which the Middle High German sich appeared was taken over as well. We have already noted that in OHG and MHG reflexives were more wide-spread than in any of the other Nord West Germanic dialects, i.e. in those dialects intransitives fulfilled the role reflexives had in OHG and MHG. Therefore, it is not surprising that under the influence of OHG and later, MHG, the number of intransitive verbs decreased in favour of reflexive forms in the evolution from Middle Dutch to Present Day Dutch.

Consequently a preliminary conclusion may be that the high number of PDE labile verbs is not a direct consequence of the presence of labile verbs in OE, but that the expansion of P-lability to encode the anticausative alternation, or the proneness of old West Germanic to prefer intransitive forms over reflexive forms, was facilitated by the absence of influence of OHG on OE. The OE reflexive system did, however, undergo some changes in later stages of the language. Particularly, inherent reflexivity decreased even further, and –self was added to the reflexively used pronouns to reinforce their status as reflexives7. The latter two changes can presumably be ascribed to the influence of Old Norse (McWorther 2002), but we will come back to this issue.

Finally, to conclude this section I will briefly discuss table 3 below that I have reproduced following Hermodsson. I have already mentioned that MD labile verbs appear in additional semantic groups and that this is normal given the date and number of the MD material. We should, nevertheless, pay attention to the fact that ‘begin/end’ verbs are absent only in OE. Hermodsson attributes the absence of labile ‘begin/end’ verbs in OE to chance because from the table it seems that this type of aspectual verbs has been central to the labile verb type of West Germanic in general. One reason for this conclusion concerning ‘begin/end’ verbs in OE is the occurrence of such verbs as anastodjan and dustodjan ‘begin’ in Gothic, which is remarkable given the antiquity and relative scarceness of the data available for Gothic. Secondly, the presence of the Frisian verb meria which can express both tr. and intr. ‘stop’ also indicates that the absence of such verbs in OE is remarkable or, as Hermodsson puts it, fortuitous, since data for Old Frisian is almost equally scarce as that for Gothic.

Table 3 ‘Overview of non-denominative labile verbs according to meaning’ (Hermodsson 1952: 311)

Meaning OHG – MHG OS OFris OE MD motion verbs (from point A to B) + + + + + ‘begin/end’ verbs + + + + ‘break’ verbs + + + + + motion verbs (on the spot) + + + + ‘open/close’ verbs + + + + ‘mingle/spread’ verbs + + ‘show’ verbs + + ‘change’ verbs + + ‘mental state’ verbs + ‘privativa’ +

Verbs with aspectual meaning in OE that could fill in the gap in Hermodsson’s data are for

example: on-ginnan ‘to begin, set to work, endeavour, attack’ (which is more frequent than be-ginnan or a-ginnan) (7) and ge-endian ‘to end/come to an end’ (8). These are by no means the only possible verbs pertaining to the class of aspectual verbs in OE; it is merely an intention to complete Hermodsson’s list and support his claim that ‘begin/end’ verbs are central to the labile type in Old West Germanic regardless of their absence in OE.

7 For a more in depth investigation of reflexive marking in English, i.e. self-strategy versus plain personal pronoun used reflexively from ME to PDE: Peitsara, Kirsti. 1997. “The development of reflexive strategies in English”. In: Grammaticalization at work (eds.) Matti Rissanen, Merja Kytö & Kirsi Heikkonen, 227-370. Berlin: Mouton de Gruyter.

10

(7) a. Ic onginne tó wearmigenne [+ gerund. / inf.] [Dict. Sax.-Lat.-Angl., by Somner 38, 4]

‘I began to warm up/feel warm’, (Lat. calesco) b. Se ðe gód onginneþ and ðonne áblinneþ ne biþ hé Godes leóf on ðæm néhstan dæge [Blickling

Homilies 21, 35 ed. by R. Morris] ‘he who begins good and then ceases, will not be God's friend at the last day’

(8) a. Oþoniél geendode his dagas [Anglo-Saxon Book of Judges 3, 11] ‘Othoniel ended his days’, read: ‘Othoniel is dead’

b. Ðá geendode se gebeorscipe [Th. apol. 18, 8] ‘then the feast came to an end’

b’. Siððan Eádgár geendode [Anglo-Saxon Reader by Sweet 106, 44] ‘since Edgar died’

Moreover, a more recent investigation of the diachronic development of labile verbs by Alan

McMillion (2006) seems to corroborate Hermodsson’s claim. In order to chart the development of labile verbs in English, McMillion divides them into four semantic groups; verbs denoting a change of state, motion verbs, aspectual verbs, verbs indicating a change of quantity and a final heterogeneous category which he calls “other”. For our argument, however, only his class of aspectuals is relevant. He suggests that aspectual verbs, i.e. “those verbs that denote the initiation, continuation, or end of an event (McMillion 2006:187)”, although relatively few in number, “became labile quite early (McMillion 2006:187)”. Table 4, taken from McMillion, contains the labilization dates for 5 aspectual verbs McMillion includes in this category.

Table 4 ‘Labilization dates for aspectual verbs’ (McMillion 2006: 187)

tr. intr. labile origin end 1000 1000 1000 from OE geendan ( <PGM: *andja) begin 1200 1000 1200 from OE beginnan continue 1340 1340 1340 from Old French continuer finish 1350 1450 1450 from Old French finiss-, present participle stem of fenir stop 1400 1530 1530 from OE (for)stoppian

Overall, we can conclude that it is not OE that deviates from the standard in historical times, but OHG which has, apart from a different morpheme for expressing reflexivity, also much more reflexive verbs than the other Old West Germanic dialects. One possible reason for this discrepancy is that the original meaning of sich and its PIE ancestor *sṷe was not reflexive proper, but anaphoric. Anaphoric use of reflexive morphemes is attested in, e.g. Homeric Greek and classical Latin where it was often used in subclauses to refer to the subject of the matrix clause. Competition between anaphoric and reflexive meaning could lie at the basis of the use of regular personal pronouns for expressing reflexive meaning in certain languages, but we will not develop this argument here since it falls beyond the scope of this paper.

3 Factors triggering P-lability

3.1 Phonological factors

3.1.1 Disappearance of OE preverbal ge-, be-

Visser (1963:127) lists 34 transitive-intransitive verb pairs for which the prefix ge- essentially indicated a valency increase, but which have lost this prefix over time (via ʒe- > ʒy- > ʒi- > y- > i- > ) thus giving rise to P-lability. McWorther (2002) following Hiltunen (1983: 92), notes that in OE a great deal of derivational morphology was shed or replaced by free morphemes. The latter, however, usually only occurred when the original affix had concrete semantic meaning, e.g. niþer ‘down’, ymb ‘around’ or for ‘up, out’. On the other hand, affixes such as ge- or be-, for which concrete

11

semantics were not readily available were shed without substitution. I will dedicate the remainder of this section to a brief overview of the long scholarly tradition on the meaning and function of the prefix ge- followed by a tentative explanation of why the meaning of ge- is relevant to our discussion.

Table 5 ‘Transitive-intransitive verb pairs with ge-’

transitive with ge- intransitive without ge- 1 geabydan to wait for abidan to remain 2 geœmtian to make empty œmtian to be at leisure 3 geœrnan to reach, gain by

running œrnan to run

4 gebirhtan to make bright birhtan to shine 5 geceapian to buy, purchase ceapian to bargain 6 gefeallan to overthrow feallan to fall 7 geferan to get by going,

reach, perform feran to go

8 geflowan to overflow flowan to flow, be fluid 9 gegalan to enchant galan to be wanton 10 gegan to overrun, occupy,

subdue gan to go

11 gegangan to overrun, occupy, subdue

gangan to go

12 gegnigan to cause to bend gnigan to descend, sink 13 gegrowan (of soil) to produce growan to sprout, flourish 14 gehleapan to leap upon, mount hleapan to leap, dance 15 gehliehhan to deride hliehhan to laugh 16 gehwelian to inflame with foul

matter hwelian to suppurate

17 gehwitian to make white hwitian to be(come) white 18 gehirsumian to obey, serve hirsumian to be obedient 19 geirnan to run through,

reach, obtain irnan to run

20 gemanigfealdian to make numerous manigfealdian to become numerous 21 geminsian to lessen, impair the

credit of minsian to become small

22 gerestan to give rest to, lodge restan to repose 23 geridan to ride over, occupy,

seize ridan to ride

24 geri(g)nan to wet with rain ri(g)nan to rain 25 gesadian to satiate sadian to be sated, get wearied 26 gespringan to reach by

spreading, get by going, cause to spring

springan to leap, burst forth, rise

27 gestandan to cause to stand, place, attack

standan to stand

28 gestinkan to perceive by the sense of smelling

stincan to emit a smell

29 geswœtan to cause to labour swœtan to sweat 30 gesyngian to commit sinfully syngian to sin 31 geþyldigian to endure, bear

patiently þyldigian to be patient

12

32 gewadan to traverse, pervade wadan to go, move, stride 33 gewarian to guard protect,

hold, warn warian to be wary

34 gewinnan to obtain by fighting, gain, conquer

winnan to labour, toil

As early as the 17th century, linguists have pondered over the meaning and origin of OE ge-. In

1659, William Somner suggested that OE ge- was equivalent to Latin cum. This proposal, however, has been discredited for several reasons. Most importantly, Latin cum and Germanic ge- do not have the same distribution and do not stem from the same PIE root8. Secondly Somner mixes up two linguistic categories; OE ge- is a preverb, whereas Latin cum is in fact a preposition. The reason why we have included Somner in this overview is because his assumption has influenced many later linguists, particularly Streitberg (1891).

