home edit page issue tracker
This page pertains to UD version 2.

iobj: either nominal object of a ditransitive verb with neutral alignment

The use of the relation iobj in Latin is inevitably marginal, if present at all, since ditransitive verbs with neutral alignment, i.e. verbs with three arguments two of which are expressed as core objects (= in the accusative case for Latin; see Malchukov 2013), represent a very limited, closed class consisting of:

and some of their derivates like edoceo ‘I teach thouroughly’ or interrogo ‘I inquire’. There are other sparse, early attestations of verbs taking two objects, but some of them might be spurious, while they are generally strongly constrained in their realisation anyway (Bennett 1914, p. 247 ff.).

The Latin language is typologically peculiar in that the class of its neutrally aligned ditransitive verbs does not include do ‘I give’ and similar terms, which instead invariably follow an indirective alignment, i.e. the recipient is expressed as an oblique argument, most often a bare nominal in the dative case, while the theme is expressed as the core object. There are also verbs like dono ‘I give as a present’ with a secundative alignment, i.e. the recipient is expressed as the core object, while the theme is oblique, e.g. in the ablative case (where we can see an instrumental nuance). It seems so that Latin’s neutrally aligned ditransitive verbs are restricted to those with a “maximal” asymmetry between the recipient and theme arguments (cf. Haspelmath 2015, §7): doceo is protoypical in this sense, since the recipient will almost invariably be an animate, specific referent, while the theme an inanimate, usually abstract entity; celo shares similar features as a verb of dispossession; rogo, flagito and posco might be regarded almost as opposites of ‘give’-verbs where transitivity is high with respect to the theme, which is also mostly an abstract referent (often expressed as a complement clause) as opposed to the usually animate and specific recipient, and, further, we notice that flagito and posco both originally represent derived forms (respectively diminutive/frequentative and inchoative) which give them a nuance of “urgence” and thus possibly heightened transitivity. All of these verbs also present indirective or secundative constructions, some of which can be seen as cases of lexical partitions (e.g. doceo aliquem de aliqua re ‘I inform someone about something’, instead of ‘I teach’). Conversely, other verbs of asking like peto ‘I ask (to get something)’ and quaero ‘I ask (to know something)’ are always indirectively aligned, possibly because the spatial analogy is stronger for them, as substantiated by the recipient being introduced by the prepositions ab or ex ‘from’, and not by the dative case as for do (cf. again Haspelmath 2015, §7).

Some commented examples of these constructions follow.

Miramur nos et pariter credimus osculatique mensam rogamus nocturnas ut suis se teneant dum redimus a cena \n We-wonder us and evenly we-believe kissed-and table we-ask nocturnal that to-its-own self that-they-keep while we-come-back from dinner
i?obj(rogamus,nocturnas)
ccomp(rogamus,teneant)
i?obj(we-ask,nocturnal)
ccomp(we-ask,that-they-keep)

‘We were full of wonder and faith, and we kissed the table and prayed the Night-riders to stay at home as we returned from dinner.’ (Perseus phi0972.phi001.perseus-lat1.xml@782, Satyricon 64.1, Petronius)

et docebat eos in parabolis multa et dicebat illis in doctrina sua \n and was-teaching them in parables many and was-saying to-those in doctrine his-own
i?obj(docebat,eos)
i?obj(docebat,multa)
obl(dicebat,illis)
i?obj(was-teaching,them)
i?obj(was-teaching,many)
obl(was-saying,to-those)

‘And he taught them at length in parables, and in the course of his instruction he said to them’ (PROIEL 10478, Jerome’s Vulgate Mark 4)

petia ipsa de terra in integrum mihi eas dedisti \n piece same off earth in integral to-me them you-have-given

obj(dedisti,eas)
obl(dedisti,mihi)
dislocated(dedisti,petia)
obj(you-have-given,them)
obl(you-have-given,to-me)
dislocated(you-have-given,piece)

‘that same portion of land - you have given me those integrally’ (LLCT test-s26)

