Hardcover
-
PICK UP IN STORECheck Availability at Nearby Stores
Available within 2 business hours
Related collections and offers
Overview
Product Details
| ISBN-13: | 9780521847896 |
|---|---|
| Publisher: | Cambridge University Press |
| Publication date: | 08/25/2005 |
| Series: | Cambridge Classical Studies |
| Pages: | 298 |
| Product dimensions: | 5.83(w) x 8.86(h) x 0.94(d) |
About the Author
Read an Excerpt
Cambridge University Press
0521847893 - Expressions of Agency in Ancient Greek - by Coulter H. George
Excerpt
1
PASSIVE VERBS AND AGENT CONSTRUCTIONS
Beginning students of Ancient Greek soon learn that the agent of a passive verb is marked with the preposition ὑπό followed by the genitive.1 Then, of course, the exceptions come to light. The most common of these is the dative of agent, which, for the beginner at least, may be explained away as occurring with perfect passives and -τέος verbals. Later, however, one comes across other irregularities, notably the use of prepositions other than ὑπό+G. The conditions that motivate these apparently anomalous agent markers have not yet been satisfactorily explained. The aim of this book is to do so.
I begin with an introductory chapter that lays a theoretical foundation for the work and discusses the reasons why these passive-with-agent constructions (PACs) occur in the first place. In Chapter 2, I move on to Homer, as the Iliad and Odyssey represent the earliest texts that are syntactically complex enough to have PACs. Because the Homeric data are so different from the later evidence - in particular, these constructions are far less common in Homer - they are best dealt with separately. Next, in Chapter 3, I look at the dative of agent. It is relatively well understood already - it occurs most prominently with perfect verbs - but one question in particular deserves further treatment: When do perfect verbs take ὑπό+G rather than the dative one might otherwise expect? Then, Chapter 4 treats the central issue I shall examine, namely, the conditions motivating the use of prepositions other than ὑπό+G as agent markers in the classical prose authors. Because these conditions were different for poetry than for prose, I reserve discussion of tragedy and comedy for Chapter 5. Finally, in Chapter 6, I trace the development of agent constructions in post-classical Greek, ending with the Byzantine Greek of Digenis Akritis.
Passive voice
If this work is to explain the conditions that influence what agent marker is used in a PAC, it must first set out guidelines for determining what constitutes such a construction. Essentially, such guidelines must be formulated so as to identify the two components of a PAC, a passive verb and an agent. Now the idea of a passive verb at first glance would seem fairly straightforward to most speakers of English, who will be familiar with the passive voice of their native language. If a verb occurs in the passive voice, then the subject of the verb is not the agent, but rather the patient of the action of the verb. Broadly speaking, such an account is sufficient to describe the voice system of English, and, indeed, it is the opposition between active and passive that has dominated the Anglo-American literature on voice. But, while it will be the focus of this study as well, the situation is more complicated in Greek, where, in addition to the active and passive, there is also a middle voice, the general function of which is to indicate that the effects of the action described by the verb in some way affect the subject of the verb.2 Additionally, it will be crucial to maintain a distinction between the form and function of a voice. For, as the Greek passive was only ever partially independent of the middle, the two could quite easily be confused: a middle form can have a passive function and vice versa.3 Such contamination is not surprising in light of the overlap between the functions of the middle and passive.
Now the voice opposition most prominent in theoretical linguistics is the contrast of active and passive voice, as proponents of transformational grammar have taken it up as an example of a transformation exhibiting noun-phrase movement in the shift from deep to surface structure.4 This approach is sensible, because it is, in most cases, easy to trace the relation between a passive sentence and its active "prototype." In essence, the passive voice rearranges the relations of the verb to its core nominals (that is, the subject and object), in particular indicating a reduction in the verb's valence (that is, a transitive verb becomes intransitive). Generally, this process involves the demotion of the subject of the unmarked construction (usually the agent) to an oblique relation in the marked construction, typically accompanied by the promotion of the object (usually the patient) to the subject relation.5 Diagrammatically, with brackets indicating optional elements:
| Subj | Vb-Act | Obj | Subj | Vb-Pass | [Obliq] | ||
| | | | | → | | | | | |||
| Agt | Pat | Pat | [Agt] |
The English passive fits in well with this schema:
Achilles killed Hector → Hector was killed [by Achilles]
The primary functions of the passive follow naturally from the syntactic remappings it occasions. First, the passive can be used when it is advantageous for the patient of the action to be the grammatical subject, either pragmatically, because it is a narrative theme, or syntactically, so that it may serve as a pivot. As an example of the first, Palmer offers, "The child ran into the road. He was hit by the car" (1994: 136). The second is illustrated by the frequent use of passive participles in many different Indo-European (IE) languages. As Jamison notes, "A passive participle is of far more use in speech [than a finite passive verb], for it provides a more concise and elegant means than a relative clause of embedding into a matrix clause any clause whose object would be coreferential with a noun phrase in the matrix clause" (1979b: 203, italics hers). Secondly, a speaker can use the passive to avoid naming the agent, because it is so obvious as to be unnecessary, or because it is unknown, or even to obfuscate the responsibility for an action. Not all languages, however, are able to express the agent of a passive verb. Latvian provides the textbook IE example of a passive that cannot construe with an agent.6 Still, Greek can express the agent, and it is precisely the variety of its agent constructions that makes it so interesting.
