Early English causative constructions and the “second agent” factor [1]

Brian Lowrey
University of Picardie - Jules Verne


This paper examines the types of complement constructions associated with the principal periphrastic causative verbs in Old English. It seeks to show how complement selection is determined, to some extent at least, by semantic factors, and by some of the causative universals identified by Song (1996). Complementation patterns occurring with the implicative causatives, for instance, are different to those associated with common non-implicative manipulative verbs. And among the causatives themselves, individual verbs tend to select different kinds of complement, associated with different types of causative meaning. One parameter that proves to be particularly relevant to the choice of both verb and complement is agentivity, not only that of the causer, an element that has often been highlighted in previous work, but also that of the causee, the “second agent,” and the extent to which causer and causee interact. It is argued here that this factor plays an important role in the distribution of causatives in Old English, as well as driving some of the changes which have taken place within the causative group. Indeed, this paper shows that three different causative verbs, (ge)don, make and cause, have undergone much the same type of evolution, at different stages in the history of English, each in its turn becoming increasingly associated with fully agentive contexts.

1. Introduction

My intention in this paper is to examine the manner in which the principal periphrastic causative verbs in Old English (OE) combine with different construction types to produce causative meaning, in the light of the causative universals described by Song (1996) and building on work in Lowrey (2010a) and (2011). I shall try to determine which semantic factors have a role to play in this process. In particular, I shall be concerned with agentivity and with what I call here the “second agent” factor, the status of the causee (the complement subject), and the extent to which the latter is presented as interacting with the causer (the subject of the causative verb).

2. Agentivity and causation

Agentivity has long been recognised as an important element in the expression of causation in general. Previous work has tended to concentrate on the agentivity (or otherwise) of the causer, and causer agentivity does indeed prove to be highly significant in determining which causative verb will be used and with which type of construction in a number of languages. For example, in Present-Day English (henceforth PDE), cause is less likely to appear with an agentive causer than make (Lowrey 2002: 73–83), and in modern Dutch, as Kemmer & Verhagen (1994: 143) point out, the distribution of the two main causatives, laten and doen, can be explained at least partially by the same factor, laten being used almost exclusively with agentive causers.

Causer agentivity surfaces frequently in the different cognitive models of causation put forward in the literature. For Delancey (1985: 5-6), an agentive causer permits the creation of what he calls a “two-stage chain of causation” in which “a decision on the part of the agent to perform an act causes the performance of the act, which in turn causes an event external to the Agent” (1985: 5). The presence of this initial, agentive stage in the causative process is, Delancey contends, marked either lexically or structurally in a number of languages. Delancey’s model of agentive causation is summarised in Fig. 1 below:

Act of volition Action Event

Figure 1. Agentive causation, Delancey (1985: 5)

This model, as we shall see, also inspires that suggested by Song (1996) in her cross-linguistic investigation of causative constructions. Both models, however, highlight the volition of the causer, and say little about that of the causee. It will be my contention here that the status of the causee in analytical causative constructions also needs to be taken into consideration, particularly when he/she is also an agent. Cole (1983) shows how the presence of an agentive causee influences factors such as case-marking on the complement noun phrase (NP) in causative constructions across a number of languages. The same parameter plays an important role in OE in the selection of both causative verb and complement type, as I shall show in the following section.

3. OE causatives

I shall start by identifying the main OE causative verbs and complement types, as well as those that only occur infrequently, with which I shall not be concerned in this paper. Verbs such as sendan, gespanan, or geniedan, the latter two with a manipulative sense, are all used occasionally in OE with causative meaning, as in the following examples:

(1) Ic Beda Cristes þeow 7 mæssepreost sende gretan ðone leofestan cyning Ceolwulf
‘I Bede, Christ’s servant and priest, have greetings sent to Ceolwulf, the dearest of kings 
(Bede, 2: 1) [2]
(2) … þæt he eac swylce Eorpwald Eastengla cyning Rædwaldes sunu to þone gespeon, þæt he forlæt þa idelnesse deofolgilda
‘…that he also induced Eorpwald, king of the East Angles, son of Raedwald, to give up the futility of devil-worship’
(Bede, 140: 26)
(3) he geniedde ægþer ge Perse ge Læcedemonie þæt hie gebetton þa burg þæt hi ær tobræcon
‘He forced both the Persians and the Lacedaemonians to repair the city that they had previously sacked’
(Orosius, 98: 26)

In addition, the verb lætan is sometimes used, usually in the V+I construction (see 3.1), as a causative rather than a so-called “permissive” verb (with the meaning of PDE let). [3] An example can be found in (22), below. It does not become frequent, however, until the end of the OE period, when it begins to supplant causative hatan (Lowrey 2011).

Finally, the verb habban is also very occasionally used as a causative, with a passive complement (Kilpiö 2010: 43-44). A (probable) example is given in (4) below:

(4) 7 æðelwald sæt binnan þæm ham mid þæm monnum þe him to gebugon, 7 hæfde ealle þa geatu forworht in to him, 7 sæde þæt he wolde oðer oððe þær libban oððe þær licgan
‘And Athelwald sat within the town with the men that submitted to him, and had all the gates shut upon him, and said that he would either live or die there’
(Parker Chronicle, 900)

However, as Kilpiö points out, habban remains a somewhat marginal member of the causative group in OE (he reports only 19 or 20 instances in the whole of the Dictionary of Old English Corpus). I shall not, therefore, say any more about it here, referring the reader instead to Kilpiö (2010) or to Łęcki (2010) for detailed discussion. [4] The bulk of the burden of expressing causation, throughout much of the OE period, is shouldered by two verbs, hatan and (ge)don, and it is on these elements that I shall concentrate in this paper.

