CC BY
26
Acta Linguistica Petropolitana. 2022. Vol. 18.1. P. 226-245 DOI 10.30842/alp23065737181226245
The verb areskein in Ancient Greek: Constructions and semantic change
Silvia Luraghi
University of Pavia (Pavia, Italy); silvia.luraghi@unipv.it
Abstract. The paper discusses the diachrony of the semantics and syntax of the verb apsoKsw areskein. This verb means 'repair, make amend, appease, please' in Homeric Greek, in which it mostly features the aorist stem, and always indicates an inchoative situation. In Homer, it takes an argument structure construction that contains a human first participant encoded in the nominative, and possibly a human second participant encoded in the accusative. Co-occurring dative constituents had inanimate referents and encoded the semantic role of instrument. A single occurrence from Hesiod shows an alternative construction with a partitive genitive in the place of the instrumental dative, which conforms to the construction of verbs of satiation. In Herodotus a new construction appears with a human participant encoded in the dative. The verb apsoKsw areskein occurs most frequently in the present stem, hence indicating a state. The dative constituent encodes an experiencer and qualifies as first argument in the construction, based on semantic and pragmatic evidence. In such construction the stimulus is generally encoded in the nominative and agrees with the verb, but impersonal occurrences with no nominative constituents are also attested. Attic prose writers feature two constructions containing either a dative or an accusative experiencer accompanied by a nominative stimulus. Both the dative and the accusative experiences qualify as first argument in most cases. Beside the rise of new argument structure constructions, the verb apsoKsw areskein also undergoes a semantic change and shifts to the domain of experience. This shift was at its onset in Homeric Greek, in which most occurrences had the meaning 'repair', 'make amend', that did not refer to an experiential situation. This is also shown by the different distribution of aspectual stems in Homer (mostly aorist; no present occurrences) and Herodotus (mostly present).
Keywords: experiential domain, constructions, non-canonical constructions, mental and emotional states.
Acknowledgment. Research for this paper was carried out in the framework of the project Dipartimenti di Eccellenza 2018-2022 (Ministry of University and
© Silvia Luraghi, 2022
Research) and was funded by the Italian Ministry of Education and Research grant n. 20159M7X5P Prin 2015 Transitivity and argument structure in flux.
Глагол areskein в древнегреческом языке: конструкции и семантические изменения
С. Лураги
Университет Павии (Павия, Италия); silvia.luraghi@unipv.it
Аннотация. В статье обсуждается семантика и синтаксис глагола ареакеп' ar¿sкein. В гомеровском греческом данный глагол означал 'исправлять, улаживать, умилостивлять, угождать'; он неизменно является показателем инхоативной ситуации и употребляется в конструкции, в которой первый участник обозначается номинативом, а второй участник может обозначаться аккузативом (если речь идет о человеке) или дативом (в случае неодушевленного участника). У Геродота появляется новая конструкция, в которой второй участник-человек обозначается дативом. В аттической прозе гомеровская и геродотовская конструкции представлены вперемежку, так что не всегда очевидно, который из двух именных компонентов конструкции характеризуется как основной участник.
Ключевые слова: экспериенциальность, синтаксические конструкции, неканонические конструкции, ментальные и эмоциональные состояния.
1. Introduction
In this paper I describe and discuss the contractions of the verb ápé-oKeiv aréskein 'make amend, appease, please' in Ancient Greek. I analyze the use and the constructions of the verb in texts ranging from Homer to Classical Attic-Ionic. In recent literature [Baródal et al. 2012; Viti 2017], this verb has been argued to belong to the experiential domain, corresponding to English like, and to offer evidence for a non-canonical construction with either a dative or an accusative experiencer subject. In fact,
taking a diachronic perspective, one can see that during the timespan covered by Ancient Greek texts the verb has undergone semantic and syntactic changes. I argue that the meaning 'like' only appears after Homer, and that the verb likely did not belong into the experiential domain in origin. Non-canonical constructions are likewise post-Homeric.
The paper is organized as follows. In Section 2 I introduce the notion of non-canonical subject and discuss some research on this topic. In Section 3 I analyze the Greek data. Section 4 contains the conclusion.
