I. The Passive in English

1.1 Introduction: Voice The passive is a part of a wider concept - the concept of 'voice*. Therefore, one may firstly try to find out what the term 'voice' means. 1.1.1 The Origin of the Term 'Passive' The passive is a part of the voice system in English, which has attracted attention of grammarians and linguists for long. Mustanoja (1960) states that the primitive Indo-European had two voices, active and middle. The passive is a later development. The middle disappeared as the passive gained ground. According to Mustanoja (1960), the ultimate origin of the passive voice is obscure. Bloomfield (1963) goes back to Latin and points out that there are two types of construction - one with an actor-action construction: amat 'he (she, it) loves', and one with a goal-action construction: amatur 'he (she, it) is loved'. Similarly, Hockett (1958) states that two of the inflectional categories that Latin verbs have are 'voice' and 'aspect'. There are two voices, 'active' and 'passive', and there are two aspects, 'imperfective' and 'perfective'. According to Myers (1966), Latin has one set of active inflections called the 'active voice', and an incomplete set of inflections, such as 'vocor', pieced out with verb phrases such as 'vocatus sum', called the 'passive voice'. Myers (1966) rightly points out that English does not have two comparable sets of inflections, but the term 'voice' is still used. In the study of the grammar of English, these terms were initially borrowed fi-omth e study of Latin. 1.1.2 Definitions of Voice Gorrell and Laird (1953) define voice as a distinction in verb forms between the active, in which the subject of the verb is the actor, and the passive, in which the subject of the verb receives the action. Without using terms like 'agent' and 'recipient', Myers (1966) simply states that constructions like 'I call' are said to be 'active', and constructions such as 'I am called' are said to be 'passive'. Marckwardt (1964) states that in contrast to Latin, all the forms of the English passive are periphrastic in form. Todd and Hancock (1986) state that in traditional grammars, are classified according to tense, voice and mood, largely because such distinctions are relevant to Latin verbs. They define 'active voice' as the term used to categorise sentences, and verb phrases where the subject of the sentence is the agent or instigator of the action. Quirk et al. (1972) and Quirk et al. (1985) define voice as a that makes it possible to view the action of a sentence in two ways, without any change in the facts that are reported in the sentence. 1.1.3 The Active-Passive Alternation Very often voice is not defined independently, but defined only through definitions of the active and the passive voices, or simply illustrated by giving examples of both the constructions. Huddleston (1984) illustrates the system of voice with one example each of the two voices, and states fiirther that the unmarked member of the pair is said to be in the active voice, and the marked one in the passive voice. The terms 'active voice' and 'passive voice' are based on the role of the 'subject- referent' in clauses that express an action. Thus, for Huddleston (1984), voice is a general linguistic term that is applied to systems yielding such variation in the semantic roles associated with the subject of the sentence, with a concomitant variation in the form of the VP. He applies the term both to the system and to the VP system. Kroeger (2004) states that the opposition or alternation between active and passive is the most common type of voice system among the languages of the world. According to Shibatani (1988), voice is "a mechanism that selects a grammatically prominent syntactic constituent - subject - fi-om the underlying semantic function ('case' or 'thematic roles') of a clause" (p. 3). Shibatani (1988) fiirther states that in the prototypical passive form, a patient fimctions as a subject. The classical languages like Greek and Sanskrit also have the middle voice, similar to the passive voice, which is a later development in Indo-European languages. According to Mustanoja (1960), while an active clause describes an action fi-om the beginning to the end, i.e., from the initiating force (subject) to the goal (object), in a passive clause the action is traced back fi"omth e end to the beginning. According to Arce-Arenales et al. (1994), the difference between the passive and 'middle diathesis' is that in the latter, the 'agent-subject-topic' also exhibits the status of an affected entity. (See Arce-Arenales et al., 1994 for more details about the distinction between the active, 'middle diathesis' and the passive.) 1.2 The Passive in Relation to the Active in English The passive voice is often defined and described in relation to or in contrast with the active voice. This section is about the relationship between the active and the passive in English. 1.2.1 Various Views on the Active and the Passive Broadly, there seem to be two kinds of approach to the active voice. One is that the active voice is any clause or sentence in which the agent of the action is the subject. Only some of such sentences can be turned into the passive. Berk (1999) is of the view that active sentences are unmarked, and if a sentence is not specifically marked passive, it is technically active. Leech et al. (1982) also take the view that where there is no passive voice auxiliary, the verb phrase is in the active voice. But a majority of the grammarians seem to define the active voice only in relation to the passive. Asher (1994) refers to the active voice in terms of the active- passive alternation in English, For Biber et al. (1999), it is only transitive verbs that are in the active voice, which is usually the case for active verbs. Further, they mention that transitive verbs can also occur in the passive voice. Quirk et al. (1972) describe the two types of voice as two ways of viewing the action of the sentence, without change in the meaning. Quirk et al, (1985), too, hold the same view. Jacobs and Rosenbaum (1968), Ouhalla (1994), Leech and Svartvik (1975) and Morenberg (1997) also write about the passive in relation to the active, with the view that they are svnonvmous. Halliday (1985) has a different view to offer regarding the active voice. For him, a sentence that is not in the passive is not necessarily in the active. A clause with no feature of 'agency' is neither active nor passive, but 'middle'. One with agency is non-middle or 'effective' in voice. An effective clause is then either active or passive. 1.2.2 Semantic Roles of Participants in the Active and the Passive Most grammarians also agree that the active and corresponding passive constructions have usually the same meaning. Kroeger (2004) maintains that the two forms have the same basic meaning and the same participants filling the same semantic roles. Cattell (1969) states that every native speaker of English feels that there is some relationship between active and passive sentences. Even if Jespersen (1933) is of the view that one and the same idea can often be expressed in the two different ways, he also adds that they are not synonymous in every respect, which shows that it is "not superfluous for a language to have both turns and thus be able to shift the point of view. As a rule the person or thing that is the centre of interest at the moment is made the subject of the sentence, and therefore the verb is in some cases put in the active, and in others in the passive" (p. 120). In Radford (1997) and Radford (2004) also we find similar views on identical thematic (i.e. semantic) roles in the two voices. Mihailovic (1967) is of the view that active and passive clauses which do not fulfil this condition do not stand in a transformational relationship. Downing and Locke (2002) distinguish between the two types of voice on the basis of the terms the 'Agent' and the 'Affected'. Leech et al. (1982) mention the view traditionally held that the 'grammatical' subject of the passive is the 'logical' object of the active clause. Schibsbye (1970) also uses the terms 'logical subject' and the 'logical object'. Bloor and Bloor (1995) use the terms the 'performer of the action', the 'beneficiary' and the 'goal of the action'. Gorrell and Laird (1953) point out that usually (i.e. in an active construction) the agent or the actor is mentioned first in a sentence. When the agent or actor is less important than the action or the result, or the actor is unknown or should not be mentioned, or continuity requires that some word other than the actor appear first, the passive construction is used. But there are some grammarians have rejected such a notional definition of the active and the passive. (For example, Huddleston, 1984.) Huddleston (1976) does admit that the subject of the active voice has the role of the 'actor' or the active participant and that of the passive voice has the role of the patient or the passive participant. But he asserts that this difference in the semantic role of the 'surface subject' does not serve to define active and passive sentences. He points out that sentences such as "Everyone knows the answer" and "The answer was known by everyone" do not express the action of an actor on a patient. They are classified as active and passive, not due to semantic roles, but because they show the same syntactic contrast as in the pair "John attacked Peter" and "Peter was attacked by John". 1.2.3 Level of Analysis of Active or Passive Another point worth considering here is the level at which the distinction between the active and the passive is viewed by different grammarians. Jespersen (1933) considers the verb in the active or in the passive. Leech et al. (1982) consider the verb phrase in the active or in the passive, and also consider the passive at the clause level. Huddleston and Pullum (2002) analyse the passive at the clause level. Mihailovi6 (1967) uses the term 'passive' at the level of verbal groups and also at the clause level. Schibsby (1970), Halliday (1985), Berk (1999), Lester (1971) and Jacobs and Rosenbaum (1968) refer to passive sentences as opposed to active ones. Quirk et al. (1972), Quirk et al. (1985) and Huddleston (1984) state that the active-passive relation involves two grammatical levels - the verb phrase and the clause. 1.2.4 The Passive as Rearrangement of Elements Very often, the passive is described in relation to the active construction, in terms of 'rearrangement' of elements that may be called the subject and the object of the active verb or the actor/experiencer/agent of the action and the patient/affected/goal of the action. Huddleston and Pullum (2002) are of the view that active and passive clauses differ in their syntactic form, but they are very similar in meaning. In the active construction, the subject is aligned with an active role, with the role of agent. In the passive construction, the subject is aligned with a passive role, the role of patient. The object of the active appears as the subject of the passive, and the subject of the active appears as the complement of the preposition by. They refer to the by-phiase as 'the internalised complement', as it is an internal complement of the passive VP. The internalised complement is generally optional. Asher (1994) mentions similar differences between the active and passive constructions. Such a formula of active-passive correspondence has been given by many other grammarians. (For example, Quirk et al., 1972 and Quirk et al., 1985.) Kilby (1984) makes the relation between the active and the passive a part of his definition of the passive. If the subject is interpretable in the same way as the object of the corresponding active verb, the construction fulfils one of the two conditions that the passive meets. Morenberg (1997) describes the passive as movement of the core object to the fi-onto f the verb. Radford (1997) and Radford (2004) give four main properties that differentiate passive sentences from their active counterparts. First, passive sentences generally require the auxiliary be. Second, the main verb in passive sentences is in the passive participle (i.e. past participle) form, which is generally homophonous with the participle form. Third, passive sentences may contain a by-phrase in which the complement of by plays the same thematic role as the subject in the corresponding active sentence. Fourth, the expression which serves as the object of an active verb surfaces as the subject in the corresponding passive sentence. Leech et al. (1982) state the passive rule as follows.

SNPI PMV"0NP2 ^ SNP2----PpassMv-- {\p =by + NP,} Kroeger (2004) describes passivisation as primarily a realignment of grammatical relations, involving the 'demotion' of the agent of the action from active subject to passive oblique (what is usually called the Ay-phrase), and also normally involving the promotion of the patient from active object to passive subject. Jespersen (1933) believes that the person or thing that is at the centre of interest at the moment is made the subject of the sentence. Likewise, Curme (1931) is of the view that in normal narrative, the modifier of the verb (i.e. the subject) is usually important, and in sentences with an important modifier of the verb the active is the natural form of statement. On the other hand, the passive is often more appropriate when the verbal activity is prominent. According to Schibsby (1970), if both the agent and the patient are expressed, the passive sentence covers the same reality as the corresponding active one. When the agent is not specified the conception of this logical subject is sometimes so vague that the content of the passive form is close to a corresponding intransitive form of the verb. Thus, "the door was opened" may have roughly the same content as "The door opened." Chomsky (1957) uses the passive as an effective way for showing the limitations of Phrase Structure grammar. He excludes the passive from basic structures (the 'kernel') and claims that it is derived by transformation. Among the traditional grammarians, Jespersen (1924) is one of the first to realize the 'transformational' nature of the relationship between the active and the passive. Chomsky (1957) puts forth the transformational rule in the following way. "If Si is a grammatical sentence of the form NP1-AUX-V-NP2, then the corresponding string of the form NP2 - Aux+be+en - V - by+NPi is also a grammatical sentence. For example, if John - C - admire - sincerity is a sentence, then sincerity - C+ be+ en- admire -by + John (which... becomes "sincerity is admired by John") is also a sentence" (p. 43). This transformational rule is called T-Passive. It is an optional transformation. When applied to the active kernel strings which are generated by the phrase structure rules (PS rules), an active sentence is transformed into a passive one. Chomsky reformulates the rule T-Passive later, because there are some problems regarding some of its implications. One is that the active and the passive have the same deep structure, and therefore the same meaning. Another is that any acceptable active sentence with an object in it can have an equally acceptable passive sentence derived from it. It is evident that most grammarians agree that the semantic roles/thematic roles/arguments (or whatever other labels they may be given) are central to their analysis of the active-passive relation. The only point of difference between the traditional grammarians and post-Chomskians is, as expressed by Kroeger (2004), - how does the ^ammai produce these changes? At the end of this section about the active as in relation wdth the passive, Cooray's view (1967) may be referred to. He mentions the tendency to take the active voice sentence as the basic linguistic pattern where a transitive verb is used, the passive construction being a variation of it meant for occasions when the object of the verb needs emphasis. He points out that, as Jespersen (1933) observes, over 70 % of the passive sentences in English contain no mention of the active subject, the by- adjimct being left out, because it is not called for. Therefore, Cooray (1967) is of the view that passive sentences cannot be treated as the active sentence patterns in reverse order and that it would be more realistic and helpfiil to both students and teachers of English as a second language to treat the passive construction as a linguistic pattern that stands in its own right, independently of the active, and depending for its use on the context.

1.3 The History of the Passive in English Wyld (1914), as quoted in Devi (1991), makes four chronological divisions of English: i. Old English: end of 7* century A.D. to 1050 A.D. ii. Early transition English: 1050-1150 A.D. iii. Middle English: 1150 - 1400 A.D. iv. Modem English: 1400 A.D. - present day Following Devi (1991), the first two taken together will be referred to here as the Old English period. 1.3.1 Old English Traugott (1972) is of the view that in Modem English, the passive can be considered a special kind of 'subjectivalization', one in which the Patient or Goal is subjectivalized and the Agent is not. In Modem English such a construction is automatically marked by the passive auxiliary be + past participle and by the prepositional phrase 'by + Agent'. In Old English, however, passivisation was more complex. It involved subjectivalization of a Patient or Goal if the Patient or Goal received the accusative case marker in a non-passive (active) sentence. Traugott (1972) writes about the interesting history of the passive in the following words, "...it did not always require a be auxiliary in Germanic. Gothic had a passive verbal inflection, not a be auxiliary. In OE a relic of the inflection is to be found in one verb hatte, a passive verb meaning 'is/was called'" (p. 82). Old English had two non-finite forms - i. The infinitive, and ii. The participle There were two kinds of participle. Participle-1 differed from Participle-2 through voice and tense. Participle-1 was present and active, Participle-2 denoted a state resulting from a completed past action - past participle. The latter had an active meaning when formed from intransitive verbs, and had a passive meaning when formed from transitive verbs, e.g. gehoften (fought). According to Quirk and Wrenn (1955), with a single exception. Old English verbs showed only active voice inflexions. The exception, they state, is hatan (call) which, besides having a vocalic active preterite het, had a consonantal preterite hatte, which was passive and which was used both for the present and the past. Later, Old English lost the passive inflection of the verb, so the verb had only one voice, i.e. the active. For the passive, the two auxiliaries, beon (be) or weorSan (become), along with the past participle forms of the transitive verbs, were used to denote a state, resulting from a completed action. Quirk and Wrenn (1955) are of the view that to some extent there was a distinction of aspect involved, beon being used in durative expressions and weorSan in perfective expressions, Traugott (1972) argues that the be form of the passive seems to have originated as a restructuring of adjectival phrases. In Modem English we find sentences like the following, which are regarded as ambiguous. 1. The shop was closed. It has two possible meanings - somebody performed the act of closing the shop, or the shop was already in the state of having been closed. The former is unanimously recognised as passive by all grammarians. There is a difference of