Half a century later Benson (1701) posits that ge- is in fact devoid of meaning. However absurd this might seem, it is nonetheless a step in the right direction. Traditionally, linguists attribute to ge- transitivizing (Visser 1963) and, more recently, perfectivizing (Mitchell & Robinson 1986) or resultative (van Kemenade & Los 2003) meaning. In other words, the absence of any lexical meaning is an important prerequisite for positing abstract grammatical meaning. In 1891 Streitberg was the first to develop this argument. He insisted that the original meaning of Gothic ga-, which Streitberg assumed was zusammen ‘together’, had bleached in OE and that its function was to perfectivate the verb to which it was attached (Trobevšek Drobnak, 1994).

However, in a milestone publication about the OE preverbal ge-, Lindemann (1970) empirically refutes two important claims, namely that (i) the prefix ge- is meaningless and that (ii) it monosemically expresses perfective aspect. Firstly, Lindemann argues that scholars such as Streitberg drew inaccurate conclusions because of incorrect interpretations of accusative forms that followed the verbs prefixed with ge-. The intransitive gan, for example, could be followed by an accusative when it was prefixed with ge-, yielding gegan. However, this accusative does not necessarily make the verb transitive, Lindemann suggests that these accusatives may as well be interpreted as accusatives of direction or goal. Secondly, the romantic aspiration of certain linguists to equate the Germanic verbal system with that of the Slavic languages, in which aspect presumably plays an important role, was an important incentive for treating ge- as a perfectivizer. More recently, however, Eythórsson (1995) has developed the argument in favour of perfectivity independently of dubious analogies with Slavic.

Although Lindemann (1970) refutes the assumption that ge- makes the action telic, he does not claim that ge- is entirely devoid of meaning. His alternative proposal is that OE ge- expresses direction which is either terminative (i.e. perfective and telic) or non-terminative (i.e. imperfective and atelic); consequently, verbs prefixed with ge- are not exclusively transitive. OE preverbal ge- can therefore, be translated in PDE with ‘at, on, to, toward, be-’, ‘out, forth, away’, but not with ‘together’ nor can it be left untranslated. Finally, most recently van Kemenade & Los (2003) have proposed that ge- expresses resultativity; for a complete overview and more recent empirical attempt at a survey of the distribution of OE preverbal ge- see McFadden (2013) who seems to agree with van Kemenade & Los (2003) that ge- expresses resultativity.

Not only ge- was shed in OE, also other prefixes, most notably be- and for-. In PDE some verbs with prefix be- and for- are still in use, e.g. ‘bereave’, ‘behold’, ‘bemoan’, ‘bedazzle’, ‘beseech’, ‘bewitch’, ‘forbear’, ‘forbid’, ‘forgather’, ‘forgive’, ‘forget’. However, McWorther notes that for- and be- are no longer productive, i.e. there is no such verb as “forenglish” meaning “to Enlgishify” (McWorther 2002: 234), whereas in Afrikaans or even Yiddish these prefixes remain productive. Furthermore, some of these verbs, especially those with be-, pertain to an elevated literary register.

Similarly Visser argues that be- disappeared because verbs prefixed with this prefix fell into disuse, e.g. bebark, bedwell, bechirp, beflow, befly, begaze, beglide, beglitter, bego, behoot, beleap, belie, bemew, berain, beride, berow, beshite, beshriek, besit, bescramble, bescratch, besparkle, beswink; OE. befaran,

8 Brugmann reconstructs Goth. ga- from PIE *ghō. (Pokorny: *g hō behind, towards/enclitic (emphatic ) particle) while Latin cum derives from PIE *kom meaning ‘at, by, with, alongside’.

13

betihtlian, betimbran, bewitan, bewlatian (Visser 1963: 134). He suggests that the transitive use of these verbs in their unprefixed form today is due to an ‘open space’ in the grammar which, he says, is comparable to pull and drag chains in the realm of phonology. McWorther (2002: 234), however, rightly points out that instead of assuming that the prefixed verbs as a whole fell into disuse, it was only the affix that did so, precisely because the simplex verbs are still in use, i.e. bark, lie, mew, rain, etc. If this is true then Visser’s hypothesized pull or drag chain is superfluous. Be this as it may, with the disappearance of ge-, be-, for-, English shed an important means for marking valency distinctions. Consequently, PDE P-labile verbs are more often than not monosyllabic verbs, whereas the other Germanic languages, especially German, held on to polysyllabic verbs and thus avoided homonymy of transitive and intransitive verbs (Abraham 1997: 27).

In sum, English has shed more of its derivational morphology than its West Germanic sisters, only in Scandinavian the cognate of ge- was lost quite early as well (cf. section 3.4). Consequently, many intransitive forms became identical to corresponding transitive forms thus yielding P-labile verbs which are mostly, like other regular English verbs, monosyllabic. Moreover, it would be difficult to determine in these cases whether intransitivation or transitivation is involved, especially since recent investigations have pointed out that ge- might not have been a perfectivizing or transitivating affix at all. In other words, if ge- did not make the action transitive, then its disappearance could not have triggered P-lability. In any case, the 34 verb pairs mentioned above are too few in number to have had decisive influence in the rise of P-lability in English.

3.1.2 Intransparency of causatives with –jan suffix

In the older Germanic dialects transitive weak verbs could be derived from intransitive strong verbs by means of the suffix –jan. Apart from deverbal derivations, -jan could also derive denominatives from nouns and factitives from adjectives, e.g. OE gladian ‘to be/make glad’ in which the Proto-Germanic –jan suffix is still visible as –i-. Nevertheless, authors such as van Gelderen (2011) have argued that in OE, especially in late OE, the –i- suffix was no longer transparent, i.e. speakers ceased to interpret it as a causativizer. Moreover, during the Middle English period the opposition -an/-ian was neutralized; verb pairs such as swefan ‘to sleep’ and geswefian ‘to put to sleep’ - ME sweve; dwellan ‘to err’ – dwellian ‘to lead astray’; heofan ‘to lament’ – heofian ‘to bewail’; meldan ‘to make a statement’ – meldian ‘to declare’; wealcan ‘to toss, roll, move round’ – wealcian ‘to roll up, curl’ PDE ‘walk’. However, it must be noted that verbs ending in –ian and –an do not always respectively denote transitive and intransitive verbs since the inverse is also possible; of course this does not influence the eventual labile outcome, but it is worth mentioning9.

Similarly, in cases where –jan caused i-umlaut, i.e. fronting and raising effect on the root vowel in the preceding syllable, the difference between the original intransitive non-fronted vowel and the new fronted vowel of the causative counterpart was often levelled out. Consequently, the transitive and intransitive verb became identical. Again it is in these cases difficult to ascertain whether transitivation or intransitivation is involved. Consider for example Visser’s examples (1963:131): sincan (intr.) – sencan (tr.) ‘sink’, þyncean (intr.) – þencean (tr.) ‘think’, scrincan (intr.) – screncan (tr.) ‘shrink’, slincan (intr.) - *slencan (tr) ‘slink’, springan (intr.) – sprengan (tr.), stincan (intr.) – stencan (tr.). Visser argues that over time –i- and –e- sounds were no longer phonologically distinct, instead they became allophones of /i/ and later were uniformly spelled as i.

