Latin queo ‘I can’ and Sanskrit yā - ‘to be possible’: a parallel development?

: This study challenges the alleged parallel between Latin queo ‘I can’ and Sanskrit yā - ‘to be possible’. Though sharing some similarities (e.g., motion as a source domain, negative polarity contexts), they exhibit significant differences. Queo is personally constructed and typically conveys participant-related possibility, while yā - constructions are linked to agent demotion. The passive meaning of yā - in personal constructions and its possible connection to modality are also briefly addressed.


Introduction
Addressing the origins of the Latin verbs nequeo 'I cannot' and queo 'I can' as derived from eo 'to go, to come', Osthoff (1898) posits a parallel in the semantic development of the Sanskrit verb yā-'to go'.He also suggests translating Sanskrit na yānti vaktum (cf.example 9 below) with Latin dici nequeunt 'they cannot be expressed ' (1898: 179).This alleged parallel is later mentioned by Schokker (1969Schokker ( -1970: 22): 22), who investigates the origins of the New Indo-Aryan passive construction featuring jānā (from Sanskrit yā-).The aim of this paper is to reassess this parallel to determine the extent to which Latin queo and Sanskrit yā-exhibit a similar semantic development and a similar syntactic behaviour.While both verbs show a clear development of the modality of possibility from motion constructions, they are deeply different from the semantic and the syntactic points of view, as shown below.The following section presents recent advances in research on queo and nequeo (hereafter referred to as (ne)queo) and outlines their semantic and syntactic profile.Section 3 describes the semantic and syntactic profile of yā-and compares it with that of (ne)queo.The emergence of the passive meaning of yā-in personal constructions and its possible connection to modality are also briefly addressed.In the conclusion (4), further lines of investigation are proposed.

FRANCESCA DELL'ORO
Latin queo 'I can' and Sanskrit yā-'to be possible': a parallel development?
2. Semantic and syntactic profile of (ne)queo With regard to the origin of the Latin verbs queo 'I can' and nequeo 'I cannot', the current etymological explanation posits that nequeo derives from the univerbation of the conjunction and adverb neque '(and) not' with the impersonal form it/itur 'it goes' from eo 'to go, to come' 1 .Subsequently, on the model of negative vs positive pairs (with and without the negative preverb ne-), such as ne-scio 'I don't know' : scio 'I know', nequeo was reinterpreted as the negated form (i.e., as ne-queo) of a positive form queo which was newly created 2 .In this form this explanation traces back to Brugmann (1904aBrugmann ( : 668, fn. 1, 1904b: 64, fn. : 64, fn. 2), who suggests starting with the meaning «es geht irgendwie nicht» 'somehow it does not go' 3 .Brugmann's view (neque + eo) corresponds to the etymology currently accepted and reported in reference etymological dictionaries of Latin: de Vaan (2008: s.vv.queo and nequeo), Ernout & Meillet (1959[2001]: s.v.queo), Walde & Hofmann (1938-1956: s.v. queo).
The early history of Latin (ne)queo has been recently revised.Dell'Oro (2023) has challenged Brugmann's view that an impersonal construction is at the origin of (ne)queo.In fact, the impersonal construction is not reliably attested in the pre-classical period.In the passages where this construction might have occurred, the fragmentary condition of the text makes it impossible to determine definitively whether the usage is personal or impersonal.Consider, for instance, the following fragment from Caecilius Statius' comedy 4 : ( On the basis of philological data and typological parallels, Dell'Oro (2023) suggests that nequeo did not develop from impersonal constructions, but from personal ones.Below, I outline the main points of her argument.
1. On the semantics of eo, cf.Nuti (2016).2. According to Moussy (2001: 486), the direction of derivation between queo and nequeo remains a matter of debate.Interestingly, Moussy notes that Cicero considered nequeo to be derived from queo.This insight from Cicero highlights how ancient Latin speakers might have perceived the relationship between the two verbs from a synchronic standpoint.However, I harbour some reservations regarding the reliability of using this data to reconstruct the early history of queo and nequeo.3. The core of this explanation was already suggested by Osthoff (1896).The form nequeo is thought to contain a negation (ne), an instrumental form or an indefinite adverb -on this point cf.already Bréal (1889: 128) who suggests quī 'how, by which' -and the verb eo.Notably, Brugmann's addition of «irgendwie» may echo the etymological suggestion made by Osthoff. 4. For a discussion of this and the other two ambiguous pre-classical passages, cf.Dell'Oro (2023: 188-189).5. Unless otherwise specified, the editions of the Latin texts cited in this study are taken from the Classical Latin Texts database, accessible at https://latin.packhum.org.With regard to the linguistic glosses, I follow the Leipzig glossing rules.I have added the following abbreviations not present in the Leipzig system: GER = gerund and / = verse division.