8 opinion about the latter. It is often called 'stative' or 'statal passive'. In Old English the stative form was distinguished by adjectival inflections. Old English did not have progressive or perfective passives. But passives with perfective meaning were expressed by the passive auxiliary with the past tense and those with progressive meaning were expressed by the passive auxiliary with the non- past tense. The perfect with beort (= be) occurred only with intransitive verbs and the passive with beon occurred only with transitive verbs. As a resuh, there was no cause for confusion between the perfect and the passive. Old English also had passive constructions without an auxiliary, but the verbs that were used for the purpose were uninflected. Modem English also has a few instances of this type. For example, 2. This house is to let. 1.3.2 Middle English The changes in the grammar of Middle English occurred due to the reduction of inflections. The first change to take place was the loss of Old English passive with weordan (= become). Mustanoja (1960) finds that in Middle English, the prepositions expressing agency were by, from, mid, of, through, and with and that unambiguous cases with by indicating the agent of a passive verb were rare until the end of the 14* century. Old English had no progressive passive. The use of passival forms of the following type is found fi-omth e beginning of the fourteenth century. 3. The house is on building. Jespersen (1938) points out that the passive construction "the house is being built" is an innovation dating from the end of the eighteenth century. In Old English, only direct objects could be promoted to the subject position. The same is found in Middle English, too. But a new development was that with the loss of inflections, the earlier indirect objects were now treated as direct objects. They could be promoted to the subject position of the passive sentence. Mustanoja (1960) states that a construction like "I was given a book" became a feature of English usage in the 15* century. During the late Middle English period, have + past participle began to be used more fi-equently than Perfective be + past participle. The reason was that the perfective be + past participle was sometimes ambiguous with the passive be + past participle and the resultative stative be + past participle such as the following. 4. The door was closed. Such ambiguity was rare in Old English, because resultative statives were adjectival, as stated above. In early Middle English the perfective auxiliary have + past participle came to be used in with the be + past participle passive auxiliary as in the following. 5. The wronges that han been doon to yow The wrongs that have been done to you "the wrongs that have been done to you" (Traugott, 1972, p. 145) Chomsky introduces the term 'pseudo-passives' for what is usually called prepositional passives. From the following sentence, the passive is created in such a way that the preposition stands out at the end of the syntactical imit. 6. He slept in the bed. 7. The bed was slept in. In Old English, prepositional passives were not used. Mustanoja (I960) observes that the type "he was laughed at began to appear around 1300, but remained rare until the end of the 14^ century. Gaaf (1930) states that in Old English the place of the preposition was not yet fixed. However, in the transition period between Old and Middle English the latitude allowed to the preposition disappeared to a large extent. In Middle English, Old English verbs such as on locian were re-analysed into loken on {look on). On is a constituent of a , rather than a constituent of the following NP. 1.3.3 Modern English An important feature of Modem English is the expansion of verb groups, particularly since the eighteenth century. For instance, in the passive, in place of is taken of the earlier periods in Modem English, we have the following altematives in the present day English, as stated by Jespersen (1909-49b). i) is taken ii) is being taken iii) has been taken iv) gets taken

10 In the beginning, continuous aspect lacked active/passive distinction. Therefore, there could be sentences like the following. 8. The man is building a house. (Active) 9. The house is building. (Passive) But such a sentence could create ambiguity. Therefore, by the end of the eighteenth century, an overt progressive passive construction of the following type emerged. 10. The house was being built. According to Traugott (1972), one of the main developments involving the progressive is the emergence of an overt passive progressive construction. In Elizabethan English, be and have with intransitive verbs were used to form passives. With some intransitive verbs, it is still possible to use be or have. For example, 11. He is gone. 12. He has gone.

1.4 The Passive in English! This section is about various views on the passive in English. While referring to various views on the relation between the active and the passive, definitions of or views on the passive have inevitably been mentioned above. Those which have been mentioned need not be repeated here. 1.4.1 Various Approaches to the Passive The passive in English has been defined in the so-called traditional grammars chiefly in two ways - semantically (in terms of the semantic roles of the subject/object) and syntactically (the auxiliary and main verbs at the VP level, and the change of positions of subject and object at the clause level). Some grammarians define the passive using both the semantic and syntactic aspects. To mention an example of the semantic approach, Gorrell and Laird (1953) describe the passive as a construction that enables one to put some word other than the actor or the agent of the action in the first position of the sentence. (See Leech and Svartvik, 1975 and Downing and Locke, 2002 for similar views.) Gorrell and Laird (1953) also connect the active-passive constructions with the arrangement of the Given and the New information. The Given information is put in the initial position and the New information gets the end-focus. The grammarians of the 'functional

11 grammar' also focus on the semantic aspect of the passive. Halliday (1985) defines the active as one having the Agent as the subject, and the passive as one having the Medium as the subject. According to him, the passive is used for making the Medium the subject, and to make the Agent either 'late news' by putting it at the end, or implicit, by leaving it out. Some grammarians mention both the semantic and syntactic aspects in the definition/description of the passive. Jespersen (1924) makes a distinction between 'notional' and 'syntactic' categories of the active and the passive. In a sentence like "The book sells well" we find an example of the notional passive. "Jill is loved by Jack" is notionally as well as syntactically in the passive. Bloor and Bloor (1995), who follow Halliday's fimctional approach (1985), mention both the semantic and syntactic aspects of the passive. They describe a passive clause as one in which the verb phrase includes some form of the auxiliary be and a past participle. Following the functional approach, they analyse the active and passive sentences in terms of 'Actor', 'Beneficiary' and 'Goal'. Gramley and Patzold (1992) and Asher (1994) are among others who combine the semantic and syntactic aspects in their description of the passive. Christophersen and Sandved (1969) begin to describe the passive by analysing four types of be + past participle construction. 1. He was interested. 2. He is gone. 3. He was caught in a traffic jam. He was bom in 1950. 4. He was rescued by a French mountaineer. The car was stolen a few days ago. In construction 1, there is the verb (not the auxiliary) be, and also a past participle used as an adjectival. Construction 2 differs from 3 and 4 in that it consists of an intransitive verb. Christophersen and Sandved (1969) begin with a formal approach of classifying be + past participle constructions, but they seem to combine it with a semantic approach. The meaning of construction 2 is 'active', because the performer of the action denoted by the verb is also the subject of the sentence. In 3 and 4 the meaning is passive, because the performer of the action is not the subject of the sentence. On the contrary, the subject denotes what they call 'the undergoer' of the action. The difference between 3 and 4 is that the former has no corresponding active

12 form, whereas the latter has; but this will be taken up in the section on the types of the passive. In grammars written after and following Chomsky, too, a combination of semantic and syntactic aspects is often found. Jacobs and Rosenbaum (1968) offer a typical transformationalist view that in the passive transformation there is an interchange of noun phrases. Sinclair (1972) argues that the Prime Mover (of the action) in the Deep Structure cannot be the subject of the passive. There are also some grammarians - traditional and modem - who define the passive only in formal or syntactic terms. Cattell (1969) is of the view that in a sentence such as "The boy fears the teacher", one can hardly say that the teacher is the receiver of any action. The same is true of the passive form "The teacher is feared hy the boy". Cattell (1969) asserts that the difference between the active and the passive is primarily one of form and structure, rather than that of meaning. Huddleston (1976) asserts that the difference in the semantic role of the 'surface subject' does not serve to define active and passive sentences. He classifies active and passive sentences, not on the basis of semantic roles, but with the help of the syntactic contrast between them. Huddleston (1984) also rejects what he calls the notional definition of the active and the passive. Dagut (1985) demands a teaching grammar with focus on the semantic and pragmatic fimctions of aspects of grammar like the passive. He points out that the original transformational generative grammar that developed from Chomsky (1957) was syntactic in nature, and the "'purity' of the original formulation" in Chomsky (1957) has declined to much "semantic contamination" (p. 3). Dagut (1985) holds the view that the transformational grammar, being syntactic, and not semantic or pragmatic, is not a teaching grammar. He adds that whether the passive is derived fi-om the active or not, and whether or not the agentless passive results from agent- deletion rule is irrelevant for a teaching grammar, though it is surely of considerable interest for a theoretical grammar. There are some grammarians who take a purely formal view of the passive construction, without following the generative approach, but they do not reject it overtly. The formal definition of the passive that Quirk et al. (1972) accept, though it is modified in Quirk et al. (1985), is that it contains be + the past participle (the -ed form) in the verb phrase. Basically it comes from Svartvik (1966), and it is followed

13 in some later works on the passive like Femalld (1977), Stein (1979), and Granger (1983). 1.4.2 Views on the Passive - Chomsky and After 1.4.2.1 The Transformational Models As mentioned above, Chomsky (1957) excludes the passive from basic structures (the 'kernel') and claims that it is derived by transformation. His passive rule implies that any meaningful and correct sentence in the kernel with an object could be transformed into the passive. It is also implied that the active and the passive have the same deep structure. However, Chomsky (1965) argues that the active and the passive must be derived from different underlying structures. Chomsky (1965) states that verbs generally take manner adverbials freely,bu t there are some verbs, such as cost, owe, resemble, and weigh, which do not take manner adverbials freely. They also do not imdergo the passive transformation. He concludes that if a verb does not take a manner adverbial freely, it cannot be passivised. However, Lakoff (1970) rejects Chomsky's theory (1965) about the passive transformation and manner adverbials. He points out that the latter's analysis does not account for such verbs as know, believe, consider, think, see, hear scad perceive. These verbs may not take manner adverbials freely, but may vmdergo the passive transformation. He suggests a 'rule featiires' solution to account for excluding some middle verbs like resemble from the operation of T-Passive. According to his rule features solution, the lexical entries of such verbs will be marked with a rule feature [- PASSIVE]. The original transformational model does not satisfactorily account for agentless passives, or what are also called 'short passives' as in the following. 13. The Indian team was defeated. It cannot be claimed that the above sentence is synonymous with 14. The Indian team was defeated by the South African team. It is said in that model that 13 is derived from 15 by 'ellipsis rule'. 15. The Indian team was defeated by someone. The deleted element 'by someone' (or 'by something') is considered pronominal representative of the general category HUMAN (or NON-HUMAN) nouns, and therefore does not violate the 'recoverability condition', which states that only recoverable elements can be deleted. But the deleted phrase "by the South African

14 team" cannot be pronominal representative of either of the general category nouns HUMAN and NON-HUMAN. Freidin (1975) also points out the weakness in the method used for excluding verbs like cost, resemble and weigh from the operation of the passive transformation. But he objects to LakofTs solution (1970) by pointing out that it does not explain v^^hy such verbs are exceptions. 1.4.2.2 The Lexical Theory Transformationalists advocate enrichment of the transformational component and simplification of the base. Another approach is to extend and enrich the base and simplify the transformational component. Those who hold this view are 'lexicalists'. A) Freidin's Approach Freidin (1975) proposes a 'lexical-interpretive approach' to passives. He states that a grammar that captures the active-passive relation must include a rule that accounts for the synonymy of active-passive pairs. He suggests that the synonymy of actives and passives can be specified in terms of the semantic equivalence of predicates and the semantic functions associated with them. Any governs a particular set of semantic functions which is determined according to the conceptual structure of the semantic class of the predicate. He mentions functions such as 'agent', 'theme', 'source' and 'goal'. He states the following rule of semantic interpretation which makes active and passive sentences as synonymous - "Si is synonymous with Sj if (a) the predicates of Si and Sj are semantically equivalent - in which case it follows that the predicates govern the same semantic functions - and (b) each semantic function is filled by the same lexical material for each predicate" (p. 392). B) Wasow's Analysis Wasow (1977), as referred to in Devi (1991), argues that some participial forms are but they differ both distributionally and semantically fi-omth e syntactic passives. 16. The closed window has never been closed. (participial ) 17. ? The window closed has never been closed. (syntactic passive) There is semantic contrast between the two sentences. In the adjectival construction, it is implied that a window can be built closed already. But, as for the latter, a window can't get closed unless someone closes it.

15 Wasow (1977) argues that passive participles which are adjectival because they are prefixed with 'un-' or immediately preceded by 'seem', must be formed by a lexical rule whereas the verbal passives would be transformationally derived. However, later, he revises his theory, and states that structure-preserving transformations should be eliminated in favour of rules internal to the lexicon. C) Bresnan's Lexical Theory: Bresnan (Occasional Paper #7) puts forth a lexical theory of grammar. According to her, rules of grammar defined in the formal system of grammatical representation provide each sentence of a language with dual representations consisting of a constituent structure ('c-structure') and a functional structure ('f- structure'). The c-structure represents the constituency of the sentence which is phonologically interpreted. The f-structure is the representation of its meaningful grammatical relations, and it is semantically interpreted. Relating the lexical- functional structiire to the different grammatical c-structures through the mediation of logical argument structiire is the foundation of the lexical theory. Grammatical relations are lexically encoded by assigning grammatical functions to the predicate argument structures of lexical items. The predicate argimient structure positions may be identified with semantic roles such as agent, theme and instrument. The grammatical function assignment then associates these semantic roles with grammatical functions (subject, object, etc.). Bresnan (Occasional Paper #7) adds that the semantic roles of the grammatical fimctions associated with a predicate may be changed by lexical rules, such as the passive. According to her, passivisation can be universally characterised as follows. (SUBJ) L> 0 / (OBL) (OBJ) |-> (SUBJ) The fimctions SUBJ, OBJ, OBL belong to the inventory of imiversal functions that includes SUBJ(ect), OBJ(ect), OBL(ique), COMP(lement), etc. The passive changes a transitive lexical form whose subject is agent and whose object is theme to a grammatically intransitive lexical form (i.e. one which lacks an OBJ function). (a)L (SUBJ), (OBJ) -• AGENT THEME

16 (b)L ((OBL)/0, (SUBJ)) AGENT THEME In the passivised lexical form (b) above, the subject function is now correlated with the THEME role of L, and the optional oblique function is correlated with the AGENT role. Applying this universal passive rule to English, Bresnan (Occasional Paper #7) formulates the passive in English as follows. Functional change: (SUBJ) —•0/(BY OBJ) (OBJ) ->• (SUBJ) Morphological change: V —• V[part] The morphological effect of passivisation in English is to alter the base verb to a participle. In the lexical theory, most of the transformational rules are eliminated. Only a few local rules like agreement rules are retained in the grammar. The base component contains a set of P.S. rules and a greatly enriched lexicon. The P.S. rules directly generate all base sentences of the language. The deep and surface structures are identical. 1.4.2.3 The Government-Binding Theory The transformational models are developed into a new model of grammar, which is called the Government-Binding theory of grammar. In this theory, rules are explained as interaction of principles and lexical properties rather than existing in their own right. The passive is also treated as the interaction of various principles and sub-theories like X-bar theory, projection principle, theta theory, case theory and Binding theory rather than as a single rule of transformational approach. Chomsky (1982a), as referred to in Devi (1991), puts forth a general principle of movement called 'Move- a', which is defined as 'Move any category anywhere'. The transformational approach to the passive includes pre-verbal NP-postposing and post-verbal NP-preposing. In GB analysis, the passive sentence has a d-structure in which the NP occurs after the verb and is then moved into the subject position. GB has two levels of syntactic representation, d-structure before movement and s-structure after movement. It links soirnds (i.e. phonetic form) and meanings (i.e. logical form) in the following way.

