Introduction

Data studied

Even within such a coarse-grained classification, there were problems in deciding on the most appropriate category for some of the papers in view of their cross-disciplinary character: in cases of doubt the information given in the journal as the authors' department was taken as decisive.

The average length of the English abstracts is 88 words (range 13-255 words), while that of the Portuguese abstracts is 91 words (range 18-235 words). Although there are some instances of divergence between the English abstract and the Portuguese abstract, for example where material is included in one but not the other, and one example of divergence so extreme that the English and Portuguese abstracts point to exactly opposite conclusions (possibly as a result of one abstract but not the other attempting to take account of a critical referee's report), in the great majority of cases the two summaries can be regarded as equivalent texts, whether as close translations or as paraphrases.

The (A)VS Structure in Brazilian Portuguese

1. É notório o interesse mundial dos economistas agronomistas e pecuaristas no desenvolvimento da bacia amazônica . . .

2. Além da caracteristicas referidas, merecem especial atenção a originação do aparelho estomático e a presença de fibras do mesófilo . . .

Fronted active verbs are in the data, however, much less frequent than fronted passive verbs. Two forms of the passive are found in Portuguese. One is cognate with the passive in English, being constructed with the auxiliary ser and the past participle of the verb. The other, the so-called Reflexive Passive is constructed with the reflexive clitic -se but is passive in function (Siewierska 1984 pp. 162-185). These will be referred to as the be-passive and the se-passive respectively. It so happens that two otherwise very similar abstracts by the same author provide a convenient illustration of the functional equivalence of these two forms (fronted passives printed bold):

3. 	BE-PASSIVE		SE-PASSIVE
   a)	Neste trabalho são apresentados os procedimentos computacionais para o cálculo de derivadas sucessivas do inverso de uma função, derivadas sucessivas de uma função elevada a um expoente real e derivadas da raiz quadrada de uma funçao, através do conhecimento dos elementos do triângulo de Pascal, utilizando-se a fórmula de Leibnitz.	b)	Neste trabalho apresenta se um procedimento computacional para o cálculo de derivadas sucessivas da divisão de duas funções, incluindo o cálculo e derivadas da forma indeterminada 0/0, através do conhecimento dos elementos do triángulo de Pascal, utilizando-se a fórmula de Leibnitz.
   	São apresentadas também as listagens dos subprogramas FORTRAN correspondentes.		Apresentam-se também as listagens dos subprogramas FORTRAN correspondentes.

The following table shows figures from the data for the numbers of fronted versus non-fronted passive verbs in free finite clauses with non-clausal subjects from the abstracts in Ciência e Cultura and from an equivalent amount of text from the weekly news magazine Veja:

	Ciência e Cultura			Veja
	Fronted	Non-fronted	Total	Fronted	Non-fronted	Total
BE-PASSIVE	44	90	124	6	39	45
SE-PASSIVE	47	4	51	9	0	9
TOTAL	91	84	175	15	39	54

The following conclusions follow from these figures:

1. In both the abstracts and in Veja the choice between fronted and non-fronted passive verbs is salient only with the be-passive. The data indicate that in written Portuguese se-passives are normally fronted, the four instances of non-fronted se-passive in the abstracts being only marginally classifiable as passives. In other words, it is only when reflexive constructions are fronted that they are unambiguously interpretable as passive (see Garcia 1975 for a similar finding for Spanish).

2. Comparison of the abstracts with the texts from Veja shows these forms of the passive to be over three times more frequent in the former than in the latter (one occurrence for every 52 words against one for every 170 words). This finding is consistent with the relative frequency of the English passive in academic texts when compared with genres such as journalism, and is presumably to be explained in part by the greater impersonality of the agentless passive.

3. Among the passives there is a higher proportion of fronted verbs in the abstracts than in the text of Veja (52% versus 28%). This observation suggests that in addition to the general linguistic factors that favour fronting in some contexts over others, there may be a further genre-specific function of fronted passive verbs in Portuguese academic abstracts. This is a point to which we shall return later.

