Abstract

This article presents a study of the evolution of the passive structure ser + past participle in Mexican Spanish during the Colonial Period, following the frequency and character of the structure previous to its exportation from the peninsula (based on a control corpus from the end of the XV century) and throughout the three centuries leading up to the independence in 1810. The study purports to show a trend of diminishing frequency in the use of the passive structure with ser, manifested overall in the language. The attempt at ascertaining the period when the changes took place, has yielded statistically significant data pointing to the XVI century as the crucial turning point in the demise of this structure. The paper further describes the structure both morphosyntactically and semantically, so as to aid the understanding of its function in the language, and offers a hypothesis of how and why the change took place. The findings are also compared to other approaches to the passive voice structure in Spanish.

1. Introduction

Although the first impression upon reading a medieval Spanish text may result in the belief that, comparatively, the passive structure with ser had died out by the time Spanish spread to the Americas, a statistic analysis of its frequency of occurrence in texts shows that the decrease is not significant over the period spanning the XII to XV centuries, but is remarkable during the first century of the Colonial Period. It is important to ascertain the modalities of this decrease in usage, given that historical changes often emerge as alternations in frequency of occurrences, and not just as structural modifications resulting in loss or acquisition.
The present study focuses therefore on the XVI century, the period of more dramatic decline in usage of the form: control corpora from medieval Spanish and the subsequent centuries of the Colony (XVII, XVIII and beginning of the XIX centuries) where analyzed in order to isolate the period of major change. The main corpus itself is composed of a great variety of documents from the XVI century: official letters from Cortés to the Royals of Spain, private letters from the colonial settlers to their families in Spain, wills, laws and official decrees, theatrical pieces, historical writings, records of business transactions amongst others. The control corpora also include prose fiction and the first newspaper articles from the beginning of the XIX century (Yucatán region). Most documents come from the highland region of central Mexico, i.e. from the area now comprised in the states of México, Puebla, Guerrero, Michoacán and the Federal District (Mexico City ).
The variety of documents was aimed at proving that the decrease in frequency of use of the passive structure was a tendency manifested throughout all registers of the language, or at least those that could be monitored by means of written documents. It was also important to observe any stylistically marked differences, wherever possible. Although there are some minor stylistical differences, they are due to various motivations: they may depend on the type of document analyzed, since more elaborated, narrative prose may recur to the passive as one of its many stylistic devices or on individual preferences; thus a writer like Cortés, whose knowledge of Latin is well attested, seems to favour the structure well above the average of his contemporaries, whereas Diego de Ordáz, an army captain writing in the 1520's, never employs it unless it be in set, lexicalized phrases. These differences account for a high standard error, but do not skew the overall tendency towards obsolescence that the structure unndergoes.
Sampling of documents was set at 10,000 words per document or per group of documents, basing this count on the statistical significance of the trend followed by the curve in documents where passives were noticed to be scarce. After 10,000 words the average frequency of passives did not vary significantly, so the cutoff point was established on samples of that size. Of course, this holds all the more for documents where passive structures are more frequent.
The significance of a diachronic analysis of the demise of this structure lies in the fact that it allows to ascertain changes in the form and function of the passive, aside from the quantitative evaluation of its frequency of use. Thus, by noticing changes in the sintagm itself, such as: presence and type of participants, characterizing them both semantically and morphosyntactically, possible intervening elements between auxiliary and past participle, introduction of the agent by different prepositions, amongst others, it may be feasible to venture hypotheses about the modality and the reason for the significant change in the use of the passive with ser. This tendency is still manifested nowadays ­ or only partially reversed by newspaper language and scientific articles ­ as noticed by Gili Gaya (1991), when he speaks of the "repugnance" of the language for the use of this structure.