17 d-structure t movement

PF component LF component (PF - Phonetic Form; LF - Logical Form) LF represents syntactic meaning. "By the phrase 'logical form' I mean that partial representation of meaning that is represented by grammatical structure" (Chomsky, 1979, p. 165, as quoted in Devi, 1991, p. 78). Chomsky (1981) proposes a well-formedness condition, which he calls the 'theta-criterion'. This condition states that any position filled by an argument must be assigned exactly one semantic role, and semantic roles can only be assigned to positions that are filled by arguments. Since the subject position does not contain an argument, no thematic role is assigned to it. This makes it possible for the object NP to move there. Movement of the object is made obligatory by the case filter, which requires all phonologically realized NPs to have abstract case. Passive participles are assumed not to assign case, and so the object must move to the subject position. General conditions on movement ensure that the subject position is the only available position. In this way, the central properties of the passive are derived from the interaction of general conditions. Chomsky's analysis (1981) regards the semantic role assigned to the subject as suspended in passives. However, Asher (1994) argues that there is evidence that the semantic role of the subject is present in the passive. Certain modifiers that are sensitive to the presence of a subject argument are able to appear in passives, as in the following. 18. The students were purposefully arrested. 19. The students were arrested to avoid a fuss. Evidence like this, he argues, shows that the subject role is assigned to an 'implicit argument'. Haegeman (1991) summarises the properties of passivisation according to GB theory in the following way. i) The verb is affected.

18 ii) The external theta role of the verb is absorbed. iii) The structural case of the verb is absorbed. iv) The NP which is assigned the internal theta role of the passive verb moves to a position where it can be assigned case. v) The movement of the NP is obligatory in view of the case filter. vi) The movement of the NP is allowed because the subject position is empty.

1.5 The Types of the Passive There is a lot of disagreement about the passive regarding what can be called a passive, which usually boils down to a disagreement as to what are acceptable as the types of the passive. The disagreement is also about how the types of the passive are accounted for. Therefore, various approaches to the types of the passive are dealt with in this section. 1.5.1 The Passive Meaning 1.5.1.1 Notional Passive Jespersen (1924) makes a distinction between 'notional' and 'syntactic' categories of the active and the passive. In a sentence like "The book sells well" we find an example of the notional passive. "Jill is loved by Jack" is notionally as well as syntactically in the passive. 'Notional passive' has its verb in the active voice, but it is sometimes regarded as a kind of passive, as it is passive in sense. Similarly, Jespersen (1909-49a) gives examples of 'activo-passive' use of some verbs - act, compare, cook, cut, draw, drink, eat, feel, tell, etc. Scheurweghs (1959) also mentions that in certain set phrases, the active forms are found with a passive meaning, as in the following. 20. A new adaptation oiLittle Dorrit is now playing. The passive construction of the above would be as follows. 21. A new adaptation oi Little Dorrit is now being plaved. Likewise, Berk (1999) also mentions constructions that are 'active in form and passive in meaning'. 22. Cotton >va5/ie5 well. Berk (1999) points out that in such sentences the subject is the patient (i.e. affected), not the agent. These sentences also do not have the verb morphology required for the passive. He mentions that these constructions are sometimes called 'pseudo-passives'.

19 But he also reminds us that a true passive must be passive in form. Gramley and Pfitzold (1992) also discuss active constructions with passive meaning. 1.5.1.2 Ergatives and Middles Asher (1994) states that English has two types of construction that are similar to passives in that a direct object of the deep structure appears as a subject in the surface structure. An interesting feature of both the constructions is that they are not signalled by any morphological mark on the verb. They are 'ergatives' and 'middles'. The following is a pair of an active and an ergative respectively. 23. John cooked the goose. 24. The goose cooked. The following pair shows the contrast between the active and the middle. 25. John z^amrec? the floor. 26. The floor/?am^ei/easily. The object of the corresponding active construction becomes the subject in both ergatives and middles, and the subject of the active disappears. But ergatives and middles are much more restricted than what Asher (1994) calls the 'core passive'. 1.5.1.3 Lexical Passive According to Palmer (1965a) and Palmer (1965b), there are several types of active sentences that are both semantically and syntactically like the passive. Under the term 'lexical passive', he includes two categories, one of which is like what has been referred to as 'notional passive'. The other category that Palmer (1965a) and Palmer (1965b) include in the lexical passive is that of 'adverbial passives'. These are with verbs that can be used in a passive sense, as in the following. 27. These shirts wash well. Such uses of the active in the passive sense are adverbial in that they usually occur with . 1.5.1.4 Unaccusatives Ouhalla (1999) gives the following examples. 28. John broke the vase. 29. The vase broke. He argues that there is a class of verbs in English which enter into a ' alternation' of the type illustrated in the two sentences quoted above. The verb break in these sentences has two 'arguments' (i.e. semantic roles) - an agent realised as the

20 subject and a theme role (i.e. the affected) realised as the object. In the second example, however, the verb break has only the affected, in this case realised as the subject. Ouhalla (1999) writes that in the first instance the verb is transitive, and in the second one, 'unaccusative'. (See Ouhalla, 1999 for further details about the differences between imaccusatives and middles.) 1.5.2 Statal/Actional Passives The term 'statal passive' refers to a construction like the following, in which a state is described, not an action. 30. My work is done. This construction does not have any direct active counterpart, such as the following. 31. I i/o the work. The term 'actional passive' refers to a construction that refers to an action, not a state, and that has a direct active counterpart. The following is an example. 32. The work has been done. For a long time, there have been debates about the statal and actional passives in English, and whether constructions without an active counterpart can be called passive or not. Jespersen (1909-49b) refers to the terms 'static' or 'statal passive' and 'kinetic' or 'actional passive' used by Curme (1931). Huddleston (1984) and Mustanoja (1960) use the terms 'statal' and 'actional' passives. Similarly, Huddleston and PuUum (2002) make a distinction between 'stative' and 'dynamic' passives, and relate the two to the terms 'verbal' and 'adjectival' passives. Gramley and PStzold (1992) use the terms 'statal' or 'stative' and 'dynamic' passives. They state that passives most fi-equently appear with the auxiliary verb be, but with be, the constructions are often ambiguous, as they can denote the occurrence of an act or the result of such an occurrence. Passives with have or get as auxiliary verbs are not ambiguous. 1.5.2.1 Passive of Being and Passive of Becoming Jespersen (1909-49b) explains that in sentences like "His bills are paid", sometimes the present result of a past action implied in the participle is predominant (i.e. "he owes nothing now, he has paid the bills") and sometimes the element of present represented by are is predominant (i.e. "he pays his bills regularly"). In the former, the 'state' of the bills having already been paid is highlighted; in the latter, the 'action' of paying the bills is presented in the passive. Jespersen (1909-49b) calls the former 'the passive of being' and the latter 'the passive of becoming'.

21 1.5.2.2 Statal Passive and Classification of Verbs Feraalld (1977) states that the EngUsh passive can be divided into two major types. One is a fairly clear-cut category often classified as actional passive. It consists of verbs which admit a direct active transformation and often have an agent or potential agent. The other is a much less clearly defined category; it has various subdivisions, the most common of which is 'statal passive'. Femalld (1977) refers to the views of Palmer (1965b) and Huddleston (1971). Palmer (1965b) analyses the statal passive as a form of be (having its fiill verb value) plus a past participial adjective. Femalld (1977) refers to Huddleston's explanation (1971) of statal passives. According to Huddleston (1971), non-dynamic verbs, such as know are unequivocally statal; a small group of dynamic verbs, such as carry never lend themselves to a statal interpretation; and the rest, a large majority of dynamic verbs, such as finish in "It was decided also to have this president elected immediately even before the rest of the constitution was finished", are often ambiguous. 1.5.3 Passives with no Active Counterpart or 'Pseudo-passives' The term 'pseudo-passive' has been used by different grammarians with different meanings. Here, it is referred to as a category having no active counterpart. Sinha (1973), as referred to in Femalld (1977), argues that there are some subcategories within statal passives, the most important of which are locational verbs (e.g. situate) and ceremonial verbs (e.g. establish). In their statal sense, these verbs often do not have an active counterpart. (See Sinha, 1973 for fiirther details.) There are some models of analysis of be + past participle construction in English, and different models show different types of relationship between the 'true passives' and 'pseudo-passives'. A) Mihailovic's Classification Mihailovic (1967) divides be + past participle constructions into passives and pseudo-passives. Mihailovic (1967) states that while clauses with passive verbal groups are in contrast to clauses with active verbal groups, clauses with pseudo- passive verbal groups are not in such a contrast. She calls 'pseudo-passive' all those 'BEVEN' groups (i.e. be + past participle) which do not stand in opposition to active verbal groups. To sum up her analysis of pseudo-passives, she lists the possible relationships between the active constmctions and the be + past participle constmctions in the following manner.

22 1. The transitive active construction is in a transformational relationship with the passive construction. 2. The intransitive active construction is in free variation with the pseudo-passive construction. The following examples illustrate this. 33. We went to before we married. 34. We went to London before we were married. 3. The intransitive active construction has no pseudo-passive equivalent. 4. The pseudo-passive construction has no active equivalent, as in the following example. 35. We are not supposed to smoke here. B) Christophersen and Sandved's Classification According to Christophersen and Sandved (1969), the be + past participle construction falls into various sub-groups illustrated by the following. 1. He was interested. 2. He is gone. 3. He was caught in a traffic jam. 4. He was rescued by a French mountaineer. The car was stolen a few days ago. According to Christophersen and Sandved (1969), the construction of Type 1 has the past participle used as an adjectival. Construction 2 has an intransitive verb. Of the four types of be + past participle construction, Christophersen and Sandved (1969) recognize only 3 and 4 as passives. They further write that constructions of Type 4 have a definite relationship with active sentences. Sentences of Type 3 do not have any active counterpart. They add that sometimes the construction in 3 is called a 'pseudo-passive' and the one in 4 is a genuine example of the passive voice. 1.5.4 Adjectival Passives As has been noted above, Christophersen and Sandved (1969) do not accept the construction of the type "He was interested" as passive. They argue that there is a superficial similarity between Type 1 and other types. In Type I, be is not an auxiliary verb, and the past participle is used as an adjectival. This is an unequivocal position. But some others' views are varied in this regard. Kroeger (2004) makes a distinction between the verbal and adjectival uses of the passive participle. He holds the view that adjectival passives are derived from verbal passives. Asher (1994) considers verbal passives, which have active transform, as 'core passives', and presents

23 adjectival passives as one of the 'related constructions'. The following are some views on adjectival passives in detail. 1.5.4.1 Semi-passives Gramley and Patzold (1992) treat the adjectival construction as 'the semi- passive'. They argue that get, become, grow, feel, seem and some other verbs are less of passives than '' followed by a participial adjective. They claim that this can be confirmed by the fact that an intensifier such as very, awfully or extremely may precede the participle, as in "I got (very) tired". They conclude the point with the following remark. "They resemble the passive, but are not usually regarded as true passives" (p. 122). (Emphasis mine.) This position taken by Gramley and Patzold (1992) shows the ambiguity involved in this approach to adjectival constructions. 1.5.4.2 Formal and Non-formal Passives Kulkami (1981) makes a distinction between 'formal' and 'non-formal' passives. Formal passives are those which have an active counterpart and have been transformationally derived from the actives. He gives the following examples of non- formal passives. 36. Bill got hurt. 37. Bill remained hurt. The above description points to what is usually called the adjectival passive. A further complication in the matter is that he also includes in 'non-formal passives' those sentences that are usually termed as 'notional passive' or said to be passive in sense. The following is an example. 38. The door opened. 1.5.4.3 Central Passives, Verbal and Adjectival Passives Huddleston (1984) recognizes only those passives that contain the agent phrase as the most central kind of passive. He adds that there are various other passive constructions which cannot be derived from an active counterpart. They bear a less direct or systematic relation to actives. Though this is true, it is remarkable that Huddleston (1984) recognizes as the most 'central' kind of passive that construction which is not very frequent, compared to agentless and other passive constructions, as indicated by Jespersen (1933), Quirk et al. (1972), and empirically proved by Svartvik (1966). However, he incorporates among the passives all those constructions containing a passive use of a past participle. As can be seen, a drawback of this

24 approach is the tautology involved in it - the passive is a construction that consists of a passive use of a past participle. Among the types of the passive that he identifies, adjectival passives are also incorporated. According to him, the former of the following is a verbal passive, while the latter an adjectival one. 39. The vase was broken by Tim. 40. The vase was already broken. The following is ambiguous between the two. 41. The vase was broken. There is ambiguity in Huddleston's analysis (1984) because of the use of two different terms for describing the same phenomenon - 'adjectival' and 'statal'. (See Huddleston, 1984 for ftirther details.) 1.5.4.4 Adjectives and the Past Participle Forms of Verbs Huddleston and PuUum (2002) draw attention to the large-scale overlap between adjectives and the past participle forms of verbs, and observe that as the verb be can take complements represented by either of these categories there is a significant resemblance between them. This leads to an ambiguity between a verbal passive and a complex-intransitive clause containing an adjectival passive as predicative complement. Huddleston and Pullum (2002) add to the complexity of the problem by making a distinction between a passive clause and a clause containing a passive. According to them, an adjectival passive, having a statal interpretation, is a passive only in a derivative sense. The clause that contains an adjectival passive is a complex-intransitive construction. Thus, according to them, the clause "They were very worried" is not itself a passive, but it merely contains a passive. They assert that passives in the strict sense are always verbal. Huddleston and Pullum (2002) also give grammatical tests to distinguish adjectives from verbs. (See Huddleston and Pullum, 2002 for further details.) Ouhalla (1999) gives the following as an example of the adjectival passive. 42. The island was uninhabited. It is passive in the sense that its internal argument (i.e. object) surfaces as the subject of the sentence, on a par with verbal passives; and it is adjectival because it shows properties that are usually associated with adjectives. To sum up the above discussion on statal and adjectival passives, sometimes 'statal' refers to a state rather than an action, and sometimes it is equated with adjectival passives as opposed to verbal ones. Adjectival passives are statal in nature.