Other verbs that coalesced are OE brinnan/birnan/bernan/beornan – bærnan (< Goth. brannjan) ‘burn’, būgan – biegan/bygan ‘bow’, meltan – mieltan/myltan ‘melt’. Particularly, coalescence of intransitive and transitive verb forms of this kind happened during the Middle English period because of the ambiguous use of the OE grapheme u which came to represent both the OE /ū/ and Old French /y/ (Visser 1963).

Many Germanic languages today still show traces, albeit marginally, of this i-umlaut, including English: lie (intr.) – lay (tr.), sit (intr.) – set (tr.), drink (intr.) – drench (tr.), fall (intr.) – fell (tr.), rise

9 In fact, Hermodsson (1952: 95-96) notes that already in Gothic verbs ending in jan existed that were exclusively intransitive, e.g. anananþjan, ‘have audacity’, anakumbjan ‘sit’, sainjan ‘be late’, which is probably due to the impossibility of univocal reconstruction of the –jan suffix, i.e. it goes back to several different PIE forms; on the origin of jan see K. Brugmann’s Grundriss der vergleichenden Grammatik der indogermanischen Sprachen, 2nd edition, volume 3, p.236.

14

(intr.) – raise (tr.). That the intr. verb is in these cases the base verb from which the trans. verb is derived can be proven if we look at the past participles of these verbs: past participles or simple past forms of the intr. verbs show ablaut (e.g. ‘Today I drink’ but ‘Yesterday I drank’) which is typical of strong verbs, whereas the trans. verbs have dental preterites and past participles (e.g. ‘Today I drench my pancakes in syrup’ but ‘Yesterday I drenched my pancakes in syrup’) which is typical of weak forms (van Gelderen 2011: 115). An illustration of the use of the suffix -jan and its consequences in 7 modern Germanic languages is given in Abraham (1997): for matters of convenience I have reproduced Abraham’s table and added data for OE. Table 6 ‘Umlaut causatives’ (“Paradigma der Umlautkausativa”, Abraham 1997: 14)

English intr. =tr.

crack melt wake up drown

Old English intr. springan myltan/mieltan wacian

drincan / druncnian

tr. sprengan meltan wæcnan drencan/ fordrincan

German intr. (zer)springen (er) schmilzt aufwachen ertrinken tr. (zer)sprengen (er) schmelzt aufwecken ertränken

Dutch intr. stukspringen smelten ontwaken verdrinken tr. laten springen (doen)10 smelten wekken verdrinken

Danish intr. springe smelte vågne drukne tr. sprænge lade smelte vække drukne

Icelandic intr. springa bráðna vakna drukkna tr. sprengja bræða vekja drekkja

Norwegian intr. springe smelte våkne drukne tr. springe smelte vekke drukne

Swedish intr. springa smälta vakna drunkna tr. sprang smälta väcka dränka

Frisian intr. springe smelte, rane wekker wurde ferdrinke, fersûpe

tr. springe/ barste litte

rane, smelte (litte)

wekker meitsje drinz(g)je, fersûpe

Faroese intr. springa smelta vakna drukna tr. spreingja smelta vekja drukna

3.2 Morphosyntactic factor: dative-accusative case syncretism The well-known case syncretism of the dative and accusative (in favour of dative forms, both for

nouns and pronouns) which began in late OE and was generalized in ME may have caused confusion between direct objects and indirect objects. Consequently, some intransitives that took dative complements expressing recipients or beneficiaries, e.g. andswarian ‘answer’, bugan ‘obey’, cidan ‘chide’, derian ‘injure’, foryrnan ‘forerun’, helpan ‘help’, heorcnian ‘harken’, hyran ‘obey’ and þancian ‘thank’ of Anglo-Saxon origin, but also ME Norman French loans, e.g. avail, agree, displease, fail, please, profit, serve and vail, were increasingly interpreted by speakers (and hearers) as transitives. In other words, the loss of the distinction between dative and accusative case marking on IOs and DOs respectively created functional uncertainty which in some cases resulted in the conversion of intransitive verbs + indirect object into transitive verbs + direct object.

To support his claim that dative complements lost their status as beneficiaries or recipients and were successfully reanalysed as direct objects Visser refers to passivization, since this is a well-established diagnostic for objecthood. Therefore, Visser concludes that if the former dative

10 Brackets have been added by me, as a native speaker of Dutch the author of this paper feels that the causativizing auxiliary doen ‘make’ is, at best, optional, thus the Dutch smelten can be considered as labile.

15

complement occurs as subject of a passive clause there is sufficient evidence for assuming that transitivation has taken place. However, Visser does acknowledge that this type of change does not happen overnight, but that an intermediary step would consist of marking the former dative complements by means of prepositions like to or at. Consider the following examples (Visser 1963: 127-129):

Table 7 ‘Reanalysis of dative complement as DO: cidan and derian’

Dative complement

[Ælfric, Hom. I, 156, 10]11 Seo menigu (þe eode beforan ðam Hælende) ciddon ðam blindan ‘The men (who were going before Jesus) chided the blind man’ [Metres of Boeth. (Krapp) 4, 36]12 Hio ful oft dereð unscyldegum ‘she often injures the guiltless’

intermediary PO

[Ælfric, Hom. I, 152, 17] Ða men (þe beforan þam Hælende ferdon) ciddon ongean ðone blindan ‘The men (who were going before Jesus) chided against the blind man’ [1588 Shakesp., L.L.L. IV, iii, 132]13 You chide at him [a1325 Prose Psalter liv, 3]14 Hii were derend to me in ire [Psalm 55,3] ‘for they bring down suffering upon me in their anger’

complement in zero or neutral case (>DO)

[1430 Lydg., Chron. Troy II, xii]15 Ye shall heare anone (=at once) how that he chit The quene Heleyne [c1400 Maundev., (Roxb.) iii, 9]16 nane schal dere oþer, ne let oþer to see

subject of passive clause

[1387 Trevisa, Higden (Rolls) VII, 35]17 þere Dunstan was strongliche despised and i-ched no passive attested with derian, it was replaced by ‘hurt’/’injure’ of French origin

Similarly, intransitive verbs requiring a complement in the genitive may have undergone the

same change as described above, provided that the complement in the genitive was formally identical to an accusative form or an indeclinable particle, e.g. þe (relative pronoun). For example: Beowulf, 1014, Fylle gefǽgon, ‘they rejoiced in the plenty’ and Ælfric’s Old English Genesis, 43, 4, 8, We bicgaþ þa þing ðe we beþurfan, ‘we will buy the food that we need’ (Visser 1963: 129). However, Visser notes that verbs taking a genitive complement such as gefægnian and beþurfan became obsolete very early; therefore, it is highly unlikely that this type of verb would have had significant influence of the process of transitivation described by Visser.

A crucial point, however, is that this kind of reanalysis does not give rise to P-lability. Instead, the verbs andswarian, bugan, cidan, derian, foryrnan, helpan, heorcnian, hyran and þancian, show A-lability, also commonly referred to as ‘absolute use of transitive verbs’, and not P-lability. The reason why I have chosen to include case syncretism in my overview of possible factors triggering P-lability is that on purely theoretical grounds it is possible that in some cases P-lability did in fact arise. In any case, it is somewhat unfortunate that Visser does not consistently differentiate

11 The Homilies of the Anglo-Saxon Church by Ælfric, translated by Benjamin Thorpe. 12 Krapp, G. P., ed. The Paris Psalter and the Meters of Boethius. (Anglo-Saxon Poetic Records; vol. 5.) New York: Columbia U. P., 1932; pp. 153–203 13 Love’s Labour’s Lost is an early comedy by William Shakespeare. 14 The Earliest Complete English Prose Psalter edited by Karl Daniel Bülbring, 1891 15 John Lydgate’s Troy Book is a translation of 'Historia destructionis Troiae' by the Italian writer Guido delle Colonne 16 The Travels of Sir John Mandeville printed for Roxburgh Club, 1889 17 John Trevisa translated Ranulf Higden’s Polychronicon from Latin to English.

16

between A-lability and P-lability throughout his description of ‘transitivation in the history of English’.

Although Visser’s grammar has been a useful source for Old and Middle English data throughout this paper, we must bear in mind that, according to today’s standards, Visser’s treatment of the concept ‘transitivation’ lacks appropriate nuances. We will come back to this issue in section 4 where we will show that it could be useful to differentiate transitive from non-transitive, on the one hand, and transitive from intransitive on the other hand.