2. Dative experiencers and non-canonical subjects
Constructions involving dative experiencers common in several Indo-European languages, such as Italian mi place, German mir gefällt or Russian mne nravitsja, all meaning 'I like', with a dative first person pronoun and a verb inflected in the third person singular, have recently been described as containing a non-canonical dative subject (see, e.g., [Barödal, Eythórs-son 2009] and several other publications by Barödal and her associates). Evidence for such constructions has been adduced from virtually all Ancient Indo-European languages, especially for experiential verbs ([Barödal et al. 2012; Viti 2017] among others). [Viti 2017] pointed out that it is exactly the meaning 'like' that tends to show non-canonical subjects most consistently across the Indo-European languages and cross-linguistically.
In most Ancient IE languages, dative experiencers with 'like' verbs (and, depending on the language, with other experiential verbs) may occur in sentences that contain a nominative stimulus, or in sentences that contain an infinitive, as in the Latin examples (1) and (2).
(1) quod tibi lubet, idem mihi lubet REL.NOM 2sg.dat like.PRs.3sG same.NOM 1sg.dat like.PRs.3so
'What you like, I like as well' (Plaut. Most. 1, 3, 138).
(2) non libet mihi deplorare vitam NEG like.PRS.3sG 1sG.dat complain.INF.PRS life.Acc
'I don't like to complain about life.' (Cic. de Sen. 23.84).
In example (1), nominative stimuli, quod and idem, occur alongside the dative experiencers tibi and mihi and trigger verbal agreement: for this reason, they might qualify as subject, even though other subject properties are arguably shown by the dative argument. In (2), on the other hand, the dative experiencer is the only NP in the sentence, and the verb takes a dependent infinitive. This type of construction is traditionally considered impersonal.
In cases in which both a dative experiencer and a nominative stimulus occur, subject properties other than verb agreement must be taken into account. A list of subject properties has first been provided by [Keenan 1976]. They include coding properties (besides nominative case marking and verb agreement, the position in the sentence is also included), behavioral properties, including the possibility to govern control infinitives, conjunction reduction and various cross-reference properties, reflexive control, and semantic properties, including agency, autonomous existence and selectional restrictions [Keenan 1976: 324].
Coding properties characterize the nominative constituent (the stimulus) in occurrences such as (1), but it needs to be remarked that the role of position is hard to assess in free word order languages. Behavioral properties have been shown to shed little light on subjecthood in Ancient IE languages, except Germanic, and remain inconclusive for Old Indo-Aryan, Latin, Greek, and Slavic [Barddal, Eythorsson 2020: 259-260]. Semantic properties seem more helpful, because the experiencers are generally more agent-like than the stimuli, as the former, but not the latter, also undergo selectional restrictions, being necessarily sentient hence animate (mostly human). Still, in specific occurrences stimuli may be more subject-like than experiencers. Consider examples (3) and (4) from Italian.
(3) A Giovanna piace Maria e la vede
to Giovanna like.PRs.3so Maria and 3sg.acc see.PRs.3so
sempre volentieri. always gladly
'Giovanna likes Mary, and she (i.e. Giovanna) is always happy when she sees her.'
(4) Giovanna e Paola piacciono a tutti e Giovanna and Paola like.PRs.3pL to all.PL and
hanno molti amici. have.PRs.3pL many.PL friend.PL
'Everybody likes Giovanna and Paola, and they (i.e. Giovanna and Paola) have many friends.'
While in (3) the experiencer PP a Giovanna is in initial position (a property of subjects in Italian) and controls the elliptical subject in the coordinated clause, in (4) the stimulus participant, Giovanna e Paola, may well be considered more prominent than the experiencer tutti 'everybody', both because of referential properties (individuation), and because it stands in initial position and controls the elliptical subject in the coordinate clause. Hence it qualifies as first argument in the argument structure of the verbpiacere [Luraghi 2020: 43-44].
Among 'like' verbs that allow 'impersonal' constructions in Ancient Greek, Viti [2017: 375] mentions apeoKeiv areskein and avSaveiv handanein, both usually translated as 'please'. Examples are (5) and (6).
(5) Tauxa ^peoe 091 nrneeiv tauta Srese sphi poieein dem.acc.pl please.AOR.3sG 3pl.dat make.PRs.iNF
'All this they agreed to do.' (lit.: 'it pleased them to do these things') (Hdt. 8, 19, 2).
(6) Kai 091 aSeiv to npooraxepra ^r|K£Ti
kai sphi hadein id prosotero meketi
and 3pl.dat please.iNF.AOR art.acc further neg
rcXeeiv pleein
navigate.iNF.pRs
'And they preferred to sail no further.' (Hdt. 3, 45, 1).