25 according to Huddleston and PuUnm (2002); but statal passives are not necessarily only adjectival. The terms 'statal passive', 'adjectival passive', 'pseudo-passive', 'notional passive' and 'non-formal passive' are all mixed, overlapping and also create a lot of confusion. 1.5.5 Unpassives Siegel (1973) analyses unpassives separately. Svartvik (1966) makes a general category of passives with compound verbs having past participles, but not infinitive forms. This category includes impassives, as will be explained in the section on Svartvik (1966) later. Kroeger (2004) makes a distinction between the derivation of passives with verbs that have corresponding active verbs with un- and that of passives with verbs that have no such corresponding active verbs. The following example of the passive has no active counterpart, as there is no corresponding active form of the verb. 43. Our products are untouched by human hands. Such examples show the properties of adjectival passives, according to BCroeger (2004). He argues that the participial forms as in the example above must be derived by adding un- to adjectival passives. These participial forms with un- can only be used as adjectives and not as verbs. According to Siegel (1973), unpassives are "characterised by the presence of an auxiliary verb (either be or go), by the presence of a past participle prefixed with un-, and by the optional presence of an agent phrase" (pp. 301-302). Siegel (1973) offers a detailed analysis of unpassives and considers the various possibilities regarding how they may be derived. (See Siegel, 1973 for further details.) 1.5.6 Agentful / Agentless Passives Another important criterion that is often used for classifying passives is the presence or absence of an agent phrase. Usually the passive or the active-passive relation is illustrated with examples of passives containing agent phrases. Though it may be pointed out later that the agent phrase is optional, the discussions on the passive almost invariably begin with examples of what are called 'agentful' passives. To name a few, Bach (1974), Bloor and Bloor (1995), Downing and Locke (2002), Greenbaum (1996), Huddleston (1984), Huddleston and Pullum (2002), Kilby (1984) and Quirk et al. (1972) do the same. This may create the impression that the agentful passives are the basic passives.

26 As mentioned above, the grammatical subject of the active can be optionally present in the passive construction in an agent phrase (often referred to as b)H- Prep. Phrase). Givon (2001) observes that in most languages, the agent is excluded from the passive clause by a rule of grammar, but in relatively few languages such as English, agentful passive clauses may be grammatical, but their text frequency is low. Dagut (1985) gives the following as the 'cardinal' uses of the passive - 1) without any by- element; 2) with an added 6y-element denoting an animate agent; 3) with an added by- element denoting an inanimate force. He adds that it is a statistically confirmed fact that the 'agentless' use of the passive is the commonest use. It is so common that it can be regarded as the essential 'meaning' of the form. Svartvik's study (1966) is a major source of the statistically confirmed conclusion that the agentless passive is the commonest type of passive in English. Other studies like Femalld (1977) and Granger (1983) also support the conclusion in a general way. In their analysis of sentences and clauses in scientific English, Huddleston et al. (1968) also find that the majority of passives in their corpus do not contain an agent, but fiirther point out that in only some of them (especially where the agent is the writer or experimenter himself/herself) there is an identifiable agent that can be inferred. The remaining ones have no identifiable agent, with verbs such as distribute, base, reduce and divide used in scientific writing. They would be like non-agentive passives at the bottom end of the passive scale discussed by Svartvik (1966). (See section 1.8.2 below.) Jespersen (1933) and Palmer (1965b) also make a similar observation about the high frequency of agentless passives in English. 1.5.6.1 Agentive and Non-agentive Passives Kulkami (1981) makes a broad classification of passives into agentive and non-agentive passives. Here, the agentive passive is what is usually called the agentfiil one. The sub-types of non-agentive passives are - 'agentless', 'obvious-agent deleted', 'obscure-agent deleted' and 'discourse passive'. The agentless passive is that which has an instrument rather than an agent as its 'logical subject'. In the following sentence, according to Kulkami (1981), the 'obvious-agent deleted' type is found. 44. His mother could not be disturbed to notice what he was doing. Kulkami points out that in the above sentence, the agent phrase 'by him' is so evident that it is not expressed. When the agent can be inferred in general terms on the basis

27 of our knowledge of the world, it is 'obscure-agent deleted passive', as in the following. 45. Never in my life have I seen such a thing as when the explosion was produced. To introduce the next sub-type of non-agentive passive, Kulkami (1981) quotes Halliday's view (1970) that the basic unit of language in use is not a word or a sentence but a 'text'. The word this in the following example represents a presupposed element, which can be found in the immediately preceding sentence. So, the sentence needs to be considered in the light of the preceding one, i.e. in terms of discourse analysis. He calls this type 'discourse passive'. 46. This is chiefly directed against existing inequalities of wealth and income and the serious evils with which they are associated. Kulkami's classification (1981) is useful to the extent that it is centred on the nature/location of the agent. But there are fuzzy boundaries found in this analysis, too, as in the discussion on statal/adjectival/pseudo-passives etc. A striking feature of Kulkami's classification (1981) is that adjectival and statal/pseudo-passives are absent in it. He does not account for them at all. In a way, his classification is clearer, but that is owing to the fact that he has left out areas that are disputable. 1.5.6.2 Huddleston's Central and other Passives As mentioned above, Huddleston (1984) takes the agentful as the most central type of passive, and includes the agentless passives among the other types. He believes that the agent is a freely omissible element of clause structure, and there are no cases where the rules of require an agent to be present. He refers to the conclusions of studies that indicate that about 75% to 80% of passives are agentless. He argues that an agentless passive cannot be derived from an actual active clause, as "Max was killed'' does not encode the same information as "Someone killed Max". 1.5.6.3 Full Passives, Long and Short Passives, Truncated Passives Cattell (1969) refers to the agentive passive as 'the full passive'. Often the agentful and agentless passives are called 'long' and 'short' passives respectively. (See Biber et al., 1999.) Like Huddleston (1984), Huddleston and Pullum (2002) are of the view that short passives have no exact active counterpart. Some grammarians, like Culicover (1976), use the term 'truncated passives' for agentless or short passives.

28 Kilby (1984) mentions the fact that whether the short passive is derived from the long passive or not is a matter of controversy. But he asserts that one would be mistaken in thinking that the long passive is considered a more 'normal' type of passive than the short one. He refers to the study of the frequency of the passive conducted by Svartvik (1966) and also by himself, and points out that short passives are used with the frequencyo f 70 to 80%, the variation being determined by the types of language use - imaginative prose, informative prose, academic writing, informal conversation, etc. 1.5.7 The Passive Scale A few grammarians have found that the passive is not a uniform category, but there are types of passive which exist on a scale. In this section. Quirk et al. (1972 and 1985) and Palmer (1965b) will be considered. Quirk et al. (1972) give a formal definition of the passive -be + past participle of the main verb - which includes sentences like the following. 47. The violin was made by my father. 48. This conclusion is hardly justified by the results. 49. Cosil has been replaced by oil. 50. This difficulty can be avoided in several ways. 51. We are encouraged to go on with the project. 52. John was interested in linguistics. 53. The modem world becomes more and more highly industrialized and mechanized. Quirk et al. (1972) add that although these sentences satisfy the formal definition of the passive, they show a variety of 'voice relationships'. The types of the passive that they classify are as follows. 1 Agentflil passives: Sentences 47 & 48 in the above list have a direct passive-active relation. The difference between the two is that the first has a personal agent, whereas the latter has a non-personal one. Sentence 49 has two possible active transforms - 54. Oil has replaced coal. 55. [People in many countries] have replaced coal by oil. 2 Agentless: Sentence 50 illustrates the most common type of passive, which has no expressed agent. It is the agentless passive. 3 Quasi-passives: Sentences 51 & 52 represent a mixed class. Its members have both verbal and adjectival properties. The adjectival properties are the potentiality for (a)

29 coordination of the participles with adjectives, (b) adjectival modification with quite, rather, more, etc, (c) replacement of be by a lexically 'marked' auxiliary like feel, seem, etc. In the adjectival use of the past participle, the occurrence of 6y-agent is rare, not impossible. In sentence 52 above, in is used as an agent phrase preposition. It is called a 'quasi-agent'. There are several other prepositions, too, which can introduce such 'quasi-agents', such as about, at, over, to and with. 4 Non-agentive passives: Sentence 53 above has no active transform or possibility of agent addition, since no agent is conceived of The participles like industrialized have adjectival values. This class also includes 'statal' passives as in the following. 56. The house is already sold. In the corresponding active, there would be a change of aspect, i.e. shift to the (has sold). This notion of 'passive scale' given by Quirk et al. (1972) is based on Svartvik's model (1966) of the passive. It will be dealt with in detail in the section on his model, (See 1.8 below.) Quirk et al. (1985) modify the model of the passive to some extent. They include only the agentive and agentless among the central (true) passives. They exclude the quasi-agentive and non-agentive ones from the 'central' or 'true' passives on the 'passive gradient'. The reason for excluding these four sentences is that they do not have a clear correspondence with an active verb phrase or active clause, and are increasingly remote from the 'ideal passive' (of agentftal nature). Quirk et al. (1985) classify sentences classified in the earlier model with the label 'quasi-agentive' as 'semi-passives'. Those which are called 'non-agentive' in the earlier model are now termed 'pseudo-passives', as it is only their superficial form of verb + -e^ participle that recommends them for consideration as passives. As compared with their earlier model, the revised model of Quirk et al. (1985) displays an ambiguous position of recognising or not recognising certain constructions as passives. The ambiguity lies in including adjectival participle constructions, statal constructions and constructions without active transform among passives, but excluding them in a way by calling them 'semi-passives' and 'pseudo- passives' and considering them apart fi^omth e 'central' passives. Palmer (1965b) expresses the view that there are combinations oibe plus the past participle, which are clearly not 'true passives'. He finds the following three classes apart fi-omth e true passives.

30 i) Pseudo-passives: Palmer (1965b) finds some past participle forms to be wholly adjectival, as in the following. 57. The problem seems complicated. 58. John seems prepared to help. Some forms show no adjectival ftinction though there are no corresponding active forms, as in the following. 59. The house was situated in the country. 60. You aren't supposed to do that. There are some dubious cases where a corresponding active form might be thought possible. 61. The library is intended to be used. 62. Your trust was misplaced. ii) Semi-passives: These are past participle forms that appear to have corresponding actives, yet they show adjectival features. iii) Statal Passives: Here, the past participle forms are essentially perfect in meaning and refer to a resultant present state. Although these appear to have something in common with the pseudo-passives, they differ from each other to a large extent. The pseudo-passives are lexically restricted in the sense that they cannot be freely formed from any verb. But there are no such restrictions on statal passives. Yet even statal passives have degrees of'adjectiveness'. Like some others. Palmer (1965b) also makes a distinction between true passives on the one hand, and pseudo-passives, semi-passives and statal passives on the other hand. Various features of these categories overlap. 1.5.8 Prepositional Passives or 'Pseudo-passives' The term 'pseudo-passive' has been used in different senses by different grammarians. It has been mentioned before that the notional passive is also called pseudo-passive. Quirk et al. (1985) use the term in the sense of statal passive. Palmer (1965b) uses it to describe passives with adjectival features. But Chomsky (1965) uses it in a very different sense - in the sense of what are called 'prepositional passives'. A construction like the following is called a prepositional passive. 63. Indian food is often spoken of as rich, which is to say fatty and spicy meaning chilU hot. (/C£-/A^D:S2A-051#1:1:A)

31 The passive is usually connected with transitivity, as the active verb the object of which is made the subject of the passive has to be transitive. But one of the advantages of Chomsky's analysis (1965) is supposed to be that it can explain even the passivisation of an intransitive verb followed by a preposition with an object, which becomes the subject of the passive. It may be helpful to quote the syntactic definition of the prepositional passive given by Couper-Kuhlen (1979). NP + {can/must/will/...} + {be/get} + en + Prep + {by NP) There are various views on this process, as explained below. 1.5.8.1 A Historical Perspective on Prepositional Passives Gaaf (1930) is of the view that a prepositional object is a direct object, as spoke to in "Nobody spoke to him" can be regarded as a syntactical unit equivalent to a transitive verb. Gaaf (1930) finds that in Old English the place of the preposition was not yet fixed and it is sometimes impossible to distinguish between a preposition and an . In the transition period between Old and Middle English the latitude allowed to the preposition as regards its place in the sentence disappeared to a large extent, at least in prose. He fiirtherobserve s that the intimate connection that was felt to exist between a passive transitive verb and a preposition probably played an important role in the origin of the passive of intransitive verbs accompanied by prepositions. The process was accelerated by the loss of inflectional endings. 1.5.8.2 Christophersen and Sandved's Views Christophersen and Sandved (1969) point out that instead of a transitive verb plus an NP functioning as the direct object in the active sentence, we may have a verb plus a preposition or an adverbial followed by an NP. They find another variety of the prepositional passive, in which a verb is followed by an NP plus a preposition and the whole group is followed by another PP. An example is given below. 64. There are, however, some exceptions and counter exceptions which have been taken note of hy Panini. {Kolhapur Corpus, J 34,980-990) Christophersen and Sandved (1969) also observe that to be transformed into a passive construction, a phrase must be of regular occurrence, such as look after, tamper with, and take care o/and not of an ad hoc . 1.5.8.3 The 'Affectedness' of the Object of the Prepositional Phrase It has often been argued that an important constraint on the prepositional passive is that the object of the Prepositional Phrase should be viewed to be affected

32 in a certain way in order to become the subject of a passive sentence. Huddleston (1984) gives the following example. 65. Her hat had been sat on. He argues that this passive construction is acceptable, as her hat is significantly affected by the act of sitting on it, whereas the following may not be acceptable. 66. The stone bench had been sat on. Similarly, Huddleston and Pullum (2002) are of the view that passives of this kind are felicitous only if the verb phrase indicates either a significant property or a change in a significant property of the subject-referent. Kilby (1984) also writes about affectedness as an important criterion. 1.5.8.4 The Criteria of Proximity and Semantic Relations According to Hudson (1990), it is important that the preposition must not be separated from the verb. 67. Consequently a crowd demanding a judicial inquiry into the firing also turned violent, was fired upon and four more persons were killed. (Kolhapur Corpus, K 10, 1170-1180) 68. *Consequently a crowd demanding a judicial inquiry into the firing also turned violent, was fired bullets upon and four more persons were killed. The latter is not acceptable, as the verb^ire is separated fi-omth e preposition upon. Among the semantic criteria, Hudson (1990) mentions that though some adjunct-like prepositions are allowed in prepositional passives, passivisation is impossible if the stranded preposition refers to a time or reason. The following is an example. 69. * Supper was slept after. 1.5.8.5 The Idiomatic Nature of Prepositional Passive Kilby (1984) points out that the prepositional passive is acceptable whenever there is an idiomatic relationship between the verb and the preposition. The latter of the following is a doubtfiil case, as it is not idiomatic in nature, whereas, being idiomatic, the former is acceptable. 70. That conclusion was independently arrived at by several investigators. 71. *The stadium was simultaneously arrived at by several competitors. He further states that the 'idiomaticity' is about the relationship between the verb and the preposition, not the preposition and the object NP.

