Language Contact Effects on Verb Semantic Classes: Lability in Early English and Old French

12 Language Contact Effects on Verb Semantic Classes: Lability in Early English and Old French Luisa García García (University of Seville; [email protected]) Richard Ingham (University of Westminster; [email protected]) Copyright: This is an Author Accepted Manuscript version of the chapter by the same name and authors, published in Medieval English in a Multilingual Context. Current Methodologies and Approaches, edited by Pons-Sanz, Sara M. and Louise Sylvester, 2023, pp. 343-375, Palgrave MacMillan reproduced with permission of Springer Nature Switzerland AG. The final authenticated version is available online at https://doi.org/10.1007/978-3-031-30947-2 Abstract Old and Modern English differ sharply in the prevalence of lability, the extent to which verbs alternate between transitive and intransitive frames (McMillion 2006, van Gelderen 2011). Such alternations are attributed by Pinker (1989) and Levin (1993) to membership of semantic classes: change-of-state/position verbs are mostly labile in modern English, whereas destroy-verbs and psychological verbs are not (Levin 1993). This study investigates how far verb semantic class membership was criterial for lability in older stages of English. It appears to have been a gradable phenomenon in Old English: among change-of-state/position verbs, lability was far from negligible (García García 2020), a very limited number of psychological verbs are found to have been labile, and destroy-verbs were non-labile. Middle English members of the latter two verb classes were also analysed and it was found that in both classes some expansion took place in the extent of lability within their members. We propose that this occurred under the influence of the corresponding Old French verbs, also analysed in this article. Lability in Old French has been argued by Ingham (2020) to have had a decisive influence on rendering the change-of-state/position verb classes labile in Middle English. It is shown here to have been present in earlier Old French with some psychological verbs and to a certain extent with destroy-class verbs, but to have been declining in the period of maximum French influence on Middle English. It is concluded that contact influence has had a large-scale effect on the lability of English verb classes (especially change-of-state/position verbs) where Old English already had substantial numbers of exemplars to act as ‘bridgeheads’ for the developing syntactic trend, but not where Old English lacked them, as with psychological verbs and destroy-class verbs. Lasting contact influence appears to have combined with, and been favoured by, existing predispositions within the language. 1 Introduction A very large body of research bears out how significant contact with other languages has been as a driver of changes in English over the centuries. Yet linguistic change does not always succeed. For one reason or another, an innovation that is initially taken up by a speech community, or some members of it, may fail to generalise and oust its competitor forms. Research into so-called ‘failed changes’ has been gaining some traction in recent years. Postma (2010) discussed two cases: the temporary replacement of hem ‘him’ in a Middle Dutch border dialect by the innovative lexical reflexive sick, borrowed from Low German, 1 and the temporary extension of do-support to nonemphatic declarative clauses in Early Modern English. He noted that the maximum point of the failed innovations was connected in time with a strong rise of the successful innovations that accompanied them: these were another reflexive form in the Dutch dialect case, and the rise of dosupport in other clause types. Bacovcin (2017) studied the short-lived V - to NP - NP construction with dative verbs in Middle English, which he accounted for as a temporary by-product of successive changes in the Middle English Case system. At the level of lexis, Sylvester et al. (forthc.) have researched lexical variation in Late Middle English, showing that numerous innovations, especially loan items, did not succeed in displacing established lexemes. In this study we consider changes to the argument structure of English verbs in two semantic domains where contact influence may well have made some impact, but not sufficiently to alter their character long-term. Importantly, the innovative syntax which did not survive with these verbs emerged at a time when, as found by Postma (2010), the change in question was becoming successful in comparable linguistic domains. The phenomenon in question was the ability of a verb to appear in morphologically unmarked form in both causal-transitive and noncausal-intransitive frames, as discussed more fully in the next section. 