Considering the earliest attestations of the language, the Homeric poems, however, it is striking to find that no impersonal constructions occur. Dative experiencers (attested only with avSaveiv handanein in Homer), when present, occur with nominative stimuli. Remarkably, Conti [2010] has shown that in the case of the verb ^eXeiv melein 'be a matter of concern for' and its compounds, too, impersonal constructions only started
occurring in post-Homeric Greek. In fact, impersonal constructions, though attested in Ancient Greek for a limited number of experiencer verbs, do not occur with experiential verbs in Homeric Greek, with the exception of the verb SoKéœ dokéo 'seem', a verb of appearance that can take both nominative stimuli with dative experiencers and control infinitives with dative experiencers (see [Luraghi 2020: 159-160]).
More in general, it is remarkable that Ancient Greek is not a language that favors dative experiencers even in personal constructions, as has been pointed out in Viti [2016] and Luraghi [2020: 79-80]. This is especially true of Homeric Greek. Notably, even with the verb âvôaveiv handànein it is not always the case that the dative experiencer is the most salient participant in Homeric Greek. Indeed, along with occurrences such as (7), we also find occurrences such as (8).
(7) Sixa 5é o^ioiv ^vôave PouX^
dikha dé sphisin hëndane boulé double ptc 3pl.dat like.iMPF.3sG opinion.NOM
'They favored two different opinions.' (II. 18, 510).
(8) Â^ivo^oç... Nioou ^aiôi^oç uioç... Amphinomos... Nisou phaidimos huios ... Amphinomos.NOM Nisos.GEN glorious.NOM son.NOM
^aXioxa 5è nnveXorceifl ^vôave ^tiOoioi malista dè Pënelopeiëi hëndane mùthoisi especially ptc Penelope.DAT like.iMPF.3sG word.DAT.PL
'Amphinomos, the glorious son of Nisos, especially pleased Penelope with his speech.' (Od. 16, 394-398).
In example (7) the stimulus is inanimate and the experiencer is clearly more salient and more agent-like. In (8), on the contrary, not only is the stimulus human, it also is portrayed as actively and intentionally trying to please the experiencer: it shares features of agents, and it does not seem to be less prominent than the experiencer in such a situation. Similar to the case of Italian piacere 'like' in (4), from the point of view of the argument structure of the verb, the stimulus constituent qualifies as first argument in this occurrence (see the discussion in Luraghi [2020: 241-245]).
3. The verb apscKsiv areskein
In this section, I analyze the meaning and the constructions of apeoKeiv areskein. I review data from Homer (Section 3.1), post-Homeric epics (Section 3.2), Herodotus (Section 3.3) and Attic prose writers (Section 3.4). I then discuss the findings (Section 3.5).
3.1. Homer
The verb apeoKeiv areskein is attested only in the aorist and once in the future stem in Homeric Greek, hence expressing an inchoative, rather than a stative meaning. This points to a different semantics from the one typical of verbs that indicate mental or emotional states, such as 'like'. The verb seems to have been in the process of undergoing a semantic shift toward the experiential domain: indeed, in the Iliad it only partly belongs into this domain. Out of 10 occurrences, six are from the Iliad and four from the Odyssey. When the verb takes a human second argument, this is encoded in the accusative: dative constituents occurring with this verb are typically inanimate and indicate instrument.
In the Iliad, we find in the first place two formulaic contexts. The former features two almost identical occurrences of the middle aorist with neuter pronouns as direct objects, in which the verb means 'repair', 'amend', as in (9). In the second formulaic context the active aorist infinitive occurs twice, in two identical passages, and means 'make amends'. It does not feature a stimulus, as shown in (10).
(9) dU' i0i Tafrca S' orcioQev'
all' ithi tauta d' opisthen
but come.iMp.pRs.2sG dem.acc.pl ptc later
apeooo^eQ'
aressometh' appease.AOR.MiD.1pL
'Come on, we'll amend these things later.' (11. 4, 362).
(10) ay ¿0еХю apeoai, So^evat т'
aps ethelo aresai, domenai t'
back want.PRs.1sG appease.iNF.AOR give.iNF.AOR ptc
anepetoi' anoiva
apereisi' apoina endless.Acc.PL ransom.Acc.PL
'I want to make amends, donate an endless ransom.' (Il. 9, 120 = 19, 138).