33 Ziv and Sheintuch (1981), while writing about the 'passivisation of oblique NPs' (i.e. prepositional passives), also point out that the passivisation of objects of prepositional phrases in sentences containing direct objects is seen in idiomatic expressions. 1.5.8.6 Concrete vs. Abstract Passive Subject Quirk et al. (1972) find that the acceptability of a prepositional passive can be stated in terms of concrete/abstract passive subjects. For example, the foUov^ang would be acceptable in their analysis. 72. In tuning the pitches of the svaras were apparently arrived at through the ear as in current practice. {Kolhapur Corpus, J 68, 811-830) But, Quirk et al. (1972) point out that the following is unacceptable, as the passive subject is concrete, not abstract. 73. *The splendid stadium was arrived at. 1.5.8.7 Bresnan's Lexicalist Explanation of the Prepositional Passive Bresnan {Occasional Paper #7) offers a lexicalist explanation of the prepositional passive. According to her, there is a lexical relation between the verb and its passivised subject. Thus, marched through means crossed, paid for means purchased, spoken o/means mentioned, etc. According to her theory, paid for, made fun of, etc. are also morphologically complex words in the lexicon. She adds that the lexical theory of passivisation in general explains the word-like properties of prepositional passives, whereas the transformational theory does not. 1.5.8.8 Couper-Kuhlen's Semantic-Syntactic Analysis Couper-Kuhlen (1979) makes an empirical analysis of the prepositional passive based on data provided by a large, representative corpus and supplemented by native-speaker intuition. A) Notional Configuration of Active Subject & Object Couper-Kuhlen (1979) firstly gives a notional configuration of subject and object and of passivisability in the following table. Active Subject Active Object Agentive Benefactive Agentive Comitative Agentive Goal Agentive Instrument Agentive Objective/Patient

34 Agentive Replacive Experiencer Stimulus These combinations of active subject and object are basically passivisable, according to Couper-Kuhlen (1979). B) Lexical Choice in the Prepositional Passive Another important step in the construction of the prepositional passive is that of the lexical choice. There are certain kinds of lexical choice, such as idiomatic expressions and metaphorical use of words, which may create constraints in the prepositional passive, according to Couper-Kuhlen (1979). i) Idiom Though the close association or lexical cohesion is favourable to the construction of the prepositional passive, too much of lexical cohesion can block the passive. She gives the following example. 74. John beat about the bush. 75. *The bush was beaten about... ii) Infrequent Use Just as too much lexical predictability is incompatible with the prepositional passive, so is too little lexical predictability. If a particular verb + preposition collocation is infrequent in the active, it may be tolerated in the active, but is likely to be doubtful in the passive. The following examples demonstrate this point. 76. The whole family joyed in her success. 77. ?Her success was joyed in by the whole family. iii) Metaphorical Use A passive usually occurs when lexical verbs are understood literally, not metaphorically. The following example is given to illustrate the point. 78. The chair was sat on. 79. *The committee was sat on. iv) Complex Structure Referring to Fillmore (1971a), Couper-Kuhlen (1979) makes a distinction between a verb with a simple lexical structure {come or go: motion) and a verb with a complex lexical structure {swim as in "May I swim in?": motion + manner). She argues that verbs which are strongly suggestive of manner are less acceptable in the passive than those which indicate only reaching of a goal, as in the following examples.

35 80. The street was gone across. 81. The street was hopped across. C) Choice regarding Verbal/Modal Features According to Couper-Kuhlen (1979), the choice of verbal/modal features is significant to determine the acceptability of prepositional passives. At least three major patterns of verbal/modal choice are found in acceptable prepositional passives. 1) The first pattern can be illustrated with the help of the following example. 82. A similar contract was entered into with the Great Indian Peninsula Railway Company for a line from Bombay to Kalyan at an estimated cost of + 500,000. (Kolhapur Corpus, J 42, 0050-0080) 2) A second pattern is found in the following. 83. Mysticism is the science of ultimates... the science of self-evident Reality, which cannot be 'reasoned about', because it is the object of pure reason or perception. (Kolhapur Corpus, J 53,0090-0110) 3) A third major verbal/modal pattern may be seen in the following. 84. Every rupee that was paid had been receipted for. (Kolhapur Corpus, K 35, 520) D) Information Focus The next factor deciding the acceptability of prepositional passive is the information focus. A prepositional passive is acceptable when the ^j^-phrase contains new information. If it contains given information, the passive is not acceptable. 85. The stimuli were always removed and replaced whether a change was called for by the random sequence or not. (Kolhapur Corpus, J 24, 1000- 1020) 86. ?There are, however, some exceptions and counter exceptions which have been taken note of by Panini. (Kolhapur Corpus, J 34, 980-990) In the latter, "by Panini" is not new information, as he has been mentioned earlier in the text. On the other hand, "some exceptions and counter exceptions" is new information. Therefore, according to Couper-Kuhlen (1979), it would be a doubtful case. E) Couper-Kuhlen's Grammar of Prepositional Passives The semantic-syntactic, lexical and contextual aspects of the prepositional passives comprise the grammar of the prepositional passives in the model used by Couper-

36 Kuhlen (1979). According to her, the following are acceptable, as they fulfil all the criteria mentioned above. 87. This bed has been slept in by Thomas Jefferson. 88. This bed has been slept in. 89. This bed has been slept in by someone. 1.5.9 Types of the Passive according to the VP Some types of the passive are classified according to the passive verb phrase. Some such analyses are given below. 1.5.9.1 The ^^-passive Some grammarians specifically analyse 6e-passive, as different firom any other possibilities, the main contender being the ge/-passive. This section is about the analyses of 6e-passives, where it is specifically mentioned so. The classification oibe + past participle constructions offered by Christophersen and Sandved (1969) has been mentioned above. Some views on the differences between 6e-passives and get- passives have also been mentioned. Givon (2001) classifies passives into promotional 6e-passive, promotional gef-passive and non-promotional impersonal passive, and adds that syntactic 6e-passive (which has an active counterpart) has its origin in adjectival resultative (lexical) passive. (See Givon, 2001 for further details.) Some grammarians define and describe the passive only in terms of be, and do not mention get or other passive auxiliaries at all. (See Mihailovic, 1967.) A) Frajzyngier's Aanalysis of the ^e-passives Frajzyngier (1978) does not agree that the verb BE is a part of the underlying passive construction in all languages. He refutes the view expressed by Langacker and Munro (1975) that 'stative' or 'existential' is an inherent meaning of passive sentences. Frajzyngier (1978) observes that in a number of languages there are at least two ways of marking passive voice. By making a cross-language review, he shows that Z>e-passives are by no means typical in all the languages universally. For instance, as Kulkami (1981) finds, in Marathi, it is not equivalent oibe, but equivalent of go. Whenever Z>e-passives do occur, the language also has some other means to indicate passives. 5e-passives are structurally a sub-class of nominal sentences. Frajzyngier (1978) finds that there are languages that have 6e-passive and nominal sentences, and there are languages that have a nominal sentence and no fee-passive.Bu t there are no

37 languages that have &e-passives but no nominal sentences. He concludes that nominal sentences are basic, and Zje-passives are a sub-class of nominal sentences. B) Granger's Model Granger (1983) makes a detailed analysis of the be + past participle construction in spoken English, giving special emphasis on the passive. She analyses the construction on the basis of structural, lexical and semantic criteria, and classifies it into seven categories. She refers to Svartvik (1966) and Stein (1979), who use the term 'passive' to describe all types of be (or any other auxiliary) + past participle constructions. But she uses the term 'passive' in its traditional sense, what is usually called 'the true passive', having a direct correspondence with the active. She reduces Svartvik's model (1966) of the passive to two sets of cases. 1 There is a potential active counterpart: the form is either verbal or adjectival. 2 There is no potential active counterpart. She finds three main categories of be + past participle construction (what she calls be Ved) on the basis of direct alternation with the active, indirect alternation with it and with no alternation. They are three categories, namely passives, verbal pseudo- passives and adjectival pseudo-passives. These are further subcategorised, making seven categories in all. Of these seven, only one is that of the passive as such, according to her. Passives: Passives are in direct alternation to a semantically equivalent active verbal group. They are 'agentful' if they contain an overt agent and they are 'agentless' if the agent is simply potential. Adjectival Pseudo-passives: Adjectival pseudo-passives are be + past participle combinations which do not stand in alternation to a semantically equivalent active verbal group and whose past participle form behaves like a 'central adjective'. Verbal Pseudo-passives: These are similar to adjectival pseudo-passives as regards their relationship with the active, but the past participle form displays no adjectival features. The remaining categories are 'usually passive category', 'peripheral combinations', and 'statal' combinations. (See Granger, 1983 for further details.) Granger's classification (1983) is very usefiil in that it covers all the complex and overlapping possibilities of classifying the be + past participle construction. As she herself points out, she combines the syntactic aspect of the be+ past participle

38 construction, which is central to Svartvik's classification (1966) and the lexical character of the combination underlined by Stein (1979). C) Granger's Classification of the Passive Granger (1983) finds that the passive outnumbers all the other categories of be + past participle in her corpus. It covers more than half of the be + past participle forms. She classifies passives according to the position that the passive holds in the sentence - marginal and central. i) Marginal Non-finite Passives She makes the following observations about the non-finite passives. 1 The most fi-equent of the non-finite forms in her corpus is the passive infinitive. It mainly occurs as an adjunct to a noun. 2 The passive infinitive is sometimes introduced by there. 3 It may also occur as an adjunct to an adjective or adverbial adjunct of purpose, as well as in the structure ^br + A'^+ infinitive. 4 Present (perfect) participles occur as quasi-predicative adjuncts but may also occur as fi^eeadjuncts . (See Granger, 1983 for further details.) ii) Central Passives The passive structure is central, according to Granger (1983), if it is part of the main verb phrase. Central passives can be simple (i.e. composed of a lexical verb only or a lexical verb and an auxiliary or semi-auxiliary) or complex (i.e. composed of a catenative followed by a non-finite verb form). In both the cases the passive form may be finite or non-finite. Central Passives: Complex Verb Phrases The passive form may be the catenative itself and/or the non-finite verb form tied with the catenative. Granger (1983) finds that there are four possibilities as follows. a)NP Vi pass V2 act (by NP) b)NP Viact NP V2 pass (by NP) c)NP Viact V2pass (byNP) d)NP Vipass V2 pass (by NP) Granger (1983) finds that structure (a) is the most predominant of the four complex verb phrases of central passives. The predominant non-finite form is again the infinitive.

39 Central Passives: Simple Verb Phrases According to Granger (1983), this category of passives constitutes 84% of the passive category. This category has no constraints on the active-passive alternation. She divides them further into two sub-classes - those with monotransitive verbs and those with ditransitive verbs. A majority of central passives with simple verb phrases belong to the former, and mostly have the active direct object as the passive subject. iii) Agentful vs. Agentless Passives She also classifies passives into agentful and agentless ones. She states the frequency of both of them in marginal, central (complex) and central (simple) passives. The agentless is more frequent, varying from 80% (in simple) to 91.5% (in complex). 1.5.9.2 The Ge/-passive According to some grammarians, the only serious contender for be as a passive auxiliary is get, the other verbs like look, feel, seem and become usually being understood as copular verbs. There are also some, like Christophersen and Sandved (1969), who do not take into consideration any auxiliary other than be at all, while analysing the passive. A) G^/-passive: How Common? ' "^ l\ ^«» n " Granger (1983) restricts her study of the be + past participle construction in spoken English only to be because, referring to Svartvik's study (1966), she points out that the use of the 'lexically marked auxiliary verbs' like get, become and feel is only marginal. But she does collect all the instances oiget + past participle from her corpus and classifies them. She finds that the proportion of get to be is very low. Having accepted the view that the passive has a direct alternation with the active, she finds that a majority of uses of get are not passives at all, but they are adjectival uses of get. She calls them 'non-passive' get + past participle forms. Her findings from the corpus indicate that get is used in passive only in 17% instances in spoken English. Greenbaum (1996) holds the view that get is used less commonly as a passive auxiliary. Berk (1999) states that increasingly, English speakers are using a passive construction with get as an auxiliary. According to him, written examples of the get- passive date back to the middle of the eighteenth century and the ge/-passive in written English is older than that in spoken English. However, it is more common in colloquial speech of the present day English. Palmer (1974) also observes that it is

40 more common in colloquial than in formal language and sometimes carries with it a hint of disapproval. Quirk et al. (1972) and Quirk et al. (1985) are of the view that the passive auxiliary is normally be, and its only serious contender is get. They add that get is usually restricted to constructions without an expressed agent. According to them, ^e/-passive is avoided in the formal style; and even in the informal style, it is far less frequent than the Z>e-passive. Leech and Svartvik (1975) also observe that the get- passive is found only in the informal style, and normally in constructions without an agent. It is most often used when the subject of the passive sentence suffers adversely as a result of the action. B) Ge/-passive: Passive of Becoming Morenberg (1997) states that passive sentences can suggest a sense of 'becoming', and get seems to suggest the sense of becoming stronger then be. He adds that especially in the informal language, get is used in order to emphasise the sense of becoming. Schibsbye (1970) points out that passive verbs (with be) are ambiguous, and they may denote either condition or action (change). In colloquial speech get is widely used as a passive auxiliary verb to ensure the connotation of change in such verbs. Making a distinction between the passive of being and the passive of becoming, Jespersen (1909-49b) observes that get and become are now increasingly common as auxiliaries for the passive of becoming, as the distinction between them and be is particularly useful, as in the following. 90. At that time he was not married. He got married in 1920. Jespersen (1933) incorporates get in his definition of the passive - "The English passive is formed with an auxiliary be, but often also get or become, and the second participle..." (p. 120). There are also some grammarians who accept get as a passive auxiliary with some reservations. Gramley and Patzold (1992) are of the view that at times get is less a passive than a copula followed by a participial adjective. C) Distinction between be- and ge/-passives Huddleston (1984) describes ger-passives as a type of the passive. He makes a distinction between get and other copular verbs such as become, look, remain and seem. The latter can take only adjectival passives, whereas get is recognised as an alternative to be in the verbal passive construction. He adds that get can also take

41 adjectival passives. Huddleston (1984) gives the following differences between verbal passives with be and get. (i) Get tends to be avoided in more formal styles. (ii) There are a few cases where be cannot be replaced by get, for instance, with verbs taking a non-finite complement: 91. Ed was heard to observe that... 92. *Ed got heard to observe that... (iii) Get tends to give a sense of some initiative or responsibility on the part of the subject, as in the following example. 93. Ed got arrested. Huddleston (1984) argues that it is because of this that get is more likely to be used in imperatives (where the style permits it), so that it assigns an agentive role to the explicit or implicit subject of the imperative. He makes a valid point here, as can be seen from the use of imperative constructions used in advertisements, especially in their slogans, illustrated in the examples below. 94. Get connected. 95. Get carried a^vay. D) Be and Get in Expanded Passives Huddleston and Pullum (2002) classify passives on the basis of the VP into two basic types - 'bare passives' and 'expanded passives'. A passive containing neither be nor get is a bare passive (non-finite passive). Be- and ger-passives are described as 'expanded passives'. They contain a bare passive augmented by means of a catenative verb that can carry the full range of verb inflections. They point out the area of ambiguity between verbal passives and adjectival complex-intransitive constructions, illustrated in the following examples. 96. They got killed by the hijackers, [verbal: get- passive] 97. They got very frightened. [adjectival: complex-intransitive] 98. ThQy got frightened. [ambiguous] But they add that these ambiguities arise less often with get than with be. An important feature for which ge/-passives have an advantage over be- passives is that it has an actional sense whereas be is likely to have a statal sense. E) Get: Passive or Ergative? Haegeman (1985) observes that in traditional literature on the passive the ge/-passive is usually treated as a variant of the 6e-passive and most discussions of the

42 construction categorise get as a passive auxiliary, but she shows that get lacks the properties of an auxiliary verb. i) Auxiliaries like have, be and modals allow negative contraction, but get does not. ii) Auxiliaries invert with the subject under Subject-Aux-Inversion, but get does not. iii) VP-deletion may strand auxiliaries, but get cannot be stranded. Haegeman (1985) makes use of the Government-Binding theory to explain the ^e/-passive. She begins by explaining the part of the GB theory that she uses. The verb destroy, for example, has two theta-roles associated with it - agent and theme (i.e. affected). 99. Bill destroyed the house. 'Bill' is the agent, and 'the house' the theme. Haegeman (1985) uses Burzio's analysis (1981) of ergative verbs. In an ergative construction like the following, the external theta-role of Agent is not assigned. 100. The boat sank. lOljf- -^ ec -<—sank the boat NP VP NPi Here, 'ec' standslbr 'empty category'. Semantically, passivised and ergative sink are similar, but not identical. In passives the agent may be present, but it cannot be present in ergatives. 102. The boat sank (*by the enemy). 103. The boat was sunk (by the enemy). According to Haegeman (1985), get is ergative in the following and does not assign Agent to its subject. 104. His girlfriend got invited (to all the parties). She concludes that the ge/-passive is partly a passive, through the presence of the passive morphology on the participle; but it is also an ergative construction with get as an ergative verb. 1.5.10 Other Classifications Some classifications of the passive are based on the structure of the passive verb phrase, the auxiliary verb used in it, the type of object that follows the active verb, etc.