2 Lability 1 Sick was ousted by another reflexive form, sich, which succesfully replaced hem. According to Haspelmath’s (1993) typology, languages possess a variety of causativisation and anticausativisation strategies involving verbs of change of state or location, 2 including P(ATIENT)- lability, where the PATIENT or UNDERGOER argument can appear either as subject or direct object and the verb can adopt a noncausal or causal sense without alteration of the verb form, e.g.: (1) a. The branch broke. (noncausal-intransitive) (1) b. The man broke the branch. (causal-transitive) (2) a. The ship sank. (noncausal-intransitive) (2) b. The pirates sank the ship. (causal-transitive) It is found most notably in the semantic sub-classes of change-of-state verbs and change-of-location verbs, as illustrated above in (1)-(2), respectively, where the positionally alternating argument denotes the entity undergoing the relevant change. Lability in Modern English has been wellstudied descriptively and theoretically (e.g. Visser 1963; Pinker 1989; Levin 1993; Rappaport and Levin 1995; McMillion 2006). The semantic criteria model of Pinker (1989) — (see also Rappaport & Levin 1995) — catalogued verbs on the basis of semantic constraints on their ability to alternate . Thus, verbs such as break involve a change of state which could occur independently of the action of an agent. Verbs such has cut, however, do not alternate because their semantics necessarily involve the action of an agent. In Modern English lability is virtually absent from the semantic class of verbs with psychological meaning (henceforth psych verb) (Levin 1993, van Gelderen 2011, Alexiadou 2016). The semantic role of EXPERIENCER (roughly, the entity undergoing the psychological process) can appear either as subject or direct object of a psych-verb clause, e.g.: (5) John feared a lightning strike. (6) A lightning strike frightened John. However, the verbs in question do not allow both constructions: (7) *John frightened a lightning strike. 2 ‘Causativisation’ is the process by which a noncausal verb is rendered causal, as in ‘laugh’ > ‘make laugh’. ‘Anticausativisation’ turns a causal verb into non-causal, as in ‘scare (so.)’ > ‘get scared’. (8) *A lightning strike feared John. Lability is also absent from the class of destroy-verbs (Levin 1993), even though their meaning involves a change of state in the PATIENT argument which otherwise permits lability (cf. (1)): (9) The storm wrecked the ship outside the harbour. (10) *The ship wrecked outside the harbour. Following the criteria-based approach, Levin (1993) observed that destroy-class verbs differ semantically from change-of-state verbs in that the latter encode some specific information about the resulting state of the entity that is e.g. broken or sunk (cf. (1)-(2) above), whereas destroy-class verbs encode simply the fact of the entity’s destruction. In the Middle English period, the position seems to have been interestingly different from Present-Day English. It was noted observationally by one of the present authors that some verbs with psychological meaning, as well as certain verbs with meanings similar to ‘destroy’, also showed lability. These observations prompted the questions pursued in this research, namely, (i) how widespread lability was among Middle English verbs belonging to these semantic classes, (ii) whether their Old English forerunners were also labile, and (iii) whether contact with French might account for any changes that occurred, as with change-of-state/position verbs. 3 Lability in earlier English: Previous research The literature on lability in Old English is somewhat limited but adequate for a basic understanding of the situation. The first consistent contribution to the study of the argument structure of Old English verbs (known to us) is Hermodsson (1952), which contrasted the valency of a large number of verbs in Old High German, Old Frisian, Old English and Old Saxon. Hermodsson (1952: 210) pointed out the radical spread of labile verbs in English, when compared to other Germanic languages. He concluded from that comparison that lability is an inherited trait that originates in the Ingvaeonic language group, comprising Old Frisian, Old English, Old Saxon (and Middle Dutch) , and that is clearly visible in the Old English period. He acknowledged, though, like Visser (1963: 99), the abundance of intransitive verbs in Old English, compared to Present-Day English. McMillion (2006) claimed that Old English is not especially labile. 