FRANCESCA DELL'ORO An in-depth analysis of the Early Latin texts clearly shows that, where it is possible to verify whether there is a syntactic subject, nequeo and queo are always used personally.They convey almost exclusively dynamic modality, primarily participantrelated possibility.'Dynamic' refers to a type of modality characterising an event or state in terms of possibility or necessity 6 .Dynamic modality can refer to someone's abilities (participant-inherent dynamic modality, cf. 2), to their possibility of acting or being in a specific situation (participant-imposed dynamic modality, cf. 3) and to possibilities inherent to the situation (situational dynamic modality, cf.4).The nomenclature and definitions are based on Nuyts (2016).It is important to stress that precisely distinguishing between these three sub-types can be difficult.
( The impossibility of restraining someone, here the slave Gripus, is presented as general.It is not related to a specific individual capable (or incapable) of affecting the state of affairs (henceforth SoA), nor is it contingent upon specific circumstances that enable or restrict the participant's actions.
While both participant-inherent and participant-imposed modality usually imply a volitional participant with an agentive or agent-like role, situational modality often involves no participant.If there is one, it is usually not relevant.In pre-classical Latin, 85.88% of the attestations of (ne)queo express participant-related dynamic modality.Note that the personal construction of (ne)queo aligns well with that of eo as a verb of self-motion.Importantly, the Latin data fit current typological knowledge about the development of participant-related possibility modality from verbs of motion in motion-cum-purpose constructions.In such constructions, the main participant who is voluntarily moving (in order to) accomplish something has developed into the main participant about whom possibility is predicated.
The use of (ne)queo in the passive form or associated with a passive infinitive is rare 8 .It is also worth noting that the distribution of negative and positive forms is as follows: Nequeo 42.37% Queo within the scope of any form of negation 44.63% Queo 12.99% 7. All passages from Plautus' comedies are taken from the edition by W. de Melo.This will not be specified further.8. (Ne)queo passive constructions usually express situational modality and, due to the absence of an agentive or agent-like participant, function similarly to impersonal constructions (cf.also below).However, this fact does not support the traditional hypothesis that (ne)queo originated from impersonal constructions, as (ne)queo constructions are personal.
This data shows that in absolute terms, queo is more frequent than nequeo.However, queo is typically used in contexts of negative polarity, making negated queo an alternative to nequeo.Contexts with positive polarity are comparatively rare 9 .
The paper that provides the above-mentioned data and results concerning (ne)queo does not deal with the Sanskrit parallel suggested by Osthoff (1898).The primary aim in this paper is to address this issue.Additionally, the alleged Sanskrit parallel will enable a brief examination of the relation between motion, passive meaning, and the emergence of modality.
Before dealing with the modal meaning of the verb yā-and its constructions, it is worth highlighting two points.Firstly, it is important to note the long-lasting -already Vedic -tendency for Sanskrit to employ motion verbs as semi-copular verbs in various constructions.Unlike the copula, a 'semi-copula' adds some elements of meaning, such as aspectual nuances, that the copula alone does not convey 10 .Sanskrit motion verbs such as i-'to go, to walk, etc.' and car-'to move one's self, to go, to walk, etc.' evolved into semi-copular verbs early on.When combined with non-finite forms such as present participles, these constructions may lose their original lexical meaning of motion and instead express aspectual meanings related to habituality and continuity (cf., e.g., Whitney 1896: 394-395, and more recently, Grieco 2023a, 2023b 11 ).This point is strictly intertwined with the second one.
