versión impresa ISSN 0716-5811
Lit. lingüíst. n.17 Santiago 2006
Literatura y Lingüítica N° 17, págs: 303-324
Lingüística: artículos y monografías
Corpus linguistics at the service of English teachers
Leonardo Juliano Recski
This paper contemplates how corpus evidence might be used to address teachers' questions about English grammar and suggests that corpus linguistics has an important role to play in raising their awareness to linguistic features and patterns. The article surveys a range of grammatical questions posted by EFL/ESL teachers on four Orkut communities devoted to the teaching and learning of English. It concentrates on three specific types of questions: synonymous lexical items which function differently and are reported to be difficult to teach and explain; linguistic evidence that contradicts the prescriptive grammar rules that teachers have been taught during their education; and collocations that teachers attempt to explicate. In attempting to address the teachers' questions, corpus evidence is used to offer possible explanations. It is suggested that the use of corpus data in addressing these questions is not only convincing but also leads to discoveries of patterns and meanings which might not be found in other reference materials such as grammars and dictionaries.
Keywords: corpus linguistics; EFL/ESL teachers' questions; linguistic description; language awareness
Este artículo contempla cómo la evidencia del corpus lingüístico podría ser usado para responder preguntas de los profesores de inglés respecto de la gramática inglesa, y sugiere que el corpus lingüístico tiene un rol importante que jugar para incrementar la advertencia respecto de los patrones y aspectos lingüísticos. El artículo describe una selección de preguntas en torno a gramática inglesa enviadas por profesores de inglés como idioma extranjero y como segunda lengua en cuatro comunidades Orkut dedicados a la enseñanza y aprendizaje del idioma inglés. El artículo se concentra en tres tipos específicos de preguntas: ítemes léxicos sinónimos que funcionan de manera diferente y se reportan como difíciles para enseñar y explicar; evidencia lingüística que contradice las reglas de la gramática prescriptiva, la cual se ha enseñado a los profesores de Inglés por siempre en su formación, y expresiones de uso frecuente que los profesores intentan explicar. Al intentar responder las consultas de los profesores, se usó evidencia recogida de un corpus lingüístico para ofrecer posibles explicaciones. Se sugiere, con este estudio, que el uso de corpus lingüístico para fundamentar las explicaciones, no solamente es convincente, sino que además conlleva a descubrimientos de patrones y significados que podrían no hallarse en otros materiales de referencia, tales como libros de gramática y diccionarios.
Palabras claves: corpus lingüístico – profesores de inglés como lengua extranjera y como segunda lengua – descripción lingüística – conciencia lingüística
1. Corpora and linguistic description
Before the existence of corpora, linguistic description relied very much on native-speaker intuition and introspection. Native-speakers normally describe what they know about language, or what they perceive language to be, rather than how language is used. The easy accessibility of huge bodies of naturally occurring texts on the computer has made it possible for us to test the robustness of linguistic descriptions which were based on introspection and elicitation, and to gain new insights into language structure and use. It has helped us to gain a better understanding of how language is actually used rather than how language is perceived to be used. Examining specific instances of language use gives us insights into how language works which would never have been obtained by simply introspecting about the language system. Such insights result in our construing the linguistic system in a different way. So far, corpus-based studies have focused on four main types of description and analysis: lexical collocation by examining the frequency and context of occurrence of linguistic items (see for example, Sinclair 1991; see also Kjellmer's (1994) dictionary of collocations based on the Brown Corpus), syntactic patterning based on co-occurrence of grammatical word-class tags, genre analysis based on the co-occurrence of groups of linguistic items and processes (see for example, Biber, 1988), and discourse structure and cohesion in spoken and written English (see for example, Carter and McCarthy's Spoken English Corpus at the University of Nottingham - e.g. Carter & McCarthy, 1997) (see Kennedy, 1998, for a sound summary of corpus-based studies). The findings of the above studies, particularly word-based studies, have important implications for second or foreign language teaching, as we shall see in the next section.
2. Corpus-based studies and ESL / EFL teaching
In EFL and ESL situations, learners do not have the same amount of exposure to the target language as they do in L1 situations. Therefore, it is safe to presume that in most cases they are unlikely to acquire the language efficiently without systematic guidance on linguistic forms. By focusing on words which have a high frequency of occurrence and by concentrating on the usual rather than the exceptional, teachers can help learners acquire the language more efficiently, especially at elementary and intermediate levels. The findings of corpus analysis can be used as a basis for selecting and sequencing linguistic content, as well as for determining relative emphases. A number of studies have observed discrepancies between corpus findings and the selection of and emphasis given to linguistic content in ESL and EFL textbooks and curriculum. As early as the sixties, George (1963, cited in Kennedy 1998: 283) studied a corpus of English that was based on written texts and found that the highest frequency of occurrence of the simple present is not to indicate habitual or iterative actions, such as "I go to school by bus every day" (5.5%), but rather the actual present, such as "I agree with you" (57.7%) or neutral time, such as "My name is Mary" (33.5%). His findings converge with a more recent grammar of English compiled by Mindt (2000) based on corpora totaling 240 million words of spoken and written English. Mindt found that the three prototypes which make up the majority of all cases of the present forms of verbs are the extended present, the actual present and the timeless present. This is contrary to the emphasis given to the habitual present in most ESL and EFL textbooks as the major function of the simple present.