End-weight and Communicative Dynamism

End-weight is the well-known phenomenon that long and/or syntactically complex elements tend to be placed after shorter and/or less syntactically complex elemnets (see for example Quirk et al., 1985, pp. 1231-1232). The significance of end-weight for the subjects in the abstracts is shown most strikingly by their relative length, fronted subjects being on average 7 word long and non-fronted subjects 14 words long. An even clearer measure is given by the proportion of subjects four or fewer words long: 44% for fronted subjects against 7% for non-fronted subjects. As to verbs, the significance of end-weight is shown by the fact that all verbs forming minimal predicates (ie where the predicate consists only of the passive form of the verb) are fronted, while all those constructed with a modal verb such as poder and/or a preposition phrase closely linked to the verb are non-fronted.

Communicative Dynamism (CD) is used here in the sense defined in work within the Prague theory of Functional Sentence Perspective by linguists such as Firbas:

By CD I understand a property of communication, displayed in the course of the development of the information to be conveyed and consisting in advancing this development. By the degree of CD carried by a linguistic element, I understand the extent to which the element contributes to the development of the communication, to which, as it were, it 'pushes the communication forward'. Thus if examined in its non-marked use, the sentence He was cross could be interpreted in regard to the degrees of CD as follows. The lowest degree of CD is carried by He, the highest by cross, the degree carried by was ranking between them. (Firbas 1972).

1. The degree of CD of an element is associated with the extent to which the element is context-dependent or context-independent. Thus in the sentence quoted by Firbas, the anaphoric pronoun He carries the lowest degree of CD because it is entirely dependent on the preceding context. The terms context-dependent and context-independent are preferred to the more familiar terms 'given and 'new' as they emphasise the linguistic rather than the psychological basis of the distinction (cf. the use of the term 'context-construable' in Rochement and Culicover 1990, p. 20).

2. The degrees of CD are established without reference to the linear arrangement of the message. However, it is central to much work in FSP that the unmarked choice is, as in Firbas's example, for left-to-right arrangement to reflect the degrees of CD from lowest to highest.

3. By allowing for varying degrees of CD the approach permits greater descriptive flexibility than does Halliday's insistence on a binary distinction at the ranks of clause, sentence and paragraph of theme/rheme (linear arrangement) and topic/comment (given/new) (Halliday 1985). As Halliday himself has recognised, that insistence carries penalties: for example, it is impossible within his framework to make any statement about linear ordering or topic/comment structure within a clause following a thematic adjunct unless the ad hoc decision is taken to regard such adjuncts as rank-shifted clauses (Halliday 1985, p. 61).

4. The relationship referred to earlier between the degree of CD of an element and the weight of that element is that, other things being equal, one would expect context-dependent elements to require less specification, and thus to receive less weight, than context-dependent elements. The limiting case for reduction in weight of context-dependent elements is, of course, that they should disappear altogether: thus in Portuguese Estava furioso is a normal rendering of the English sentence quoted ealier, the marking of the verb for person and number making an anaphoric pronoun redundant.

It is clear that in the abstracts fronted subjects in general carry a lower degree of CD than do non-fronted subjects. For example, in free be-passive clauses 21 of the fronted subjects but only 2 of the non-fronted subjects carry explicit anaphoric reference to the preceding text.

Another factor that seems to affect the distribution of CD and thus the decision whether to front the verb is the presence or absence of an initial adjunct. Pontes (1987, p. 111) quotes without further discussion the claim of Cunha (1976) that the presence of such an adjunct is one of the features that favour the placing of the verb before the subject in Portuguese, The abstracts provide some support for this claim: among the 124 free clauses containing be-passives with non-clausal subjects, nearly one half of the fronted verbs but only one in ten of the non-fronted verbs occur after initial adjunct. This tendency of Portuguese may be compared with the Germanic languages such as German, Dutch and Norwegian in which post-adjunct inversion is obligatory, and also with English which retains traces of such a rule in structures such as "In the doorway stood an old man" and "Only yesterday did I hear the news".