3 In recent research into valency in the history of English, Ottosson (2013) and van Gelderen (2011), following Hermodsson, also trace back Present-Day English use of labile coding to Old English, while noting at the same time the much higher number of labile verbs in Present-Day English. More specifically, van Gelderen (2011: 122) links the rise of lability in Old English to the loss of the causative formation. This link has been studied by García García (2020), who points out that, since the causative formation is by far the main valency-changing mechanism in Proto- Germanic, the decline of its transitivising function in favour of labile coding in Old English demonstrates the impetus of labilisation already in this early period (2020: 170). This has been confirmed by a recent corpus study that shows that the number of labile verbs increases significantly both in token and type in late Old English with respect to early Old English (García García and Ruiz Narbona 2021). Note that most of the Old English causatives that become labile denote a change of state or change of position: e.g. byrnan ‘to burn (intr.); to burn (sth.)’, gecelan ‘to be cool; to make cool’. The number of labile verbs massively increases over time. Ingham (2020) has shown that a great expansion of lability took place in Middle English: less than 20% of native verbs surviving into Middle English in the semantic areas of change-of-state and change-of-position were found to have been labile in Old English, while in Middle English two-thirds of them were attested as labile, with an even greater proportion of verbs in these semantic areas borrowed from French. He argued that intense French influence in the later medieval period was responsible for the large-scale adoption of lability by Middle English. Medieval French, unlike the modern language, made extensive use of lability in those same areas, so the case for seeing French contact influence at work in the large-scale extension of lability at the very time when the influence of French on Middle English generally was at its height (thirteenth and fourteenth centuries) appears solid. Arising from that research, we wish to consider how far contact with French might have affected the argument structure of verbs in other semantic areas where lability could have, but did not, become dominant. The two lexical domains to which the present study is devoted are psych verbs and destroy-verbs (Levin 1993). Psych verbs in earlier English have received considerable attention in recent years from a number of authors, notably Allen (1995), Moehlig-Falke (2012), 3 He came to this conclusion by applying Haspelmath’s (1993) questionnaire of thirty-one inchoative / causative verb pairs to Old English. However, the reliability of such questionnaires for extinct languages has been rightly disputed, because of the absence of native speakers who can confirm the most usual gloss (see van Gelderen 2011: 115). Van Gelderen (2014) and Miura (2014), particularly focusing on developments that led to rearranged linking of semantic roles to grammatical functions. Such studies have often highlighted the process by which initially object-EXPERIENCER verbs such as (modern spelling) like, e.g. (11), shifted to the subject-EXPERIENCER type, e.g. (12): (11) Đam wife þa word wel licodon. (Beo 491) DET.-DAT. woman-DAT. DET.-NOM.PL. words well please-PRET.-3PL. ‘Those words pleased the woman well’ (12) The woman really liked those words. (PDE) Studies have also examined the impersonal construction with psych verbs in Old English (see (24) below) and its disappearance in Middle English and beyond. Certain authors (Guidi 2011; van Gelderen 2014; Alexiadou 2016) have commented on lability in psych verbs. Alexiadou (2016) discussed the absence of what she called the ‘causative alternation’, i.e., lability, in this semantic class; she mentioned that, whereas both psych-verbs and change-of-state verbs could take part in morphological alternations of the causative type in Old English, only psych-verbs ‘were for some reason incompatible with lability’ (Alexiadou 2016: 7). This alleged disparity is one of the points of departure of the present article. Thus, in Old English, change-of-state, psych- and also destroy-verbs participated in the causative morphological alternation (e.g. abylgan ‘to anger’ vs. belgan ‘to swell with anger’, wyrdan ‘to injure, hinder’ vs. forweorþan ‘to perish, go off’). The morphological alternation was already declining in Old English in favour of labile coding. However, all three classes were not affected by the spread of lability in the same way. The different development of verb semantic classes in pre-modern English regarding lability is the focus of our study. Argument structure in destroy-verbs has received little or no attention. Our work appears to be the first to consider lability or otherwise in the semantic domain from a diachronic perspective, and also to relate linguistic change in these domains to contact influence from French, more specifically from the variety of Old French used in England during the centuries following the Norman Conquest, and conventionally known as Anglo-French (or Anglo-Norman; Trotter 2003; Rothwell 1993). Although in some respects, especially phonology, it deviated from the continental language, in terms of syntax it remained broadly similar to it (Wilshere 1993; Ingham 2012). 