3.1 Causative constructions with hatan

The presence among the causative group of hatan, whose basic meaning is something like ‘order, command,’ might at first sight seem surprising. After all, verbs of ordering tend not to be implicative verbs, to borrow a term from Karttunen (1971). They do not fit the definition of a causative suggested by Shibatani (1976: 1-2): “the speaker believes that the occurrence of the caused event is wholly dependent on the occurrence of the causing event; the dependency of the two events here must be to the extent that it allows the speaker to entertain a counterfactual inference that the caused event would not have taken place, provided that all else had remained the same.” Ordering someone to do something does not entail that the act was carried out, unlike making someone do something, which does, as the contrast between (i) and (ii) shows:

(i) I ordered them to leave the room but they didn’t leave the room
(ii) *I made them leave the room but they didn’t leave the room

Make is thus usually accorded causative status, whereas order remains a non-implicative verb in PDE. In OE, however, hatan is used implicatively on many occasions, to express not just the giving of an order but also the fact that the act was carried out, as in the following example from the Anglo-Saxon Chronicle:

(5) Her Offa Miercna cyning het Æþelbryhte rex þæt heafod ofaslean
‘At this time Offa king of the Mercians had king Athelbright’s head cut off’
(Parker Chronicle, 792 [actually 794])

The causative construction here is expressing more than the simple giving of the order to execute Athelbright. What the chronicler is telling us is that this is the year in which the unfortunate monarch met his end. Royster (1918) was perhaps the first to insist upon the essentially causative status of hatan in OE. A series of arguments and examples in support of this idea were given in Lowrey (2010a, 2011), and I shall not labour the point any further here. Suffice to say that hatan is used as a causative especially in what Denison (1993: 165) calls the V+I (“Verb + Infinitive”) construction, as in (5), where the causee is understood. The Gricean inference that the order was indeed carried out seems to have been lexicalised to a large extent in this construction (Lowrey 2010a: 76-79). V+I typically expresses what I shall call “fully agentive” causation, where both the causer and the causee are Agents (which of course fits the basic “ordering” semantics of hatan). The embedded predicate in this construction is invariably telic, an “accomplishment” in the classification proposed by Dowty (1979). A similar tendency can be oberved in both OE (Lowrey 2010a) and Middle English (ME) (Lowrey 2002, 2010b).

The causative use of hatan is interesting with regard to the typology proposed by Song (1996), based on a sample of over 140 languages. Song identifies three basic causative types, called respectively the “AND” type, the “PURP” type, and the “COMPACT” type (1996: 17). The first two indicate a tendency for lexical items evoking either linear chronological sequences (the AND type), or purpose or intent (the PURP type) to be pressed into service as markers of causation. The third type describes cases where the causative element and the embedded verb are considered to undergo some degree of fusion, and is characterised by what Song calls “the contiguity or the compactness of V[cause] and V[effect]” (1996: 21). The COMPACT causative is presumably to be seen as a more highly grammaticalised form, although Song does not use the term herself, describing the COMPACT type rather as the “diachronic residue” of the AND and PURP types (1996: 134).

In view of Song’s typology, the use of hatan as a causative might seem less surprising, in that the notion of ordering could be linked semantically to both the AND and PURP types. After all, ordering implies a linear chronological sequence, whereby the giving of an order precedes its execution, and also implies some kind of purpose or intent: generally, one orders to be done what one wishes to see done. Hatan therefore would seem to be a likely candidate for incorporation into the causative group. An interesting question is how far it goes along the way to becoming a COMPACT causative. Certainly, there is evidence that hatan begins to evolve towards greater compacity, at least in some of its uses, as the following examples, where hatan and the infinitive of the complement are contiguous, show:

(6) … 7 ic hit het weorpan on fyr
‘ … & I had it thrown onto the fire’
(Heptateuch, Exodus, 124: 19)
(7) ic þe het feccan þæt þu mine find wirigdest
‘I had you fetched so that you might curse my enemies’
(Heptateuch, Numbers 152: 3)

The accusative pronouns hit and þe, of course, are the direct objects of weorpan and feccan respectively, the “patients” in the terminology of Kozinsky & Polinsky (1993), occuring in preverbal position, as was often the case in OE. Note, however, that they surface not just before the infinitive, but before the causative as well, exactly as if they had been analysed as arguments of a verbal complex composed of hatan and the infinitive. Although both verbs remain free morphemes, it would appear that they have begun to function as a single semantic and syntactic unit.

Song’s typology comprises varying degrees of compacity, ranging from the lexical causatives, at one extreme, where no distinction can any longer be made between the morphemes corresponding to her V[cause] and V[effect], to the causative constructions of Romance languages like modern French, where causative faire and the complement infinitive can in some circumstances be separated. Indeed, the latter show many similarities with the hatan V+I of (6) and (7). Faire and the infinitive remain free morphemes, but clitic pronouns must precede the complete faire + infinitive sequence, as in:

(8) Je la ferai réparer dès que possible
I it will do-repair as soon as possible
‘I will have it repaired as soon as possible’

Both display what Song considers to be “the essential property of the COMPACT type: the physical adjacency of V[cause] and V[effect]” (1996: 28). [5] The difference between the modern French faire causative and the hatan causative is that, whereas the order of (8) is basically the only option available to French speakers, hatan in OE also permitted another variant of V+I, as in (5), where arguments of the infinitive could indeed intervene between it and the causative. If COMPACT causatives are indeed more heavily grammaticalised forms, then the hatan causative in OE would seem to have reached a less advanced stage in the grammaticalisation process than modern French faire.

3.2 Causative constructions with (ge)don

As well as hatan, the verb (ge)don is also widely used with causative meaning, and the two verbs seem to be, to an extent at least, in complementary distribution. They select different types of complements, and are used to express different kinds of causative meaning. (Ge)don, unlike hatan, rarely occurs in the V+I construction. It is common, on the other hand, with two complement types: a “small clause” and a finite þæt clause (V+ÞÆT). Parallel to this syntactic contrast, certain semantic properties also oppose the two verbs, the most important of which appears to be agentivity.