3.3 Syntactic factors

3.3.1 Structural reanalysis of direct speech as direct objects with verbs expressing manner

of speaking

While it is true that many intransitive manner of speaking verbs have become construed with a direct object via reanalysis of the quoted words that accompanied them, especially in literary prose, e.g. ‘coo’, ‘bellow’, ‘falter’, ‘groan’, ‘insist’, ‘pule’, ‘pout’, ‘simper’, ‘smile’, ‘smirk’ or ‘titter’; I should stress that, more often than not, in these cases A-lability arises. Nevertheless, it is justified to mention this process in the context of P-lability for the more general class of sound emission verbs. I suggest that many of today’s sound emission verbs, e.g. ‘bang’, ‘beep’, ‘boom (out)’, ‘buzz’, ‘clack’, ‘clang’, ‘click’, ‘clink’, ‘jangle’, ‘jingle’, ‘sound’, ‘pop’, ‘ring’, might have become labile via the process described for the manner of speaking verbs. Example (9) illustrates the first transitive use of ‘buzz’ in 1583 found in the OED in which the originally intransitive ‘buzz18’ is used as if it was a manner of speaking verb. I propose that this might have paved the way for originally purely intransitive sound emission verbs to have become construed with an object which in turn might have given rise to P-lability.

(9) Having buzzed his venomous suggestion into their ears. (10) a. The bell buzzed. [examples taken from Levin & Rappaport (1994:67)]

b. The bees buzzed. c. The postmen buzzed the bell. d. *The beekeeper buzzed the bees.

Recently, sound emission – and for that matter also – light emission verbs have received special attention in the literature, e.g. in Levin (1993), Levin and Rappaport (1994) and Goldberg (1995) amongst others19. Verbs of sound emission, unlike the more specific manner of speaking verbs, can freely vary between animate and inanimate subjects. In my view this property is crucial for allowing P-lability. Following Levin and Rappaport, causative uses of sound emission verbs is licensed if the sound emitter is inanimate. External causation in the case of ‘buzz’ for example requires the possibility for direct manipulation of an external causer on the object. In (10c) ‘the bell’ can be caused to make a sound if someone pushes it, i.e. direct physical manipulation, whereas ‘the bees’ in (10d) do not fulfil this prerequisite and thus do not allow causative use (Levin & Rappaport 1994).

3.3.2 Competition between intransitive and reflexive forms

The common strategy for Germanic languages is to overtly mark inherent reflexivity, the only exception to this rule is English. In English typically self-directed actions such as ‘shaving’, ‘bathing’ or ‘dressing’ do not receive overt reflexive-marking (anymore), whereas in all other Germanic languages the marking of inherent reflexives is (still) common. Inherent reflexivity differs from another type of reflexivity, namely ‘literal’ reflexivity’ as McWorther (2002) calls it, and refers to actions that are typically conceptualized as other-directed such as ‘to kill’ in ‘he killed himself’, and

18 Although Visser does not dedicate a separate paragraph to verbs of sound emission, he does mention verbs such as ‘buzz’ in a more general enumeration of intransitive verbs (i.e. this list consist of no less than 506 verbs) that developed a transitive homonym over time. 19 Sound emission verbs also appear as verbs of directed motion: The bullet whistled through the window or The car roared up the driveway (Levin 1993:4)

17

are obligatorily expressed with the –self strategy in English. These two types of reflexivity are sometimes also termed intrinsic versus extrinsic reflexivity. Consider the following examples of intrinsic reflexive marking in Germanic: English ‘shave’ ‘turn’ German ‘sich rasieren’ ‘sich drehen’ Dutch ‘zich scheren’ ‘(zich)20 draaien’ Frisian ‘hy skeart him’ ‘(jin) omdraaie’ Afrikaans ‘hy skeer hom’ ‘sy draai haar om’ ‘she turns around’ Danish ‘barbere sig’ ‘dreje (sig)’ Norwegian ‘barbera seg’ ‘snu (seg)’ Swedish ‘raka sig’ ‘vrida sig’ Icelandic ‘Jon rakar sig, rakst’ ‘snúa sér, snúast’ sig = acc. – sér = dat. Faroese ‘raka sœr’ ‘snúa sœr’ Yiddish ‘goln sikh’ ‘vendn/kern sikh’

Evidence of this kind leads McWorther to conclude that inherent reflexive marking was a Proto-

Germanic feature. He notes that it commonly occurs with “motion verbs (German sich beeilen “to hurry”), psych-verbs (sich erinnern “to remember”), and verbs of social behaviour (sich benehmen “to behave”) (2002:220)”. In Old English, inherent reflexive marking was already optional (McWorther 2002, 222) and Hermodsson, as noted in the previous sections, reports that reflexives in Old English are relatively scarce21, especially reflexives denoting psychological states (1952:194). In fact, Hermodsson mentions (only) four examples of psych-verbs in OE that can be used reflexively, e.g. hine gebelgan ‘to make oneself angry’, hine forhtian ‘to be afraid’, hine sc(e)amian ‘to be ashamed’, hine fortruwian ‘to be over-confident’. The optionality of reflexive marking with these verbs is apparent after consulting the Bosworth-Toller Anglo-Saxon dictionary for gebelgan which can be used reflexively (11), intransitively (12) and transitively (13):

(11) Se wísa Catulus hine gebealg [refl. acc] [Bt. 27, 1; Fox 94, 32] [King Alfred’s Anglo-Saxon version of Boethius’ de Consolatione Philosophiae, edited by the Rev. S. Fox.] ‘the wise Catulus made himself angry’

(12) Gebulgon wið ða twegen gebróðru [intr.] [Mt. Bos. 20, 24.]

[The Gospel of St. Matthew, edited by J. Bosworth] ‘they were indignant at the two brothers’

(13) Ðæt he écean Dryhtne bitre gebulge [tr. with dat.] [Beo. Th. 4651; B. 2331.]

[The Anglo-Saxon Poem of Beowulf, edited by Benjamin Thorpe, Oxford, 1855.] ‘that he had bitterly incensed the eternal Lord’

Firstly, apart from proving McWorther’s claim that marking reflexivity was optional in OE, these

examples show how difficult it is to ascertain verb valency in OE, since based on the above examples gebelgan should be included in the list of labile verbs, but neither Visser nor Hermodsson do so. Secondly, they indicate that the loss of inherent reflexive marking may be an important factor in triggering P-lability. McWorther suggests (2002: 257) that the Old Norse speakers had “a tendency to omit the inherent reflexive pronouns in speaking English” because in Old Norse the free reflexive pronoun had weakened and transformed into a suffix that was added to the verb both for intrinsic and extrinsic reflexives, e.g. býsk ‘you arm yourself’, stask ‘he stabbed himself’, þeir setiask niþr ‘they sat down’. The use of free reflexive pronouns in ON, however, did sometimes occur, but only with

20 Brackets around the reflexive pronoun as in the Dutch example ‘(zich) omdraaien’ indicate that the reflexive pronoun is optional. In other words, brackets indicate competition between reflexive and intransitive forms. For the other languages no native speaker advice was available to me. 21 In OE approximately 195 are inherently reflexive, while in OHG there are about 483 (Hermodsson 1952:193)

18

datives and even then it was optional. Consequently, Old Norse speakers might have felt a certain redundancy in using the OE free reflexive pronouns, since in their native tongue inherent reflexivity was not or hardly ever overtly marked by means of an additional reflexive pronoun. To support this claim, McWorther adduces additional evidence from non-standard English dialects, i.e. the use of inherent reflexive pronouns “persists more robustly in areas with less Scandinavian influence (2002: 256)”, e.g. sit thee down and lay him/her down in Cheshire, Derbyshire, and Staffordshire.

However, to support our own hypothesis, i.e. that the loss of inherent reflexive marking caused P-lability, an important prerequisite is that the occurrence of reflexive pronouns should be frequent enough in the first place. It is clear that the omission of the reflexive pronoun can gives rise to intransitive constructions, but, if at the time the Scandinavian settlers arrived, reflexive pronouns were scarce already, and then it is possible that “something else” had set this evolution in motion. More precisely, as I have shown in section 2, inherent reflexives were as a rule rather infrequent in Old West Germanic dialects, except in OHG. In other words, the combination of the proneness of OE to prefer intransitive forms over reflexive ones and the influence of ON speakers might have tipped the scales for English to definitely favour intransitive forms over reflexive ones, thus paving the way for an unrestrained expansion of P-lability. While in the other West Germanic dialects where reflexives were scarce as well, it was the influence of OHG that might have tipped the balance towards the other side. After all, if Hermodsson’s data for MD is correct, we would expect that reflexive marking in this language would have been more similar to English.