4 Analysis of Old English and Middle English destroy-verbs In this section we describe and compare the argument structure of Old English, Middle English and Anglo-French destroy-verbs. First, we present a detailed account of the data collection process of the Old English and Middle English material, and then discuss the results. The data collection procedure used for Old English material was first to filter all the verbs in Bosworth and Toller’s (1898) Anglo-Saxon Dictionary (ASD) which contained in their definition ‘destroy’ or any of the other members of the ‘destroy’ class identified by Levin (1993: 239). We used the advanced search tool in the electronic version of the dictionary, which turned up 122 verbs with the word ‘destroy’ in their definition, thirty-nine verbs with ‘waste’, eleven with ‘ruin’, nine with ‘demolish’, eight with ‘ravage’, and zero with ‘wreck’. We then determined the argument frames in which the verbs were attested, consulting the citations in the ASD and, when available, in the Dictionary of Old English (DOE), completed up to letter I so far. Only four of the verbs thus selected were attested as labile, that is, were used as both causal-transitive, with the PATIENT in object function, and noncausal-intransitive, with the PATIENT as subject. These are brecan ‘to break, demolish (sth.)’ / ‘to force a way out of confinement’, slītan ‘to slit, tear, rend, destroy, waste (sth.)’ / ‘to tear, get torn’, losian ‘to lose, destroy’ / ‘to be lost, perish’ and forfaran ‘to perish, be destroyed’ / ‘to kill, destroy’. For the first three verbs, the meaning category ‘destroy’ is clearly secondary and results from the translation of specific contextual usages of the respective primary meanings, namely ‘break’, ‘tear’ and ‘lose’. OE forfaran ‘to perish, be destroyed’ / ‘to kill, destroy’ is the only labile verb in the destroy-class in Old English retrieved by our searches. Examples of its noncausal-intransitive and causal-transitive usages follow: Noncausal-intransitive: (13) a. & mycel sciphere hider com suðan of Lidwicum, & hergodon swyðe be Sæfern, ac hi þær mæst ealle syþþan forforan (ChronD 910.10) 4 ‘And a big fleet of war came hither from the south of Brittany, and they pillaged much along the Severn, but afterwards almost all were destroyed there’ Causal-transitive: (13) b. …þæt man þa sawla ne forfare, þe Crist mid his agenum life gebohte (LawVAtr 2) ‘…so that man does not destroy the soul, which Christ bought with his own life’ 4 Old English works are cited after the Dictionary of Old English Corpus. Unsurprisingly, our results closely coincide with previous findings, such as Hermodsson’s (1952: 196ff.) and Visser’s (1963: 97); neither study includes any destroy-verbs in its list of labile verbs in Old English. Additional proof for the (near-)absence of lability in destroy-verbs in Old English is supplied by the fact that causative verb pairs with this and similar meanings (‘injure’, ‘harm’, ‘kill’) are kept intact, unlike causative pairs denoting change-of-state or change-of-position, which frequently become labile, as noted above. Some examples of causative pairs around the meaning ‘destroy’ are ā-hrȳran ‘cause to fall, destroy’ vs hrēosan ‘fall, be destroyed’, wyrdan ‘injure’ vs for-weorðan ‘perish, vanish, go off’ (intrans.) or cwellan ‘kill’ vs cwelan ‘die’ (see complete list in García García 2020: 175-6). This suggests that the incompatibility of lability with destroy-verbs is probably (nearly) categorial in Old English, and not merely the result of attestation gaps. The method used to find destroy-verbs in Middle English was first to probe the Middle English Dictionary (MED) using the ‘definition and notes’ search tool so as to identify all the verbs in whose definition the words ‘destroy’, ‘waste’, ‘ruin’, ‘wreck’, ‘ravage’ or ‘demolish’ occurred. 5 Some 263 hits were obtained in this way, the majority for ‘destroy’. For each verb, the primary sense was then examined to identify whether labile syntax was attested in the citations, that is, whether the argument bearing the PATIENT role was attested in subject and also in object function of the same lexeme without morphological alteration. This procedure required some refinement in order to arrive at appropriate target verbs: the MED adopts a very wide range of definiens terms, seemingly in order to give an idiomatic translation of particular uses of lexemes in the citations provided, e.g. ‘destroy’ is proposed as a translation of voiden and misleden respectively, in the following citations: 6 (14) … in caas hir baner… were voided or rased. (Vegetius 89/4) ‘In case their banner were to be destroyed or cut down’ (15) Icham adred, Þurth his folye he is misled. (Arthur & Merlin (Auch.) 4988) ‘I am afraid that through his folly he is destroyed’ 5 Rarer words listed by Levin (1993: 239) such as decimate and obliterate were left aside as it was expected that one of the five terms selected would suffice to generate the corresponding Middle English lexis. 