2. Decline of the passive with ser + past participle

The frequency of appearance of a passive structure with ser would be about 0.006, i.e. one passive per 166 words on average, in the Mio Cid (1140 ca.). This number declines steadily throughout the Middle Ages, until Spanish is brought to the New World. The point on the graph (Fig. 1) marked as 1490 ­ thus an idealized reference point in peninsular Spanish before it spread to America ­ is perhaps not completely neutral as to frequency of the structure in written texts of that period, since it stands for the reading given by a theatrical piece, F. de Rojas' La Celestina, and therefore mimicks spoken dialogue to a certain extent: passives may have been used more sparingly than in a narrative text. It can be noticed, in fact, that some of the documents from the Colony (notably Cortés' letters to the Spanish Royals, for instance) show a higher frequency of use of the structure than was registered even in the XII century. In fact, in many of the graphs, a distinction will be drawn among four groups of data: La Celestina represents Spanish at point 0 of the analysis, before it was exported to America; Cortés represents the prototypical and conservative literary use of the passive structure, while still belonging to the early part of the XVI century; other miscellaneous documents from the first half of the century were grouped together (in the graph "DLNE 16.1"), as well as comparable documents from the second half of the century ("DLNE 16.2"). In the first graph, the data from 38 documents are shown.

However, the linear regression shows that a significative tendency towards a change in frequency was registered throughout the century, although the standard error varies considerably from document to document. The overall tendency towards a lower frequency of use in the second part of the century is shown by the fact that two thirds of the documents in that period are below 0.003, i.e. one passive every 333 words approximately, and the majority of those are even below 0.002, i.e. one passive every 500 words. Remarkably, ca. one third of the documents in the second part of the XVI century show no use of the passive at all and in the following two centuries, the frequency lowers and steadies itself around values of one passive every 800 words (XVII/XVIII, i.e. 0.00125 if it were graphed) to 2500 words (late XVIII and early XIX century, 0.0004), except for entire texts where it is never used.
The graph in Figure 1 should perhaps be visualized as a window of a larger graph showing a logistic type function (Kroch 1989) where the tendency is constant in the centuries XII, XIII, XIV and XV, possibly decreasing steadily but not drastically, then the data within the XVI century record a sharp decline in the frequency which is then steadied in the following centuries and tends to zero. The change in Colonial Spanish can therefore be ascribed to the XVI century; it is interesting to notice that, comparing these with Sepúlveda's findings (1988), peninsular Spanish was more conservative in presenting the same tendency a century later, i.e. during the XVII century.

3. Function of the passive structure with ser

When analyzing the passive structure from a functional grammar perspective (Siwierska 1991), which bases its theory of syntax on semantic and pragmatic functions of the elements in the clause, it is important to notice that the predicate frame, i.e. the arguments in a transitive event, remain semantically unchanged. What changes in the switch from active to passive voice is only the perspective, the pragmatic salience of the participants involved in the clause. What characterizes a transitive event (Givón 1994; also cf. Dik 1989, Foley and VanValin 1984) is the presence of: an agent, which, in a prototypical transitive clause, involves a volitional, controlling participant, actively responsible for the event (salient cause); a non volitional, inactive, non controlling patient registering the changes of state of the event (salient effect); a certain type of verbal modality, encoding a compact non durative event, completed and realis, thus the noticeable disfavour for imperfective tenses, negative or subjunctive clauses.
Thus, the passive with ser should be seen as an implementation of this shift in salience of the arguments. In these terms, it could either denote an upgrading of the patient, by promoting what is syntactically an object to subjecthood; or a demotion of the agent, pushed from syntactic subject to the periphery of a prepositional phrase or suppressed altogether. Typically (Shibatani 1985) the passive is defined purely in terms of suppression of the agent, whereas Givón (1981) assigns three main functions to this structure: topicalization of a non­agent, suppression of the agent and making the verb less active.
It is important to consider exactly what functional ground this structure covers in the variety of Spanish considered here, since it is well known that other passive structures were available in the language, such as the impersonal se passive and the one with estar + past participle, and competition from those structures may have played a role in the obsolescence of the ser form.

3.1 Semantic and morphosyntacic represenation of the structure and its participants

3.1.1 Presence of agent and patient

Considering that any language has more than one strategy to cover a certain functional domain, and Spanish has at least another two "passive" constructions, an analysis of the participants in a transitive event may help ascertain the function of ser + past participle.
The prototypical passive structure showing both agent and patient, traditionally called primera de pasiva, is a rare occurrence: it only appears complete in 17.6% of the instances, the average being lower (11% ca.) in the XVI century DLNE documents overall (cf. Fig. 2), but higher (over 30% of the times) in Cortés' documents, showing a markedly conservative use of the structure.
It is indeed an agent­suppressing structure: the agent is only present on average 22.4% of the times the passive is used, and again the values of Cortés are markedly higher than the average for the whole century (a mere 13% ca.) and for the Celestina, touching 35.2%.