Remarkably, the form apeoai aresai in (9) does not indicate an emotion, nor does it contain any possible experiencer. Here and in a similar passage in Il. 6, 526 the sentence that contains apeoKeiv areskein follows the description of a confrontation between two human participants, with one taking the initiative to end the confrontation and to leave more discussion for a later time. The verb indicates mutual reparation. Similarly,
(10) indicates that someone is willing to make reparation. It does not contain an experiencer, and not even a beneficiary. In fact, the verb apeoKeiv areskein in these occurrences cannot be viewed as belonging to the experiential domain, and human participants involved in the situation are encoded through the nominative, not through the dative.
The same meaning is shown, outside formulaic expressions, in (11).
(11) EupuaXo^ Se e afrcov apeooaoQra
Eurualos de he autdn aressastho Euryalus.NOM ptc refl dem.acc appease.iMP.AOR.MiD.3sG
¿neeooi Kai 5юрю, ¿nei ou ti eno^
epeessi kai dSroi, epei ou ti epos
word.DAt.PL and gift.DAt as neg indef.acc word.Acc
KaTa ^oipav eeinev A^Ktvoe Kpexov
kata moiran eeipen Alkinoe kreion
according right.Acc speak.AOR.3sG Alcinous.voc powerful.voc
toi yap еую
toi gar ego ptc ptc 1sg.nom
tov ^eivov apeooo^ai ю<; oi)
ton xeinon aressomai hos su
dem.acc stranger.Acc appease.Fut.MiD.1sG as 2sg.nom
KeXeúeu;... ^ícpeót; o Sií ^oi S<BKa<;
keleúeis... xípheós hó dé moi dokas order.PRs.2sG sword.Acc rel.acc ptc 1sg.dat give.AOR.2sc
ápeooá^evo^ énéeooiv
aressámenos epéessin
please.PTCP.AOR.MiD.NOM word.DAT.PL
'And let Euryalus make amends to the stranger himself with words and with a gift, for the word that he spoke was in no wise seemly ... Lord Alcinous, I will indeed make amends to the stranger, as you order me. ... This sword which thou hast given me, making amends with gentle speech.' (Od. 8, 396-397, 401-402, 414-415).
In (11), ápéoKeiv aréskein takes an accusative second argument that refers to a human being. In principle, accusative participants, too, might be taken as non-canonical subjects, as in Latin impersonal verbs of the type mepudet 'I feel shame'. However, in (11) the nominative constituent qualifies as first argument: in the second part of the passage, it refers to the participant which ranks highest on the individuation hierarchy, the first person singular. The state of affairs is presented taking Eury-alus as its starting point, and not the second human participant, which is referred to once with a demonstrative, and once with an epithet, ^eivov xemon 'the stranger'. In the final part of the passage, the stranger is reported as speaking, and praises Euryalus for making amend. The aorist participle ápeooá^evo^ aressámenos here does not take any other nominal argument except for the instrumental dative énéeooiv epéessin 'with words'.
In (12) the verb more clearly belongs into the experiential domain, and means 'please', 'appease'.
(12) áAl' exi Kai vuv ppaZró^eoQ' ra Kév
all' éti kai nün phrazómesth' hós kén but yet and now show.suBj.PRs.MP.1pL so ptc
^iv ápeooá^evoi neníQra^ev
min aressámenoi pepíthomen
3sg.acc please.PTCP.AOR.MiD.NOM.PL persuade.suBj.AOR.1pL
ôœpoioiv t' àyavoîoiv erceooi xe
dôroisin t ' aganoîsin épessi te
gift.DAT.PL PTC mild.DAT.PL word.DAT.PL PTC
'Let us think of how we may persuade him, pleasing him with kindly gifts and with gentle words.' (Il. 9, 111-113).
In (12) the subject, not overtly expressed, is the first person plural, a participant from the top of the individuation hierarchy that shares properties of agents rather than of stimuli, including the possibility of manipulating an instrument, here ôœpoioiv dôroisin 'presents' and erceooi épessi 'words'. The second human participant is portrayed as the target of the first participant's action, and can be taken as the second constituent in the construction.