While some constructions with yā-do not convey modal meaning, they licence passive meaning.Cf. vināśaṃ yāti 'he goes to destruction, i.e., he is destroyed' (Monier-Williams 1899: s.v.yā-), where the construction features an abstract noun 12 .9. For a more detailed account of nequeo and queo in the pre-classical period, cf.Dell'Oro (2023).10.Cf.: «The main difference between constructions containing a copula and those containing a semicopula is that the semi-copula can never be left out without changing or affecting the meaning of the resulting construction.In other words, the semi-copula adds an element of meaning to the construction in which it occurs, whereas the copula does not» (Hengeveld 1992: 35).It is important to note that, though the literature on copulas is extensive and studies on copulas often touch upon the status of semi-copulas and that of auxiliaries (see Moro 2006, 2010 3 , Pustet 2003, Creissels 2017, Arche et al. 2019, among  others), the investigation of semi-copulas remains significantly less developed (cf., e.g., Schmitt 2005).In keeping with this trend, the specific relationship between semi-copular constructions and the emergence of modality and related domains is still inadequately explored (cf.Sansò, Giacalone Ramat 2016, and, e.g., Lauwers, Duée 2011, regarding the emergence of the closely related notion of evidentiality).11.For yā-, cf.Grieco (2023: 236 with example 22).12. Cf. Renou (1961: 495) about the aspectual value of motion constructions with abstract nouns, and Schokker (1969Schokker ( -1970: 4-6) : 4-6) for the diachronic development.Cf. also Grieco (2023a: 246 with example 44).
With reference to the examples collected by Schokker (1969Schokker ( -1970)), yā-constructions that licence modal meaning also convey passive meaning.This could reflect a bias due to the main focus of Schokker (1969Schokker ( -1970)), i.e., the emergence of passive voice.While this is not the aim of this study, a new corpus-based diachronic investigation of the available data could shed new light on this issue 13 .Surprisingly, Schokker (1969Schokker ( -1970) ) does not explicitly notice or comment on the association between passive voice and possibility.Yet, the connection is strong and relevant from the historical point of view.Typological studies have shown that modality can develop from motion constructions -as in the case of Latin (ne)queo (cf.Dell'Oro 2023: 9-10 with references) -as well as passive constructions ('potential passive' in the terms of Haspelmath 1990: 33-34).Furthermore, motion constructions can evolve into passive constructions, featuring or not modal meaning (cf.Kuteva et al. 2019: 95-96, 314-315, and, with reference to necessity, cf. also Bourdin 2014).Therefore, with respect to the pathway followed by Sanskrit yā-, the question arises whether the emergence of modality has been mediated by the passive meaning of the construction or the two meanings emerged independently.As will become apparent from the analysis of the passages, it is currently not possible to reach a definitive conclusion on this issue.
Though the details of the diachronic development are not entirely clear (Bubenik 1998: 134), it is interesting to note that modern Indo-Aryan languages use the construction 'go + past participle' to express passive meaning.The 'go'-verb in this construction is a reflex of Sanskrit yā-, as shown by Schokker (1969Schokker ( -1970)), who provides a comprehensive overview of the attested constructions from Sanskrit to the modern Indo-Aryan languages.Cf., e.g., in Hindi the construction 'ja + past participle' 14 : ( 13.Further investigation into an enlarged corpus is necessary to understand the emergence of both the passive and the modal constructions.14.Cf. Montaut (1991: 130) and Kachru (2006: 176-178).15.Linguistic glosses have been harmonised.
'Will the letter be written by Renu? i.e., Will Renu be able to write the letter?' (from Kachru 2006: 176 16 ) Sansó and Giacalone Ramat (2016: 14-18), drawing on Bourdin (2014), suggest that the modal meanings in the modern Indo-Aryan constructions 'go + past participle' arose from passive voice.With respect to the focus of this paper, i.e., Sanskrit, it is important to note that the most relevant constructions are those with the infinitive and those with the gerund 17 , as will be apparent in a moment.Therefore, whatever the exact reconstruction of the grammaticalization pathway of the 'go + past participle' construction attested by the modern Indo-Aryan languages, it does not seem possible to draw a direct line of continuity.However, it seems undeniable that not only the modal and passive yā-constructions, but also the aspectual constructions featuring other motion verbs, such as i-and car-, paved the way to the emergence of the Modern Indo-Aryan analytic passive construction.This diachronic issue, which needs to be further investigated, is beyond the scope of this paper.