1) agent preservation / Cortés #83

Y puesto que hubo algunas contradiciones... entre un hijo bastardo del señor natural de la tierra, que HABÍA SIDO MUERTO POR MUTEZUMA...

"And since there was some disagreement... between an illegitimate son of the lord of that region, who had been killed by Moctezuma..."

2) agent suppression / DLNE 16.1 #17

...y en los tormentos no confesó saber dello nada y FUE CONDENADO por el escándalo y otros delitos a pena de muerte y esecutose la sentencia, no embargante que apeló.

"...and during the torture he did not confess that he knew anything about it and was sentenced to death for the scandal and for other crimes and the sentence was carried out in spite of his appealing."

This seems to go against the claim made by Raquel Hidalgo (1994) for modern Spanish, that the structure is agent­preserving: her data did not show a significative difference between preservation and suppression of the agent. On the other hand, in a comparison with the estar + past participle and the se forms (which are clearly agent suppressing), the canonical form where the agent can be maintained, but need not to, is the form with ser; nonetheless, in the data analyzed for the present study, agent suppression is registered significantly more often than its preservation.
The presence of the patient is overwhelming, varying from ca. 60% to 90%, but averaging 77.6% of all occurrences of the structure, which seems to indicate that another function of this kind of passive is patient promotion. On the other hand, Hidalgo maintains (ibid.) that there is not enough evidence for patient promotion in her data, which are drawn from contemporary peninsular Spanish.
From this first analysis it can therefore be gleaned that the passive with ser, at the moment previous to its drastic decline, had both a patient promoting and an agent suppressing function.

3.1.2 Types of agent and patient

In accordance with the prototypical nature of the agent as a volitional initiator of the action, it should not surprise if the tendency is for it to be an animate NP: this is what is predicted by theories of attention flow and saliency in discourse, for instance, Siwierska's personal hierarchy (1988) foresees that any of the following elements are respectively more likely to be subjects/agents or topics of the action expressed by the predicate than those that follow them:

1st < 2nd < 3rd person human < higher animals < other organisms < inorganic matter < abstracts

Thus it can be evinced from the graph (Fig. 3) that in those instances when the agent is expressed, it tends to be human (or animate at least ­ the latter part of the century may show a spurious tendency towards inanimate agents due to the type of documents chosen for DLNE 16.2). Since the appearance of the agent is marked, within a marked construction, it should not surprise if it tends to be expressed by a full noun phrase (NP) rather than by a pronoun: if the agent needs to be set off somehow in a structure that is usually agent suppressing, it will not be rendered by a pronoun, which would refer back to some participant still salient in discourse, but by a full NP, as a way of embossing the relevance of it being mentioned in such a position.
It is more difficult to draw any conclusions about the morphological and semantic nature of the patient. Overall, the patient is also tendentially human, but this may be due to the fact that topics of clauses in general may tend to be human, be they in active or passive clauses, in accordance with Siwierska's hierarchy (1988). It also does not seem to vary significantly in form, except that it is more common to find clausal patients (i.e., noun phrases modified by a relative clause) than it is to find clausal agents (cf. Fig. 4).

3) clausal patient / DLNE 16.2 #7

...para que no le SEA DICHO lo que dize a los injustos rreyes que ipse rregnaverunt...

"...So that what he says to the unjust Royals that reign there won't be said to you ..."

It has, however, been noticed elsewhere (Hidalgo 1994, Sepúlveda 1988) that the passive structure chosen whenever the participants are inanimate is the se structure; despite the discrepancy caused by the Celestina, it may then be stated that the prototypical ser passive presents animate (human) participants, especially as far as the agent is concerned.

3.2 Correlation between the nature of the agent and the preposition introducing it

Figures 5 and 6 show an interesting comparison between the prepositions that introduce the agent, de and por. The percentages of occurrence presented significative differences once passed through a chi­square test: it was noticed then, that de tends to be used more often with pronouns and por with full NP's, de is associated with animate agents and por with inanimate ones, perhaps a heritage from the Latin preposition denoting an instrumental function or the cause of the event, rather than the agent.

4) de and animacy / Cortés # 44,

Se halló el pueblo... como si nunca HUBIERA SIDO HABITADO de persona alguna.

"The village was found... as if it had never been inhabited by any living soul."