43 1.5.10.1 Kruisinga's Classification of the Types of the Passive Manickam (1983) gives Kruisinga's classification (1931) based on the natxire and function of elements following the verb in an active expression. There are five types of the passive in his classification. (i) V - O type: If an object follows an active verb, it can be turned into the passive subject. This type is called the direct passive. (ii) V - PP type: If the verb in the active sentence is followed by a prepositional phrase, the object of the preposition can be passivised. (iii) V - O - O type: If an active sentence has two objects following the verb, any one of them can passivise. Earlier, there was a tendency to make the direct object the passive subject, retaining the indirect object in its original position. In such cases, the indirect object takes the preposition to before it - 105. They offered him a UGC grant. (Active) 106. A UGC grant was offered to him. (Passive) But, as Jespersen (1933) observes, during the last few centuries there has been a growing tendency to make the indirect object the subject in the passive, the reason being the greater interest that we have in persons than in things. Therefore, now it is more common to have constructions like the following. 107. He was offered a UGC grant. (iv) V - O - PP type: When the object of an active sentence is followed by a prepositional phrase the object can be made the subject of the passive, leaving the prepositional phrase to form a close group with the verb. It has been dealt with in the previous section. (v) V - O - Pred. Adjunct type: When the object is followed by a predicative adjunct or a complement, only the object passivises, but not thfe complement. The following is an example. 108. He was called a fool 109. *A fool was called him by me. 1.5.10.2 Double Passive Manickam (1983) refers to double passives as another type, which is formed in the following way. 110. We attempted to form a new plan. This sentence has two verbs, attempted and to form; and both are transitive. One of them can be passivised as in either of the following.

44 111. ^e attempted a new plan to be formed. 112. It was attempted to form a new plan. The verbs attempted and to form in the former and the latter examples respectively are still active, and they are, in turn, passivised, in which case the object of the infinitive is made the subject of the passive, as follows. 113. A new plan was attempted to be formed. 1.5.10.3 Passive of Experience Curme (1935), as referred to in Manickam (1983), analyses this type of the passive. It represents the subject as experiencing something. The past participle serves as an object complement, predicating something of the subject of the main verb, which is usually have or get. 114. Last week, I had my foot hurt.

1.6. Functions of the Passive in Englisli It has been felt that the active and the passive sentences mean almost the same. Just as this made grammarians and linguists study the relationship between the two, they also tried to find out why and when the passive is used. As mentioned earlier, Jespersen (1933) states that though the active and the passive constructions almost convey the same information, they are not exactly synonymous, which shows that it is not superfluous for a language to have both the types of construction. When attempts were made to find out what prompts the user to choose the passive rather than the active on a specific occasion, grammarians had various views on the fimctions of the passive in English. 1.6.1 Common Observations about the Functions of the Passive There is some agreement about the general functions of the passive. The following are some representative views. 1.6.1.1 Views of Various Grammarians 1) Jespersen (1933) makes the following observations about the reasons for the use of the passive. i) The active subject is unknown or cannot be easily stated. In the following example, no particular persons are identified as the agent. 115. Rajputs and Maithilis are also not considered a big force. (Kolhapur Corpus, A 27,1420-1430)

45 ii) The active subject is self-evident from the context. In the following, the makers of the ad can be easily guessed as the active subject or agent. 116. No clue was offered in the ad; perhaps it is a trade secret. {Kolhapur Corpw^, A 23, 0120-0130) iii) Jespersen (1933) is of the view that there may be "a special reason (tact or delicacy of sentiment) for not mentioning the active subject; thus the mention of the first person is often avoided, in writing more frequently than in speaking" (p. 121). In the following sentence, a remark is made about making efforts, without specifying who should make the efforts. 117. Efforts must be made to ensure that larger attendance does not result in lowering of the standards. {Kolhapur Corps, B 17, 930-940) iv) When the active subject is mentioned, the reason for using the passive is that more interest is taken in the passive subject than the active one. 118. In Andra Pradesh about ninety-two applications were received by the government. {ICE-IND:S\B-051 #79:1 :B) v) Another reason for using the passive with the agent mentioned in it is that it may facilitate the connection of one sentence with another. 119. Similar results have been obtained by Gupta & Broota (1975) in a study on the judgment of apparent vertical. (Kolhapur Corpus, J 24, 1490-1500) 2) Curme (1931) observes that the passive is often employed to avoid mention of the participants, and also to avoid a change of subject. 3) According to Christophersen and Sandved (1969), the passive is the normal construction when the agent is unknown or difficult to specify. It is used also when it is necessary or desirable to leave the agent unspecified, for instance in impersonal style such as scientific writing. The following is an example of impersonal style. 120. This is dealt with in Applied Economatrics. (Kolhapur Corpus, ill, 0730) Christophersen and Sandved (1969) also mention that the choice between the active and the passive depends upon where it is desired to place the centre of interest. The subject of the sentence attracts the main interest. For example, 121. All Indian religions have a common spiritual source: even Christianity and Islam as lived by the Indian people are profoundly influenced by the state of mind which is peculiar to India. (Kolhapur Corpus, J 22, 0020- 0040)

46 4) According to Berk (1999), the passive performs two important functions. One is that it maintains the topic of a discourse. The other reason for using the passive is "the egocentrism of human beings" (p. 120). We are likely to make a human being the subject of a sentence even when there is no topic to be maintained. For example, 122. The elderly teacher, Mrs. Joshi. was afflicted by leprosy ten years back. {Kolhapur Corpus, A 26, 0580-0590) 5) Bloor and Bloor (1995) point out that the choice between the active and the passive is determined by the considerations of the topic of the discourse. To maintain the topic, whichever type of voice is more convenient, is used. 6) Roberts (1967) mentions the general tendency to believe that when other things are equal the active kernel sentence is better than the passive transform, as the active is a "stronger, more direct, more straightforward way" of saying things, whereas the passive is "roundabout, wordy and weak" (p. 163). He adds that other things may not be equal. According to him, the existence of the passive fulfils some stylistic needs. It permits one to bring the object of the active into prominence. Roberts (1967) also states that the passive is opted for when the agent of the action is unimportant. 7) According to Dovming and Locke (2002), the following are the reasons for using the agentless passive, in particular. i) The agent is unknown, although implied. ii) It has already been referred to, directly or indirectly. iii) It may be understood fi-omth e context, but is considered irrelevant. iv) The implied agent is 'people' or 'one', the passive expressing a general statement. v) The speaker wants to highlight the predicator, as in the following. 123. So the concept is changing, those days certain things spoken in Indian English were consideredvn:ong. iICE-IND:SlA-02U2^:\:A) vi) Either out of politeness, to avoid blaming someone else or to avoid taking the blame oneself, the speaker wants to hide the origin of the action. 124. How can our States develop themselves if they are denied a fair price for this? {Kolhapur Corpus, A 04, 0550-0560) 8) Dagut (1985) makes the following observations about the functions of the passive. i) The passive, when considered an agentless function, takes its rightful place in the whole system of 'impersonal' constructions in English (other constructions being those introduced by someone/one/we/you/they/people).

47 ii) In case of agentful passives, the subject of the active utterance is given the more prominent position of rheme or comment in the passive in the form of a by- phrase. iii) The agentless passive has the function of making the action or event denoted by the verb the prominent part of the sentence. iv) Function (iii) is secondary to, and derived from, the function of making it possible not to mention the agent. 1.6.1.2 End-focus and End-weight According to Quirk et al. (1972), the principles of end-focus and end-weight are important reasons for using the passive. The principle of end-focus demands that the new or important information should be put at the end of a sentence. The principle of end-weight demands that the complex or weighty part of a sentence should be put at the end. The passive is "a means of reversing the 'normal' order of 'agentive' and 'affected' elements, and thus of adjusting clause structure to end-focus and end- weight" (p. 943). Though it has been mentioned above that himian beings are chosen as the subject of a sentence, it is not the only consideration. The following example is contrary to that principle. 125. The reservation idea was first mooted by the Backward Classes Commission, headed by Kaka Kalelkar in 1955. (Kolhapur Corpus, A 01, 0710-0721) Probably, the reason for this choice is the principle of end-focus. As can be seen from this example, there are views that focus two opposite considerations. According to one view, the passive is used to focus the affected by making it the passive subject. On the other hand, according to another view, the passive performs the function of putting the new information at the end, thereby following the principle of end-focus. Examples quoted above show that the choice depends on what factor is more important in a particular sentence. The following is an example of end-weight, the Z>y-phrase indicating the agent being longer than the passive subject. 126. That department was started by this professor called professor Ranjit Gill. (/Ci:-/A©:S1A-013#210:1:B) Dovraing and Locke (2002), too, mention that if the agent provides new information, it is focused on by using the passive clause consisting of a by-phrase.

48 Other motivations for using the passive, according to them, are the principles of end- focus and end-weight, which are closely linked. "New participants introduced onto the scene of discourse need to be described and defined more fully than known ones" (p. 253). 1.6.1.3 Mihailovic and Hill on the Principle of Prominence in the Passive The different views on how the passive functions for making a particular part of the sentence or clause prominent are mentioned above. The following is a detailed explanation of a specific case of disagreement between Mihailovic (1963) and Hill (1964). Mihailovic (1963) argues that the claim made by grammarians like Sweet (1930) about the passive being a device to give prominence to the active object is wrong. The assumption behind such a claim is that the initial position in itself carries prominence and emphasis. But, she points out that in a language where the normal word order of the statement is Subject-Predicate, the initial position of the subject does not make it particularly prominent. She argues that the functions of passive constructions in which the active subject is 'suppressed' and of those in which it is present in the form of a 6y-phrase should not be dealt with under one heading, as they have different functions. In case of the former, the syntactic and semantic requirements which determine the transformation of an active sentence into a passive are the need to get rid of a subject which adds no significance to the sentence and a desire for objectivity of attitude. On the other hand, in a passive utterance that includes a Z>v-phrase denoting the agent, which adds significant meaning to the sentence and cannot be suppressed, the choice of the passive is a device to give prominence to the agent. The primary stress given to it makes it very prominent. Hill (1964) argues that neither Sweet (1930) nor Mihailovic (1963) is right regarding his or her views on giving or not giving prominence to the active object by making it the subject of the passive. Hill (1964) demonstrates that the logical subject (the thing about which the information is being given) and the logical predicate (the information which is being given about the logical subject) are not necessarily the same as the grammatical subject and the grammatical predicate. For example, in "The sky is blue" (in reply to the question "Which of the things you can see now is blue?") the sky is the grammatical subject, but blue is the logical subject. Devices such as stress are important in signalling the logical subject and logical predicate. As far as the active and the passive are concerned, it is possible to make either the front

49 position or the end position in the sentence the most emphatic, and both the active and the passive are found with both of these ahematives. 127. Active: ("What happened to John?") 128. A snake 'killed him yesterday. 129. Passive: ("What happened to John?") 130. He was killed by a 'snake yesterday. 131. Active: (What was it that killed John yesterday?") 132. It was a 'snake that killed him yesterday. 133. Passive: ("What was it that John was killed by yesterday?") 134. It was a 'snake that he was killed by yesterday. What Hill (1964) wants to show is that it is possible to make the initial position or the end-position prominent equally well, and the choice of the active or the passive has nothing to do with it. 1.6.1.4 Functional Sentence Perspective While explaining the functions of the passive, Kilby (1984) employs the term 'Functional Sentence Perspective' used by Prague School linguists. According to 'FSP', sentence-initial position tends to be used for words or phrases which are 'given' or predictable from previous linguistic material, or from the context of the utterance, while 'new', less predictable information tends to appear later. These positions are called 'theme' (not to be confused with 'theme' in the sense of the affected in post- transformational grammar) or 'topic' and 'rheme' (from the Greek word meaning 'what is said) or comment on this topic respectively. What comes early in a sentence is its 'theme', and what follows is the 'rheme'. As subjects usually come at the beginning of sentences, they also normally designate the theme. But sometimes the object is the theme. By using the passive the object of the active can be given the thematic position at the beginning of the sentence. The rheme usually gives new information and the theme coincides with given information. Kilby (1984) clarifies that the theme does not have to be given information. It is possible to have new information in both the theme and the rheme. Likewise, the rheme may not necessarily be new information, not mentioned earlier at all. But usually, the theme coincides with given information, and the rheme with new information. He states the following considerations. i) Definite NPs have a strong tendency to be the theme and indefinite ones tend to be the rheme.

50 ii) The rheme tends to be lengthier than the theme. iii) Human beings are more likely to be the topic of conversation than inanimate objects. The following findings from Kilby's corpus (1984) consisting of 122 full passives are significant as far as functions of the passive are concerned.

Subject NP Agent NP Given 73% 20.5% New 8% 79.5% Other 19% 0 Definite 69.5% 64% Indefinite 11.5% 36% Animate 16% 50% Inanimate 65% 50%

The subject NPs have a stronger tendency to be 'given' and to be definite. The agent NPs tend more fi-equently to be 'new' information. The subject NPs are more frequently inanimate. Agent NPs have a tendency to be animate and inanimate with equal frequency. 1.6.1.5 Findings of Biber et al. Biber et al. (1999) write a corpus-based grammar of the spoken and written English as used by native British and American speakers. They classify finite passives into two classes - long passives (i.e. containing an agent phrase) and short passives (i.e. without an agent phrase). They give their findingsabou t the average length of the subject and that of the agent phrase on the basis of the data collected from their corpus. The findings show that the subject of a long (i.e. agentful) passive is firequently short, and the agent phrase is frequently longer than the subject. Biber et al. (1999) classify subject and agent phrases according to their information status - given, given/new and new. They introduce the intermediate category of'given/new' as they are of the view that it is required to handle multi-word phrases which include elements of both given and new information. Their findings show that about 90% of the agent phrases bring in new information.