6 All Old English, Middle English, and Anglo-French citations are referenced in the short forms given in the respective dictionaries. The search also provided numerous hits where the ‘destroy’ meaning arose only thanks to a phrasal use, e.g. s.v. putten in the combination put down, as in: (16) They wolde putt downe mannes reason in fulfillyng of their worldly desires. (Chartier, Treat. Hope 59/20- 21)‘They would overturn human reason while fulfilling their worldly desires’ The great majority of hits did not correspond to items with ‘destroy’, ‘ruin’, and so on as the primary meaning (taken as being mentioned in the first sense listed) and were therefore culled. Thirty-six verbs remained, some of them labile, e.g.: (17) a. Right as wormes shende a tree, Right so a wyf destroyeth hir housbonde. (Chaucer CT.WB 376) ‘Just as grubs destroy a tree, so does a wife destroy her husband’ (17) b. Sonne myght there none discende Lest the tender grasses shende. (RRose 1400) ‘No sun could reach down there lest the tender grass would be destroyed’ (18) a. Ye han nat anientissed or destroyed hem. (Chaucer CT.Mel 2438) ‘You have not eliminated or destroyed them’ (18) b. With þe lest winde..Þe note anentis..& all to noʒt worthis. (Wars Alex.3302) ‘With the least wind the fabric of a spider’s web is destroyed and is nothing’ Table 12.1 below lists all the verbs belonging to the destroy-class and their valency in Middle English. NB: No lexeme meaning ‘wreck’ was obtained that was not already identified by the use of the probe ‘destroy’, leaving five meaning categories: ‘destroy’, ‘waste’, ‘ruin’, ‘ravage’ and ‘demolish’. Table 12.1: Labile and non-labile Middle English verbs in the destroy-class in MED Labile Non-labile anienten, forfaren, perishen, shenden, spillen, tinen: ‘destroy’ astruien, awerden, corrupten, diliʒen, destroien, fordon, forspillen, forwirchen, outenden, squachen, stroien, teren, tocasten, todreosen, toshenden, tospillen, towasten: ‘destroy’ westen: ‘waste’ adwinen, aswinden awesten, forclingen, iwesten towesten, wansen ‘waste’ amerren, defeten: ‘ruin’ hypothesis of French influence as a reason for this development, by examining lability in psych verbs used in the French of England. 6 Anglo-French psych-verbs In Old French lability was found by Heidinger (2011) to have been common in verbs of change-of- state/position (see also Hatcher 1942), but his study did not cover psych verbs. Research into the syntactic behaviour of these verbs in insular French (Anglo-French) was accordingly undertaken for the purpose of the present enquiry. The reverse look-up facility of the AND was used to mine data on the basis of Levin (1993) in the same way as described above for Middle English. In this way a total of seventy-six psych verb lexemes was obtained. Depending on their argument structure patterning, citations were categorised into those taking only an object EXPERIENCER, 8 those taking only an object EXPERIENCER with an impersonal subject, those taking only an EXPERIENCER subject, and those that were labile, e.g. respectively (30), (31), (32) and (33a)‒(33b) below: (30) E, s'um le desturbast, ne seüst parfiner. (Becket 2365) ‘And had someone disturbed him, he would not have managed to finish’ (31) De vostre maladie me peise. (Fabliaux 21.276) ‘I am troubled by your illness’ (32) "Sire Reis, fait li il, bien devum doluser.” (Becket 155) ‘(My) Lord King, he said to him, “We must feel great sorrow”’ (33) a. De la voiz del tuen tuneire espowenterunt. (Oxf Ps. 1.103.8) ‘By the voice of your thunder (they) will (be) frighten(ed)’ (33) b. É Deu chalt pas sur les Phistiens tuná é forment les espoentá. (Quatre Livres, p. 15) ‘And God quickly thundered over the Philistines and greatly frightened them’ The results of the analysis illustrated in (30)‒(34) above are shown in Table 12.5: 8 Passive forms, e.g. il fu esbai (‘he was surprised’) were ignored, as they were analysable in a stative sense as consisting of estre ‘be’ + adjectival past participle, and so did not necessarily attest to a transitive use. Table 12.5: Psych verb categories as attested in AND entries (twelfth to fourteenth century) Subject EXP. only Object EXP. only Labile total 38 (50.7%) 20 (26.7%) 17 (22.6%) 75 As can be seen, verbs in the labile group were in a minority, though at 22% not a negligible one. Effectively, Anglo-French psych-verbs were divided roughly between those taking a subject EXPERIENCER only, and those taking an object EXPERIENCER, some of which allowed lability, while some did not. The last attestation dates of each construction involved in lability were also recorded, and these showed an interesting temporal progression. Verbs