Latin (Indo‑European, Italic)

Some verbs have passive morphology, but are active by syntactic and semantic criteria.

1 Background

1.1 Syntax

The morphological passive has a number of different functions. What we can call the true passive is formed from transitives, resulting in object promotion:

The agent, if expressed, is marked by the preposition a(b) 'by'.

There is also an impersonal passive construction (with a dummy 3rd person subject), which can be formed by intransitive verbs as well as transitive:

Alongside the true passive function, passive morphology may also represent a middle voice (Baldi 1976; Flobert 1975: 37 characterizes this as 'extrinsic' versus 'intrinsic' passive), i.e. some kind of reflexive:

Note that this middle function of the passive is lexically restricted.

As a kind of valency-reducing operations, passivization typically yields an intransitive verb. However, there are some ditransitive verbs that take two accusative objects, and their passives are transitive:

1.2 Morphology

The passive is morphologically heterogeneous. The range of variation can be seen by looking at the present, present infinitive and perfect forms (in this case, of a 1st conjugation verb):

Note that the perfect passive forms are periphrastic; the lexical stem is treated as an adjective, and declined for gender and number of the subject.

2 Deponency in Latin

In the class of verbs known as deponents, the passive forms have the value of actives, while the corresponding active forms are lacking.

2.1 Morphology of deponents

Though Latin deponents are often characterized as having the form of passives, this is an oversimplification. In the table below, the (abbreviated) paradigm of a normal transitive verb is compared to that of a deponent; in both cases, only 3rd person singular or masculine (nominative) singular forms are given. Note the division into A, B, C, D, discussed below.

In sections A and B of the paradigm, the deponent verb has only passive forms, with active value. Note that in section B, the result is a past active participle, a value not found among normal transitive verbs.¹ In section C, the normal transitive paradigm has only active forms. The deponent verb likewise has the active forms. In section D (the gerundive), it's not clear whether the form should be descried as functionally active or passive, because of the peculiar syntax associated with it.²

2.2 Deponents as a form ~ function mismatch

Since there is no obvious single correlate of 'having the value of an active', it is worth reviewing in what ways the deponent verbs pattern with actives rather than passives.

Syntactically, deponents may be transitive, like active verbs, while passives are intransitive (with the exception noted above):

Semantically, in the phenomenon that we can term 'voice attraction', the auxiliary verbs coepi 'begin' and desino 'cease' match the voice of the main verb (or at least tended to do so for a certain period in the history of Latin). ³ However, the correspondence is not based directly on morphological voice: only true passives and impersonal passives (Kühner 1955: 677) induce passive morphology on the auxiliary:

Morphologically, the shape of the deponent paradigm can be taken as evidence against the argument that passives and deponents both realize the same underlying value -- i.e. that whatever value is expressed by the inflectional passive (be it passive or reflexive) is lexically inherent in deponents (Wackernagel 1926). First, observe that a deponent verb makes use of both passive and active forms, without this having any effect on the verbs argument structure. The value of the inflectional passive, on the other hand, is necessarily tied to its morphology: any alternation between active and passive morphology within the paradigm is invariably correlated with some alternation in argument structure. Second, one might claim that deponents do display an active ~ passive alternation, provided one regards the gerundive as passive, which is a disputed question (see footnote 2). If the gerundive is construed as passive, then one would need to recognize two distinct motivations for passive morphology, a true inflectional passive (the gerundive) and an inherent, lexical middle (elsewhere).

There is one part of the paradigm where the behaviour of deponent verbs is unique, namely the perfect participle. In non-deponent deponent verbs, the perfect passive participle lacks an active counterpart. One might then expect, given that the past passive participle form of deponent verbs should be construed as active, that this would yield a morphosyntactic value (perfect active participle) not otherwise present in the verbal system. This is true, but not without a qualification: through the Classical period, they could not govern an accusative object. For example, the deponent verb confiteor 'confess' may normally take an accusative object, but as a perfect participle only appears without one:

In later periods , accusative objects were permitted, but remained rare:

It should be stressed that there does not appear to have been any restriction on the use of the perfect participle of deponent verbs, be they (potentially) transitive or not, only on their participation in an overtly transitive construction.

3 Diachrony

The origin of the Latin deponent class goes back to Proto-Indo-European, so it is worth reviewing the relationship of the two systems. PIE had two-voice system, active and middle. The Latin voice system shows two important innovations. First, the original middle has largely become specialized as a passive. Second, the perfect forms in PIE had no corresponding middle; these forms were innovated within Italic, using what was orignally a verbal adjective.