Song’s “Goal → Event → Result” model, inspired by Delancey (1985), once more insists upon the agentivity of the causer at the Goal stage. The important factor in my OE corpus, on the other hand, where all the causers are Agents, seems to be the status not of the causer but that of the causee. [6] (Ge)don appears whenever the causee is not an Agent. As might be expected, the complement NP in the case of small clause complements is generally a Theme:

(9) Uton nu gehyran ond geþencean hwæt he dyde, ond mid hwy he us freo gedyde
‘Let’s now listen and think about what he did, and how he set us free’
(Blickling, 58: 21)

However, it also proves to be the case of the great majority of subject NPs of a finite verb in a þæt complement clause:

(10) 7 heo gedyde mid hiere lare þæt ealle Mæcedonie wæron þæm cyninge wiðerwearde
‘And through her teaching she made all Macedonians hostile to the king’
(Orosius, 148: 6)
(11) Ond he swa cwæð: Gedoþ þæt eow sy mete gearo on minum huse
‘And so he said, see to it that there is food ready for you in my house’
(Blickling, 26: 15)

Sometimes, the complement NP is, theoretically at least, an Agent. Such cases, however, are comparatively rare, and even here the agentivity of the causer often appears somewhat weak:

(12) Æfter minum deaþe God cymð to eow and deþ þæt ge farað of þison lande to þam lande þe he swor Abrahame
‘After my death God will come to you and see to it that you leave this land for the land that he promised to Abraham’
(Heptateuch, Genesis 88: 24)
(13) Ne nim þu wif of hira cynne þinum sunum, þe læs þe hig gedon þæt þine bearn singion on heora godas
‘Do not take a wife for your sons from among their people, lest they cause them to sin before their gods’
(Heptateuch, Exodus 127: 13)

Some would analyse examples such as these as cases of “indirect causation.” The term needs to be used with caution, however, as the notion of (in)direct causation has proved notoriously difficult to define (see, for example, the definitions proposed by Baron 1974, or by Kemmer & Verhagen 1994). The relevant parameter in (12) and (13) might be, rather, the absence of what Terasawa (1985: 133) calls the “ACT ON” factor: no direct manipulation is implied of one Agent by another in order to bring about the situation. [7] Interestingly, as my translation suggests, these are cases where in PDE (at least in the author’s own native, southern British Isles, variant of English) cause would probably be preferred to make in order to avoid such an inference.

Ikegami (1981), too, observes that (ge)don can appear with small clause and þæt complements, although his paper focuses more specifically on early ME. Ikegami suggests (1981: 89) that the causative use of (ge)don with a small clause (and, later, with an infinitive complement) may have evolved from (ge)don V+ÞÆT, although, as he himself recognises, this is somewhat speculative. Ikegami gives no attested examples in support of his position. My own findings show that there is at least some evidence to suggest that the opposite may be true, and that (ge)don V+ÞÆT may in fact have evolved from the small clause construction. In a text such as Orosius, for instance, causative (ge)don is particularly common with a type of small clause comprising a Theme NP and a prepositional phrase (PP) headed by to, as in (14) and (15):

(14) þa dyde he him eac þa ricu to gewealdon
‘He also brought the kingdoms under his sway’
(Orosius, 114: 28)
(15) … oþ Romane gefengon Somnita cyning 7 heora fæsten abræcon 7 hie to gafolgieldum gedydon
‘… until the Romans took the king of the Samnites and sacked their fortress and made them tributaries [and them to tribute put]’
(Orosius, 122: 24)

Examples such as these suggest a link with the “put” sense of (ge)don, the idea of “moving” something towards a state or result. One can quite easily imagine how the þæt construction might have evolved from small clauses of this type, especially in the light of examples such as the following:

(16) hi Læcedemonie mæst ealle awestan 7 hi to þon gedydon, þæt hi hi selfe leton  ægþer ge for heane, ge for unwræste
‘They destroyed almost all the Lacedaemonians, and caused them to consider themselves both abject and weak [and them to it put, that they themselves considered…]’
(Orosius, 98 :22)

Contexts such as (16) could have served as a kind of “bridge” between the small clause and þæt constructions. The þæt clause in (16) is the complement of a pronoun, þon, whose contribution in semantic terms is limited, to say the least. The PP in this kind of context could easily have been felt to be redundant, and ultimately disappeared, the þæt clause being reanalysed as the complement of (ge)don. All of this remains very speculative, of course, and difficult to prove. If such a derivation occurred, it must have done so at a very early stage. (Ge)don already occurs both with a small clause and with a þæt clause in comparatively early OE texts. Moreover, Royster (1922: 334-5) points out that both “nominal object plus a predicative adjective” constructions as well as finite clauses appear under the causatives don in Old Saxon and tuon in Old High German, as well as with (ge)don in OE, suggesting a common Germanic origin. Timofeeva (2011: 99) notes no change in the frequency of (ge)don V+ÞÆT between early and late OE, whereas we might to expect to see some kind of increase if it were a recent innovation, as it spread to new contexts.

3.3 Causatives and manipulatives in OE

Complement selection also distinguishes the causatives hatan and (ge)don from the traditional manipulative verbs, which entail some kind of manipulation of one entity by another with a view to bringing about a situation, but without implying that the situation necessarily occurs. Since it would obviously be beyond the scope of this paper to examine the complements all the manipulative verbs used in OE, I shall take two common OE manipulatives and compare them with the causatives. These are bebeodan (‘order, bid’), theoretically close in meaning to hatan, and biddan (‘ask’). Typically, these verbs tend to occur in a V+NP+ÞÆT construction, in which the matrix verb has two internal arguments, the þæt clause and an intervening, independent NP. This NP and the subject of the finite clause are generally co-referential, as in (17) and (18):

(17) Uton feallan to ðære rode, and þone Ælmightigan biddan þæt he us ahredde wið þone modigan feond þe us afyllan wile
‘Let us fall down before the Cross, and ask the Almighty to deliver us from the arrogant devil who wants to bring us down’
(St Oswold, vol. II 126: 19)
(18) Forþon se apostolica papa bebead Theodore biscope, þa he from him ferde, þæt he him on his biscopscire gerisne stowe foresege
‘For the apostolic pope ordered bishop Theodore, when he left him, to provide a suitable place for him in his diocese’
(Bede, 256: 32)