(A)VS Structure and Stereotypical Argument in Academic Abstracts

There is another sense in which the fronted verbs in the abstracts may be context-dependent, and that is that they depend on the context of assumptions shared between writer and readers as to the sort of information contained in academic abstracts and the way that information is structured. With one exception, discussed below, all the verbs are either indicative - that is, they refer to what is stated in the paper, or they are informative - that is, they refer to what was done in the research reported in the paper (see St. John, 1983, for a fuller discussion of these terms). Thus, among the quoted examples, 3a and b, 14, 18, 20, 21, 24, 29, 30, 31 and 32 are clearly indicative, while 8, 10, 11, 12, 13, 17, 19, and 25 are informative. In Brazilian abstracts written in Portuguese, as in British abstracts written in English, the indicative/informative distinction correlates with choice of tense, the present being used for indicative statements ('what is in the paper') and the past for informative statements ('what was done in the research'). The following table shows some of the verbs that are fronted to make indicative and informative statements in the abstracts:

The centre column shows those verbs which are, on the evidence of tense selection, used in the abstracts to make both indicative and informative statements: the fact that the results of an experiment are analysed reports the contents of the paper, but that they were analysed reports one of the procedures undertaken in the research. No doubt with further data other verbs - eg investigar and fazer análise - might move over to the centre column. What is striking about the table is the close correspondence between Portuguese and English in the lexical realisation of indicative and informative statements, the only apparent anomaly being the classification of descrever as both indicative and informative (compare examples 9 and 23). The equivalent English verb 'describe' would normally be used to make indicative statements only.

When we turn to the non-fronted be-passives, we find that in addition to the indicative and informative statements found with fronted passives, there are also statements that are neither indicative nor informative in the senses defined above, but give for example the background to the work done (4 and 5) or an evaluative comment on the conclusions reached (6):

4. Esta metodologia é baseada na oxidação do ácido ascorbico catalisada por ions cúpricos, com consumo de oxigênio.

5. Ulceras herpéticas tróficas - ou úlceras metaerpéticas como às vezes são chamadas - são definidas clinicamente como lesões superficiais ovóoídes em uma córnea profundamente hipostésica.

6. Na verdade, as três tendências aqui destacadas poderiam ser melhor classificadas como tendências em técnicas de aplicações de matematica, isto é, matemática como uma linguagem, do que propriamente como uma interpretação da natureza.

The only fronted passive verb used to make what appears to be a background or evaluative statement is found in one of the 9 papers in the field of Social Science:

7. Na análise dos fatores que propiciaram o surgimento e expansão do processo de industrialização no Brasil, tem-se acentuado bastante as mudanças exógenas que são, realmente, as fundamentais para freá-lo ou acelerá-lo.

The facts summarised above suggest that in Poruguese academic abstracts the fronting of passive indicative and informative verbs act as a signalling system which 'exposes the bones' of the abstract, and places the information with the highest degree of CD - that is to say, exactly what is in the paper or what was done in the research - in its most natural position at the end of the clause. For example, in the following (complete) abstract the four fronted informative verbs articulate a stereotypical argument in scientific papers and abstracts alike of Purpose (Estudou-se . . . ), Materials and Methods (Foram constituídas . . . ) and Results (Observou-se . . ., observou-se . . . ) and throw into dynamic prominence the context-independent information in the weighty non-fronted subject noun-phrases as to what was studied, what was constituted,and what was observed.

8. Estudou-se o comportamento de Drosophila sturtevanti em culturas puras e mistas. Foram constituídas populações de culturas puras de D. sturtevanti, D. simulans e D. immigrans e de culturas mistas da primeira espécie separadamente com as duas outras. Observou-se, em culturas puras, um mecanismo de controle de progênie, em D. sturtevanti, relacionada com a densidade parenta. Nas culturas mistas observou-se interferência de uma espécie na progénie da outra, com prejuízo para ambas as espécies.

In the following abstract each of the three fronted indicative se-passive verbs that articulate the argument shows a different syntactic context for fronting: weighty non-clausal subject, empty anaphoric suhject, and weighty clausal subject:

9. Descreve-se um método cinético capaz de distinguir entre alguns modelos da bomba de sódio, e testa-se com mecanismos encontrados na literatura. Conclui-se que o método é um critério simples, útil na distinção entre expressões cinéticas e, em casos favoráveis, na avaliação de parâmetros cinéticos.

(A)VS Structure and Interlanguage Strategies

Strategy 1: (A)SV

10. Analisou-se relações de dopamina cerebral com as funções motoras.

    The relations between dopamine and motor functions were analysed.