A case in point that illustrates the scarcity of reflexives in OE is the Old English epic poem ‘Beowulf’ which was presumably written somewhere during the 8th and 11th century in a West-Saxon dialect. In this long poem of approximately 3000 metrical verses there are only 12 instances of reflexive constructions (verse numbers of these instances found in Hermodsson 1952:194):

(14) a. héold hyne syðþan - fyr ond fæstor (inf. healdan) [142]22

‘he held himself then – further and safer’

b. Wit unc wið hronfixas werian þóhton (inf. werian) [541]

‘we ourselves against whales planned to defend’

c. Nó ic mé an herewæsmun hnágran talige (inf. talian)

guþgeweorca, þonne Grendel hine; [677-8]

‘Of force in fight no feebler I count me,

in grim war-deeds, than Grendel deems himself

d. Hylde hine þa heaþodeor, hleorbolster onfeng (inf. hyldan) [688]

he bent himself down and the pillow received his cheek

e. Flod blode weol --- folc to sægon --- (inf. weallan) [1422]

‘The flood welled bloody --the folk stared at it—‘

f. syðþan he hine to guðe gegyred hæfde. (inf. gegyrian) [1472]

‘when he himself for war had equipped.’

g. Reste hine þa rumheort; reced hliuade (inf. restan) [1799]

‘rested him then, the large-hearted man; the hall towered’

h. hwæðre he hine on folce freondlarum heold (inf. healdan) [2377]

‘yet he maintained himself among his people with friendly instructions’

i. þæt ðá áglaécean hý eft gemétton: (inf. gemetan) [2592]

‘that the fierce enemies again met each other:’

j. hyrte hyne hordweard· hreðer aéðme wéoll (inf. ahyrdan) [2593]

‘the hoard-guardian heartened himself, his breast swelled with breath’

k. sigora waldend þæt hé hyne sylfne gewræc (inf. gewrecan) [2875]

‘victories' Ruler, that he avenged himself’

l. ac hé hyne gewyrpte þéah ðe him wund hrine (inf. gewyrpan) [2976]

22 Online version of Beowulf accessed via: http://www.heorot.dk/beowulf-rede-text.html

19

‘but he recovered, though the wound had touched him’

After scrutinizing these 12 examples, it is clear that, just as with the previous example gebelgan,

one verb could occur in more than one construction in Old English, i.e. there was competition between reflexive use and intransitive use23. At this point it is unclear to me whether differences in meaning existed between the intransitive and reflexive use of a verb and if so what these differences would imply, e.g. restan has also been attested intransitively; léofa Béowulf hafast þú geféred þæt þám folcum [dat. pl.] sceal Géata léodum ond Gár-Denum sib gemaénum ond sacu restan [Beo. 1857] ‘Beowulf dear: you have achieved that to both these people peace shall be common, to Geat-folk and Danemen, and that the strife shall be suspended’. The same is possible for gewyrpan (<gewyrpte) and weallan (<weol), however, in (e) weol seems to be used intransitively rather than reflexively, thus Hermodsson apparently unrightfully mentions this example here. A final prove of how difficult it is to determine verb-valency in Old English is shown by the verb hyldan. According to Bosworth-Toller dictionary hyldan can be used transitively, in which case the object is often a pronoun that is coreferential (=reflexive), and intransitively, e.g. Hié tó gebede hyldon, ‘they bent down to pray’. In any case, it should be stressed that this kind of variation is not unusual for older, non-standardized stages of languages, but it does make investigations into the rise of lability extremely difficult.

3.3.3 Voice syncretism: the development of passive and perfective morphology

English is not only unique when it comes to the vast expansion of P-lability; it also has some other features that make it stand out from its Germanic sisters. One such factor that will receive special attention in this section is the absence of variation in auxiliary selection in PDE. Variation in auxiliary selection between be and have is frequent in both Romance and Germanic languages, although it seems that the phenomenon is overall more persistent in Germanic than it is in Romance.

Apart from English, Swedish is the only Germanic language that has extended the use of have for all intransitives; whereas all the other modern Germanic languages still formally distinguish between two types of intransitive verbs by means of different auxiliaries in perfect tenses. Roughly we can say that be is favoured by unaccusative verbs, while have will occur more frequently with unergatives24. Among the Romance languages, however, several languages have generalized have for

In Spanish for example, haber ‘have’ was uniformly used to form perfectives by the 17th century; other Romance languages that generalized cognates from Latin habere are Portuguese, Catalan spoken in Barcelona, the Sicilian and Tarantino dialects in Italy, Rumanian and also Walloon (Sanchez Arroba 2010:611).

The situation in OE regarding auxiliary selection, on the other hand, is similar to that in most contemporary Germanic and Romance languages with auxiliary selection. According to McFadden & Alexiadou (2010: 391-392):

“Both the periphrasis with have and that with be had their pre-OE origins in stative resultatives, the former with transitives (like I have my bags packed), the latter with intransitives (like My bags are arrived). In the course of OE and early ME, these resultative constructions grammaticalized to become perfects—constructions with anterior temporal-aspectual meanings not restricted to resultativity. In the process, have spread to unergative intransitives like work (which, being atelic, could not have formed a resultative), while be was established as the norm with unaccusatives. Later in ME, around 1350, have began to replace be with unaccusatives. This process was gradual, involving a period of variation

23 McMillion argues that the last verbs to shed inherent reflexive marking were verbs with prototypical “middle” semantics, i.e. verbs that are typically conceptualized as self-directed. In other words, the first verbs to shed the reflexive morphology were non-core members of the “middle” verbs (McMillion 2006). 24 I use the terms unaccusative and unergative here as they were originally defined by Perlmutter (1978) in his ‘unaccusative hypothesis’, in which he proposes that the subject of an unaccusative intransitive verb is in fact an underlying object that has been promoted to subject position, whereas the subject of an unergative intransitive verb is an original subject.

20

lasting several hundred years, where the relevant verbs could appear with either have or be. The former appeared earliest and most consistently in modal and irrealis contexts, past and infinitive perfects, and clauses with iterative or durative semantics. From this point onward, be lost more and more ground to have, finally disappearing by the end of the nineteenth century”. [emphasis mine, JS]

Nevertheless, despite the difference in auxiliary selection, the resultative periphrastic construction in Old and Middle English with have and be had in common that they were both active in voice and that they referred to (or at least implied reference to) “past or anterior eventualities” (McFadden & Alexiadou, 2010:391), i.e. they were much like perfects, e.g.:

(15) a. I am come as ʒe bade me [Mirk’s festial, 1415]

‘I have/*am come as you asked me.’ b. he haþe foghten wyth þe fend [Mirk’s festial, 1415] ‘He has fought with the enemy.’

From this it follows that the perfect periphrasis with certain intransitives was formally identical to passive structures which were also formed with be and which implied an external agent. Visser argues that the resultative be periphrasis with originally intransitive verbs might have been interpreted by speakers as passives of transitive verbs, thus triggering P-lability. However, I will argue that some of the examples that Visser provides are clearly perfective and thus do not provide suitable environments for transitivation to take place, since perfectives do not imply reference to an external agent.

(16) To haue bene in welth and hartis blys, And now to be dekeit and in wo. [1513, Douglas,

Æneis XI, Prol. 148] (17) Fifteene hundred Foot, fiue hundred Horse Are march’d up to my Lord of Lancaster [1597,

Shakespeare, Henry IV, II, I, 187] (18) How that myn Egle fast by Was perched hye vpon a stoon [1379, Chaucer, House of Fame,

1991] ‘(I noticed) that my eagle was perched high upon a rock nearby’

In (16) the occurrence of a perfective have structure in the first part of the clause seems to favour

a perfective reading in the second part of the clause as well. Similarly, I suggest that march in (17) should be interpreted as perfective as well because originally intransitive motion verbs like march typically select be in perfective contexts. Although content wise (17) could include an external causer, i.e. someone who ordered soldiers to march to the Lord of Lancaster, we will agree with the OED which explicitly quotes example (17) to illustrate the obsolete use of intr. march in perfect formations with the auxiliary be. The first unambiguous (non-passive) tr. use of march in the OED is: “[Shall we] unsweare faith sworne, and on the marriage bed Of smiling peace to march a bloody hoast? [1623, Shakespeare, King John, III, i, 172]”. Whether or not the transitive use of march came into existence via passive interpretations first remains unclear, although on the basis of current evidence such an evolution of ‘march’ cannot be excluded.