Summing up, the verb àpéoKeiv aréskein in Homeric Greek is only marginally a verb that can be seen as belonging into the experiential domain. In fact, the formulaic nature of the occurrences in which the verb means 'repair', 'amend' and does not take a second argument points toward a development of the verb that started shifting to the experiential domain during the time span covered by the Homeric poems. It takes a construction that involves a nominative and an accusative constituent, often accompanied by a dative. Of these, the nominative must be taken as first argument: it always refers to a human participant actively involved in the situation. The accusative constituent is the second argument, likewise human, targeted by the action of the first participant. The dative constituent must be taken as third argument. It is always inanimate and indicates an instrument. Hence, constructions exhibited by this verb are NomAcc and NomAccDat, with the nominative always to be taken as first argument.
3.2. Post-Homeric epics
Post-Homeric epics offer evidence for a single occurrence of àpéoKeiv aréskein in (13). The passage is remarkable, because in the place of the instrumental dative we find a partitive genitive.
(13) aï 5è ^pévaç eùx' àpéoavxo aï^axoç
hai dè phrénas eût' arésanto haimatos
DEM.NOM PTC soul.ACC when satisfy.AOR.MID.3PL blood.GEN
àvôpo^éou androméou human.GEN
'And when they had satisfied their souls with human blood.' (Hes. Sh. 255-256).
This occurrence, which remains isolate, shows an extension of the construction of verbs of satiation to àpéoKeiv aréskein. Notably, verbs of satiation may take either the partitive genitive or, similar to àpéoKeiv aréskein in other occurrences, the instrumental dative [Luraghi 2020: 101-103]. In this occurrence, we can observe the extension of the NomAccGen construction typical of verbs of satiation to àpéoKeiv aréskein.
3.3. Herodotus
In Herodotus a new construction emerges, in which àpéoKeiv aréskein occurs with a dative constituent denoting a human participant. The verb most often occurs in the present (imperfective) stem, more suitable for a verb indicating a mental state: more specifically, 15 out of 19 occurrences feature the present stem, and only four the aorist. Let us consider examples (14) and (15).
(14) ejxoi 5è ai oai ^eyaXai emoi dè hai sai megalai 1sg.dat ptc art.nom.pl poss.2sg.nom.pl great.NOM.PL
eùxuxiai oùk àpéoKouoi
eutukhiai ouk aréskousi fortune.NOM.PL NEG like.PRS.3PL
'I don't like these great successes of yours.' (Hdt. 3, 40, 2).
(15) evxe Sii oi ^ X®Pn "npeoe eite dé hoi hë khôrë érese either ptc 3sg.dat art.nom land.NOM like.AOR.3sG
eice Kai aUra^ ^Qe^nce noi^oai touto
eite kai allos ethelese poiesai touto or and otherwise want.AOR.3sg make.iNF.AOR dem.acc
'Either because he found the land to his liking, or because for some other reason he desired to do so.' (Hdt. 4, 147, 4).
In both examples, the verb apeoKeiv areskein indicates a positive attitude of a human participant, the speaker (^oi moi) in (14) and a third person that has already been previously introduced in (15), toward some entities. In particular, in (14) the speaker expresses his dissatisfaction with some circumstances, ai oai ^eyaXai eucuxiai hai sai megalai eu-tukhiai 'your great successes', while (15) refers to a situation in which a concrete referent, ^ x®Pn he khOre 'the land', is the object of wish and desire of a human being.
In both passages, the dative denotes an experiencer, and qualifies as first constituent in the construction, as it refers to a more salient participants than the one referred to by the nominative NP. As for other subject properties, example (15) is especially enlightening, as it features two coordinated clauses, in which the experiencer is both the dative argument of apeoKeiv areskein and the subject of ^Qe^noe ethelese 'he wanted', which is not overtly realized but would normally be encoded in the nominative. If we look at the wider context from which this example is taken, we can see that both the dative oi hoi and the null subject in the coordinated clause refer to a participant which is the topic of this particular stretch of discourse: 'Cadmus son of Agenor landed at the place now called Thera during his search for Europa; and having landed, either because he (oi hoi) found the land to his liking, or because for some other reason he (0) desired to do so, he left on this island his own kinsman Membliarus'. This example also shows one of the four occurrences of the aorist stem. In this passage, the experiencer is depicted as suddenly experiencing the rise of a favorable attitude triggered by the stimulus.
Out of 19 occurrences of apeoKeiv areskein in Herodotus' Histories, nine are similar to (14) and (15), one contains the impersonal construction shown in example (5), also with a dative experiencer. Four other
occurrences also contain a dative experiencer and a nominative stimulus, but they feature a middle verb form as in (16).