51 They mention that traditionally, the passive has been associated with formality and impersonal style. Accordingly, the passive is the most frequent in academic prose and the least frequent in conversations. Short passives do not contain agent phrases, and the agent is most typically human. Therefore, short passives denote impersonal style. This is found in their different frequencies in different registers. They are most frequently used in academic prose, next to that in news. 1.6.1.6 Sinha on the Indispensability of the Passive Sinha (1974) refers to the general opinion that the active is basic as compared with the passive. He refers to Orwell's opinion (1966) that one should never use the passive where one can use the active. He argues that Orwell (1966) does not want to condemn the passive, but he wants to criticise the use of vague expressions that hide the real political intentions. Sinha (1974) quotes a passage in which Orwell (1966) himself uses a series of the passive - "Defenseless villages are bombarded from the air, the inhabitants driven over in the countryside, the cattle machinegunned, the hut set on fire with incendiary bullets: this is called 'pacification'. Millions of peasants are robbed of their farms and sent trudging along the roads with no more than they can carry: this is called'Xrsaisfer of population' or 'rectification of frontiers'" (p. 631), Even if rhetoricians are of the view that the passive makes the expression weak, this passage highlights the point the author is trying to make. Sinha (1974) argues that if we try to put these sentences into the active, "we not only face the problem of finding out who did Orwell accuse of bombing villages and so on, but also of evaluating the impact of the presence of these agents in surface sentences" (p. 632). Sinha (1974) refers to Stanley's opinion (1971) that the passive permits the deletion of pertinent information, and his view is that Stanley (1971) fails to appreciate that in certain contexts the passive conveys a piece of pertinent information more effectively than the active. The active often tells us more than what is required. Sinha (1974) refers to Grice's 'cooperative principle' which asks us to make our contribution as informative as is required. Repeating what is evident, the active may be less preferable than its passive counterpart. According to Sinha (1974), the theme or topic that the speaker has in mind in a continuous discourse is a vitally important factor that determines the choice of the active or the passive. He suggests that we should "begin with the idea of a chain of themes or topics and find out how they are going to be related. The relative importance of themes in a discourse will determine which transformations will apply

52 in a context, which may, for instance, make the passive optional in a particular context" (p. 638). He further states, "There is no reason to deny that the active and the passive have transformational relation in abstract syntax and there are several reasons to conclude that the active is more basic than the passive, but in a particular discourse, because of special semantic and syntactic requirements, sometimes the passive is the only choice" (p. 639). 1.6.1.7 Stem's Model Stein (1979) accepts, to a large extent, Svartvik's definition (1966) of the passive as a be (or any lexically marked auxiliary commutable with it) + past participle. She concentrates on the functions of the passive in her corpus-based study. She has a new explanation of the fimctions of the passive to offer. She uses a graphical representation for resultative process and activity verbs. ... I 1 \ ^ • XI X2 X3 X4 XI is used in order to refer to the beginning. The extralinguistic order of stops called at in the course of the development of a process or an activity is not necessarily the same at the linguistic level. For instance, in "He gave the boy a book", he is XI, the book is X2 and the boy is X3. But that is not the order at the linguistic level, where the arrangement shows "XI—X3—X2". The fiinctiono f the passive in English consists in representing a resultative process as not having its beginning in XI (i.e. the agent). Stein (1979) asserts that the basis for the positional change of the NPs is not an information focus device but the fact that the process or activity does not begin in XI. Another element takes the normal position of XI. This change in the word order of the noun phrases can then be exploited for information focussing purposes. She states that the information focussing devices are "derivable from the function of the passive, but they are not the function of the passive itself (p. 135). But h is very difficult to find how this model of analysis is different from the commonly made observation that the passive brings the affected to the initial position. 1.6.1.8 Granger's Analysis of the Passive as a Textual, Pragmatic and Stylistic Variable Granger's analysis (1983) of the be + past participle construcfion in English has been explained earlier. In the same work she studies the functions of the passive extensively, with the help of findings from a corpus. She points out that the functions

53 of the passive have been studied so far by considering it a textual and pragmatic variable, but it has not been considered a stylistic variable. She begins by asserting the need to study the functions of the passive. She is of the view that transformationalists have failed to explain the passive satisfactorily because all the different versions of transformational grammar suffer from the same weakness: "a high level of abstraction and formalization with no (or little) empirical support" (p. 48). She adds that according to functionalists, the reason for the transformationalists' failure is that they have attempted "to describe in a purely formal way a relationship which is mainly functional" (p. 49). A) The Passive as a Textual and Pragmatic Variable Granger (1983) summarises the textual approach to the passive by identifying three main factors in the use of the passive. One is the focalising of the actor (i.e. giving the actor or agent the end-focus), which is applicable to agentful passives. The second is the 'zeroing' of the actor, which is applicable to agentless passives. The factor applicable to both the agentful and agentless passives is the selection of the affected as theme (the 'non-actor as the unmarked theme'). The pragmatic approach is concerned with agentless passives and the reasons for omitting the agent. When the agent is omitted without any serious loss of information, it can be called the 'innocuous passive'. The reasons for omitting the agent in innocuous passives are purely textual. They are linked with the way in which the speaker organises his discourse. But sometimes the agent is omitted as a part of the speaker's strategy of giving less than the required amount of information. Such passives are called 'insidious'. B) The Passive as a Stylistic Variable Granger (1983) finds that the passive has not been studied as a stylistic variable property. The only study of the passive used in various registers is that of Svartvik (1966). But, according to her, a weakness of Svartvik's study (1966) is that the major determining factor in the frequency of the passive is a distinction such as that between informative and imaginative prose rather than a difference of subject matter or difference between spoken and written language. But Granger's corpus- based study (1983) gives different findings. Taking 'field', 'tenor' and 'mode' as the dimensions of the use of language, she finds that low passive uses share the following stylistic features.

54 i) The field is unrestricted and primarily personal. It involves the speakers as 'private persons' with their own thoughts, feelings and experiences, not as 'public persons' (such as politicians and lecturers), ii) The texts with low fi-equencyo f the passive display a spontaneous mode, iii) Such texts also show an informal personal tenor. Heavy passive users display a restricted field (which is primarily non-personal or factual); and they are non-spontaneous and formal. Putting these three factors together. Granger (1983) makes a single factor - colloquial or non-colloquial style. Colloquial ^ [+ personal] ^ [- personal] Non- Style [+ spontaneous] • [- spontaneous] colloquial [+informal] • [-informal] Style

Low passive use goes together with colloquial style, whereas high passive use goes together with non-colloquial style and this holds true whether the text is spoken or written. Any change in one of the three features shown in the diagram above may bring about a change in the frequency of the passive. Granger's study (1983) shows that it is a very detailed analysis of the frequency of the passive and the style used. It throws light on the reasons why the frequency of the passive varies irrespective of spoken or written medium. For instance, oration shows ['-personal'] and ['-spontaneous'] features; and therefore, in spite of the spoken medium, there is a very high frequency of the passive in oration. The same is true of discussions, too. Therefore, Granger's analysis (1983) and her findings are useful for studying the functions of the passive.

1.7 Frequency Studies of the Passive in English In the last few years, the frequency of the passive has been an object of study for grammarians who have extensively studied the passive in English. The frequency has been counted in order to find out the relationship of the passive with the active, to find out the relative frequencies of agentflil and agentless passives and also to analyse the functions of the passive. Apart from those who especially describe the passive in respect of its frequency, there have often been references to the relatively low fi-equency of agentful passives and relatively high frequency of agentless ones. (See Jespersen, 1933 and Quirk et al, 1972.) But there are also some major studies that

55 have given corpus-based statistics of the frequency of the passive or of some of its features. The advantage of these works is that they describe the passive as it is actiially used, on the basis of findings from their corpora. Svartvik (1966), Granger (1983) and Biber et al. (1999) give statistics of the frequency of passives in various registers of English. 1.7.1 Svartvik's Study of Voice in English Svartvik (1966) analyses passives collected from texts belonging to various registers, such as science, news, speech and novels. Though his study is devoted to voice, its findings are especially important as regards classification of the types of the passive and the frequency of the passive in English. His corpus indicates that the frequency of the occurrence of agentive passive (i.e., agentflil and agentless passives taken together) clauses per thousand words is as high as 32 in a text written in the register of science at one end, and as low as 3 in television advertisements at the other end. In printed news, the frequency per thousand words is 18.6. In novels, it is 5.2 at the lowest and 11 at the highest. Svartvik (1966) draws a conclusion that the passive is comparatively more frequent in the written rather than spoken medium and in informative rather than imaginative prose. He also counts the firequencyo f agentive passives, quasi-agentive ones and non-agentive ones in texts from various registers. Of the agentive ones, those with animate agents are 10% in number, the ones with inanimate agents 8%, those with Janus agents 2% and agentless passives are 80% in number. It is the last of these figures that has clearly indicated that the passive in English is predominantly of the agentless type. Svartvik (1966) also gives statistics of other features, such as the proportion of various verbal groups. For instance, the verbal groups with the or simple past tense make 54% of the agentive passives, and next to that in number are those with modal verbs (30%). He also gives the percentage of each passive clause type in agentive passives. Passive transitive clauses make 73% of passive clauses, passive semiditransitive 9%, passive ditransitive 3%, passive semifactitive 5% and passive factitive 10%. He also gives the average length of the subject (i.e. number of words in the subject of the passive) and the average length of the agent phrase. The findings firom the corpus show that agents tend to be longer than subjects. All these findings, based on a study of the passive which is perhaps the first of its kind in English, throw light on the use of the passive in English and also on various important features of the passive in English.

56 1.7.2 Fernalld's Survey of the Passive in English Feraalld (1977) follows Svartvik's definition (1966) of the passive. She analyses the passives in some college texts on various subjects to find out the frequency of the various types of the passive according to Svartvik's classification (1966). She finds that agentive passives in her corpus make 15% of the passives, agentless 71%, quasi-agentive 2% and non-agentive 12%. The following are some of her conclusions. i) Agentful passives are relatively low in frequency, which can have significant implications for the teaching of the passive in English, ii) The written texts on physics and transcribed lectures on it show the same frequency of the passive, which is a finding in accordance with Granger's conclusions (1983) about the functions of the passive stated above, iii) The high frequency of the passive is found not only in science but in other subjects as well. In fact, the passive is more frequent in history, economics and sociology texts than physics texts. Femalld (1977) is of the view that "studies which attribute a higher percentage of passives to scientific writing than to other subject matter may have to re-examine their findings" (p. 65). iv) Though passives are probably transforms of the actives, there are very few which can be transformed directly into the active without additional changes. 1.7.3 Kilby's Findings Though Kilby (1984) does not count the frequency of the passive in his data, he gives statistics of an important feature of the passive - the nature of the subject and agent NPs. In order to analyse the functions of the passive, Kilby (1984) finds out the nature of the subject and agent NPs in his corpus consisting of 122 fiill passives. He considers three factors - Given/New, Definite/Indefinite and Animate/Inanimate. Among subject NPs in his corpus, 73% have Given information, and 79.5% of the agent NPs have New information. This shows the function that the passive performs. Kilby (1984) also gives us a list of verbs and their co-occurrence with prepositions in prepositional passives that he finds in his corpus. 1.7.4 Granger's Findings about Passive Frequency In order to study the functions of the passive and the passive as a stylistic variable. Granger (1983) studies the frequency of the passive in the texts belonging to

57 various registers from her corpus. Like Svartvik (1966), she gives the frequency of the occurrence of passive clauses per one thousand words. The lowest frequency (1.98 per one thousand words) is found in conversations and the highest (8.88 per one thousand words) in oration. The frequency is 7.5 in discussions. Granger (1983) also gives the findings regarding the proportion of agentfiil and agentless passives. The overall proportion shows 18% of agentftil and 82% of agentless passives. The frequency of the agentless passives is 65% at one end (in commentary) and 94% at the other (in conversations). As explained above, these statistics help Granger (1983) draw conclusions about the relationship between the frequency of the passive and colloquial style. 1.7.5 Frequency Analysis of Biber et al. Biber et al. (1999) find that passives are most conamon in academic prose - occurring about 18,500 times per million words (i.e. 18.5 occurrences per one thousand words). Their occurrence is 12 per thousand words in news. The occurrences of short and long passives, be- and ger-passives, non-finite passives in various registers are also counted. Biber et al. (1999) also give a list of the most commonly used verbs in the passive, with their relative frequency in various registers - conversations, fiction, news and academic prose. For instance, make, give, use, find and see are most frequent in academic prose. Like Svartvik (1966), Biber et al. (1999) also count the number of words in subject and agent phrases. Like Kilby (1984), they also give the statistics of the 'givenness' in subject and agent phrases in three categories - Given, Given/New and New. 1.7.6 Stein's Frequency Analysis Stein (1979) gives corpus-based statistics of the occurrence of various verb phrases in the passive across registers. For instance, the overall percentage of finite verb phrases in the simple present and simple past is 68.65. Modal + be + past participle has the percentage of 15.53. The remaining types of passive verb phrase are less common than these two. Stein (1979) also gives the statistics of the occurrences of non-finite passive verb phrases, among which be + v-ed is the most common. She also counts the number of various modal and non-modal auxiliaries in finite passive verb phrases. These detailed descriptions of the finite and non-finite passive verb phrases and of the auxiliaries in finite passive verb phrases is a major advantage of her work.