Finally, example (18), which is also given in the OED to illustrate the transitive meaning of perch in passives, e.g. “to be settled on a perch”, is a bit trickier. However doubtful the interpretation of this type of clauses with be and past participle may be, I think it is clear that in (18) again no passive-agent is implied. Moreover, the OED explicitly states that intr. use is often used of birds sitting on a perch, therefore it seems more likely that this example predicates the resultant state of the eagle, not as a result of having been perched (i.e. put on a perch) by someone, but as a result of the bird having perched itself; thus again a perfective reading instead of passive is preferred.

Visser is undoubtedly right in suggesting that formal identity between passives and intransitive perfectives had occasionally triggered P-lability and that the ambiguous semantics of past participles when used attributively also contributed to this development, i.e. a progressed state could equally receive two different interpretations. Nevertheless we should keep in mind that this cannot have been the main reason for the elevated number of labile verbs in English. First, have

21

progressively started to replace be from a very early period and we know that with have no passive transitive interpretation was possible. Secondly, all other Germanic languages except Swedish did preserve both be and have in perfect tenses, but have not developed a tendency for P-lability anywhere near to what we observe in PDE.

However, in the domain of voice there is one additional feature that sets apart English, namely the absence of an eventive “become” verb in the passive next to the stative “be”. Widely accepted in the literature is that the beon /wesan-passive has always served to express both so-called ‘stative’ and ‘eventive’ passives, even in Gothic (van der Wal 1986), while OE weorthan was exclusively reserved for eventive passive readings. Nevertheless, according to McWorther, weorthan was already in decline in the OE period and had disappeared completely at the beginning of the Middle English period:

“Modern English has innovated the marking of passive with get (He got hit) and have (He had his hair cut), but are pragmatically constrained, encoding especial activeness on the part of the subject, with the have passive essentially a causative. […] “Overall, English remains the only Germanic language without a lexical item dedicated exclusively to expressing a pragmatically neutral manifestation of the passive (McWorther 2002: 239)”.

McWorther, however, does not relate the absence of a special form for ‘eventive’ passives to the rise of P-lability. (Nor any of the other changes in morphology or syntax that involve the loss of marking valency that he discusses in his article, e.g. disappearance of preverbal ge- and be- and the loss of inherent reflexives, the loss of be to mark intransitivity in perfect tenses).

While it is indisputable that the disappearance of weorthan makes English stand out from its sisters, we should be careful to link this to the rise of P-lability directly. It might be tempting to assume that the absence of weorthan created a gap in the English verbal system and that, subsequently, this gap was filled up by anticausative structures like the vase broke. However, the semantics of the anticausative construction do not seem to fit the semantics of passives entirely. Eventive passives always, although often not explicitly, imply an external agent, whereas anticausatives are neutral in this respect. Utterances like the vase broke can apply to situations where an external agent broke the vase - which incidentally seems more logical here - but also to situations where the breaking of the vase was self-instigated, i.e. some inherent property of the vase, i.e. quality of the ceramics, age, etc., caused it to break. In my opinion, it would be odd if we were to assume that a highly specific structure such as the one with weorthan came to be replaced by another structure that is essentially vague on a crucial point such as agentivity.

Moreover, according to Fryd (2009) the rise of get-passives co-occurs with the decay of weorthan, thus suggesting that the highly specific get-passives compensate for the loss of weorthan. It seems appropriate to add here that passives with werden in German or worden in Dutch, i.e. true eventive passives, necessarily imply an external agent. Anagnostopoulou (2003: 6-7), following Kratzer (2000), states that in German eventive passives are incompatible with a reflexive action, whereas stative passives are not, e.g.:

(19) a. Das Kind war gekämmt

The child combed its own hair or someone else combed the child’s hair. 'The child was combed'

b. Das Kind wurde gekämmt The child became combed 'The child was combed (by someone)'

(20) a. De man is geschoren The man shaved himself or was shaven by someone else. ‘The man is shaved’ b. De man wordt geschoren door zijn kapper/*door zichzelf. The man became shaved ‘The man is shaved’

22

This evidence further corroborates our argument that a structure like the vase broke which is essentially neutral when it comes to voice (cf. Davidse 1992) would be a replacement for an eventive passive structure that seems to favour a reading implying an external agent is highly unlikely.

3.4 External factors: Old Norse influence It is possible that a number of changes that make English unique among its contemporary

Germanic sisters can in fact be ascribed to the influence of Old Norse25, i.e. inherent reflexive marking in section 3.3.2. Starting from the late 8th century up until the 11th century, ON speaking Danes, who mainly settled in the northeast area called Danelaw, and later also ON speaking Norwegians, who settled in the northwest intermingled with the original Anglo-Saxon population who spoke a variety of northern Old English dialects. An important difference with the Norman Conquest is that the Scandinavian settlers, who were by and large an illiterate group of seafaring warriors, did not impose their native language on the autochthonous population through administration or legislature from above. On the contrary, the Scandinavian settlers adopted – or at least tried to adopt – Old English, since they were effectively included in the lower layers of the population.

Consequently, the socio-linguistic context during the period of the Scandinavian settlement was much more appropriate for contamination of Old English under the influence of ON than during the Norman Conquest. McWorther adduces several pieces of evidence both historical and linguistic. We have already mentioned the most important historical facts, i.e. time, place and type of invasion, nevertheless we must bear in mind that it is not possible to know every detail of this invasion, e.g. it is unclear what the exact number of Scandinavian invaders actually was. From a linguistic point of view, however, profound Scandinavian influence on the English language can hardly be denied. Apart from thorough lexical borrowing, i.e. some very frequent verbs (give, take, want, etc.), nouns (knife, horse, husband, etc.), adjectives (awkward, weak, flat, etc.) and even pronouns (they, their), there has, according to McWorther, also been influence on morphology and syntax.

In his article McWorther describes ten changes with respect to which English deviates from the general ‘Germanic typology’; an overview of these changes is given in table 8 below. Six of these changes can be ascribed directly to features that were present in Old-Norse; the remaining four changes McWorther attributes to the more general consequences of language contact and adult bilingualism. No less than four changes (i.e. the ones indicated in bold in table 8) described in this paper that might have triggered P-lability in English are directly related to ON. Therefore, if McWorther is correct, it would seem that it was the Scandinavian invasion which has thoroughly reshaped the face of the English language.

In order to understand how ON exactly influenced OE, I should first briefly explain some theoretical background. The rationale that underlies McWorther’s reasoning in this article crucially relies on the notions of simplification and complexification. More specifically, McWorther suggests that English is underspecified, i.e. less complex, in certain areas of the grammar than the other Germanic languages. Usually languages, not just Germanic languages, out-balance each other in terms of complexity in the different areas of their grammar, since the basic assumption within linguistics is that each language is approximately equally complex. English on the contrary is clearly exceptional in that it is the least complex or least Germanic language within the Germanic branch. Note, however, that this does not mean that English does not have unique features lacking elsewhere, e.g. English is the only language that has do-support thus in this area of the grammar English ‘overspecifies’ where the other Germanic languages ‘underspecify’.

25 Of course, ON is not the only language that has influenced OE, but it is the only language that could have had such a profound influence as to bring about far-reaching syntactical changes like the extraordinary expansion of P-lability. McWorther (2002) provides linguistic evidence contra Old French, Celtic or Old Low Dutch influence.