(16) ^alioxa tSV ^vnox^prav ^peoKovxo
malista ton mnestSron ereskonto
most art.gen.pl suitor.GEN.PL like.iMPF.MP.3pL
oi an' AO^verav arciy^evoi
hoi ap' Atheneon apigmenoi
art.nom.pl from Athens.GEN.PL arrive.PTCP.PF.NOM.PL
'Among the suitors he liked best those that had come from Athens.' (Hdt. 6, 128, 2).
Summing up, the constructions shown in Herodotus are DatNom or simply Dat in the case of the impersonal construction, with the dative experiencer functioning as first argument in the construction and the nominative stimulus, if present, as the second argument.
The Homeric construction with a nominative and an accusative human participants, possibly accompanied by an instrumental dative, shown in (11), does not occur in Herodotus. However, in four occurrences we find a passive verb, that roughly provides a passive counterpart to the Homeric occurrences, as in (17).
(17) Paoxleucov Se XKUQECOV O SKUII^ basileuon de Skutheon ho Skules reign.PTCP.PRs.NOM ptc Scythian.GEN.PL art.nom Scyles.NOM
Siavrn ouSa^ra^ ^peoKeTO XKuQiKfi
diaitei oudamos eresketo Skuthikei
way_of_life.DAT neg please.iMPF.MP.3sG Scythian.DAT
'Though being king of Scythia, Scyles was in no way pleased with the Scythian way of life.' (Hdt. 4, 78, 3).
Note however, that, as verbs that take the dative or the genitive can passivize in Classical Greek [Conti 1998], one can also hold the construction in (17) as the passive of the DatNom construction that we found in other occurrences.
3.4. Attic prose writers
Plato's Dialogues provide evidence for both constructions, with the dative and with the accusative, as shown in (18) and (19).
(18) orcoTepra ouv ooi, ^v S' eycb, hopoteros oun soi, en d' egO, which_way ptc 2sg.dat say.AOR.1sg ptc 1sg.nom
apeoKei;
areskei? please.pRs.3sg
'Which (method) you prefer?' (Rep. 1.348b).
(19) nocepov touto ouk apeoKei oe;
poteron touto ouk areskei se?
whether dem.nom neg please.pRs.3sg 2sg.acc
'Does this not please you?' (Crat. 433d).
In (18) the experiencer is a second person pronoun, ooi soi, and the stimulus is an adverb orcocepro^ hopoteros 'which (of two) way', while in (19) we find again a second person pronoun encoding the experiencer, this time in the accusative oe se, and a nominative stimulus, touto touto 'this'.
Other Attic prose writers, similar to Herodotus, show a preference for dative experiencers, as shown by Attic orators Demosthenes and Isocrates in (20) and (21) respectively.
(20) tcov allrav ev ti aSuvaTrav, ton allon hen ti adunaton, art.gen.pl other.GEN.pL one.Acc indef.acc impossible.GEN.pL
oio^evrov eivai to toI<; anavcrav
oiomenon einai to tois hapanton
think.pTcp.pRs.GEN.pL be.iNF.pRs dem.acc art.dat.pl all.GEN.pL
cporcoi^ apeoKeiv
tropois areskein type.DAT.pL please.iNF.pRs
'While the others think it one of the impossible things to please men of every type.' (Dem. 61, 19).
(21) ei yap xoi^ ^ev Si' apex^v
ei gar tois men di' areten
if ptc art.dat.pl ptc through virtue.Acc
Sropea<; eiXn<pooiv ^peoKov
doreas eilephdsin Sreskon
reward.Acc.PL receive.PTCP.PF.DAT.PL like.iMPF.lso
'If those who have received rewards for their virtues like (me).' (Isoc. 15, 96).
In Thucydides we find both constructions, as shown in (22) with the accusative oe se 'you' and in (23) with the dative xouxoi^ toutois 'those ones'. However, it must be pointed out that the construction with an accusative experiencer is infrequent: it only occurs in (22). A limited number of occurrences with the passive seem to provide a passive counterpart to this latter construction, rather than to the construction containing a dative, as shown in (24).
(22) ei oüv xi oe xouxcov äpeoKei ei oün ti se touton areskei
if ptc iNDEF.NOM 2sg.acc DEM.GEN.PL please.PRs.3sG
'If any of these pleases you...' (Thuc. 1, 128, 7).