These appear to be three-place structures, in which the complement NP is both an independent argument in its own right and also an integral part of a third, clausal argument. For the sake of simplicity, I have used the same term, V+NP+ÞÆT, to describe the structures illustrated by (17) and (18), although they are not, strictly speaking, identical. With biddan, the independent NP is in the accusative, with bebeodan in the dative. Nonetheless, these verbs, as far as complement selection is concerned, form a striking contrast not only with (ge)don, but also with hatan, as we can observe if we compare the distribution of the four verbs across individual texts. The relevant figures for the Heptateuch and for the Bede and Blickling texts are as follows:


Table 1. Occurrences of biddan, bebeodan, hatan, and (ge)don with finite and infinite complements, by complement type (Heptateuch)

13 [8]

Table 2. Occurrences of biddan, bebeodan, hatan, and (ge)don with finite and infinite complements, by complement type (Blickling)

3 [9]

Table 3. Occurrences of biddan, bebeodan, hatan, and (ge)don with finite and infinite complements, by complement type (Bede)

Allowing for the expected degree of variation, the picture that emerges from Tables 1–3 is a largely consistent one, the verbs concerned being distributed in similar fashion over all three texts. (Ge)don occurs more frequently, compared to the other verbs, in the Heptateuch, but always in the same V+ÞÆT construction. The tables also hint at some degree, at least, of diachronic stability, much the same distribution pattern occuring in Bede and Blickling (two O2 texts, according to the Helsinki Corpus dating system), as in the later O3 Heptateuch. [10] Tables 1–3 confirm the extent to which hatan is a frequent verb in OE, occurring in a variety of constructions, including V(+NP)+ÞÆT and Visser’s (1973) VOSI. [11] In these instances, it functions for the most part as a straightforward manipulative verb. However, it is also particularly common in the V+I construction, where it generally functions as a fully-fledged causative (Lowrey 2010a, 2011). Both biddan & bebeodan, on the other hand, are absent from V+I.

As for (ge)don, we can observe that it too is absent from the V+I construction, and displays a marked preference for finite complements. These are not quite the same finite complements as those of manipulative biddan, or bebeodan, however. Firstly, (ge)don only occurs in V+ÞÆT structures, never in V+NP+ÞÆT. In other words, it only functions with what appears to be a two-place argument structure, like the so called “pure” causatives (Terasawa 1985: 134-35) such as PDE cause. Secondly, the clausal complement of biddan and bebeodan always contains either a finite verb in the subjunctive, as in (17) and (18) above, or an infinitive preceded by a (pre-)modal, as in the following examples:

(19) …se ealdorman (…) bebead his burðegne þæt he gebringan sceolde into his gebeorscipe þa foresædan Iudith
‘The prince ordered his chamberlain to bring the aforementioned Judith to his feast’
(Judith, 361: 239)
(20) þu bæde me leof þæt ic sceolde ðe awendan (…) þa boc Genesis
‘You asked me, dear lord, to translate (…) the book of Genesis for you
(Heptateuch, Prefatio 3: 2)

(Ge)don, on the other hand, could be (and often was) followed by an indicative, especially in later OE:

(21) 7 Drihten wæs mid him 7 gemiltsode him 7 gedyde þæt þæs cwearternes ealdor him wærþ swiþe hold
‘And God was with him & was merciful to him & caused the chief jailor to become kind to him’
(Heptateuch, Genesis 69: 8)
(22) Læt hi ealle fordon, 7 ic gedo þæt þu hæfst tyn þusend punda to þinum mydercum
‘Have them all destroyed, and I shall see to it that you have ten thousand pounds in your coffers’
(Esther, 143: 129)
(23) Ic eow geseo and do þæt ge weaxað
‘I shall see you and cause you to multiply’
(Heptateuch, Leviticus 136: 11)

I have excluded from this comparison preterite forms, where the morphological distinction between subjunctive and indicative is often blurred.

If nothing else, greater compatibility with the indicative provides further evidence for the causative status of (ge)don. Song (inspired by Givón 1974: 270) considers that “subjunctive tends to occur in the complements of non-implicative, weak manipulative verbs” and that it is often used to indicate “nonfactuality” (1996: 57). Presumably, therefore, one could interpret the use of the indicative with (ge)don as a mark of the verb’s causative status, a marker of implicativity and factuality, and the expression in the syntax of a clear semantic distinction between (ge)don and the manipulatives.

4. Changes within the causative group: into ME and beyond

The second agent parameter, which determines to a large extent the distribution of hatan and (ge)don in our corpus, also plays an important role in some of the changes which take place within the English causative system, where there exists a tendency for causatives initially used to express non-agentive causation to evolve towards fully agentive contexts.

In fact, changes among the causatives are relatively few until the very end of the OE period. Unfortunately, the major change, the disappearance of hatan from causative contexts, takes place just as the old West Saxon standard begins to collapse, which makes its causes and its consequences difficult to trace. Nonetheless, we do know that causative use of hatan declines rapidly (although hatan continues to function as a non-implicative, manipulative verb) as of the middle of the 11th century (Lowrey 2011 and forthcoming). In very late standard OE texts, hatan is replaced by lætan in what appears to be a straightforward lexical change. As we move into the ME period, and regional forms come to the fore, a variety of verbs take its place. In the West and in the South, for instance, let becomes the “central” causative, commonly used in fully agentive contexts. Notably, it becomes frequent in the V+I construction, from which hatan has all but disappeared (Lowrey 2011). In Northern and North-Eastern dialects, the same role is filled by gar, which also occurs, in agentive contexts, in the V+I construction:

(24) Þei garte bringe þe mere sone / Skabbed and ful iuele o bone
‘They had the mare, all scabby and in poor condition, brought straight away’
(Havelok, ll. 2504-05)

Gar also appears in VOSI constructions with agentive causers and (explicit) causees where, much like PDE make, it can express coercive meaning:

(25) Þer-on he garte þe erl suere / Þat he sholde yemen hire wel,
‘He made the earl swear thereupon that he would take care of her properly’
(Havelok, ll. 190-91)

What happens in the Eastern dialects is of particular interest, in that it shows how the whole causative sub-system is affected by the loss of hatan. Evidence in the final continuation of the Peterborough Chronicle (the easternmost of all the Chronicles, and the only one to continue on into the 12th century) shows that don has undergone a considerable semantic shift. It appears henceforth with agentive causers and causees:

(26) Þerefter þe biscop of Wincestre Henri þe kinges brother Stephnes spac wid Rodbert eorl 7 wyd þemperice 7 suor heom athas (…) 7 sæde heom þæt he uuolde iiuen heom up Wincestre 7 dide heom cumen þider
‘Afterwards Henry, the bishop of Winchester, king Stephen’s brother, spoke with earl Robert & with the emperess & swore oaths to them (…) and told them that he would give Winchester up to them and had them come there’
(Peterborough Chronicle, 1140 )

It, too, can express coercive meaning in the presence of two Agents, much like gar in the Northern dialect:

(27) Was it noht suithe lang þer efter þatte king sende efter him 7 dide him gyuen up þæt abbodrice of Burch 7 faren ut of lande
‘It wasn’t long afterwards that the king sent for him and made him give up the abbey of Peterborough and leave the country’
(Peterborough Chronicle, 1132)

If we follow the evolution of (ge)don from OE into Eastern ME, a pattern starts to emerge, comprising several stages. I shall assume that causative (ge)don is used initially with small clause complements, where the NP is a Theme. This use is then extended to finite complements where the causee is also a Theme, or at least non-agentive. (Ge)don remains, in both constructions, strictly neutral with regard to Agent-Agent interaction, and is opposed in this respect to hatan, the agentive causative. Finally, it spreads to fully agentive and even coercive contexts, filling the gap left by hatan. The ME form, don, begins to appear in the V+I construction in Eastern texts at the beginnining of the ME period: [12]

(28) On alle wise he fandeð hu he mu3e gode weorkes letten, oððe mid ofðanche and mid sarinesse and unbleðeliche hes don don
‘In every way, he tries [to find] how he may hinder good works or have them done with regret, with sorrow, and unhappily’
(Vices & Virtues, Of Sorinesse 1: 4)

The path followed by (ge)don is interesting in at least two respects. Firstly, the transition from minimally to maximally agentive and even to coercive contexts corresponds nicely to one of the scales of “causative transitivity” defined by Hollmann (2005). Hollmann’s objective is to use this notion to explain why some PDE causatives can be passivised more readily than others, which does not concern us directly. However he draws up a series of scales in order to measure transitivity, one of which, “volitionality and agency” calls upon the very parameters discussed here. Using the classification of causative types proposed by Talmy (1976, 1985b, 1988), as well as parametres suggested by Croft (1991), Hollmann proposes the following hierarchy of causative transitivity parametres:

inducive ˂ volitional ˂ effective ˂ physical

Hollmann considers that “[p]hysical causation is the least transitive, as both A [the causer] and O [the causee] are inanimate. The inductive type, conversely, is the most highly transitive, featuring as it does an animate causer and causee. Volitional and affective causation are somewhere in between, both of them having one mental and one inanimate participant” (2005: 205). (Ge)don, from this perspective, appears to move right across the scale, from minimally transitive to maximally transitive contexts. The factor which determines, at least in part, which causatives can be passivised also plays a key role in shaping the evolution of one of the principal OE causatives.

More interesting still, perhaps, is the fact that the evolution of causative (ge)don bears a striking similarity to that of causative macian/make. Ikegami (1981: 91) concurs with Royster (1922: 353) when he states that “in any significance, macian is an infrequent word in Old English writing.” It only comes to be used as a causative in late OE where, like (ge)don, it first appears with causative meaning in small clause complements in which the complement NP is a Theme. Examples can be found in Ælfric’s prose:

(29) and swaðeah Paulus siððan forestop Stephanum on Godes gelaðunge mid menigfealdum geswincum, þone ðe he ær ehtende martyr gemacode
‘…but nevertheless Paul, through many labours, afterwards preceded Stephen, whom he had previously made a martyr through persecution, in God’s congregation.’
(Parable of the Vineyard, 82: 22)

Once again, much like (ge)don, it too begins to appear with a finite þæt complement:

(30) Ge habbaþ us gedon laþe Pharaone and eallum his folce and gemacod þæt hig wyllað us mid hyra sweordum ofslean
‘You have made us hateful to Pharaoh and to all his people, and made them want to slay us with their swords’
(Heptateuch, Exodus 96: 14)

As the presence of (ge)don with a small clause complement in the same sentence suggests, the two verbs would presumably have come into competition at this stage. Macian seems to be filling the void left by don as it begins to shift towards increasingly agentive contexts.

As we advance into early ME, and infinitive complements become increasingly common, maken continues to function as an agentively neutral causative:

(31) for idelnesse makeð mon; his mon-scipe leose
‘For idleness causes a man to lose his valour’
(Brut, l. 12434)

Only at a later stage does it spread to contexts involving an agentive causer, and increasingly, to fully agentive contexts (Lowrey 2010b and forthcoming), where it too starts to express coercive meaning. According to the OED this finally occurs as late as the 16th century, though this is perhaps a little over-cautious. There exists clear evidence that by the end of the 14th century, the make causative could already express coercive meaning:

(32) He yaf me al the bridel in myn hond, / To han the governance of hous and lond, And of his tonge, and of his hond also; / And made hym brenne his book anon right tho
‘He gave me free rein to govern his house and his land, and his tongue and his hand too, and [I] made him burn his book straight away’
(Chaucer, Canterbury Tales, Fragment D ll. 813-16)

