A TYPOLOGY OF NON-CANONICAL PASSIVE STRUCTURES: PRELIMINARY STUDY OF BORDERLINE CASES Lorena Barrera Fernandez and Carlos Galego Guitán Universidad de Santiago de Compostela Abstract: The following paper presents a discussion of the discourse factors affecting the choice of non-canonical passive structures, namely passive sentences with stranded prepositions, the become passive and verbal vs. adjectival passives, placing special emphasis on the differences among them. The adopted perspective is synchronic, exploiting a collection of spoken and written Present-Day English texts: the International Corpus of English in its British component (hereafter ICE-GB). We identify 4180 passive examples in spoken language and 6010 in written texts, and we classify them into groups. Illustrative points and explanations are provided in a qualitative and quantitative manner, and discursive-cognitive principles are also taken into account. Our aim is twofold. On one hand, we will attempt to examine the syntactic and discursive reasons that justify the use of this passive variant in actual discourse. On the other hand, we will try to gain insights into the present state of affairs in the world of syntax and discourse analysis according to leading studies in the field (D. Biber et al. 1999, A. Downing and P. Locke 2002, A. Hewings 2005, R. Huddleston and G. Pullum 2002 and B. Jonhstone 2002). Finally, we will bring to light some remarks gathered from previous research in order to draw a number of concise and clear-cut conclusions. Keywords: non-canonical passive structures, passive sentences with stranded prepositions, the become passive and verbal vs. adjectival passives, discursive-cognitive angle. 1. Introduction In section 2, we will offer a brief presentation of the Corpus employed in our study to familiarize the reader with its structures and data collection procedures. Section 3 will define, describe and characterize borderline cases of passive sentences according to the current major research. Section 3 will also provide our primary results based on the assumption that, whenever there is a grammatical reason to use a passive clause, the reason is reflected in discourse. In section 4, we will conclude by providing some practical reflections deriving from the previous research. 2. The Corpus Sidney Greenbaum set out to exploit discourse phenomena in corpora and committed himself to the creation of the International Corpus of English, a research process that took place in the late 1980s. ICE-GB by G. Nelson (1998) samples the English used by educated British adults in the 1990s. Adult refers to those people aged eighteen or over. The term educated describes the speakers and writers in the Corpus, not their English. British refers to speakers and writers who are native to England, Scotland or Wales. All samples in the Corpus date from 1990-1993. ICE-GB contains a million words of speech and writing that have been fully tagged and parsed. The most highly represented genre in ICE-GB is spoken language (3000 texts) followed by 1
written language (2000 texts). Our analysis of passive constructions in ICE-GB covered a total of 4180 examples in spoken language and 6010 in written texts. 3. Non-canonical passive types: passives with preposition stranding, the become passive and verbal vs. adjectival passives Apart from the well-known passive types that use be and get as auxiliaries, we must differentiate any other types which could be said to establish a binary opposition in the syntax of the clause in which one is the unmarked counterpart (the active clause) and the other is marked (the passive clause), as A. Downing and P. Locke (2002), M. Á. Gómez-Gonzalez (2001), M. Halliday (1985) all point out. Constructions in English, as in other languages, are ranked on a scale of markedness according to their frequency in speech. T. Givón (1993: 178) refers to this as discourse distribution. In other words, the more marked a construction is, the less frequently it is used. Speakers generally rely on structures that occur in ordinary speech and are easy to produce. T. Givón (1993: 178) calls the complexity of marked constructions structural complexity and adds another argument to account for markedness, namely cognitive complexity. The latter argument asserts that the more marked a structure is, the longer it takes to be invoked by the mind. Therefore, its processing takes more time than that of a corresponding unmarked structure. One such type is the passive sentence with preposition stranding. Although the use of prepositional passives with stranded prepositions has been increasing until PDE, in ICE they are hardly registered. These prepositional passives are a type of passivizable form often found with prepositional verbs or in structures where the preposition is lexically determined by the verb. The close lexical attachment between verb and preposition makes their relationship comparable to a single verb: the Complement of the preposition is treated like a DO. The lexically related elements must be syntactically adjacent, i.e. there cannot be an object between them. Such an object can exist in the case of Phrasal verbs, however. Let us observe an example