5) por and inanimacy / Cortés # 171

Crean VV. RR. AA. por cierto que esta batalla FUE VENCIDA más por voluntad de Dios que por nuestras fuerzas.

"Your Royal Highnesses should certainly believe that this battle was won through the help of God, rather than by our means."

If favouring animate agents over inanimate ones can be ascribed to the overall tendency in transitive events for the initiator to be human, or at least animate, it should be noticed that more often, the agent is expressed by a whole noun phrase; this is in accordance with discourse factors that contribute to the functional load of the passive (Quirk and Greenbaum 1973:410). There is a tendency to place new information towards the end of the clause (end­focus): since the structure is usually agent­suppressing, occurrence of the agent is highly marked and when it is expressed, it is mostly new information; it is therefore not surprising that the agent is tendentially expressed by a full noun phrase: if it were a pronoun, it would be bound in reference and therefore to information from previous clauses. The second tendency that may be useful to understand why agents are placed where they are, and tend to be full noun phrases, is the one that reserves final position for the more complex part of a clause (end weight). Therefore, given that syntactic structure can be manipulated, although argument structure of predicates does not change, the passive in question is a way of reversing the canonical order of the arguments in an active clause, in order to draw attention to specific parts of information, while observing the discourse principles of end­focus and end­weight.

6) agent and end­focus/end­weight / DLNE 16.1 # 48

... como a esta Nueva España llegué, ize rellación a vuestra magestad... segund lo que pude sentir y alcançar desas cosas y segund SOY INFORMADO de personas de mucho crédito y de los más antiguos conquistadores... y pobladores de la tierra...

"... as I reached the New Spain, I wrote a report to your majesty... according to what I got to hear and understand of those things and according to how I was informed by very respected people and by the earliest conquerors... and settlers of the land..."

4. Modality and possible cause of the change

4.1 Two tendencies: formulas vs adjectives

It should be remarked that some verbs seem to be used more often than others in the passive with ser. Thus lexemes such as hacer ("to do"), informar ("to inform"), loar ("to praise"), obligar ("to oblige"), recibir ("to receive"), servir ("to serve"), figure much more prominently in the data base than others. While in certain cases we find a fully fledged passive with these verbs,
7) Cortés # 67

De este Gerónimo de Aguilar FUIMOS INFORMADOS que los otros españoles... se perdieron...

"By this Geronimo de Aguilar we were informed that the other Spaniards... had gotten lost..."

in other cases, it is difficult not to dismiss the instance at hand as merely formulaic, fixed and lexicalized, rather than forming part of the active use of the language; such is the case, for example, with the mostly expunged verbs bendecir ("to bless") and maldecir ("to curse"); they were kept as actual passives whenever either the patient or the agent were expressed in full.

8) DLNE 16.1 # 11

...mas a Dios le plaze que assi sea, SEA él BENDITO, y en esto no quiero más dezir...

"..but it pleases God that it should be so, may he be blessed, and I'll say no more about this..."

9) Celestina # 104

...que MALDITO SEA el diablo y mi pecado...

"...cursed be the devil and my sin..."

Typically, however, these forms are agent­less. For the most part they are reverential forms used to address figures of great religious or political authority, thus they are often used for God and the Royals. It is also significative that those authors, such as Ordáz, who very rarely use the passive at all, retain the use of these formulaic passives: if they use a passive, it is most likely to be in one of these formulas. It is clear that such formulas have a limited usage and are to be found in certain specific kinds of prose texts only. Since the passives one finds, especially after the XIV century, are often connected to such formulas, it is clear that when the formulas stop being employed, a significant decrease in the use of the form will be recorded.
Another taxonomic problem that may shed some light on the reasons and modalities of the change is presented by those forms whose past participle may be used as an adjective, thus honrar ("to honour") is a case in point:

10) honrar as verb, DLNE 16.1 # 39

...hiso que se hiziesen onrrar por él (agent) con su lucto en sant Francisco...

"...he set it up so that they would be honoured by him through his mourning at St. Francis' (church)..."

11) and as adjective (ibid.)

... una muger honrrada que tenía su marido en compañía de don Hernando...

"... a respected woman whose husband served under don Hernando..."

but in the following case, it is difficult to make a decision:

12) Cortés # 58

...diciéndoles como todas estas partes... eran de Vuestra Alteza, y que los que quisiesen ser sus vasallos, SERÍAN HONRADOS...