Two different causative verbs, then, at different stages in the history of English, both begin at the non-agentive end of the causative spectrum, before moving all the way across to the opposite end, where they are associated with agentive causers and causees and even with coercive meaning. If we advance still further in time, we notice that a very similar phenomenon affects the verb cause. The first recorded instances of the use of cause as an analytical causative date from the beginning of the 14th century. In its earliest uses, much like (ge)don and macian before it, cause is typically associated with non-agentive contexts, as in the following example from the Confessio Amantis:

(33) And every gladschipe, as I finde, / Is confort unto loves kinde / And causeth ofte a man to spede
‘And every delight, as I find, is a comfort to love’s nature, and often causes a man to prosper’
(Gower, Confessio Amantis, Liber Quintus III l. 4795)

The subsequent evolution of cause is particularly interesting. By the early 16th century, it too appears frequently in fully agentive contexts, as in the following example:

(34) And for thys cause (as a goodly continent prince clene & fautles of himself, sent oute of heauen into this vicious world for the amendment of mens maners) he caused the bishop of London to put her to open penance, going before the crosse in procession vpon a sonday with a taper in her hand.
(More, Richard the Third, 53: 27)

Cause, with a passive complement, even comes to be used for a short period as a possible replacement for the V+I construction, when the latter is lost at the beginning of the 16th century (Lowrey 2002: 380-83), as the following examples show:

(35) Scilla, for the malignitie that he hadde towarde Marius, caused the heedes of a thousande and seuen hundred of the chiefe citezins of Rome to be striken of, and brought to hym fresshe bledyng and quicke
(Elyot, Boke Named the Governour, 19: 4)
(36) Whan Philipp Demetrius sonne, was about the Citie of Scio, and had layed siege to it, he caused to be proclaymed, that what ever bondemen woulde forsake the Citie and flee to him, he promised them liberty and their maisters wives
(Hoby, Book of the Courtier, 24: 12)

Of course, cause ultimately fails to establish itself as an agentive causative, and, once make has become the central causative, reverts to something like its original role, although the mechanisms of this process lie beyond the scope of this paper.

5. Conclusion

I have tried to show how certain types of complement construction combine with the various causative verbs in OE to express different types of causative meaning. Although one finds, as might be expected, some degree of variation in the choice of verb and of complement, there nonetheless exists in OE a relatively high degree of correlation between the choice of causative, the choice of complement type, and a number of semantic factors. Implicative causatives in OE tend to appear in different syntactic environments to manipulative, non-implicative verbs. Causative hatan shows a clear preference for the V+I construction, while in its non-causative uses the same verb is common with in VOSI or finite constructions. Causative (ge)don is used almost exclusively with a small clause or a finite complementation. I have also tried to show here how one of the most significant factors involved in determining the choice of verb and of complement is the agentivity not only of the causer but also of the causee. The same parameter, as well as going some way to explaining the distribution of OE causative constructions, also proves to be a significant factor in subsequent changes which affect the causative group, during the ME period and beyond.


[1] I should like to thank the anonymous referees for all their very helpful comments, criticisms, and suggestions. Any remaining errors, omissions, or inconsistencies are, of course, my own.

[2] Under the system adopted here, for OE texts (except the Anglo-Saxon Chronicle, where references are given by year), the figures appearing after the name of each text refer to the page and line numbers, separated by a colon, in the edition quoted in the Primary Sources.

[3] It only makes sense to speak of “permission” when both the subject of let and that of the infinitive are animate Agents. The term suggested by Talmy (1985a: 301) and (2000: 424-25) to define the concept of letting, “cessation of impingement,” the idea that the causer brings about the complement situation by ceasing to prevent it from arising, is perhaps more accurate.

[4] See also Visser (1973: §2118), whose earliest examples are from Middle English texts.

[5] Song simplifies matters somewhat by insisting upon the sole continguity parametre in seeking to define her COMPACT type. The idea of causative “fusion” raises a number of issues, especially from a syntactic perspective, which have been debated at great length in the litterature on causation: amongst others, the extent to which the argument structures of V[cause] and V[effect] merge, the transition (or not) from a biclausal to a monoclausal structure, or the marking of the causer, causee and patient arguments. Rather than reopen this particularly complex debate here, I shall refer the reader instead to Comrie (1975, 1976), Kayne (1975), Harbert (1977), Aissen (1979), Burzio (1986), Kozinsky & Polinsky (1993), and Abeillé, Godard & Miller (1997), to name but a few of the authors who raise the question of the syntax of “fusional” causative constructions in different languages, and from different theoretical perspectives.

[6] The corpus used in this study consists of the OE prose texts listed in the ‘References’ section. A number of Middle English and Early Modern English texts will also be cited here, references to which are given at the end of the paper.

[7] Following Huddleston (1984: 143-4), I use this term to designate both events and states appearing in the complement of the matrix verb.

[8] To simplify matters somewhat, I have included among the 13 occurences of bebeodan V+NP+ÞÆT three in which the matrix verb is in the passive.

[9] Again, in order to simplify the presentation, I have included under this heading one VOSI-type construction with a passive complement, one VOSI[to] construction (in which to precedes the infinitive), and one construction in which the matrix verb is in the passive, followed by a to- infinitive.

[10] According to the Helsinki Corpus classification, by date of origin : O2 corresponds to the period from 850-950, and O3 to the years 950-1050.

[11] VOSI: ‘Verb + Object / Subject + Infinitive,’ term used by Visser (1973) to describe AcI type patterns, where the accusative NP could be either the subject of the infinitive or both the subject of the infinitive and the direct object of the higher verb.

[12] Don V+I, of course, has long been considered the precursor of periphrastic do and, ultimately, auxiliary do. Ellegård (1953) derives periphrastic do directly from causative do V+I, while Denison (1985: 52-55) suggests that the same construction may initially have had a wider, perfective meaning.


Primary Sources, corpora, and dictionaries

Old English texts

Apollonius of Tyre: Goolden, P., ed. 1938. The Old English Apollonius of Tyre. Oxford: OUP.