of this passive type: (1) The first requirement is catered for by tons of packages. (B, S1B-076#45:1:B) 1 As this instance clearly illustrates, the preposition is closely linked to the verb in the sense that it is needed to form the passive voice without omitting it in the shift from the active to the passive clause. Such a shift entails a structural reanalysis whereby the preposition for enters into constituency with the verb catered rather than with the following NP: from V [P + NP] into [V+P] NP. In other words, through reanalysis, the preposition is dissociated from its NP Complement and becomes strongly associated with the verb, thus forming a single syntactic unit. As a consequence, the NP following the preposition in the active clause of (1) tons of packages is considered a verbal Complement and, therefore, can be promoted to Subject position. Passive forms with preposition stranding pose as the unmarked term in the thematic system, and their marked counterpart is preposing. This is true even though they also represent the marked pattern in the voice system shown above. The former is typically found in informal styles whereas the latter tends to be used formally and for emphasis as in other allosentences. R. Huddleston and G. Pullum (2002: 630-1) cite a list of circumstances in which preposition stranding is inadmissible. As far as the spoken part of ICE is concerned, the verbs that appear with greatest frequency are: designed to and agreed upon (parliamentary debates), referred to, operated on (legal cross- 2
examinations), dealt with, catered for (business), covered up and subjected to (spontaneous commentaries), thought of, taken care of (unscripted speeches), preferred to (legal presentations), dealt with, gone into and paid for (broadcast talks). In writing, the only verbs used with this constructions are approved of (AD), dealt with (H), spoken to (SL) and catered for (B). Passive forms with preposition stranding are truly scarce in all text-types within the Corpus. B and SP (5.9% each) comprise the highest percentages in the spoken medium. The latter constitutes an informal field, and its topics cover common subjects in which preposition stranding is more likely to occur. Business transactions (B) deal with the trading market between two shoppers, including robberies that may take place. Observe the following two examples below: (2) Then comes Seagram then being trapped in. (SPC, S2A-005#166:4:A) (3) There he was subjected to <, ><,>. (SPC,S2A-011#6:1:A) Both sentences are framed within the category of spontaneous commentaries that, as the name indicates, are spontaneous, unprepared and represent unconscious thoughts or opinions of the speakers. Thus, this text-type is characterized as informal, and that is why the highest percentages of stranded prepositions are registered here. Stranded prepositions are far less common in written form. The highest percentage corresponds to SL (1.57%) and NAWSSC (1.6%), both of which are illustrated in the examples below: (4) So later on I phoned to complain to the police who after all had told me she wouldn t be spoken to. (SL, W1B-007#46:2) (5) This episode will be dealt with later, in its proper place. (NAWSSC, W2B-015#61:1) In example (4) the event must be important since the police are involved. In (5), the Subject presumably has to assume his/her responsibility in facing an episode. The correlation lies in the fact that both SL and NAWSSC are informal registers. That is why the forms in question pose as the most commonly used. In addition, the felicity of these passive types depends on whether the action depicted expresses an outstanding change or feature of the Subject (D. Biber et al. 1999 and R. Huddleston and G. Pullum 2002: 630). In example (4), there is an action in progress (speaking) that has been initiated by the Subject. Similarly, example (5) presents an ongoing action and a change from the need to deal with the episode to actually dealing with the episode. Furthermore, there is a principle in English discourse asserting that the most important part of information (whatever the speaker wishes to highlight) should remain at the end of the clause. As such, language users seek to emphasize the verbal action above all else, which is why they use the passive voice rather than the active counterpart (M. Celce-Murcia and E. Olshtain 2001, A. Hewings 2005 and B. Johnstone 2002). Aside for the arguments set forth above, there seem to be no apparent reasons for the tendency to use passive sentences with preposition stranding. Statistics prove to be very low for all the domains, and it is worth remembering that stranding usually appears more in the oral form than in the written counterpart for the cases mentioned throughout this section. We must not forget that stranded prepositions still embody a borderline case, and sometimes it is easier for language users to resort to canonical passive constructions that employ be and get as auxiliaries. The become passive is another type of passive structure worth noting. It bears a resemblance to the get passive but poses as an alternative to it. The main difference is that, 3