with meanings involving ‘sadness’, ‘fear’, ‘enjoyment’, ‘delight’, ‘trouble’, ‘astonishment’, ‘dismay’ and ‘shame’ were attested in the twelfth century as labile, appearing not only in EXPERIENCER object examples but also attesting to the EXPERIENCER subject non-reflexive construction: (34) Esjoirat Jacob e eslecerat Israel. (Camb Ps 18.XIII.10) ‘Jacob will rejoice and Israel will rejoice’ (35) Ne nul homme ne coroucera ové celuy qui la portera. (Lapid 281) ‘And no man will get angry with the one who shall carry it’ Examples of this construction were found in Old English, as noted above, with verbs of often directly comparable meanings. However, with nine of the seventeen Anglo-French verb types collected, the non-reflexive subject EXPERIENCER construction no longer occurred in insular French after the twelfth century. By the thirteenth century, the majority of verb types previously labile now used the reflexive subject EXPERIENCER construction, e.g.: (36) Quant homme se esjoit de suen mal. (Mirour Egl 24.17) ‘When a man rejoices in his evil’ (37) Car li jugges se couroussent volunters quant om apele de eus. (Lett de Rois 51) ‘For the judges easily grow angry when one appeals against them’ In the fourteenth century, hardly any were still attested as labile with the non-reflexive subject EXPERIENCER construction as in (33a)‒(33b) above. Thus, Anglo-French was tending to lose lability in this verb domain by the time that the period of intense French contact influence on English is encountered, roughly 1220‒1400. French influence on English favouring lability would hence not be expected on the scale that took place with change of state / location verbs, where a great number of Anglo-French verbs continued to be labile throughout this period (Ingham 2020). In short, insular French would not have provided a strong model for psych-verb lability in Middle English. There was already some use of lability in Old English, which was not massively increased by contact with insular French, unlike with verbs of change-of-state and change-of- position. The much weaker incidence of lability in the psych-verb domain is in line with the hypothesis that argument structure in Middle English was most open to contact influence from French during the period of greatest lexical borrowing from that language. Lability among Anglo- French verbs of change-of-state and change-of-position remained very common throughout that time, whereas as we have seen among psych-verbs it declined quite sharply. Indeed, French influence may even, by the thirteenth century, have operated as a brake on the use of lability with Middle English psych-verbs, favouring instead the subject EXPERIENCER + reflexive construction. 7 Conclusions Our analysis of the two verb semantic domains investigated here has enabled some nuanced conclusions regarding the role of external and internal factors on language change, at least as regards verb argument structure. It was found that in the destroy-verb class a certain number of items became temporarily labile in Middle English, whereas Old English had almost no exemplars among destroy-verbs. The finding that a good proportion of Anglo-French destroy-verbs were found to be labile appears not to have been decisive: any contact influence, attested by the Middle English citations showing lability, was clearly not lasting. Thus, although there may have been some external pressure from French towards lability, the language lacked an internal basis for lability in that semantic area. In the psych verb semantic domain, Middle English made considerably more use of lability. Here the same two factors were found to be important. Old English already had some exemplars, which we would see as having constituted a kind of ‘bridgehead’, from which lability could expand its scope, in a similar way to what transpired with change-of-state / position verbs, and here a substantial number of verbs became labile in Middle English. However, the model that French could have constituted in this domain occurred too early for contact influence to have played a decisive role; by the time bilingualism (and thus contact influence) became established among the educated members of the English speech community, lability among French psych verbs was already very recessive. Whereas in Anglo-French nearly a quarter of psych-verbs were found to be labile, the non-reflexive subject EXPERIENCER construction gave way to the subject reflexive EXPERIENCER construction after the twelfth century. Nevertheless, we would judge from our findings that the combined effect of contact influence from Anglo-French and the existence of lability in a fair number of Old English psych-verbs provided the language with the momentum to develop lability somewhat further in this semantic domain than with destroy-class verbs. In both semantic domains, the temporary rise of