(23) xouxoi^ Se oük ^peoKe xa
toutois de ouk Sreske ta
DEM.DAT.PL PTC NEG like.IMPF.3sG ART.NOM.PL
rcpaooo^eva
prassdmena
make.PTCP.PRs.MP.NOM.PL
'They did not approve of these proceedings.' (Thuc. 5, 17, 2).
(24) o ^evxoi Ai%a^ oüxe ^peoKexo ho mentoi Likhas oute eresketo ART.NOM ptc Lichas.NOM neg please.iMPF.MP.3sG
aüxoi^ ecpn xe
autois ephe te
DEM.DAT.PL say.IMPF.3sG PTc
'Lichas was not pleased. He told them.' (Thuc. 8, 84, 5)
Comparison of examples (18)-(23) seems to indicate that the two constructions do not convey any special semantic difference, and it is rather a matter of choice by different authors to favor one construction over the other.
As for the structure of the constructions shown in examples from Attic prose writers, we can at least observe that in (18), (19), (22) and (23) the experiencer ranks higher on the individuation hierarchy than the stimulus: in examples (18), (19), (22) we find the second person singular, i.e. one of the participants of the speech act, while in (23) we find a demonstrative that refers back to groups of people (some of the allies of the Spartans), hence continuing the topic of the preceding sentence. As stimuli we find entities that are not only inanimate, but also characterized by a low degree of individuation, such as neuter pronouns as in (19) and (22), or abstract circumstances, such as ca npaooo^eva ta prassomena 'what was being done' in (23). Notably, however, in (21) not only the experiencer (those who have received rewards for their virtues) but also the stimulus (not overtly realized) is human, and it outranks the experiencer in the scale of individuation, being the first person singular. In (20) the stimulus is abstract, and even if the experiencer is scarcely individuated (coi^ anavcrav Tporcoi^ tois hapanton tropois 'men of all types') it remains more salient.
Example (18) from Plato does not contain a nominative stimulus alongside the dative experiencer: the stimulus is indicated by an adverb. This comes close to an impersonal construction such as the one in (6). In general, however, occurrences of the impersonal construction, in which apeoKei areskei governs an infinitive without any other nominal constituents besides the dative experiencer, quoted in lexicons from inscriptions or late authors, have limited parallels in classical prose writers.
3.5. Discussion
From the data discussed in this section one can see a clear semantic development from the meaning 'repair', indicating an act of reciprocal appeasement, to 'make amends', with a shift to the experiential domain and the development of the meaning 'please someone'. The events denoted
by the verb are always inchoative, as indicated by the aorist stem. After Homer, one can observe, on the one hand, the emergence of a shift to the domain of sensations with the extension of the construction of verbs of satiation, also with inchoative meaning, which however remained confined to post-Homeric epics. On the other hand, starting from Herodotus one finds the development of the stative meaning 'like' mostly in connection with the present stem of the verb, accompanied by a new construction that features a dative experiencer. This constituent shows subject properties, and qualifies as first argument in the construction. Occasionally, it can also occur as only nominal constituent in the sentence (so-called impersonal construction).
The developments outlined thus far can be represented as in Figure 1.
Meaning
repair ^ make amends ^ appease, please ^ satisfy Construction
Nom NomAcc(Dat) NomAcc(Dat) NomAccGen
DatNom Dat
Homer Hesiod Herodotus
Figure 1. The semantic and syntactic development of apeoKeiv areskein
like
In Figure 1 I used arrows to connect the meanings in Homer and Hesiod, but I separated the development in Herodotus with a dotted line, indicating that this was a completely new meaning and a new construction loosely related to the earlier ones.
In Attic prose writers, both a construction with a dative experiencer, as in Herodotus, and one with an accusative experiencer, as in Plato and Thucydides, are attested. The meaning of the two constructions seems to be very close: notably, the verb can indicate states in both cases, as shown by the occurrence of the present stem also with the accusative in (19) and (22). Not only the dative, but also the accusative seems to have subject properties. However, it must be pointed out that in some occurrences such as (21) it is possible that the nominative argument is considered the subject. In addition, passive occurrences such as (24) show that the argument that can be passivized qualifies as subject. Based
on comparison with active occurrences in Thucydides, it is likely that in a possible active counterpart one would find an accusative rather than a dative. Tentatively, one can conclude that Attic prose offers evidence for three or possibly four constructions: DatNom (possibly most frequent), NomDat, NomAcc and AccNom, even though a wider research on a larger corpus is needed to reach a better understanding of the distribution of all potential constructions.