58 1.8 Svartvik's Model of the Description of the Use of the Passive in English Svartvik's model (1966) is described in this independent section, as it is the model selected for the analysis of the use of the passive in Indian English in this work. Svartvik (1966) has collected 28 texts from spoken and written English. They belong to the following registers - conversations, discussions, television advertisements, radio news, press news, editorials, sports reports, popular and learned writings on science and arts, plays and novels. The whole corpus consists of 3, 23, 000 words. As mentioned by Granger (1983), before Svartvik (1966), there are no corpus-based studies of the passive with such a big corpus, consisting of texts taken from so many registers. Svartvik's study (1966) has a broader area to cover, viz. voice in the English verb. The present work is restricted only to the passive voice in Indian English. Svartvik's analysis (1966) of the passive clause covers every element of the passive clause, such as the form classes of words in the subject NP, the gender of the nouns in the subject NP, and adjuncts in the passive clause. For the present work, only a part of the model developed by Svartvik (1966), as can be found in his chapter on the use of the passive, is used. Therefore, only the relevant part of his model is described here. 1.8.1 Svartvik's Definition of the Passive Svartvik (1966) summarises the voice relationship in two statements - (i) there is a relation between the active and the passive voice, and passives, being less central than actives, are to be treated as derived from actives; and (ii) there is not a one-to-one relation between the active and the passive voice, and considering the passive simply as a derivative of the active will not yield a good grammatical description. He states that there are a number of reasons for the lack of agreement among grammarians as to what constitutes an English passive. The name 'passive' itself is partly responsible for the disagreement. Not all grammarians claim that the subject of a passive construction must necessarily be the 'sufferer' of the action. However, they have some such requirement in mind, when they preoccupy themselves with concepts of'action' as opposed to 'state'. Svartvik (1966) argues that the weakness of any such definition is that it excludes many related constructions. The line of argument that he takes is that syntactic relationships can or should be expected to be multidimensional rather than binary and that, in order to find and describe this

59 network of relations, it is necessary to keep the scope of the passive wide. Giving examples of all the constructions that have already been quoted above as accepted by Quirk et al. (1972) as passive constructions (see 1.5.7), Svartvik (1966) states that whatever the differences in meaning, they are all considered passives in his framework. All those sentences have one formal feature in common - they all have be (or auxiliaries commutable with be) and a past participle. He accepts this as the definition for his study of the passive in English. 1.8.2 The Passive Scale and Voice Relation Svartvik (1966) classifies the passive into seven types, as follows. i) Class a: We shall call this type 'animate agent passives'. The agent phrase is present in the clause, and the agent is animate. ii) Class p: We shall call this type 'inanimate agent passives'. The agent phrase is present, and the agent is inanimate. iii) Class p/7: We shall call this type 'Janus agent passives'. Sometimes prepositional iy-phrases may be interpreted either as agents, in which case they are class p passives, or as adjuncts, in which case they are class y (i.e. agentless) passives. The following is an example. 135. It advises a wise man to straighten his mind as an arrow is straightened by a fletcher. (Kolhapur Corpus, D 10, 1660-1670) In one interpretation, the fletcher straightens the arrow (therefore an inanimate agent passive), and in another interpretation someone straightens the arrow with a fletcher (making it an agentless passive). The above three types have a direct transformational relationship with the active, as they can be converted into the active without any further changes. iv) Class y: This type will be called 'agentless passives'. Svartvik (1966) observes that this is the central and most frequent passive type. It has no expressed 6y-agent, but may have direct agent extension with subsequent systemic potential active transformation. The word 'direct' denotes that agent extension and active transformation are possible within the same tense. For example, 136. Scientific progress is misused creating nuclear bombs & not for betterment but to extinguish the lovely & beautifiil God's creativity. {ICE- /iV£):WlA-002#94:2)

60 It is possible to give an agent extension such as by man to the above sentence. Subsequently, it is also possible to make an active transformation in the same tense, as in the following. 137. Man misuses scientific progress... v) Class 5: This type is called 'quasi-agentive passives'. It holds an intermediate position on the passive scale, as it has both adjectival and verbal features. Its verbal character is found in its potential transformation into an extensive active clause, as in the following example. 138. And I must say that all have been impressed with the fresh view points which he has been presenting. {ICE-IND:S2A-026# 14:1:A) It can have the following active transform. 139. The fresh view points which he has been presenting have impressed all. The adjectival nature is seen in its potential transformation into an intensive active clause, and in its ability to take coordination with adjectives, qualification, and lexically marked auxiliaries such as feel, become, seem and get. The following is an example of potential transformation into an intensive active clause. 140. Ajay, I am worried about Priti. {Kolhapur Corpus, K 31, 0490) It has the following potential transformation into intensive active clause. 141. "Ajay, Priti makes me worried." In the two examples given above, the prepositions used in agent phrases are with and about respectively. Such prepositions other than by are called 'quasi-agents'. That- clauses and /o-clauses can also fimction as agents. This class has two sub-classes - 81 (also called 'attitudinal passives') and 52 (also called 'emotive passives'). The following are examples of the attitudinal passive. 142. He was fully entitled to his opinions against birth control. {Kolhapur CorpM5, J 31, 0440-0450) 143. We feel inclined to ask how long this lovely lamb is being slain. {Kolhapur Corpus, D 04, 1290-1300) Unlike emotive passives, the attitudinal passive cannot have the following kind of active transformation. 144. ""His position made him entitled to... The feature of lexically marked auxiliary {feel in this case) can also be seen in 143. An attitudinal passive can have a 67-agent. Emotive passives, on the other hand, have 'quasi-agents' such as prepositional phrases containing in/at/with etc. or a to- or that-

61 clause. The possibility of active transformation is less normal in attitudinal and emotive passives as compared with agentive passives explained above. vi) Class 8: This type is called 'non-agentive passives'. Svartvik (1966) finds it the most multifarious class of all. In most non-agentive passives, no agent is conceived of Therefore, "it is no longer realistic to consider these clauses in terms of voice transformation potential" (p. 135). For instance, 145. The intermediate type of parenchyma cells are formed in the same manner as has been described for phloem. (Kolhapur Corpus, J 11, 1480- 1500) No agent that forms the cells is identified here. Some non-agentive passives are such that an agent is identifiable, but an active construction is not possible, as in the following example. 146. Drinking of water having more than 15 ppm of fluorides is said to cause dental as well as skeleton forms of fluorosis. (Kolhapur Corpus, H 10, 1610- 1620) It is possible to guess, in general terms, who says it, but the following active transform is not possible. 147. *We say the drinking of water having more than 15 ppm of fluoride to cause... Some non-agentive passives are not exactly non-agentive, but their direct active transformation is not possible. The active transformation has to be accompanied by a change of aspect. 148. Hence it is true that though religion is an abstract thing & it is invented by man and the same man is the one who made different religions and they are the ones who distributed men into various classes, but they are the ones who are always fighting against each other, though they know that religion is not createdh\ God. (JCE-IND:W\A-003#53:\) Even if there is an agent phrase in the passive clause, a direct active transformation is not possible. It is possible only after a change of aspect. 149. *.. .man invents it.. .though they know that God does not invent religion. It is possible to have an active transform after a change of aspect - has invented. This is what is sometimes called 'statal passive'. vii) Class ^: This type is called 'compound passives'. These are almost like the 'impassives' explained earlier. Svartvik (1966) makes a general category of passives

62 with compound verbs having past participles, but not infinitive forms, which includes unpassives. The compound verbs are morphologically isolated, in the sense that they do not have active coimterpart. 150. Catchwords and slogans, programmes and pamphlets, constitutions and manifestos, have no intrinsic and practical value unless the mind of the ruler is untainted by selfishness. (Kolhapur Corpus, J 39,0430-0460) There is no such verb as untaintand therefore, an active transform like "* selfishness untaints the mind of the ruler" is not acceptable. The above example shows that it is not impossible for a compound passive to have an agent phrase. The following example shows that not all compound passives are unpassives. 151. "No instnmient", he asserts, "is self-tuned and that the harmonizing of it is the prerogative of sense-perception is obvious and requires no proof" (Kolhapur Corpus, J 68,1300-1320) This example also shows the adjectival character of compound passives, as explained by Kroeger (2004). Compound passives are a minor type, and Svartvik (1966) does not give the statistics of its frequency. Svartvik (1966) observes that at one end of the passive scale, passives are transformationally related to active clauses (transitive, ditransitive and factitive). He argues that as we proceed down the scale, "it becomes increasingly difficuh to consider the passive in terms of a transformational voice relation, since this is being gradually replaced by a different relation which we call 'serial'. ...The concept of serial relation will help us to account for many voice features which cannot be explained otherwise" (p. 159). At the bottom end of the scale, in non-agentive and compoimd passives, there is no possibility of direct active transform. In between the two extremes of the scale there are intermediate classes with mixed properties. The agentful passives (animate and inanimate agent passives) have all the properties required for a direct active transform. The Janus agent passive is a link between agentive and agentless passives, as it has two possible interpretations - an inanimate agent passive and an agentless passive with an adjunct other than the agent phrase. The agentless passive can have a direct active transform, but not as direct as the agentful passive, since the potential agent has to be guessed. It is not overtly present, as in agentful passives. Quasi-agentive passives can either have an active extensive transformation or an active intensive transformation. Svartvik (1966) finds a cline in the class of quasi-agentive passives. At one end, they are like inanimate agent

63 passives, as they sometimes have a by-agent phrase. At the other end, they may also be similar to non-agentive passives. As w^e proceed from the agentful passives to non- agentive passives, the possibility of having an active counterpart goes on decreasing. A non-agentive passive does not have an active counterpart at all or an active transform requires a change of aspect. 1.8.3 Passive Clause Types Svartvik (1966) finds three main passive clause types - passive transitive, passive ditransitive and passive factitive. He writes that although these names are "unsatisfactory in that they are modelled on the names of their corresponding active clauses and hence imply, misleadingly, that there is always a direct active-passive relationship, it seems nevertheless useful to have an indication in easily convertible terms of potential clausal voice relation, provided the transformational constraints are also borne in mind" (p. 139). He further subdivides passive transitives into clauses which have and those which do not have a complement-like prepositional adjunct in close collocation with the verb. The clauses which do not have such collocation retain the label 'passive transitive' and the others are further distinguished according to their internal extensive (in case of 'passive semiditransitive') and internal intensive (in case of 'passive semifactitive') clause relations. As only agentive and agentless passives can have direct active transform, Svartvik classifies passive clauses only in case of agentive and agentless passives. No passive clause type is mentioned in case of quasi- agentive and non-agentive passives. As Svartvik's definition (1966) of the passive, classification of passive types and notion of the passive scale are accepted for the analysis of the passive in Indian English in this work, his method of classifying passive clause types is also followed. Passive Transitive: This is the most frequent of the passive clause types. It is illustrated in the following example. 152. The classical methods of arriving at scientific truth were rediscovered in Europe... (/C£-/M):W1A-002#20:1) There are also two minor sub-types of passive transitive, which may not be dealt with at this point. (See Svartvik, 1966 for details.) Passive Semiditransitive: The following is an example of a passive semiditransitive clause.

64 153. He was a lawyer then judge and he can be compared for that matter with Henry Fielding who was a lawyer and then a judge and it was his pastime. iICE-IND:S2A-022# 131:1: A) The difference between a passive ditransitive clause (like "I was given a reward") and a passive semiditransitive one is that of the preposition (like with in the above example) that is absent in the ditransitive clause. The collocation between compare and with in this example fulfils the condition stated by Svartvik (1966). Passive Semifactitive: The difference between passive factitive (like "I was called a fool") and passive semifactitive is that of as, which is absent in passive factitive but present in passive semifactitive, as in the following. 154. If the eighth standard becomes a terminal point in place of the tenth standard, the one great advantage will be that those who now drop out at that level can be described as pass-outs and not drop-outs! (Kolhapur Corpus, B 01, 0350-0380) Passive Ditransitive: A passive ditransitive clause is one the active counterpart of which has two objects. It has two varieties, depending upon whether the animate object or inanimate object is made the subject of the passive clause. The animate (i.e. indirect) object of the active is more commonly the subject of a passive ditransitive clause, as in the following. 155. Every donor is given a receipt. (/C£:-/A^Z):S1B-080#105:1 :C) Passive Factitive: The potential active counterpart of a passive factitive clause, as in the following example, is an intensive clause. 156. "In referring to underdeveloped countries students were commonly expected to be 'deskmen' not soiling hands. {Kolhapur Corpus, C 01, 1060- 1070) It may be noted here that Svartvik's classification (1966) of passive clause types almost coincides with Kruisinga's classification (1931) of types of the passive (see 1.5.10.1). As Svartvik (1966) has this classificafion of types of the passive clause and also another classification of types of the passive, his model can cover more features of the passive. 1.8.4 Advantages and Drawbacks of Svartvik's Model The following are the advantages of Svartvik's model (1966) of the analysis of the passive.

65 i) An important advantage is his comprehensive definition. The complications created by various other classifications of the passive (as shown in 1.5) - statal/adjectival/pseudo-passive/non-agentive etc. with overlapping features - are resolved by his definition of the passive and his notion of the passive scale. Some grammarians do not accept constructions other than those verbal ones with active transform, thereby ignoring a considerably large section of be + v-ed constructions. Some others want to accept statal/adjectival/non-agentive constructions as passives but keep them apart fi-om 'true passives'. (See section 1.5.) As Femalld (1977) puts it, "...Svartvik's concept of serial relationship accurately describes the ambiguous relationship between past participles which are verbs and past participles which are adjectives" (p. 57). She further summarises his notion of the passive scale in the following way. "He concludes that all passives have an underlying active form, and that this form can be derived either by direct transformation, or by serial relation" (p. 57). She adds that his classification "resolves the problem of having to distinguish between adjectives and verbs in ambiguous cases, and his purely formal definition of passive {be + past participle) is the most satisfying, the least ambiguous..." (p. 57). There are other comprehensive descriptions, too. Granger (1983) gives a very detailed classification of be + V-ed (what she terms be Ved) constructions. But the distinction between categories like passive, verbal pseudo-passive, adjectival pseudo- passive, usually passive and statal passive is too fine to capture at times. Granger (1983) also admits that these categories often overlap. Therefore, though very detailed and impressive, her model of analysis is practically not very useful, ii) Another strength of Svartvik's study (1966) is his findings in terms of statistical information and frequency. Earlier grammarians only observe that the agentless passive is more frequent than the agentful passive in English, but his study gives strong empirical support to that observation. His statistics regarding the use of the types of the passive clause also throws light on another important feature of the passive. He gives us a better understanding of the functions of the passive in English by giving statistics of the frequency of types of the passive, of the occurrence of definite and indefinite subject, and also that of the length of subject and agent phrases, iii) A third advantage is that Svartvik's (1966) study gives empirically supported findings about the use of the passive in actual performance. It can be a model for describing the passive in actual use in a non-native variety like Indian English, so that one can find out what to teach. Dagut (1985) stresses the importance of a teaching

66 grammar, especially for a teacher of EFL. He points out that as the learner tries to attain communicative competence, the teacher's grammar should focus on performance, especially semantic and pragmatic performance. The reason is that what is important for the learner, and therefore for the teacher, is not a particular structure as such, but the function's) which the structure performs. Therefore, for the teacher and the learner, it is not the 'derivation' of a form or structure that is pedagogically relevant, but the use to which they may be put. Dagut (1985) describes the teaching grammar as a "'performance' grammar, taking surface forms as its input and their functioning as its output, with no attempt to discover any underlying uniform system of English structure" (p. 3). Dagut (1985) argues that the agentless use of the passive is so common that it can be regarded as the essential meaning of the form, a fact which is obscured by the conversion approach. The point that is to be noted here is that in order to describe a non-native variety like Indian English, the performance of the users has to be taken as the basis for the description. Svartvik's model (1966), which is an analysis of the actual performance of using the passive, can be used for describing the passive in Indian English. If one wants to find out the similarities between the passive in native and non-native varieties of English, a model like that of Svartvik (1966) is useful. Theoretical approaches like various versions of the transformational grammar give a clear theoretical understanding of the passive and also help us understand why an unacceptable construction is unacceptable. But, as Granger (1983) points out, transformational grammars have not used any empirical support. As Singh (2005) pointed out in one of his lectures, non-native Englishes should not be described by comparing native speakers' competence with non-native speakers' performance. Therefore, Svartvik's model (1966) is useful for comparing the use of the passive in Indian English with a native variety like British English. It is not that Svartvik's study (1966) is devoid of any shortcomings. The following are some of the lacunae of his model. i) Svartvik (1966) does not even mention non-finite passives. Biber et al. (1999) and Stein (1979) give types of non-finite passives and also statistical information about the frequency of the occurrence of every type. Svartvik (1966) concentrates only on the finite passives. ii) As Femalld (1977) points out, Svartvik (1966) puts forth the notion of serial relation between active and passive, but he does not develop the point much.

67 1.9 Conclusion This chapter is aimed at malcing a brief survey of some of the important views on the passive in English and other matters related to it. The relation between active and passive has been described. Various definitions of the passive and classifications of the types of the passive have been taken into consideration. There is little agreement among grammarians about what the passive really is and what its types are. No model of analysis may be perfect, but Svartvik's model (1966) appears to have some advantages over others. For reasons given above, it is chosen for analysis in this work.

68