unlike get, it is usually not modified by adverbs denoting rapidity. Instead, it expresses an action. Some scholars assert that get expresses the result of an action. Both types deal with ingressive acts and indicate a durative and perfective aspect (L. Barrera-Fernández 2004, R. Quirk et al. 1995 and A. Siewierska 1984). Some illustrative examples are: (6)... complications can occur when the air masses become mixed up or approach from an unexpected direction. (NAWNSC, W2B-026 #59:1) (7) Anyone who questions the beliefs or actions of the leaders becomes defined first as disruptive, then as disloyal. (AWSSC, W2A-012#78:1) This passive type is mainly used by speakers when the situation they are referring to is dynamic or still in process, unlike the be passive. The become passive does not occur frequently, being restricted to less than the five percent of the whole Corpus, both in the oral and written forms. We shall now focus on other passive types, namely adjectival vs verbal constructions. Special care should be taken so as not to confuse a verbal (actional) passive with an adjectival (statal) one. Verbal passives present an auxiliary verb, namely be or get, determined by the past participle that follows. By contrast, adjectival passives are made up of the auxiliary copular be followed by an adjective. We will start by considering that not all adjectives are prototypical adjectives. In other words, not all end up in y or canonical inflectional suffixes. Some end up in en or ed inflections which are concomitant with the past participle of irregular and regular verbs respectively. This point is illustrated in several examples below: (8) The Establishment that still enforced the rule that divorced personages were not welcome in the Royal Enclosure at Ascot felt the line-up at the premiè re should not include Lynne, as she and Peter were not yet married. (NAWH, W2B-004#28:1) (9) She agrees to their plan because she is pregnant by Butch and wants the stolen money so that they can get married. (NAWH, W2B-009#122:1) (10) I mean one bloke did get married from our course. (DC, S1A-014#157:1:B ) In example (8), married can easily be identified with the past participle of the verb marry or with the adjective. The possibility of classifying married as a verb or an adjective is ruled out with the presence of get in (9) because of the sense of action it conveys. The same applies to (10). Verbal and adjectival passives do not seem conducive to being explained in terms of discursive arguments. Fortunately, there are several well-grounded syntactic criteria that help discern whether a given construction is verbal or adjectival. Some of these criteria are listed below: (i) Adjectives are gradable and may have comparative and superlative forms, but verbs are not. Furthermore, adjectives take the prefix un. In example (8), married is not gradable (*very married). (ii) Adjectives can function attributively (premodifying a noun) and predicatively (as postmodifiers). Married fulfils this criterion. (iii) Adjectives usually denote a permanent quality, whereas verbal items express a temporary process. In this case, married in example (8) refers to a process. (iv) Adjectives can occur with copular verbs other than be, namely seem, appear and become. (v) (vi) By-Adjuncts (Agentive complements) modify verbal passives. The distribution criterion is also a very good tool in helping to classify a construction as verbal or adjectival. This criterion establishes that the slot occupied by given grammatical categories can be filled equally by other items from similar categories (paradigmatic relations). In the case under consideration, a by-prepositional phrase is presupposed by the priest, and the clause could even take other Adjuncts such as in 4
church, on Monday, in spring, etc. Here it is worth remembering that Adjuncts modify the verbal phrase. Taking all these criteria into account, married in example (8) is a verbal item since it can be replaced by get. The other arguments mentioned above are also intended to account for the differences between actional and statal passives. It should be noted that the percentages of adjectival passives are very low in comparison with their verbal counterparts. In fact, the only instance that can be conducive to an adjectival interpretation is demonstrated in example (11) below. Adjectival passives can only occur with specific past participles such as broken or married and, therefore, represent less than 1% of the entire Corpus. Observe example (11): (11) Local groupings in the long term determine genealogical structure via social flux when allegiances are constantly formed and broken. (SES, W1A-011#54:1) Taking into account the above explanation, broken refers to the action of being broken, and it is a verb. The same thing happens with formed. On the other hand, other borderline cases deserve to be highlighted: (12) As they were led off the pitch, the footballers were still able to recall the name of the play in which they had been injured. (USE, W1A-004#44:1) Able is the past participle of the modal auxiliary verb can. Hence, the possibility of classifying it as an adjective is ruled out even though it means, being gifted. At first sight, many people would analyse this structure as a copulative construction with a Predicative Complement because of its semantic meaning. However, if the previous criteria are applied, able should be considered a verb. In summary, grammar is tightly blended to discourse analysis, as the passive types we have counted and characterized clearly show. Language speakers still seek simplicity and economy in language, and that is why they opt for traditional grammar constructions that they are more accustomed to. On other occasions, this low percentage of non-canonical passive types occurs as a result of conversational topics. Some topics do not lend themselves to fitting with the verbal or adjectival passive. The same holds true for a become passive or a passive with stranded prepositions. 