"...telling them that all these territories... belonged to Your Highness, and that those who would become his vassals would be honoured..."

It should therefore be stated that, while both context­based decisions and formal characteristics of the structure helped discern which items should be stored in the analyzed corpora, the ambiguity was certainly present at the time, as it still is in contemporary Spanish: La edición fue reducida ("the edition was reduced") can mean either that the edition did not comprise a large number of copies (adjective) or that an active decision to reduce the number of copies was taken by an unmentioned agent (past participle). This ambiguity may have led to a partial reanalysis of the passive with ser, as a copula + an adjectival form. Thus the patient, as subject of the copula would be present, but the agent would no longer be necessary, or even acceptable, in some forms.

4.2 Grammaticalization and discontinuity

In various examples, the structure composed by ser + past participle forms a discontinuous constituent, such that some element intervenes between the auxiliary and the lexical participle:

13) DLNE 16.2 # 43
[--- Unable To Translate Graphic ---]

... SIENDO mis hermanos Francisco Sanches y Benito Gutierres SERVIDOS...

"... being my brothers F. S. and B. G. served ..."

As can be seen from figure 7, various syntactic elements can act as "discontinuity" in this structure: most commonly either semantic participant of the passive, an adverb or adverbial phrase ("adverbial locution" in the graph). In contemporary Spanish, the structure has become welded and does not allow for anything to intervene between its constituents, so much so that the ambiguity alluded to above about la edición fue reducida, can be solved by introducing an adverb between auxiliary and past participle: la edición fue muy reducida can only mean that the edition was very limited, not, say, that it was drastically reduced; in this case, that of the verbal meaning of reducida, the adverb could only be added after the past participle: la edición fue reducida mucho.
In fact, the following four graphs intend to show an increasing lack of diversity in the elements that may intervene between the auxiliary and the lexical participle: in the Celestina passives, almost all syntactic elements marked on the graph could interrupt the structure, but by the end of the century, only patient and manner adverbs are found in that position.

[--- Unable To Translate Graphic ---]

What intervenes most often are adverbs of manner such as bien, mal etc., especially with those verbs that will later incorporate them and become compounds (cf. maltratar "to mistreat") or fixed phrases:

14) Cortés # 167

...que yo les prometía ...siendo ellos leales vasallos de v.a.... SERÍAN de mi muy bien TRATADOS...

"... I promised them...,since they were your highness' loyal vassals, that they would be by me very well treated..."

Becoming a compound verb is a sign that the structure had a fixed word order for a considerable period. This fossilization of the structure, especially with specific lexemes that eventually turn into whole formulas, and more generally the loss of syntactic independence of an item becoming a bound morpheme, may be described as a process of grammaticalization (Heine et al. 1991, Hopper and Traugott 1993). The evolution of the form may originally have corresponded to the need for an expressive way to organize information in discourse, so that the semantic role of the arguments would be respected at the cost of their syntactic function (promotion of patient to subjecthood, demotion of agent). As pragmatics becomes more tightly encoded in syntax (Heine et al. 1991:13), the structure flourishes in its fully fledged form, but patient promotion is its main role, thus the frequent suppression of the agent contributes to a partial reanalysis (Hopper and Traugott 1993:41) of auxiliary + past participle as copula + adjectival form. The kind of "linguistic cycle" described in Heine et al. (1991:213ff.) would further predict a path towards morphologization of the Spanish passive, stages of which are indeed attested in the compounded and fixed forms mentioned above (cf. maltratar, bien tratar), as well as in the formulas (ser servido). This kind of passive is, however, not as frequent as would be required for it to further proceed on the cline towards grammaticalization, where phonological erosion and further welding of the parts into morphological units would be expected. Rather, because of its loss of expressive power, once these minor functional changes have taken place, such as tendency towards adjectival reanalysis due to the lack of the expressed agent, and given the competition from the other two "passive" alternatives with estar and se, the form tends to remain as a quaint literary rarity. The loss of expressive power as the form becomes more unwieldy through grammaticalization processes, the decrease in frequency caused by the disuse of the formulaic expressions where the form was most often found and, last but not least, the competition from the other forms may have caused the rapid decline of the passive attested in Mexican Spanish in the XVI century.
Contemporary studies on the acquisition of Spanish seem to show that it is not an early form actively acquired by children (Slobin 1994:355). In terms of a "Principles and Parameters" (Lightfoot 1991) approach, another way to describe the relatively fast demise of the passive would be to say that the above mentioned factors ­ most of all, the competition from the other forms ­ cause a period of obsolescence of the structure with ser. As some fixed forms become increasingly lexicalized and others are analyzed as adjectives, the passive with ser stops being an active "trigger" for the acquisition of the structure by the younger generations; this would in turn provoke the drastic and fast decline in use of this form.