For the aforementioned ditransitive verbs, we can indeed observe two arguments expressed in the accusative case, which in Latin is the necessary condition to being a core object (while not sufficient: an argument in the accusative is not always an object). As for passivisation, even if we recognise strong tendencies in favour of either one of the objects as the candidate for the passive subject, in principle both the recipient and the theme can be selected, so there is no truly “primary” nor “indirect” object, and the choice to which one to apply iobj is ultimately arbitrary and dictated by tradition: it often falls on the recipient as the “accusative of person”, which however is the most often passivised argument, and so a more ideal candidate for obj (if this is intended to mean “primary”). In fact, simply allowing for the annotation of two objs would be the most sensible solution (and this already happens when both an obj and a ccomp/xcomp relation are present at the same time). In passive constructions, the remaining object rarely stays as such, or its realisation is limited: it is usually expressed as an oblique argument, if at all, or constrained to a (neutral) pronominal element, and this is what can be observed in all other sporadic cases where a verb sports two actual objects (Bennett 1914, p. 247 ff.). A kind of grammatical split seems at work here (Napoli 2016; cf. Haspelmath 2015, §4).

Some more or less clearly assessable examples of passivisation of ditransitive constructions follow.

huc venient qui te pervisere gliscent Parrhasii iuvenesque senes et carmina leti qui nova mirari cupiantque antiqua doceri \n hither they-will-come that you to-contemplate they-will-swell Parrhasians youngsters-and elders and songs merry those new to-behold that-they-long-and ancient to-be-taught
conj(iuvenesque,cupiantque)
nsubj(cupiantque,qui-13)
obj(doceri, antiqua)
xcomp(cupiantque,mirari)
conj(mirari,doceri)
e:xcomp(cupiantque,doceri)
e:nsubj(cupiantque,qui-13)
e:nsubj(doceri,qui-13)
conj(youngsters-and,that-they-long-and)
nsubj(that-they-long-and,those)
obj(to-be-taught, ancient)
xcomp(that-they-long-and,to-behold)
conj(to-behold,to-be-taught)
e:xcomp(that-they-long-and,to-be-taught)
e:nsubj(that-they-long-and,those)
e:nsubj(to-be-taught,those)

it. ‘qua verranno desiderosi di conoscerti giovani e vecchi parrasii, e quelli che vorranno lieti ammirare i nuovi carmi e studiare gli antichi.’

eng. ‘hither young and old Arcadians, desiring to know you, will come, and also those who will gladly long for beholding the new songs and studying the ancient ones.’ (UDante Egl-66, Eclogues III 69, Dante Alighieri)

Itaque cum in dogmatibus moralis negotii amicitiam adequari et salvari analogo doceatur ad retribuendum pro collatis beneficiis plus quam semel analogiam sequi mihi votivum est \n Therefore as in doctrines moral of-business friendship to-be-equalized and to-be-saved it-be-taught to being-retributed for-the-sake-of collected off-favours more than once analogy to-follow to-me vowed is
csubj:pass/ccomp(doceatur, adequari)
conj(adequari, salvari)
advcl(votivum, doceatur)
csubj:pass/ccomp(it-be-taught, to-be-equalized)
conj(to-be-equalized, to-be-saved)
advcl(vowed, it-be-taught)

‘Therefore, since it is a doctrine of ethics that friendship is equalized and preserved by reciprocity, it is my wish to preserve due reciprocity in making a return for the bounty more than once conferred upon me.’ (UDante Epi-246, Letters XIII 10, Dante Alighieri)

et tamen circiter parte tertia ut postea perspectum est celata atque in oppido retenta portis patefactis eo die pace sunt usi \n and nevertheless around off-part third as after-that seen-through is hidden and in town kept off-doors opened that off-day off-peace are used
nsubj:pass(celata, parte)
conj(celata,retenta)
advcl:abs(usi,celata)
nsubj:pass(hidden, off-part)
conj(hidden,kept)
advcl:abs(used,hidden)

‘and nevertheless having retained and concealed, as we afterward discovered, about a third part in the town, the gates were opened, and they enjoyed peace for that day.’ (PROIEL 53139, De Bello Gallico 2.32.4, Caesar)

There are other cases in Latin where a verb shows two bare arguments in the accusative cases, but they are all distinct from ditransitive constructions: the most frequent cases are 1) when one of two arguments is a temporal or locative expression, and so an oblique argument, 2) when one of the arguments is a secondary predication of the object, and 3) when the verb is a so-called copulative verb and qualifies the object in some way. In the first case, the apparent object cannot be the target of passivisation; in the latter two cases, the two arguments are coreferent, so the construction is different from a ditransitive one and, further, the secondary predication cannot be passivised. Still, in the latter case the treatment of the copular complement as xcomp might be seen as a particular case of double-object construction.