4. Final considerations The increasing interface between syntax and discourse analysis has encouraged us to conduct the present research. We have attempted to show that, apart from purely syntactic arguments, when a given grammatical construction is used, there is something external that can also justify its use. This external aspect lies in the discursive domain. We have ascertained that passives with stranding prepositions are seldom used by speakers and represent a very low percentage. These forms are determined by a lexical reanalysis in which the action of the verb tends to indicate an important change or characteristic of the Subject. At the same time, the End Focus principle is also at work here. This is probably because what the speaker wishes to emphasise is the action of the verb in which the preposition plays an important role. Something similar also happens with the become passive. Our study seems to clarify the fact that allosentences such as passive constructions fulfil a function in language by connecting it to the human mind. They presumably relate the 5
phenomena interwoven in language with the stimuli at work in the mind. Speakers still prefer to employ well-known traditional constructions rather than ones that are less habitual and take longer to be processed by the mind. A deeper investigation presenting new arguments that account for the use of these passive types must be conducted in order to bring the relationship between syntax and discourse analysis to a more advanced stage. Bibliography BARRERA FERNÁNDEZ, Lorena. 2004. «Get passives in ICE: a contrastive analysis to bepassives». In: Actas del III Congreso Internacional de Lingüística Contrastiva, ed. by E. GONZÁLEZ ALVÁREZ and L. IGLESIAS RÁBADE. Santiago de Compostela: Servicio de Publicaciones de la Universidad de Santiago. P. 43-9. BIBER, Douglas; JOHANSON, Stig; LEECH, Geoffrey; CONRAD, Susan; FINEGAN, Edgard. 1999. The Longman grammar of spoken and written English. London: Longman. BROWN, George; YULE, Gillian. 1996. Discourse analysis. Cambridge: CUP. CELCE-MURCIA, Marianne; OLSHTAIN, Elite. 2001. Discourse and context in language teaching. New York: Heinle and Heinle, CUP. DOWNING, Angela; LOCKE, Philip. 2002. A university course in English grammar. London: Prentice Hall. GIVÓN, Talmy. 1993. English grammar (A function-based introduction). Amsterdam: John Benjamins. GÓMEZ-GONZÁLEZ, María Ángeles. 2001. The theme-topic interface: Evidence from English. Amsterdam / Philadelphia: John Benjamins. HALLIDAY, Michael Alexander Kirkwood. 1985. Introduction to functional grammar. London: Edward Arnold. HEWINGS, Ann. 2005. Grammar and context: advanced resource book. London: Routledge. HUDDLESTON, Rodney; PULLUM, Geoffrey. 2002. The Cambridge Grammar of the English Language. Cambridge: CUP. JOHNSTONE, Barbara. 2002. Discourse Analysis. Massachussetts / Oxford: Blackwell. NELSON, Gerald. 1998. A Corpus based-study: The International Corpus of English, The British component, ICE- GB. London: University College Publications. QUIRK, Randolph; GREENBAUN, Sidney; LEECH, Geoffrey; SARTVIK, Jan. 1995. A Comprehensive Grammar of the English Language. London / New York: Longman. SIEWIERSKA, Ana. 1985. The passive: A comparative linguistic analysis. London: Croom Helm. 1 In each of the examples present in this analysis, several abbreviations referring to the text-types of the Corpus will be used. The list of abbreviations is as follows: DC stands for direct conversations, TC for telephone calls, CL for classroom lessons, BD for broadcast discussions, BI for broadcast interviews, PD for parliamentary debates, CL for legal cross examinations, B for business transactions, SPC for spontaneous commentaries, USP for unscripted speeches, D for demonstrations, LP for legal presentations, BT for broadcast talks and NBSP for non-broadcast speeches. USE stands for untimed students essays, SES for students examinations scripts, SL for social letters, BL for business letters, AWH for academic writing humanities, AWSSC for academic writing social sciences, AWNSC for academic writing natural sciences, AWT for academic writing technology, NAWH for non academic writing humanities, NAWSSC for non academic writing social sciences, NAWNSC for non academic 6
writing natural sciences, NAWT for non academic writing technology, R for reportage, AD for administrative writing, H for humanities, PW for persuasive writing and CW for creative writing. 7