Bede: Miller, T., ed. 1890. The Old English Version of Bede’s Ecclesiastical History of the English People, EETS OS 95, 96, 110, 111. London: Trübner.

Blickling: Kelly, R., ed. 2003. The Blickling Homilies. London & New York: Continuum.

Boethius: Sedgefield, W., ed. 1899. King Alfred’s Old English Version of Boethius, De Consolatione Philosophiae. Oxford: Clarendon Press.

Esther: Lee, S. D., ed. 1999. Ælfric's Homilies on Judith, Esther, and the Maccabees, Online edition. http://users.ox.ac.uk/~stuart/kings/main.htm.

Heptateuch: Marsden, R., ed. 2008. The Old English Heptateuch and Ælfric’s Libellus de Veteri Testamento et Novo (vol. I), EETS 330. Oxford: OUP

Ine’s Laws: Jebson, T., ed. 1995. Laws of Ælfred and Ine, Online edition. http://www8.georgetown.edu/departments/medieval/labyrinth/library/oe/texts/prose/laws.html.

Judith: Lee, S. D., ed. 1999. Ælfric's Homilies on Judith, Esther, and the Maccabees, Online edition. http://users.ox.ac.uk/~stuart/kings/main.htm.

St Oswold: in Skeat, W., ed. 1881–85 & 1890–1900. Ælfric’s Lives of Saints, (2 vols.), EETS OS 76, 82, 94, 114. Oxford: OUP

Orosius: Sweet, H., ed. 1883. King Alfred’s Orosius (vol. I), EETS 79. Oxford: Trübner.

Maccabees: Lee, S. D., ed. 1999. Ælfric's Homilies on Judith, Esther, and the Maccabees, Online edition. http://users.ox.ac.uk/~stuart/kings/main.htm.

Nativity of the Innocents: “Natale Innocentium Infantum”. Ælfric, The Homilies of the Anglo-Saxon Church, ed. by B. Thorpe, 1844, vol. I. London: Ælfric Society.

Parable of the Vineyard: “Septuagesima Sunday”. Ælfric, The Homilies of the Anglo-Saxon Church, ed. by B. Thorpe, 1846, vol. II. London: Ælfric Society.

Parker Chronicle: Jebson, T., ed. 1996–2006. MS ‘A,’ XML Online Edition of the Anglo-Saxon Chronicle. http://asc.jebbo.co.uk/#.

Peterborough Chronicle: Jebson, T., ed. 1996–2006. MS ‘E,’ XML Online Edition of the Anglo-Saxon Chronicle. http://asc.jebbo.co.uk/#.

Sermon: Whitelock, D., ed. 1966. Sermo Lupi ad Anglos. New York: Appleton Century Crofts.

St. Cuthbert: Thorpe, B., ed. 1846. Ælfric, The Homilies of the Anglo-Saxon Church, vol. II. London: Ælfric Society.

St Edmund: Skeat, W., ed. 1881–85 & 1890–1900. Ælfric’s Lives of Saints, (2 vols.), EETS OS 76, 82, 94, 114. Oxford: OUP.

St. Mary of Egypt: Magennis, H., ed. 2002. The Old English Life of St. Mary of Egypt. Exeter: University of Exeter Press.

State of Learning: Sweet, H., ed. 1871. Preface to King Alfred’s West-Saxon Version of Gregory’s Pastoral Care, EETS 45, 50. London: Trübner.

Other texts

Brut: Brook, G. L. & R. F. Leslie, eds. 1963, 1978. La3amon: Brut, EETS OS 250, 277. London, New York & Toronto: OUP.

Chaucer, Canterbury Tales: Benson, L., ed. 1988. The Riverside Chaucer. Oxford: OUP.

Elyot: Bear, R., ed. 1997. Online edition of Elyot’s The Boke Named the Governour, 1531, Renascence Editions, University of Oregon (pdf format). https://scholarsbank.uoregon.edu/xmlui/bitstream/handle/1794/681/governour.pdf?sequence=1.

Gower: Confessio Amantis. Macaulay, G., ed. 1900–01. The English Works of John Gower, EETS ES 81–2. London: Keegan Paul, Trench, Trübner.

Havelok: Skeat, W., ed. 1868. The Lay of Havelok the Dane, EETS ES 4. London: Trübner.

Hoby: Bear, R., ed. 1997. Online edition of Hoby’s The Boke of the Courtier, 1561, Renascence Editions, University of Oregon (pdf format). https://scholarsbank.uoregon.edu/xmlui/bitstream/handle/1794/671/courtier.pdf?sequence=1.

More: Lumby, J., ed. 1883. More’s History of King Richard the Third. Cambridge: CUP.

Vices & Virtues: Holthausen, F., ed. 1898. Vices and Virtues: a Soul's Confession of its Sins with Reason's Description of the Virtues: A Middle-English Dialogue of About 1200 A.D, EETS 89, Online edition: University of Michigan.

Corpora and Dictionaries

DOEC: diPaolo Healey, A., ed. 2004. The Dictionary of Old English Corpus. Toronto: University of Toronto. http://tapor.library.utoronto.ca/doecorpus/.

The Helsinki Corpus of English Texts. 1991. Compiled by Matti Rissanen (Project leader), Merja Kytö (Project secretary); Leena Kahlas-Tarkka, Matti Kilpiö (Old English); Saara Nevanlinna, Irma Taavitsainen (Middle English); Terttu Nevalainen, Helena Raumolin-Brunberg (Early Modern English). Department of English, University of Helsinki. http://www.helsinki.fi/varieng/CoRD/corpora/HelsinkiCorpus/.

OED: Simpson, J. & Weiner, E., eds. 1989. The Oxford English Dictionary, 2nd edition (20 vols.). Oxford: Clarendon Press.

Secondary sources

Aissen, J. 1979. The Syntax of Causative Constructions. New York: Garland

Abeille, A., D. Godard & P. Miller. 1997. “Les causatifs en français: un cas de compétition syntaxique”. Langue française 115: pp 62–74

Baron, N. S. 1974. “The structure of English causatives”. Lingua 33: 299–324.