11. Investigou-se a influência da organização do contéudo de termodinâmica na estrutura cognitiva do aluno em um curso de física geral na UFRGS.

    The effect of content organisation on the student's cognitive structure was investigated in an introductory college course in thermodynamics at UFRGS.

There are also in the data a number of instances where the 'normalising' strategy produces an English text that is awkward and unidiomatic to the extent that the linear arrangement fails to reflect the scale of Communicative Dynamism. This is particularly likely to occur when a weighty subject is combined with a light verb: that is to say, precisely those conditions that favour a fronted verb in Portuguese:

12. Foram constituídas populações de culturas puras de D. sturtevanti, D. simulans e D. immigrans e de culturas mistas da primeira espécie separadamente com as duas outras.

    Populations of pure cultures of D. sturtevanti, D. simulans and D. Immigrans and of mixed species of the first species separately with the other two were constituted.

13. Foram estudados os efeitos de luz, de temperatura e dos tegumentos na germinação de sementes de limão-cravo (Citrus limonia, Osb).	The effects of light, temperature and the presence or absence of the seed coat on limao-cravo (Citrus limonia) seed germination have been studied.

Example 14 displays an especially striking example of subject-predicate imbalance which the writer appears to have tried, unsuccessfully, to solve by dividing the meaning of incluíram-sebetween the fronted 'Besides' and the non-fronted 'are given':

14. Incluíram-se observações sobre o canal deferente, saco "espermatofórica" e glandula androgênica, além dos caracteres sexuais secundários desta espécie.

    Besides a description on the deferens ducts, spermatophoric sac and androgenic gland, whit observations on the secondary sexual characters of this species, are given.

Strategy 2: Nominalisation

a)

A description is given of the deferens ducts, spermatophoric sac and androgenic gland, together with observations on the secondary sexual characteristics of this species.

b)

Observations are made/included on the deferens ducts, spermatophoric sac and androgenic gland, and also on the secondary sexual characteristics of this species

There are in the abstracts only a couple of examples of nominalisation to minimise linear dislocation. The first exploits the dual classification already noted of Estudar/'Study' to convert the informative statement of the Portuguese text into an indicative statement in the English text:

15. Estudou-se o comportamneto de Drosophila sturtevanti em culturas puras e mistas.

    This is a study of Drosophila sturtevanti in pure and mixed cultures.

A less successful example involving nominalisation is 31, discussed below.

Strategy 3: (A)VS structure in English

16. Were studied only those trophic ulcers that complicated a first attack of dendritic keratitis.

17. Em função da distribução geográfica dos centros de investigação, foram selecionadas 15 instituções altamente representativas, em cada uma das quais se relacionam as linhas de pesquisa qualitativamente desenvolvidas, nos setores fundamentais de química analitica, inorgânica, orgânica e física-quimica.

    In view of the geographical distribution of research centres, were selected 15 outstanding institutions, assumed as being representative, and in each one are exposed the principal lines of investigation in the fundamental sectors of analytical, inorganic and physical chemistry.

18. Neste trabalho são classificados e analisados os geradores de energia a partir do vento. Também estão incluídas as equações básicas para o cálculo de energia do vento e predições da possível energia gerada por cataventos.

    In this paper are classified the various types of wind mills analysed and commented on. Also are included the basic equations for wind energy calculations and equations of power generated by windmills.

19. Somente em células das paratireóides foi notada certa reação para a fosforilase, G-6-PA e F-1, 6-PA.

    Only in the parathyroid gland cells was noted a certain reaction to phosphorylase, G-6-PA and F-1, 6-PA.

An explanation of these examples that has some intuitive appeal is that the initial adjunct has the effect of 'hiding' the VS structure: it does not stand out as starkly in 17-19 as it does in 16. An alternative and more likely explanation can be found in the tendency already noted for initial adjuncts to be associated with fronted verbs in written Portuguese. An approximate rule for Portuguese would be on the lines: 'If there is no initial adjunct, the verb is sometimes fronted: if there is an initial adjunct, it is often fronted'. This could form the the basis for the induction of an interlanguage rule for English by means of an equivalent reduction of both adverbs of frequency: '(Observation shows that) if there is no initial adjunct, the verb is rarely/never fronted: (therefore) if there is an initial adjunct, the verb must sometimes/occasionally be fronted'.