Latin does not seem to have morphologically derived causative/benefactive verbs, instead relying on “biclausal” constructions making use of other lexical verbs, so that multiple objects there can always be described as depending on either verb. Many, if not all, Romance languages derived from Latin have eventually ousted those few neutrally aligned ditransitive constructions, replacing them regularly with indirective and, to a lesser extent, secundative ones (e.g. Italian l’insegnante insegna matematica ai bambini ‘the teacher teaches mathematics to the children’). This has also been extended to periphrastic causatives, as e.g. in Italian faccio leggere un libro a una mia amica ‘I make/let a friend of mine read a book’, lit. ‘I make read a book to a friend of mine’, contrasted with possible Latin amicam meam librum legere facio, where we have the relations obj(facio,amicam), xcomp(facio,legere), obj(legere,librum), i.e. the theme is an object, though of the subordinated verb (there could also be a parallel interpretation where amicam meam librum legere might be seen as an accusativus cum infinitivo, and so as a whole as a clausal object of facio ‘I make’, where amicam is the subject).

magnificentia sua me sibi ab annosis temporibus sponte sua fecit esse subiectum \n magnificence own me to-self from old off-times off-free-will own he-has-made to-be subjected
nsubj(fecit, magnificentia)
obj(fecit, me-3)
aux:pass(subiectum, esse)
xcomp(fecit, subiectum)
nsubj(he-has-made, magnificence)
obj(he-has-made, me-16)
aux:pass(subjected, to-be)
xcomp(he-has-made, subjected)

‘his nobility of soul for long years past, as he willed, made me his servant.’ (Udante Epi-13, Letters II 1, Dante Alighieri)

Still, the contemporary presence of the two relations obj and xcomp might be seen as a special case of double object, even if the relation represented by xcomp here leans more towards a serial or periphrastic verb construction than an object in a canonical sense.

About terminology and ‘indirect objects’

The use of the label “indirect”, as opposed to “direct” or “primary”, to label one of the objects in a construction like English she teaches the students linguistics is cause of confusion, since, formally speaking, it does not correspond to what is usually meant by “indirect object” in Latin and Romance languages. The latter is an argument (more precisely, a complement in a complement-adjunct distinction framework) of the verb in the genitive, dative, ablative case or introduced by a preposition; or equivalently, an argument which is not the subject of the clause and is not in the accusative case (temporal complements and similar are left outside of this discussion, since they are usually adjuncts). As mentioned before, the ‘give’-type verbs in Latin are regularly indirectively aligned, so that the recipient will be an “indirect object” in the traditional sense, i.e. a complement expressed in the dative case. This terminology has been taken over directly into English with regard to the recipient, considering only its semantic role, while, however, English often expresses this argument in the exact same way as the “regular” object, i.e. as a bare nominal or a non-nominative pronominal form. Despite this labeling, there is no way to distinguish the two objects from each other from a formal point of view, and, moreover, what is traditionally labeled as indirect object behaves more as a “regular” object than the theme (e.g. passivisation preferences).

This is therefore a case where the traditional use of a label is misleading in the universal context of UD and leads to false expectations, also being based more on semantic roles rather than on morphosyntax. Re-applied to Latin and Romance languages, it becomes synonimous of “dative argument”, but these so-called “indirect objects” in Latin and filiated Romance languages are plain and simple oblique arguments by definition, since they are expressed in oblique (i.e. neither nominative nor accusative) cases, and there is no reason in this framework to label them as anything else than obl, or obl:arg if one wants to keep the complement-adjunct distinction. Further, they never share the behaviour of “direct objects”, especially with regard to passivisation. The class of neutrally-aligned ditransitive verbs is residual in Latin, and has all but disappeared in the contemporary Romance languages.

A better terminological choice for this relation would be “second object” (sobj), or to simply dispense with such a distinction and use obj for both arguments in the constructions where two arguments are effectively both expressed as prototypical objects are.

References

iobj in other languages: [bej] [bg] [cop] [cs] [de] [el] [en] [es] [eu] [fr] [fro] [gsw] [gub] [hy] [it] [ja] [ka] [kk] [la] [no] [pcm] [pt] [qpm] [ro] [ru] [sl] [ssp] [sv] [swl] [tr] [u] [vi] [xcl] [yue] [zh]