23

Table 8 ‘Underspecification in English compared to other Germanic languages’ (McWorther 2002:247)

G Du Y Fr S N Da I Fa A OE PDE 1) inherent reflexives

2) external possessors

3) gender beyond noun

4) loss of prefixes

5) directional adverbs

6) be-perfect

7) passive become verb

8) V2

9) singular you

10) indefinite pronoun

For matters of brevity I will only concentrate on those changes that could have triggered P-

lability, namely the loss of inherent reflexives, the loss of prefixes ge- and be-, be perfects and a passive become verb. These four changes show that English is characterized by a high degree of reduction or, as McWorther calls it, underspecification. In other words, PDE cannot overtly mark eventive passives; it does not overspecify intransitives in perfect tenses by means of be; it does not overtly mark inherent reflexivity; and last but not least – with the loss of certain prefixes – it ceases to mark valency distinctions on verbs, hence a great deal is left to context. Basically, the adult ON speakers adopted OE, but did so incompletely. Incomplete acquisition of the second language and reduction are common for adult L2 speakers, even if the two languages in question, i.e. ON and OE, are fairly similar. Thus, according to McWorther, the new socio-linguistic context that arose during the Scandinavian settlement was radical enough to bring about a drastic rupture in the development from OE to ME.

4 The bigger picture In this section I will evaluate Visser’s assumption that the rise of P-lability from OE to PDE is

triggered by a “wholesale process of transitivation” (Visser 1963: 127) and compare it with Hermodsson (1952) who claims that the rise of P-lability is due to a process of intransitivation. We will in fact argue, based on labilization data found in e.g. McMillion (2006) and Kitazume (1996) that both are right. The reason why an evaluation of transitivation is in place, is because Visser is often cited in works on valency changes in the history of English (cf. van Gelderen 2011) and because Hermodsson who’s work we have also heavily relied on supposes the complete opposite, i.e. intransitivation as main source of P-labile verbs. However, if we redefine the concept of “transitivity” in the light of recent typological developments Visser’s observation that many originally intransitive verbs have become transitive is in fact related to a more general tendency which has presumably taken place in all IE daughter languages.

As indicated, Visser attributes the rise of P-lability in English to the fact that many intransitive verbs came to be construed with an object over time. In section 3, I have elaborated five of six language internal changes that constitute this process of transitivation; (i) phonological attrition of ge- and decay of be-, (ii) Phonological coalescence of pairs such as scrincan – screncan, (iii) dative-accusative case syncretism, (iv) reanalysis of quoted speech as direct objects with verbs of saying and (v) voice syncretism between OE resultative perfect with be and the passive with be. I have not dedicated special attention to the remaining change involving intransitive verbs originally construed with a prepositional object, i.e. the tendency of replacing the PO with a that-clause

24

without preposition might have led to transitivation (Visser 1963: 133-134), e.g. ‘boast of’ – ‘I boasted that I did it’, because it simply affects a very limited number of verbs.

In fact, the frequency argument can be used for the refutation of at least 2 other changes. Pairs such as screncan-scrincan were not frequent enough to set in motion the process of labilization observed in English (McMillion 2006:197); similarly verbs of speaking – or for that matter verbs of sound emission – are also relatively infrequent and their application as true verbs of speaking is mainly limited to a literary register of written prose. Additionally, the limited number of P-labile verbs in OE, be it 55 as Visser suggests or 76 as Hermodsson suggests, could not have been the source for the analogical expansion of this category, because their number was simply too small. Moreover, in section 2 we have shown that the presence of P-lability in the Old West Germanic dialects was far from exceptional.

Nevertheless, it seems that Visser’s claim of transitivation is actually justified. In OE there were more than 200 purely intransitive verbs, whereas in PDE only about 20 of those are still exclusively intransitive (van Gelderen 2011 for a complete list PDE intransitive verbs). Although transitivation was probably not as significant in the rise of P-lability as Visser purports, it is true that many IE languages, including English have become more “transitive”. However, in this context we should not oppose transitive to intransitive, i.e. the presence of two complements versus the presence of only one complement, but rather oppose it to non-transitive. The presence of transitivity in the former sense is often taken for granted, but recent investigations have pointed out that PIE was in fact not a “transitive” language in this sense, instead it is conjectured that PIE might have been an “active language”, where the basic opposition that shapes the grammar is not one versus two complements but rather the nominal distinction between active and inactive26. Thus I propose that Visser’s argument in favour of “transitivation” should be considered in the light of the broader evolution from ‘active to transitive’ that the IE daughter languages in general have undergone.

Proof of this assumption may be found in the fact that Visser mentions the occurrence of transitivation with originally ‘impersonal’ intransitive verbs as well. Particularly, some intransitive impersonal verbs that were construed with a PO, e.g. of his falsenesse it dulleth me to ryme, ‘It distresses me to talk in rhyme of his falseness’ [Chaucer’s Canterbury Tales] and ‘personal’ intransitives with a so-called genitive of cause (Visser 1963:355), e.g. Ic gefeah ðæs weorces, ‘I rejoiced in the work’ are illustrative of “transitivation” in the broader sense. The presence of similar impersonal structures in all IE daughter languages is traditionally presented as an archaic remnant from an older ‘active’ language phase (for a more comprehensive account of OE as a language with active alignment see Toyota 2005). Similarly Bauer (2000) describes the development of OE impersonal structures governing a dative such as hit happed hem [Chaucer], to impersonal structures with prepositional complement in ME and PDE it happened to him; as well as the development of OE impersonal structures, governing a dative or accusative, to transitive constructions consisting of personal subject NP in the nominative, e.g. hit hreóweþ me became ich reowe hit ‘I regret’ and me loþeþ became I loþe ‘I loathe’ (Bauer 2000: 132-133).

In other words, we should be aware that when Visser describes the “wholesale process of transitivation” that English has undergone, he possibly confuses two different phenomena and this complicates the investigation of the causes for P-lability. First there has been a tendency in English to eliminate overt marking of valency, especially through the shedding of morphological, i.e. prefixes, and morphosyntactic features, i.e. inherent reflexive marking. This type of ‘underspecification’ or reduction, which makes English unique within the Germanic, may be attributed to the imperfect acquisition of OE by a large group of native ON speakers. Secondly, there has been a more general tendency in all IE daughter languages to prefer personal transitive structures of the type ‘I like it’ over the older impersonal on-transitive structures ‘It pleases me’.

26 Since C.C. Uhlenbeck’s very short seminal note published in 1901 on PIE case syncretism between accusative and nominative in neuter nouns, many linguists had argued that PIE must have had ergative case alignment. Vaillant (1936), Kuryłowicz (1958) and Shields (1978) among others made a strong case in favour of the PIE ergativity hypothesis. However in 1980 typological evidence from other ergative languages – mainly Australian and Caucasian – made the PIE ergativity hypothesis untenable, cf. Villar (1984) or Rumsey (1987) which led Gamkrelidze and Ivanov (1995) to reconstruct PIE as an ‘active’ language.

25

Ultimately it is of course the combination of both that has shaped PDE, but it is important for the argument developed in this paper that these two different evolutions are disentangled.

5 Conclusion My investigation of possible factors causing P-lability in English has revealed that the presence of

P-labile verbs within the broader context of Old West Germanic is more the rule than the exception. In fact, OE is a very well-behaved language when it comes to P-lability; whereas OHG constitutes the exceptional case within the West Germanic language branch because it has a very limited number of P-labile verbs. The apparent reason lies in the nature of inherent reflexivity; in OE, MD, OFris and OS there is both a different morphological means to express inherent reflexivity as well as a difference in frequency. It seems that in OHG inherent reflexivity is more than twice as frequently marked overtly, moreover OHG has in the 3rd person a separate morpheme, e.g. sich for marking reflexivity, whereas the other Old West Germanic languages dispose of the regular 3rd person pronoun to express reflexive meaning. From this we may conclude that the absence of a cognate of sich constituted a first step in the direction of P-lability.

Subsequently, in section 3 we have indicate that the inherent proneness of OE to prefer intransitive forms over reflexive ones has been reinforced during the Scandinavian invasion. Following McWorther (2002), we have argued that ON speakers tended to omit the reflexive pronoun as frequently as was the case in their native language, ON, when trying to adopt OE. Furthermore, the ON speakers might have also had an influence on the disappearance of the prefixes be- and ge-, the loss of be-perfects and weorthan.

Throughout this paper, we have argued, mainly on the basis of frequency that, phonological changes alone, causing intransitive and transitive verbs to coalesce, does not carry enough weight to set the evolution towards P-lability in motion, although it is probably is a facilitating factor. The same is probably true for voice syncretism, disappearance of be-perfects and case syncretism. Furthermore these developments are not unique to English, other languages that underwent the same changes or showed the same similarities did not develop P-lability to encode anticausativity like English did. Concerning the disappearance of weorthan I would like to suggest that this is probably irrelevant in this discussion for two reasons: first, weorthan and intransitive anticausative structures crucially differ when it comes to agency, i.e. the former necessarily implies the presence of an external agent, whereas the latter is essentially vague in this respect. Secondly, the disappearance of weorthan co-occurred with the rise of get as passive auxiliary (Fryd 2009).