Strategy 4: Pro-form Insertion

20. Inicialmente é apresentado um breve histórico do desinvolvimento da pesquisa na área da química no Brasil, apontando-se os fatores positivos e negativos de sua evolução.

    It is presented initially a brief historical of Brazilian development in the chemical field, pointing out positive and negative items of its evolution.

21. Considera-se, ademais, a distribução dos cativos - segundo sexo, faixas etárias e origem - entre os dois segmentos populacionais aludidos.

    It is also considerated the composition of the slaves who belong to those two social strata according to sex, age group and origin.

22. Além da caracteristicas referidas, merecem especial atenção a originação do aparelho estomática e a presença de fibras do mesófilo, que divergem dos feixes vasculares de menor porte, terminando livremente nas aréolas, sem a companhia de traqueídos terminais.

    It must still be mentioned the presence of fibres embedded in the mesophyll which diverge from the vascular bundles without the company of the tracheid of xylem.

In the following example there are two instances of the fronted verb descreveu-se: the English text renders the first by normalising the word order to ASV, and the second by means of an inserted 'it'.

23. No presente trabalho descreveu-se o crescimento da membrana plásmica por ocasião da citocinese, através de fusão com vesiculas originadas no Golgi. Descreveu-se ainda a formação da membraba plásmica do espermatozóide a partir de membranas de cisternas do reticulo com eliminação de parte do citoplasma que rodeia as estruturas do espermatozóide diferenciadas na espermátide.

    In the present paper the growth of the plasmic membrane during cytocinesis is reported as occurring by fusion of small vesicles arised from the Golgi to the dividing cell membrane. Itw was also described the formation of the spermatozoon plasmic membrane by the endoplasmic reticulum cisternae with the casting off of the cytoplasm around the spermatazoa inside the spermatid.

While it seems clear enough that the motivation for the error lies in an attempt to reconcile the scale of Communicative Dynamism in Portuguese with the basic SV word order of English, the form it takes is rather more mysterious. Unlike the post-adjunct verb fronting examined earlier, and unlike also the tendency noted by Trévise (1986) for French learners of English to overuse inserted 'it', the error cannot be accounted for directly on the basis of any rule or combination of rules of Portuguese: thus the the following versions of 20 are as ungrammatical as their English counterpart:

*O/Isso considera-se, ademais, a distribução dos cativos . . .

a) Extraposed It

In this structure, the pro-form refers forward (cataphorically) to a clause:

24. É sugerido também que aprendizado e "estampagem" em jovens de P. cuvieri são os principais fatores no escolha dos sítios para desova em poças ou lagos temporários.

    It is suggested too that learning and imprinting in young P. Cuvieri are the main influences on their choice of sites for constructing foam nests in puddles or temporary pools.

The following example shows the writer ingeniously exploiting the extraposed 'it', rendering the nominal uma diminuição na adsorção by means of a clause:

25. Foi verificada uma diminuição na adsorção, com o aumento da massa molecular do ácido húmico usado

    It was found that adsorption decreases with the increasing molecular weight of humic acid.

Note that being derived from a clause, the extraposed pro-form cannot show number concord:

*They were found that adsorption decreases and that molecular weight increases . . .

As has been mentioned, Portuguese but not English allows 'empty anaphora', Indica in 25 cross-referring to O trabalho in the previous statement:

26. O trabalho apresenta idéias acerca do papel dada à revisão da literatura nas dissertações e tesas desenvolvidas na área de educação. Indica algumas das idéias preconcebidas e do inadequado uso desta parte de um relatório, . . .

    The paper discusses the role and place given to the chapter related to the review of literature in the thesis and dissertations developed in education. It indicates some of the inadequate ideas and uses of this part of a research. . . .

Anaphoric reference is, of course, subject to the usual rules for number concord in English and in Portuguese. Thus, substituting Os estudos for O trabalho we get:

Os estudos apresentam idéias . . . Indicam . . .

The studies present ideas . . . They indicate . . .