lability appears to have been a failed change, never going to completion as it has largely done in the change-of-state / position semantic domains. As noted, the combined influence of Old English and Anglo-French was too weak to allow lability to put down strong roots among destroy-verbs, where even in Middle English the innovation was clearly failing. With psych verbs, we have, however, noted informally from OED entries that lability continues to be attested quite often in the Early Modern period until apparently growing obsolescent in the seventeenth century. It has not been possible in this study to take detailed account of developments in the post-medieval period, which would undoubtedly merit further attention. One suggestive point that has emerged from this inquiry is that psych verb lability in the source language (French) was being challenged strongly, and crucially at the period of maximal contact influence on English, by a competitor, the reflexive construction, as discussed in Section 6. This factor of competition may have had the effect of putting a brake on the spread of lability right across the English psych verb domain, similar to Postma’s (2010) finding that a competitor form led to a failed change in the context of language-internal development in Dutch. Some limitations must be acknowledged. Dictionaries as a data source do not normally provide evidence of the relative textual frequencies of the lexemes they feature, nor of the constructions in which they appear. Useful information on the development of verb alternations could be gained by searching a Middle English text corpus, which in particular cases might show that one or other alternant never achieved more than very minimal representation, and quickly died out. It could be that in some cases the lability noted in the dictionaries consulted for this research could be hapax attestations. There again, since available Middle English texts, especially prior to the fourteenth century, are not greatly numerous and are strongly skewed towards particular text types, it could not be claimed that the frequencies to be derived from them are representative of language use at the time the English language was undergoing contact influence from French. A further limitation is that account has not been taken of possible dialectal or idiolectal variation. Conceivably, the appearance of lability taken from our sources could be a compilation of a speaker variety in which a verb was used only in one construction by one type of speaker, and only in the other construction by another type of speaker. The resources at our disposal hardly allow us to resolve this issue. Nevertheless, some broader conclusions can be drawn. The semantic criteria model of Pinker (1989) and Levin (1993) may be valid in synchronic terms, but evidently can hardly have applied in the same terms to earlier stages of the language. Some verbs which now conform to the semantic constraints on a given alternation evidently did not in the past, while other verbs have done so for a very long time, including the Old English verbs of change-of-state researched by García García (2020). The constraint now ruling out lability with destroy-class verbs may have existed in Old English, but then appears to have been relaxed temporarily in Middle English, under contact influence as we argue, only to be reinstated later. In both cases, the semantic criteria valid for Present-Day English may be seen over time as emergent, appearing via a process whereby over time more and more verbs were used according to those criteria, while those that did not fit the criteria ceased being used in ways that would have violated them. This is naturally speculative and attendant on further research into questions of argument structure diachrony. Appendix 1: Old English psych-verbs 9 OE psych verbs Valen cy OE psych verbs Vale ncy byldan ‘embold’ O geirsian ‘be angry’, ‘make angry’ L belgan ‘swell with anger’ S lāþian ‘be hateful or loathed’ S abelgan ‘cause so. to swell up in anger’ O mislǣdan ‘mislead’ O gebelgan ‘become angry’;‘anger, enrage’ L lician ‘please’; ‘take pleasure in’ (=be pleased by) O 9 Key: L = labile; O = EXPERIENCER object only; S = EXPERIENCER subject only. bliþian ‘rejoice’; ‘gladden’ L gelician ‘please’ O ābylgan ‘anger, offend’ O mislician ‘displease’ O carian ‘sorrow’ (int.); ‘take care’ (tr.) S oflician ‘displease, be displeasing (to so.)’ O cwacian ‘shake’ S lustfullian ‘rejoice, be glad’ S cweman: imp. ‘please, satisfy’ O gelustfullian ‘be delighted’ ‘to delight’ L gecweman ‘please, satisfy’ O lufian ‘love’ S miscweman ‘displease’ O lystan ‘cause pleasure or desire’ (pers., impers.); ‘desire’ L dreccan ‘afflict. oppress’ O gelystan ‘please, cause a desire’ (impersonal) O drefan ‘disturb, agitate, oppress’ O gemǣdan ‘make insane’ O (a)drǣdan ‘dread, fear’ (adrǣdan w. dat. reflex) S (be)mǣnan ‘lament, mourn, complain’ S dwellan ‘lead into error; err (intr.)’ L niþerian ‘bring low, humiliate’ O dyrfan ‘be diligent’; ‘torment’ O sceamian ‘feel or cause shame’ (impersonal) L eargian ‘be or become timid’; ‘fear’ (tr.) S forsceamian ‘be or make greatly ashamed’ (impersonal) L faegnian ‘rejoice’; ‘welcome’ S swencan ‘cause a person to labour, harass, afflict’ O (a)faeran ‘terrify’ O geswencan ‘disturb, vex, afflict’ O forhtian ‘fear’ S swincan ‘toil, labour, work with effort’ S aforhtian ‘be, become afraid, fear, dread’ S scendan ‘put to shame’ O frefran ‘console, comfort’ O sorgian ‘care, grieve’ S gæstan ‘torment’ (only 1x) O teonian ‘vex, irritate’ O gladian ‘be glad, make glad’ L tirgan ‘vex, provoke’ O gramian ‘anger (so.)’; ‘burn with anger’ (gloss) L? tregian ‘vex, annoy’ O gremman ‘enrage, provoke’ O tweogan ‘feel, cause doubt’ (impersonal)) L grimman ‘rage, vent fury (intr.)’ S getweogan ‘to doubt, hesitate’ S agrīsan ‘be afraid, dread’ S for-þencan ‘despair’ (always refl.) S gryllan ‘provoke’; ‘gnash the teeth, rage’ (gloss) L? of-þyncan ‘cause regret or displeasure’ imp. O hatian ‘hate’ S ā-þrēotan ‘be wearisome, tedious, distasteful’ S hefigian ‘become heavy’; ‘make heavy, burden’ L ā-þrȳtan ‘weary, tire out (so.)’ O gehefigian ‘become heavy’; ‘make heavy, burden’ L wlatian ‘cause a person (ac.) loathing’ (imp.) O heortan ‘hearten, encourage’ O wōrian lit. ‘wander’; fig. ‘hesitate, err’ S hreowan ‘distress’, ‘affect (so.) with compassion for so. else’, O wrāþian ‘be angry’ (only 1x) S gehreowan ‘distress’, ‘cause to regret’ O gewræþan ‘be wroth, savage’; ‘attack, molest’ (BT & BT Addenda) L hreowsian ‘grieve, lament, regret’ S wundrian ‘wonder at’; ‘make wonderful’ S behreowsian ‘regret, lament, repent, feel pity’, S weallan ‘bubble forth’ (cp. meaning wyllan ‘boil (sth.); torment, agitate (so.)’) O irsian ‘be angry’, ‘make angry’ L Totals Labile: 14 EXPERIENCER object: 34 EXPERIENCER subject: 24 Not conclusive: 2 Total 74 Appendix 2: Encoding of EXPERIENCER role, Anglo-French verbs, as attested in AND Subject EXPERIENCER Object EXPERIENCER Labile abhomer ‘abhor’ acurager ‘encourage’ contrister ‘sadden, become sad’ admirer ‘admire' agreger ‘aggravate’ corucer ‘anger, become angry’ aduluser ‘deplore’ apeiser ‘calm’ deduire ‘amuse, be amused’ affiancer ‘promise’ asuager ‘soothe’ deliter ‘delight, be delighted’ affier ‘promise’ commovoir ‘upset' rejoir ‘gladden, rejoice’ agarder ‘observe’ conforter ‘comfort’ travailler ‘torment suffer’ amer ‘love’ cuivrer ‘vex’ vergonder ‘shame, be ashamed’ aurer ‘honour’ desturber ‘disturb’ enpourir ‘frighten, take fright’ aviler ‘dishonour’ embaudir ‘embolden’ esmaier ‘dismay, be dismayed’ chericer ‘cherish’ encumbrer ‘harass’ espourir ‘frighten, take fright’ cherir ‘cherish’ enoier ‘trouble’ esbahir ‘amaze, be amazed’ confier ‘trust’ enorter ‘exhort’ effreer ‘frighten’ creire ‘believe’ entarier ‘irritate’ esjoir ‘rejoice’ criembre ‘fear’ esmovoir ‘stir up’ esleecer ‘rejoice’ dedeignier ‘disdain’ estoner ‘astonish’ heitier ‘gladden’ se desafier ‘distrust’ grever ‘afflict’ irer ‘anger’ desesperer ‘despair’ hunir ‘humiliate’ espoenter ‘frighten’ despire ‘despise’ peser ‘calm’ Subject EXPERIENCER Object EXPERIENCER Labile doleir ‘suffer’ trubler ‘trouble’ doluser ‘deplore’ turmenter ‘torment’ duter ‘fear’ endoler ‘afflict’ envier ‘envy’ esbanoier ‘amuse oneself’ esmervoiller ‘amaze’ esmirer ‘admire’ hair ‘hate’ honurer ‘honour’ joir ‘enjoy’ leescer ‘delight’ pleindre ‘complain’ preiser ‘appreciate' reduter ‘fear’ regreter ‘regret’ repentir ‘repent’ reverencer ‘worship’ reviler ‘revile’ s’asseurer ‘to feel confident’ Totals Subj. Experiencer 30 Obj Experiencer 28 Labile 17 Total 75 Appendix 3: Native (OE-origin and newly coined) psych verbs attested in MED, with Old English source verb where relevant 10 Labile in OE Labile in ME Labile in OE Labile in ME Labile ME Subject EXPERIENCER ME accenden ‘stir, become stirred’ - ✓ belwen ‘become angry’ X S agasten ‘frighten, become frightened’ X ✓ abhorren ‘abhor’ - S angren ‘anger, become angry’ - ✓ arwurðen ‘respect’ X S arghen ‘intimidate, be timid’ - ✓ bimenen ‘deplore’ X S agrisen ‘terrify, be terrified’ X ✓ caren ‘grieve’ X S blithen ‘be/make happy’ X ✓ haten ‘hate’ X S bolden ‘be/make bold’ - ✓ lipnen ‘trust’ - S dreden ‘frighten, fear’ X ✓ liten ‘delight’ - S fainen ‘be/make happy’ X ✓ mistresten mistrust’ - S feren ‘frighten, fear’ X ✓ mistrouen ‘disbelieve’ - S forþinken ‘be/make sorry’ X ✓ mornen ‘grieve’ - S gamen ‘be/make merry’ X ✓ overhouen ‘despise’ - S gladen ‘be/make glad’ ✓ ✓ overtrouen ‘mistrust’ - S grillen ‘offend, suffer’ X ✓ quaken ‘be upset’ X S hevien ‘trouble, be troubled’ ✓ ✓ reuen ‘rue’ X S iwreþen ‘anger, become angry’ X ✓ sorwen ‘feel sorrow’ X S irken ‘make/become weary’ - ✓ traisten ‘trust’ - S 10 Key: ✓ = labile, X = non-labile, S = EXPERIENCER subject only, O = EXPERIENCER object only.