I will conclude this paper by suggesting that in fact, OE is the only Old West Germanic language that has continued a tendency that appeared to be present in all other older North Sea Germanic dialects. It has done so to an extreme degree for several, possibly unrelated, reasons. Most importantly, there was the absence of a specific reflexive morpheme sich in combination with ON influence which lead to an even larger increase of intransitive structures at the expense of reflexive structures than was already the case. Moreover, the occasional confusion among speakers as to whether a certain verb form was transitive or intransitive may have facilitated this tendency even further, i.e. phonological attrition, case syncretism, reanalysis of quoted words and voice syncretism. Yet, we must bear in mind that many changes observed by Visser are in fact part of a larger overarching process of transitivation and that it is important that we do not mix up these two evolutions.

References Abraham, Werner. 1997. Kausativierung und Dekausativierung: Zu Fragen der verbparadigmatischen

Markierung in der Germania. In: Birkmann, Thomas et al. (eds.) Vergleichende germanische Philologie und Skandinavistik: Festschrift für Otmar Werner. Tübingen: Niemeyer. 13-28.

Anagnostopoulou, Elena. 2003. Participles and Voice. In: A. Alexiadou, M. Rathert & A. von Stechow (eds.) Perfect Explorations. Mouton de Gruyter.

26

Bauer, Brigitte. 2000. Archaic syntax in Indo-European: the spread of transitivity in Latin and French. Berlin: Mouton de Gruyter.

Benson, Thomas. 1701. Vocabularium Anglo-Saxonicum. Oxonii. Davidse, K. 1992. Transitivity/ergativity: the Janus-headed grammar of actions and events. In:

Davies, M. & L. Ravelli (eds.) Advances in Systemic Linguistics: Recent Theory and Practice. M. Davies. London/New York, Pinter: 105-135.

Eythórsson, Thórhallur. 1995. Verbal syntax in the early Germanic languages. PhD dissertation, Cornell University.

Fryd, Marc (ed.). 2009. The Passive in Germanic Languages. (Groninger Arbeiten zur Germanistischen Linguistik, 49.) Groningen: Rijksuniversiteit Groningen. Thematic issue.

Gamkrelidze T.V. & Ivanov V.V. 1995. Indo-European and Indo-Europeans: a Recpnstruction and Historical Analysis of a Proto-Language and Proto-Culture (part 1): text. (English version by Johanna Nichols). Berlin.

Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Strucutre. Chicago: University of Chicago Press.

Haspelmath, Martin. 1993. More on the typology of inchoative/causative verb alternations. In: B. Comrie & M. Polinsky (eds.) Causatives and transitivty. Amsterdam, John Benjamins Publishing Co. Vol 23: 87-121.

Hermodsson, Lars. 1952. Reflexive und intransitive Verba im ältern Westgermanischen. Uppsala: Almqvist & Wiksells.

Hiltunen, Risto. 1983. The decline of prefixes and the beginnings of the English phrasal verb. Turku: Turun Yliopisto

Kitazume, Sachiko. 1996. Middles in English. Word 47(2): 161-183. Kratzer, Angelika. 2000. Building Statives. Berkely Linguistic Society 26. Kuryłowicz, Jerzy. 1958. L’accuentation des langues européennes. 2nd ed. Wrocłav: Polish Academy. Levin, Beth. & Malka, Rappaport Hovav. 1994. A preliminary analysis of causative verbs in English.

Lingua 92(April): 35-77. Levin, Beth. 1993. English verb classes and alternations: A preliminary investigation. Chicago, University of

Chicago Press. Lindemann, J.W. Richard. 1970. Old English Preverbal Ge-: Its Meaning. Charlottesville, VA: University

Press of Virginia. McFadden, Thomas. & Artemis Alexiadou. 2010. Perfects, Resultatives, and Auxiliaries in Earlier

English. Linguistic Inquiry (41)3: 389-425. McFadden, Thomas. 2013. Resultativity and the development of Germanic preverbal ge- from Old to

Middle English. The 28th Comparative Germanic Syntax Workshop Universität Leipzig. October 5th, 2013.

McMillion, Allan. 2006. Labile verbs in English. Phd dissertation, Stockholm University. McWorther, John H. 2002. What happened to English? Diachronica (19)2: 217-272. Mitchell, Bruce & Fred C. Robinson. 1986. A guide to Old English. Oxford Blackwell. Ottosson, Kjartan. 2009. The anticausative and related categories in the Old Germanic languages. In:

Folke Josephson and Ingmar Söhrman (eds), Diachronic and Typological Perspectives on Verbs. Amsterdam: Benjamins, 2013.

Oxford English Dictionary. 2nd ed. 1989. OED online. Oxford University Press. 19 May 2014 < http://www.oed.com/ >

Paducheva, Elena V. 2003. Is there an “anticausative” component in the semantics of decausatives Journal of Slavic Linguistics 11(1): 173-198.

Perlmutter, D. M. (1978). Unpersonal Passives and the Unaccusative Hypothesis. Pokorny, Julius. 1959. Indogermanisches etymologisches Wörterbuch. Bern: Francke. Rumsey, Alan. 1987. The Chimera of Proto-Indo-European Ergativity. Lingua 71: 297-318. Sánchez Arroba, María Elena. 2010. Auxiliares “ser” y “haber” en los tiempos compuestos del

español antiguo. In: Deon Victoria Heffington & Marín Marín, A. (eds.). Memorias del VI Foro de Estudios en Lenguas Internacional. Universidad de Quintana Roo, Chetumal, Q. Roo. 611-622.

Shields, Kenneth. 1978. Some Remarks Concerning Early Indo-European Nominal Inflection. Journal of Indo-European Studies 6:185-210.

Somner, William. 1659. Dictionarium Saxonico-Latino-Anglicum.

27

Streitberg, Wilhelm. 1891. Perfective und imperfective Actionsart im Germanischen. Beiträge zur Geschichte der deutschen Sprache und Literatur (PBB) 15:70–177.

Toyota, J. 2005. Old English as a language with active alignment: evidence from word order. Papers from the 20th Scandinavian Conference of Linguistics, ed. by F. Karlsson, 1-18. Helsinki.

Trobevšek Drobnak, F. 1994. The Old English preverbal ge- in the light of the theory of language change as strengthening or weakening. Studia Anglica Posnaniensia. (28): 123-141.

Uhlenbeck, C.C. 1901. Agens und Patiens im Kasussystem der indogermanischen Sprachen. Indogermanische Forschungen (12) 170–172.

Vaillant, André. 1936. L'ergatif indo-européen. Bulletin de la Société de Linguistique de Paris (37):93-108. van der Wal, Maria J. 1986. Passiefproblemen in oudere taalfasen: Middelnederlands sijn/werden +

participium praeteriti en de pendanten in het Gotisch, het Engels en het Duits. Phd dissertation, University Leiden

van Gelderen, Elly. 2011. Valency changes in the history of English. Journal of Historical Linguistics 1(1): 106-143.

van Kemenade, Ans, and Bettelou Los. 2003. Particles and prefixes in Dutch and English. In Yearbook of morphology, ed. Geert Booij and Jaap van Marle, 79–118. Dordrecht: Kluwer.

Villar, Francisco. 1984. Ergativity and animate/inanimate gender in Indo-European. Kuhns Zeitschrift (97)2:167-197.

Visser, Fredericus Th. 1963. An historical syntax of the English language. Part I. Syntactical units with one verb. Leiden: Brill.

Abbreviations A Afrikaans Da Danish DO direct object Du Dutch f. feminine Fa Faroese Fr Frisian G German I Icelandic IE Indo-European intr. intransitive IO indirect object m. masculine MD Middle Dutch, 1150 - 1500 ME Middle English, 1150 – 1500 MHG Middle High German, 1050 – 1500 N Norwegian n. neuter OE Old English, 450 – 1150 OF Old French, 840 - 1340 OFris Old Frisian, 1200 - 1550 OHG Old High German, 500 - 1050 OS Old Saxon, 700 - 1100 PDE present-day English PDG present-day German PIE Proto-Indo-European PO prepositional object S Swedish tr. transitive Y Yiddish