27. There are stressed the comparative problems in study, mainly mechanisms of organic production, nutrient cycling and the composition of the zooplankton and phytoplankton. [No direct equivalent in Portuguese abstracts].

Compare 27 with the successful use of the pro-form 'there'to minimise linear dislocation in the following citation:

28. Na raça Angorá, com diet 1, não foi constada diferença entre as faixas etárias para as duas variáveis estudadas.

    There was no significant difference for both contents for Angora breed animals submitted to diet one for the different age groups.

Example 27, although clearly an 'error' in terms of the target language, represents an ingenious solution to the indeterminacy of the pro-form in 20 – 23. The advantage of the pro-form 'there' is that it is unmarked for number, and the writer manages to avoid the puzzle of whether the pro-form is anaphoric is a way that the choice of 'it' or 'they' would not.

Some light is thrown on the data described in this paper, and on the inserted 'it' in particular, by the typological distinction between subject-prominent and topic-prominent languages drawn by Li and Thompson (1976). For a subject-prominent langauge such as English the primary requirement is that a subject should be named, and that it should appear before the verb: various syntactic devices such as extraposition and passivisation are available to satisfy the secondary need to to identify the topic of the sentence. For a topic-prominent language such as Chinese or Japanese the primary requirement is that the topic should be named first in the sentence, the need to name a subject being secondary: the subject is not marked by position and there may be no subject at all or what in traditional grammar was described as a 'double subject' (ie a topic followed by a subject). In terms of the theory of Communicative Dynamism outlined earlier, in a topic-prominent language linear arrangement follows the scale of CD far more closely than it does in a subject-prominent language.

As far as Brazilian Portuguese is concerned, Pontes (1987) makes out a case for being in most respects topic-prominent, with the colloquial language showing more features of topic-prominence that the written language. If her argument is correct, one would expect the English of the abstracts to show similarities to that of speakers of otherwise unrelated languages such as Chinese and Japanese. Evidence for such a similarity is provided by Rutherford (1979) who observes that Japanese learners of English consistently overuse extraposed 'it' in statements that have the function of 'setting the scene' for the following discourse.

Strategy 5: Active for Passive

29. Neste trabalho são apresentados observações fenológicas sobre Magonia pubescens St. Hil.

    This paper reports observations about the phenology of Magonia pubescens St. Hil.

30. Estudaram-se a morfologia e a histologia do aparelho reprodutor masculino do camarão de água doce . . .

    This paper deals with the anatomy and histology of the male reproductive system of the freshwater prawn . . .

In 31 the passive Estuda-se is replaced by a complex form which is adequate syntactically but which fails morphologically ('propose' possibly being identified as a noun as a result of confusion with 'purpose') and semantically (since the paper does not merely propose a study of the composition of the slave-owners, but carries out such a study):

31. Estuda-se a composição d massa de senhores de escravos segundo seu enquadramento em dois dos estratos sociais existente no Brasil-colônia: livres e forros.

    Our main propose is to study the slave-owners composition according to their framing in two of the social strata existing in Brazilian colonial period: free people and emancipated people.

The English versions below of 3a and b both select Active for the first sentences with their weighty original subjects (44 and 34 words respectively in the English text), and accept the linear dislocation of the Passive for the lighter subjects of the second sentences (6 words). For the supplied subject of 3b the writer selects 'work' as a direct translation of trabalho, and for 3a the personal pronoun 'we':

a)

In this work we present algorithms for the numerical computation of . . . A FORTRAN listing of the algorithms is also included.

b)

This work presents an algorithm for the computation . . . A FORTRAN coding of the algorithm is included.

Although 'we' meaning the writer or writers of the paper is used in certain contexts in the body of scientific papers (see, for example, Tarone et al. 1981), there is a prescriptive tradition against its use in abstracts, and it is understandable that the writer should have experimented (not entirely successfully) with the more impersonal 'one':

32. No presente trabalho, chega-se a diretrizes para o cálculo da descontinuade associada a uma onda de choque estacionada na entrada de um gargalo fixo. Para tanto, lança-se mão da expressão que fornece a velocidade . . .

    In the present paper, one arrives to ways to compute the discontinuity related to a stationary shock wave at the entrance of a fixed bottleneck. To do so, one utilizes the expression which gives the velocity . . .