The middle voice, on the other hand, best known to linguists from its occurrence in Greek and Sanskrit, cannot be reduced to a similarly neat syntactic description.7 Unlike the passive, which has a relatively clear syntactic function, the middle can only be defined in vague, semantic terms as indicating that the effects of the action in some way accrue back to the subject.8 Consider the two sentences: ὁ Ἀχιλλεὺς φέρει τὸ δέπας and ὁ Ἀχιλλεὺς φέρεται τὸ δέπας. The difference between the active and middle sentences does not lie in so discrete a factor as the valence of the verb, for, in both sentences, the verb predicates two arguments, an agent-subject and a patient-object. Rather, the use of the middle indicates that the action affects the subject to a greater extent. Achilles does not simply carry the goblet, but rather has an additional interest in the object: he wins it. This definition of voice accounts for all the functions of the IE middle described by Wackernagel, including direct and indirect reflexives, reciprocals, and verbs of taking (the type μισ θόω "let out for hire": μισ θόομαι "hire") (1950: 124-9).9 An additional difference between the middle and the passive is that a voice like the IE middle does not seem to be subsidiary to the active: just as there are verbs that only occur in the active, so too there are verbs only found in the middle.10 Furthermore, passive forms can correspond to a middle as well as an active: αἱρεθῆναι can mean either "to be captured, taken," as a passive to the active, or "to be chosen," as a passive to the middle.11
It might seem tempting, then, to view the passive and middle voices as phenomena of quite different natures. The passive, on the one hand, always has a discrete effect on the syntax of the sentence: detransitivization accompanied by promotion of the object. The middle, on the other, need have no visible syntactic consequences: it merely emphasizes the subject's affectedness. To make a clear distinction between the two, however, would be wrong. A voice defined as vaguely as the middle can in fact be used with precisely the same syntactic effect as the passive.12 Indeed, the passive could be described as an extreme case of subject-affectedness. As proof of the middle's ability to act as a passive, one need only consider the Greek use of the middle outside the aorist and future as the standard passive formation (the type λύομαι ὑπό) or the Russian use of the reflexive as a passive (Novoe zdanie stroitsja inženerami "The new building is being constructed by the engineers").13 Accordingly, I will use the term passive in a functional sense to denote verb detransitivization accompanied by object-promotion, whether this be achieved by a morphologically distinct passive marking or through the use of a middle voice that encompasses other functions as well.
One further debate about the passive has concerned the role of the agent: is it necessary that there be some means of expressing the agent in order for a verb to be called a passive?14 To some extent, this is a trivial question, as it is simply a matter of nomenclature whether or not one defines the passive so as to include instances of detransitivization with object-promotion that do not allow the agent to be expressed. Still, Klaiman does seek to distinguish sharply between the two types of detransitivization. In her view, the passive voice suppresses or downgrades the subject-agent but does not eliminate the logical notion of an agent; accordingly, it leaves scope for the agent to be expressed by grammatical means. In the sentence The tree was felled, an agent (e.g. a lumberjack) causing the falling of the tree is implied, if not explicitly stated, and could be expressed using the preposition by. The detransitivizing middle voice, however, termed an anticausative,15 would have no logical agent assigned, as it "[expresses] spontaneous events, i.e. situations presupposing no participant's control" (Klaiman 1991: 83-4). In the sentence The tree fell, the force that brought about the action is not implied, and could not be introduced into the sentence by means of a grammaticalized preposition like by.16
But while it is valid to distinguish between two types of detransitivization - one that allows for the expression of the agent, one that does not - it is best not to align that difference with the distinction between passive and middle voice. On the one hand, there are voices like the Greek middle and Russian reflexive, best viewed as middles owing to their wide range of uses, that can detransitivize, promote the object, and express the agent with an oblique nominal. On the other hand, there are also voices like the Latvian and Arabic passive, which, as they do little more than detransitivize and promote the object, are best described as passives, but cannot express the agent. Accordingly, one should not consider the inability to express the agent to be particularly characteristic of the middle. It would be better either to call the syntax of detransitivization and object-promotion passive in all cases, noting that some passives allow expression of the agent while others do not, or to call such constructions passive only if the agent can be expressed, and anticausative otherwise. In any event, as this study is concerned with how the agent of a detransitivized verb is expressed, it is of little importance here whether detransitivized verbs that cannot be construed with an oblique agent are to be considered passive.