In Sanskrit, modal constructions featuring yā-are most commonly attested with an infinitive 18 .The construction with the gerund is also attested.With all due reservation, though the attestations are scanty and a comprehensive corpus-based investigation is still needed, it seems possible to formulate some generalisations.When the syntactic subject of yā-is an entity that can bring about the event described by the infinitive (or gerund) depending on yā-, the construction can licence both an active or a passive reading.Only the context can help disambiguate this (cf.8).When the syntactic subject of yā-refers to an entity that contextually is not supposed to bring about the event described by the infinitive (or gerund) depending on yā-, the construction licences a passive reading.When the reading is passive, there is usually also a modal reading (cf., however, example 15 below), but not when the reading is active 19 , as illustrated by the following example: Bhūteśvara.ACC king.NOM 'And thus, once upon a time, the king went to worship Bhūteśvara' (Kalhaṇa,Rājataraṅgiṇi,V,48)  20  In this sentence, the lexical value of the verb yā-is retained.( 8) Svecchopajātaviṣayo (sveccha-upajāta-viṣayo) 'pi na At_will-added-dominion.NOM then NEG yāti vaktuṁ dehīti […] īśvara-durvidagdhaḥ go.3SG speak-INF give.IMP=so king-wrong-headed.NOM 'Though his possessions come to him at his own pleasure (lit.: though he obtains dominion at will), a wrong-headed king cannot be told to give' (lit.'a wrongheaded king cannot be said: give!') (Bāṇa,Harṣacarita II,13) In principle, (8) might be interpreted as 'a wrong-headed king is not able to say'.However, since the king in question is the god of love, Kāma, it appears more appropriate to depict him as someone who cannot be persuaded to be generous rather than as someone unable to speak 21 .Nonetheless, the ambiguity persists.
With inanimate referents the passive reading is more straightforward: (  (from Renou 1978: 71) This phenomenon does not occur with yā-(cf.example ( 16) below).The active or passive interpretation depends on the context and, in some instances, the intended meaning remains ambiguous.
As seen above, Latin (ne)queo predominantly exhibits personal active constructions.In the case of the rare construction of yā-combined with the infinitive which Osthoff (1898) suggests form a formal and semantic parallel, both impersonal and personal constructions are found 23 .Impersonal constructions are exemplified in sentences ( 12) and ( 13), while personal constructions are illustrated in sentences (14) -corresponding to example (8) -, and (15).
supreme_king.VOC etayor niyamasthayoḥ these.DU.LOC observing_a_vow.DU.LOC 'O supreme sovereign, it is impossible for both of them to speak because they observe a vow'.(lit.'As they both are observing a vow, it is not possible to speak.')(Mahābhārata, interpolation 221 after II,19,30 24 )   This is a case of participant-imposed modality expressed through a syntactically impersonal construction, i.e., without a referential subject.This passage can be interpreted as conveying situational modality, as no specific participant is involved.As previously mentioned, impersonal constructions with (ne)queo are not attested with any certainty in pre-classical Latin.Consequently, the Indian impersonal construction cannot be considered to constitute a parallel to the (ne)queo constructions.With regard to the personal constructions, instead, (ne)queo and yā-shows a certain degree of similarity, as demonstrated by the following examples.
(14 = 8) Svecchopajātaviṣayo (sveccha-upajāta-viṣayo) 'pi na At_will-added-dominion.NOM then speak-INF give.IMP=so king-wrong-headed.NOM 'Although his possessions come to him at his own pleasure, a wrong-headed king cannot be told to give.(lit.'though he obtains dominion at will, a wrongheaded king cannot be told, "give!"') (Bāṇa, Harṣacarita II, 13) (15) virahaḥ soḍhuṁ kathaṁ yāsyati separation.NOM.SG bear.INF how go.FUT.3SG'How will the separation be borne?' (Somadeva, Lalitavigraharājanāṭaka in Kielhorn 1901: 9, 4) The future tense in (15) seems to influence the modal reading.This may be because (1) both future and modalised events are not actual; and (2) the modal interpretation of the yā-construction arises from implicatures and has not yet been grammaticalized.If one accepts that (15) can be interpreted as modal, then both (14 = 8) and ( 15) can be considered examples of dynamic situational modality, as there is no prominent participant in the modalised SoA.