5. Conclusion

In this paper I have shown that a change in the frequency of use of the passive ser + past participle in Mexican Spanish did occur and may be isolated as taking place during the XVI century.
The morphosyntax and the semantics of the structure seem to indicate that its function in the language involves both the promotion of the patient and the demotion of the agent, a reversal of syntactic roles which may originally have been dictated by discourse­based needs. However, the ser passive is seen as crystallizing around recursive patterns that become more unwieldy and less expressive as they become more fixed in syntax. This grammaticalization process is paralleled by the rise in frequency of competing forms, specifically the impersonal passive with se (Sepúlveda 1988), which cause obsolescence of the form, and ultimately its demise from the lively part of the language. This evolution of the passive is followed by the more conservative peninsular Spanish a century later (ibid.).

Acknowledgements

I wish to thank Concepción Company for helpful discussions and Chantal Melis for her insight, support and complicity. My thanks also to the audience at the Language South of the Rio Bravo conference in January 1995, Tulane University, for useful and constructive comments.

Bibliography

Acuña, R. 1986. Relaciones Geográficas del Siglo XVI: México. Mexico: U.N.A.M. (vol.2)
Company Company, C. 1994. Documentos Lingüísticos de la Nueva España. Mexico: U.N.A.M.
Cortés, Hernán. 1963. Cartas y Documentos. Mexico: Porrúa.
Díaz del Castillo, Bernal. 1962. Historia Verdadera de la Conquista de la Nueva España. Mexico: Porrúa.
Dik, S. 1989. The Theory of Functional Grammar. Part I: The Structure of the Clause. Dordrecht: Foris.
Foley, William A. and Robert D. Van Valin, Jr. 1984. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press.
Fox, B. and P. J. Hopper. (eds.) 1994. Voice: Form and Function. Amsterdam/Philadelphia: Benjamins.
Gili Gaya, S. 1991. Curso Superior de Sintaxis Española. Barcelona: Bibliograf.
Givón, T. 1981. "Typology and Functional Domains", Studies in Language 4(3):333­377.
Givón, T. (ed.) 1994. Voice and Inversion. Amsterdam/Philadelphia: Benjamins.
Heine, B., U. Claudi and F. Hünnemeyer. 1991. Grammaticalization. Chicago: The University of Chicago Press.
Hidalgo, R. 1994. "The Pragmatics of De­transitive Voice in Spanish" in Givón 1994, 169­186.
Hopper, P.J. and E. Closs Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press.
Kroch, A. 1989. "Function and Grammar in the History of Periphrastic do." in R. Fasold and D. Schiffrin (eds.) Language Change and Variation. Benjamins.
Lightfoot, D. 1992. How to Set Parameters: Arguments from Language Change. Cambridge, Massachusetts: M.I.T. Press.
Márquez, J. (ed.) 1986. Clamores de la Fidelidad Americana Contra la Opresión. Mexico: U.N.A.M.
Miglio, V.G. (1995). Vitalidad de la pasiva en el Español de México: Estudio Diacrónico. unpublished M.Phil. thesis, U.N.A.M., Mexico.
Rojas, F. de. [1494?]1991. La Celestina. Madrid: Cátedra (editor: D. S. Severin).
Sépulveda Barrios, Félix. 1988. La Voz Pasiva el el Español del Siglo XVII. Contribucion a su Estudio. Madrid: Gredos.
Shibatani, M. 1985. "Passives and Related Constructions", Language 61(4): 821­848.
Siwierska, A. 1988. Word Order Rules. London: Croom Helm.
Siwierska, A. 1991. Functional Grammar. London: Routledge.
Slobin, D. 1994. "Passives and Alternatives in Children's Narratives" in Fox and Hopper 1994, 341­ 364.
Quirk, R. and S. Greenbaum. 1973. A Concise Grammar of Contemporary English. San Diego: Harcourt Brace Jovanovich.