Conclusions and Further Research

Further work might throw light on three questions that this paper does not address but which deserve further investigation. The first is the effect on communication of the difficulties identified, and the extent to which they in turn affect the handling of other language features such as lexical and tense selection, the latter notably uncertain in some of the quoted extracts. The second question is how far the interlanguage rules examined – and in particular pro-form insertion – are invented anew by learners of English in Brazil and how far they are culturally transmitted. Cultural transmission of 'fossilised' interlanguage is ususally associated with situations such as that on the Indian sub-continent where English has an independent life as a widely-used Second Language in areas of public life such as trade, education and administration. While that is patently not the case in Brazil as a whole, it could plausibly be argued that there are at least two Brazilian sub-cultures within which English has a status approaching that of a second language and which could thus act as vectors for transmission. The first is the academic sub-culture which is the main producer and the main consumer of the abstracts. If the pressure towards publication is such that Brazilian academics have not only to write about their own work but also gain access to the work of their compatriots in English, a mechanism is in place that could have allowed the emergence and transmission of a distinct variety of 'Brazilian Academic English'. The second sub-culture that may act as a vector for the transmission of fossilised interlanguage is that of Brazilian teachers of English and writers of English-teaching materials. That possibility would lend point to a third question that might be addressed: how often are they the sole responsibilities of their authors, and how often are their colleagues in university departments of English asked to translate them from Portuguese or to revise them?

Acknowledgments This paper benefited greatly from preliminary discussions with Carmen Coulthard, Iria Garcia and Wilson Leffa and from comments on the first draft by Walter Costa, Tony Dudley-Evans, Julian Edge, and Jan Firbas. To all of them I extend my thanks: the remaining inadequacies remain entirely my responsibility.

References

Firbas, J. (1972), 'On the Interplay of Prosodic and Non-prosodic and Non-Prosodic Means of Functional Sentence Perspective' in V. Fried (ed.) The Prague School of Linguistics and Language Teaching (Oxford University Press)

Firbas, J. (1987), On Some Basic Issues of the Theory of Functional Sentence Perspective' Brnno Studies in English 17

Garcia, E. (1975) The Role of Theory in Linguistic Analysis: the Spanish pronoun system (North Holland)

Greenberg, J. H. (1966) 'Some Universals with Particular Reference to the Order of Meaningful Elements' in J. H. Greenberg (ed.) Universals of Language (MIT , 2nd edition)

Halliday, M. (1985) An Introduction to Functional Grammar (Arnold)

Li, C.N. and S.A. Thompson (1976) 'Subject and Topic: A New Typology of Language' in C.N. Li (ed.) Subject and Topic (Academic Press)

Pontes, E. (1987) O Tópico no Português do Brasil (Campinas, Pontes)

Quirk, R., S. Greenbaum, G. Leech and J. Svartvik (1985) A Comprehensive Grammar of the English Language (Longman)

Rochemont, M.S. & P.W. Culicover (1990) English Focus Constructions and the Theory of Grammar (Cambridge University Press)

Rutherford, W. (1987) Second Language Grammar: Learning and Teaching (Longman)

St. John, M.J. (1983) 'Summary Writing in Dissertations' (M.Sc. dissertation, University of Aston)

Schachter, J. and W Rutherford (1979) 'Discourse Function and Language Transfer' Working Papers in Bilingualism 19

Selinker, L. 'Interlanguage', International Review of Applied Linguistics X/3

Siewerska, A. (1984) The Passive: A Comparative Linguistic Analysis (Croom Helm)

Tarone, E. (1983) 'On the Variability of Interlanguage Systems', Applied Linguistics 4/2

Tarone, E. (1988) 'Variability in Interlanguage Use: a Study of Style-shifting in Morphology and Syntax' Language Learning 35/3

Tarone, E., S. Dwyer, S. Gillette and V. Icke (1981) 'On the Use of the Passive in Two Astrophysics Journal Papers' in J. Swales (ed.) Episodes in ESP (Oxford University Press)

Trévise, A. (1986) 'Is it Transferable, Topicalisation?' in E. Kellerman and M. Sharwood-Smith (eds.) Crosslinguistic Influence in Second Language Acquisition (Pergamon)