4. Conclusion
In this paper, I have tested the claim that the verb ápéoKeiv aréskein means 'like' and features dative and accusative non-canonical subjects in Ancient Greek. Through a careful analysis of the extant evidence, I have shown that this was certainly not the case in Homeric Greek (or in post-Homeric epic). In the first place, in Homer ápéoKeiv aréskein did not indicate a mental or emotional state, as shown by the occurrence of the aorist stem that denotes an inchoative situation. In the second place, semanti-cally it only marginally belonged to the experiential domain. Its original meaning was arguably 'repair', 'amend', and indicated an act of reciprocal reparation between two human beings. The meaning 'appease, please', which is also attested in Homer, features a NomAccDat construction, with the nominative encoding the first argument and denoting a participant actively engaged in an action targeting the second participant (the accusative second argument, in such occurrences an experiencer) making use of an inanimate entity (encoded in the dative). Post-Homeric epics offer evidence for the extension to this verb of the construction typical of verbs of satiation, NomAccGen.
In Herodotus a completely new construction occurs in which the verb, now conveying a stative meaning through the present (imperfective) stem, means 'like' and features a dative experiencer. As the latter can be shown to bear subject properties, one has to reckon with a DatNom construction, which does not occur in Homer. Attic prose writers show a mix of the construction containing a dative experiencer and the one containing
an accusative experiencer. The experiencer seems to bear subject properties in most cases in which it is expressed in the dative, although not all occurrences support this claim. Its status remains less clear in cases in which it is encoded in the accusative.
Abbreviations
1, 2, 3 — 1st, 2nd, 3rd person; acc. — accusative; art. — article; aor. — aorist; dat. — dative; dem. — demonstrative; fut. — future; gen. — genitive; imp. — imperative; impf. — imperfect; indef. — indefinite; inf. — infinitive; mid. — middle; mp. — mediopassive; neg. — negation; nom. — nominative; pl. — plural; pf. — perfect; prs. — present; ptc. — particle; ptcp. — participle; refl. — reflexive; rel. — relative; subj. — subjunctive; sg. — singular; voc. — vocative.
References
Barödal, Eythórsson 2009—J. Barödal, Th. Eythórsson. The origin of the oblique-subject construction: an Indo-European comparison. V. Bubenik, S. Rose (eds.) Grammatical change in Indo-European languages. Amsterdam: Benjamins, 2009. P. 179-193.
Barödal, Eythórsson 2020 — J. Barödal, Th. Eythórsson. How to identify cognates in syntax? Taking Watkins' legacy one step further. J. Barödal, S. Gildea, E. Lujan (eds.) Reconstructing syntax. Leiden: Brill, 2020. P. 197-238. Barödal et al. 2012 — J. Barödal, Th. Smitherman, V. Bjarnadóttir, S. Danesi, G. B. Jenset, B. McGilliwray. Reconstructing constructional semantics: the dative subject construction in old Norse-Icelandic, Latin, Ancient Greek, Old Russian and Old Lithuanian. Studies in Language. 2012. Vol. 36. Iss. 3. P. 511547.
Conti 1998 — L. Conti. Zum Passiv von griechischen Verben mit Gen. bzw. Dat. als zweitem Komplement. Münchener Studien zur Sprachwissenschaft. 1998. Vol. 58. P. 13-50.
Conti 2010—L. Conti. Análisis del dativo en construcciones impersonales: Los conceptos de sujeto y de semisujeto en griego antiguo. Emerita. 2010. Vol. 78. Iss. 2. P. 249-273.
Keenan 1976—E. Keenan. Towards a universal definition of 'subject'. Ch. Li (ed.). Subject and Topic. New York: Academic Press, 1976. P. 303-333.
Luraghi 2020 — S. Luraghi. Experiential verbs in Homeric Greek. A constructional
approach. Leiden: Brill, 2020. Viti 2016 — C. Viti. Areal distribution of argument marking of Indo-European experience predicates. Journal of Indo-European Studies. 2016. Vol. 44. P. 1-84. Viti 2017—C. Viti. Semantic and Cognitive Factors of Argument Marking in Ancient Indo-European Languages. Diachronica. 2017. Vol. 34. P. 368-419.