One final point remains. Most frequently it is the semantic role of agent that is mapped to the subject relation of an active sentence and consequently demoted to the oblique in the passive.17 But there are also other semantic roles that can serve as the subject of a sentence. For example, Achilles, in Achilles saw Hector, may be labeled an experiencer or a perceiver, rather than an agent, as Achilles' seeing Hector does not involve the same level of deliberate participation on Achilles' part as would, say, his striking Hector. In this work, however, the term agent will not refer in this narrow sense exclusively to the participant that is responsible for effecting an action, but rather will denote more generally the noun that would be mapped to the subject relation in a transitive sentence, whatever its more precise semantic role may be. This broader definition captures better the fact that ὑπο+G performs essentially the same function both in ὤφθη ὑπὸ τοῦ Ἀχιλλέως and in ἐπλήγη ὑπὸ τοῦ Ἀχιλλέως.
Origins and development of the passive voice in Greek
An examination of the passive voice and the expression of the agent in ancient Greek inevitably raises questions about the historical development of voice in the Greek verb. Although many problems remain unsolved, it is generally acknowledged that, by the time of classical Attic, Greek had undergone a transition from a two-voice system, with opposition between an active and middle, to a three-voice system, with the addition of a passive. The transition, however, must be regarded as incomplete, for the passive only became independent of the middle in the aorist and future.18 A look at this development should begin with Proto-Indo-European (PIE) itself.
The verbal systems of Greek and Sanskrit suggest that PIE had a two-voice system, with a primary opposition between the active and the middle.19 First of all, there is a clear historical relation between the morphology of the Greek and Sanskrit middle (e.g. present thematic third singular -εται : -ate, third plural -ονται : -ante), while the distinctively passive forms in each are clearly unrelated formations: Greek's -(θ)η- aorist (and future) passive marker on the one hand, Sanskrit's -i third singular aorist passive and -ya- present passive markers on the other. Second, the middle is used similarly in the two languages: both exhibit a reflexive middle, be it direct or indirect (cf. λούομαι (τὰς χεῖρας) and vahate "(direct reflexive) go; (indirect reflexive) marry"), a reciprocal middle (cf. διαλέγομαι and vivadate "dispute with one another"), and a dynamic middle, indicating the total involvement of the subject (πόλεμον ποιεῖν "cause a war to come about" vs. πόλεμον ποιεῖσ θαι "conduct a war," compare tiṣṭhati "stand" vs. tiṣṭhate "hold still").20 Third, some of the same verbs in both languages inflect either only in the active or only in the middle (the activa and media tantum): βαίνω/gacchati, ἐστί/asti on the one hand, ἧσται/āste, κεῖται/śete on the other. Latin too, though traditionally described as having an opposition between active and passive rather than between active and middle, provides some evidence for the contrasts of voice detailed above, for instance the direct reflexive lavari.21
But the lack of a distinct passive morpheme does not imply that PIE could not express the passive, as many languages can use the middle in this function. In both Greek and Sanskrit, the middle was often used to denote passivization.22 That Latin's inherited middle voice (admittedly morphologically different from the Greek and Sanskrit) came to be used primarily as a passive also hints that the PIE middle could assume a passive function. Additionally, several living languages provide evidence that a reflexive construction, similar in function to the PIE middle, can take on the functions of the passive: the Romance languages have a middle that can express an agentless passive (French la porte s'ouvre),23 while the Russian middle can express the passive with the agent.24
It does not appear possible, however, to reconstruct a single unified agent expression for PIE, for the daughter languages show a bewildering variety of constructions, presented most recently by Hettrich.25 This proliferation of agent expressions, it must be noted, does not prove that PIE could not express the agent of a passive verb, let alone that it had no passive. One need only consider the different agent expressions in the Romance languages (Spanish por,26 French par, Italian da) - none of which directly continues Latin ab with the ablative - to find a parallel for the replacement of a single agent expression in the mother language by a variety of constructions in the different daughter languages. Nevertheless, certain patterns do emerge among the attested IE languages: the genitive is frequent with participles, the dative with participles of necessity and perfects, and instrumental and ablatival expressions with finite verbs. Such tendencies can be followed in Greek but must not be pressed too closely. The loss of cases can drastically alter the way a language assigns functions to various prepositions and cases, as seen in the Romance languages. As syncretism reduced the PIE system of eight cases to five in Greek, it is natural to expect that agent expression might have changed as well.
© Cambridge University Press