The core of the deponent class presumably consisted of Indo-European media tantum verbs, i.e. verbs which had only middle voice forms; parallels can be found in other Indo-European languages for such verbs as fungor 'busy oneself with something', morior 'die', orior 'rise', sequor 'follow', loquor 'speak' and medeor 'cure' (Flobert 1975: 529). It is supposed that the defectiveness of these verbs was due to their having inherently middle semantics. The inherited deponents are all intransitive, transitive deponents being an innovation within Latin; note though that none of these are causatives (Flobert 1975: 529). Some observers have argued that deponents still represent a semantically motivated within Classical Latin (e.g. Wackernagel 1926, Kemmer 1993), but close inspection of the lexicon shows that the relationship is only approximate. For example Beck (2002), looks at deponents in the work of Plautus, and concludes that they describe situations in which the subject is affected, but also observes that deponent morphology is not a necessary correlate of this semantic characteristic (p. 226), and characterises it as 'functionally obsolete' (p. 277). Flobert (1975: 520) makes the important distinction between primary (inherited) deponents, such as those seen above, and derived (denominal or deadjectival) deponents, such as dominor 'rule' (< dominus 'master'), laetor 'rejoice (< laetus 'joyful'), piscor 'fish' (< piscis 'fish') or precor 'beg' (< prex 'prayer'). While primary deponents are semantically diffuse, the latter display, according to Flobert, a significant semantic and functional overlap with middle (in Flobert's terms 'intrinsic') passives (p. 398). Note that both types are productive in Latin: primary deponents are extended through prefixation, and denominal derivation of deponent verbs remains a regular process.

Flobert (1975: 588-599) reports 270 deponents in the archaic period, with a total of 884 attested across the corpus of Latin up till the 8th century, and notes that, on the average, deponents constitute 7% of the verbs in a given text. During the history of Latin there was both migration into and out of the deponent class. By the 8th century, though, deponent verbs typically had active variants in the present stem forms, and by the end of the century the synthetic passive had been eliminated, leaving traces of deponents only in the perfect.

Within the recorded history of Latin, there is evidence that passives and deponents developed morphologically in parallel, in spite of their functional difference. In Old Latin, where the distinction between passive and deponent was already established, the future imperative was indifferent to voice, the ending -tō(d) being used for both. Subsequently, the ending -tōr was innovated for both passives and deponents (Hofmann and Szantyr 1965: 289).⁴

Notes

¹ There is however a small number of active verbs whose past particple has an active resultative meaning, e.g. pōtus 'having drunk' (not 'having been drunk'), from pōtāre 'drink'.

² The gerundive is an adjective formed from transitive verbs, and matches the object in agreement features and case. As a component of the so-called passive periphrastic construction (expressing necessity), it seems fair to say that it is passive:

Outside of the periphrastic construction the situation is less clear. Though treated as an adjective modifying the object, one is not inclined to give it a passive translation:

More significantly, a gerundive construction is typically interchangeable with a gerund (syntactically a noun), where the gerund (treated as a masculine/neuter singular noun) is clearly active, in as much as it governs an accusative object:

The significance of this is that the gerund and gerundive are formally identical (the gerund = masculine/neuter singular non-nominative forms of the gerundive), and, at least diachronically, are related to each other, though which derives from which remains disputed. At least some authors (e.g. Miller 2000) argue that the gerundive should be treated as an active form.

³ Voice attraction was an innovation of the Classical period. Only one example turns up in Plautus, but then the phenomenon spread quickly, and is found regularly in the works of Cicero and Caesar, only to vanish shortly after -- e.g. it is not found in the works of Tacitus (late 1st-early 2nd century) (Hofmann and Szantyr 1965: 288).

⁴ That this need not necessarily have been the case is shown by Old Irish, where passive and deponent verbs are clearly related to each other (as well as to the Latin synthetic forms), but have diverged morphologically (so that, in fact, deponents can form passives: compare deponent active airigidir 'perceives' with the passive airigthir 'is perceived' (Pokorny 1923: 82, 84).

References

Allen, Joseph Henry and James Bradstreet Greenough. 1903. Allen and Greenough's new Latin gammar for schools and colleges (ed. by J. B. Greenough, G. L. Kittredge, A. A. Howard and B. L. D'Ooge). Boston and London: Ginn and Co. [available online at http://www.perseus.tufts.edu/]

Baldi, Philip. 1976. Remarks on the Latin r-form verbs. Zeitschrift für vergleichende Sprachforschung 90. 222-257

Beck, Ursula. 2002. La linguistique historique et son ouverture vers la typologie. Paris: L'Harmattan.

Bennet, Charles E. 1942. New Latin Grammar. Boson: Allyn and Bacon.

Flobert, Pierre.1975. Les verbes déponents latins des origines à Charlemagne. Paris: Société d’Édition “Les Belles Lettres”.

Hofmann, Johann Baptist and Anton Szantyr. 1965. Lateinische Syntax und Stilistik (Handbuch der Altertumswissenschaft 2, 2, 2). Munich: C. H. Beck.

Kemmer, Suzanne. 1993. The middle voice. Amsterdam/Philadelphia: John Benjamins.

Kühner, Raphael. 1955. Ausführliche Grammatik der lateinischen Sprache (part 2, 1, revised by Andreas Thierfelder). Hamburg: Hahn.

Miller, D. Gary. 2000. Gerund and gerundive in Latin. Diachronica 17/2. 293–349.

Wackernagel, Jakob. 1926. Vorlesungen über Syntax, mit besonderer Berücksichtigung von Griechisch, Lateinisch und Deutsch (vol. 2). Basel: Birkhäuser.