Burzio, L. 1986. Italian Syntax. Dordrecht: Riedel

Cole, P. 1983. “The Grammatical Role of the Causee in Universal Grammar”. International Journal of American Linguistics 49: 115–133

Comrie, B. 1975. “Causatives and Universal Gammar”. Transactions of the Philological Society (1974): 1–32

Comrie, B. 1976. “The Syntax of Causative Constructions: Cross-language Similarities and Divergences”. Syntax and Semantics vol. 6: the Grammar of Causative Constructions, ed. by M. Shibatani. London: Academic Press

Croft, W. 1991. Syntactic Categories and Grammatical Relations: the Cognitive Organisation of Information. Chicago: Chicago University Press

Delancey, S. 1985. “Lhasa Tibetan Evidentials and the Semantics of Causation”. Berkeley Linguistic Society 11: 65–72

Denison, D. 1985. “The Origins of Periphrastic DO: Ellegård and Visser Reconsidered”. Papers from the 4th International Conference on English Historical Linguistics. Amsterdam: John Benjamins

Denison, D. 1993. English Historical Syntax: Verbal Constructions. London & New York: Longman

Dowty, D. 1979. Word Meaning and Montague Grammar. Dordrecht: Kluwer

Ellegård, A. 1953. The Auxiliary ‘Do’: the Establishment and Regulation of its Growth in English. (=Gothenburg Studies in English, 11.) Stockholm: Almqvist & Wiksell

Givón. 1974. “Irrealis and the Subjunctive,” Studies in Language 18: 265–337

Harbert, W. 1977. “Clause Union and German Accusative + Infinitive Constructions”. Syntax and Semantics vol. 8: Grammatical Relations, ed. by P. Cole & J. Saddock, 121–149. London: Academic Press

Hollmann, W. 2005. “Passivisability of English periphrastic causatives”. Corpora in Cognitive Linguistics: Corpus-based Approaches to Syntax and Lexis, ed. by S. Th. Gries & A. Stefanowitch, 193–223. Berlin & New York: Mouton de Gruyter

Huddleston, R. 1984. Introduction to the Grammar of English. Cambridge: CUP

Ikegami, Y. 1981. “On the Development of the Causative Use of the Verbs do and make in English”. Eigo no Rekishi to Kozo: Festschrift for Professor Miyabe, 85–99. Tokyo: Kenyusha Shuppani.

Karttunen, L. 1971. “Implicative Verbs”. Language 47: 340–58

Kayne, R. 1975. French Syntax: the Transformational Cycle. Cambridge, Mass.: MIT Press

Kemmer, S. & A. Verhagen. 1994. “The grammar of causatives and the conceptual structure of events”. Cognitive Linguistics 5: 115–56.

Kilpiö, M. 2010. “Causative habban in Old English”. Þe comoun peplis language, Medieval English Mirror 6, ed. by M. Krygier & L. Sikorska, 43–65. Frankfurt am Main: Peter Lang

Kozinsky, I. & M. Polinsky. 1993. “Causee and patient in the Causative of Transitive: Coding Conflict or Doubling of Grammatical Relations?” Causatives and Transitivity, ed. by Comrie & Polinsky. Amsterdam: John Benjamins

Łęcki, A. 2010.: Grammaticalisation paths of Have in English. (=Studies in English Medieval Language and Literature, 24.) Frankfurt am Main: Peter Lang

Lowrey, B. 2002. Les verbes causatifs en anglais: une étude diachronique du moyen-anglais à l’anglais moderne. PhD thesis, université de Lille III.

Lowrey, B. 2010a. “Causative Verbs in West Saxon Old English”. Bulletin des Anglicistes Médiévistes 78, hiver 2010: 57–88

Lowrey, B. 2010b (& forthcoming). “Syntactic and Semantic Variation: Middle English Causatives and Complement types”.  Paper presented at ICEHL-16, Pecs, 2010: to appear in Hegedűs, I. & Pődör, D. (eds.), Periphrasis, Replacement and Renewal: Studies in English Historical Linguistics, Cambridge: Cambridge Scholars

Lowrey, B. 2011 (& forthcoming). “Whatever happened to hatanHatan and lætan causative constructions in early English”. Paper presented at CBDA-2, Tours, 2011

Royster, J. 1918. “The Causative use of hatan”. Journal of English and Germanic Philology 17: 82–63.

Royster, J. 1922. “Old English Causative Verbs”. Studies in Philology 19: 328–356

Shibatani, M. 1976. “The grammar of causative constructions: a conspectus”. Syntax and Semantics 6: The Grammar of Causative Constructions, ed. by M. Shibatani, 1–40. London: Academic Press.

Song, J. J. 1996. Causatives and Causation. London & New York: Longman

Talmy, L. 1976. “Semantic Causative Types” Syntax and Semantics vol. 6: the Grammar of Causative Constructions, ed. by M. Shibatani, 43–116. London: Academic Press

Talmy, L. 1985a. “Force Dynamics in Language and Thought”. CLS 21–2, 293–337

Talmy, L. 1985b. “Lexicalisation Patterns: Semantic Structure in Lexical Forms”. Language Typology and Syntactic Description vol. 3: Grammatical Categories and the Lexicon, ed. by Shopen, 57–149. Cambridge: CUP

Talmy, L. 1988. “Force Dynamics in Language and Cognition”. Cognitive Science 12: 49–100

Talmy, L. 2000. Toward a Cognitive Semantics: Concept Structuring Systems (vol. I). Cambridge, Mass.: MIT Press

Terasawa, J. 1985. “The Historical Development of the Causative Use of the Verb Make with an Infinitive”. Studia Neophilologica 57: 133–143

Timofeeva, O. 2011. “Infinitival Complements with the Verb (ge)don in Old English: Latin Influence Revisited”. Leeds Studies in English 42: 93–108

Visser, F. 1963-73. An Historical Syntax of the English Language. Leiden: E. J. Brill