In Sanskrit, similar constructions are also found with the gerund, as shown by (16).However, it is important to note that these constructions can also convey aspectual readings, as illustrated by (17).(16) (Aśvaghoṣa, Saundarananda VII, 15, ed. Johnston, 1936) In this case, too, the passage conveys situational modality, as it lacks an agentive participant.It is worth noticing the parallel between the 'gerund + yā-' construction and the 'infinitive + śak-' construction.(17)  : is being continuously separated), therefore the feeling 'this is mine' is improper with regard to a coming together that is transitory as a dream' (Aśvaghoṣa, Buddhacarita VI, 48, ed. Johnston 1936) If the formal differences with regard to the active/passive forms are set aside, these constructions closely resemble the (ne)queo constructions, when the active or passive forms of (ne)queo introduce a passive verb.Cf. examples (1) Plautus, However, as outlined above, (ne)queo shows a clear preference for the active form and for active infinitives.Moreover, the syntactic subject of (ne)queo usually refers to an entity with an agentive role.The construction typically expresses participant-related modality.Judging from the scanty Sanskrit attestations, this does not hold for yā-.Also taking into account the previous discussion about the impersonal construction, it is therefore possible to conclude that, beyond some superficial similarity (i.e., the semantic evolution from motion to possibility), the two verbs seem deeply different in their syntactic and semantic profiles.

Conclusion and outlook on future research
The comparison of the Latin (ne)queo construction with the Sanskrit yāconstruction has shown that the parallel suggested by Osthoff (1898) is misleading.Although it is possible to recognise a similar semantic development from motion to possibility, there are significant differences.The features of the relevant modal constructions are schematically presented in Table 2.The emerging picture indicates that Latin (ne)queo is a more grammaticalized form, while Sanskrit yā-also has lexical and aspectual non-modal values.With an infinitive: more recent construction that likely predates the 5 th century CE modal meanings.As observed, (ne)queo evolved from personal constructions with an agentive or agent-like participant.The emergence of the modal meaning in yāconstructions presents a less clear picture.It is apparent that the semantic development of yā-from motion to modality is tied to the emergence of the semi-copular uses of this verb (and of other motion verbs), expressing aspectual and voice values.However, due to insufficient data, it remains unclear whether the modal yā-constructions evolved from passive constructions (x 'goes' not done → x cannot be done → it is not possible to do x) or from impersonal constructions (doing x does not go → doing x is not possible → x cannot be done).It is crucial to highlight that the development of modality from impersonal motion constructions is a neglected area of the diachrony of modality and deserves further research.More broadly, the investigated case also underscores the need to explore the interplay between agent demotion/promotion and personal/impersonal constructions to better understand the emergence of certain types of modality.As seen above, both the Sanskrit construction with yā-and the Latin construction with (ne)queo convey the notion of possibility.Yet, the Sanskrit construction and the Latin one also diverge.The Sanskrit modal constructions with yā-are impersonal or convey passive meaning, while (ne)queo constructions are personal and usually convey active meaning.Moreover, the construction seems to influence the modality types that it is possible to express.Specifically, the Sanskrit and the Latin constructions tend to convey a distinct type of possibility: in the Latin construction the participant in the modalised event is usually salient, while in the Sanskrit construction it is usually the modalised event that is highlighted.These findings call for confirmation through intralinguistic studies of modality within Sanskrit and Latin.

Bibliographical references
not possible to stay without a king' (lit.'without king [it] doesn't go to stay') (Nārāyaṇa, Hitopadeśa I 9 p. 41, 3 Peterson) For the former (bonds) can be cut or broken by one's own might or the strength of friends, but the snare of love cannot be loosened except by true knowledge or hard-heartedness'.
NOM.SG therefore sleep-been.LOC.SG coming_together.LOC.SG 'And since this world is in a state of continuous separation (lit.

Table 2 .
Overview of the relevant features of (ne)queo and yā-modal constructionsThe investigation of (ne)queo and yā-has revealed that there is not a singular trajectory from motion to modality, as there exist varied motion constructions and