Holmes (1988) compared a corpus analysis and a textbook analysis of epistemic modality and found that, like most textbooks, important epistemic uses of modal verbs are under-taught and that lexical verbs expressing modality, such as appear, believe, doubt, and suppose, nouns such as possibility, tendency, and likelihood, and adverbials, such as perhaps, of course, and probably, tend to be given little pedagogical attention. Ljung (1991) compared the EFL textbooks at upper secondary level in Sweden with the Cobuild corpus and found that 20% of the most frequent one thousand words in the learners' texts did not occur in the most frequent one thousand words in Cobuild. Biber, Conrad and Reppen (1994) examined the structural options for postnominal modification and the attention given to these options in popular ESL and EFL textbooks. They found that typically more pedagogical attention was paid to finite and non-finite relative clauses than prepositional phrases as noun modifiers, in contrast with their analysis of the Lancaster Oslo/Bergen (LOB) corpus, which shows prepositional phrases as noun modifiers occurring far more frequently than relative clauses (see also Quirk et al., 1985: 1274). Kennedy (1998) observes that similar incompatibility can be found in the pedagogical focus on grammatical quantifiers such as all and every in many textbooks to indicate the concept of totality when in both written and spoken corpora totality is much more commonly lexically marked, such as entirely, completely, whole, throughout.
The above very brief summary of some comparative studies of corpora and ESL and EFL textbooks show the relevance of corpus studies to ESL and EFL teaching and learning. One of their major contributions is to provide objective quantitative evidence of the distribution of linguistic items on which the goals and content of the curriculum can be based.
3. Corpus analyses and teachers' language awareness
One area that is under-explored is the relevance of corpus linguistics to teacher education, particularly in the area of teachers' language awareness (see also Berry, 1994; Hunston, 1995). In the last decade or so, more attention has been paid to the importance of raising teachers' language awareness (see for example the collected papers in Bygate et al., 1994; Hawkins, 1999). This paper hopes to emphasize that teachers' language awareness is one area in which corpus linguistics has an important contribution to make. It examines grammatical questions that English teachers of EFL/ESL posted on Orkut communities devoted to the teaching and learning of English to seek advice, and attempts to demonstrates how empirical linguistic data which show the context and frequency of occurrence of the linguistic items in question can be a powerful tool to raise teachers' linguistic sensitivity, to help teachers question long-standing assumptions, and to gain new insights into language structure and use.
4. The Orkut1 communities of EFL/ESL Teachers
Currently, there are innumerous Orkut communities of EFL/ESL teachers on the web. Most of these communities were created in 2004 and some of them (e.g. English as a Second Language) have close to 20.000 members. The criterion I have adopted for choosing which communities to scrutinize was solely based on the number of members, since communities with a large number of participants were more likely to yield more questions. A total of four communities were investigated in search of grammatical questions posed by teachers:
In the rest of this paper, I shall examine the grammatical questions sent by teachers over a period of two years, discussing some of the ways in which corpus data could be used to help them tackle the sorts of questions they are faced with in their everyday teaching.
5. Teachers' grammatical questions and corpus evidence
An analysis of the grammatical questions posted on the Orkut communities of EFL/ESL teachers over the period of two years shows that they largely fall into one of the following six types. The first type has to do with synonymous lexical items. Some lexical items are largely synonymous but have different usage. Teachers are aware of the difference in usage but have problems explaining the difference to students, for example, tall and high. Some lexical items appear to be synonymous but teachers are not sure if they are "absolute synonyms" (Partington, 1998), for example, day by day and day after day. The second type relates to linguistic evidence that contradicts the prescriptive grammar rules that teachers had been taught when they were learners, the most frequently asked being subject-verb agreement and the use of the definite article. The third type concerns lexical collocations which teachers try to rationalize but sometimes cannot. The fourth type consists of lexical items which teachers take to be absolute synonymous but have been asked by students to explain whether there is any difference in meaning, for example, big and large, lastly and finally. The fifth type are prescriptive stylistic rules which seem to have been passed on from generation to generation but are queried by teachers, for example, the rule that one should not begin a sentence with because, and and but. Finally, the sixth type concerns lexical items which students find confusing because their translation into the students' first language is either identical or very similar, for example find and look for which will have very similar translations in Portuguese. Because of the limit of space, I shall focus on the first three types of questions. Teachers' questions will be cited, followed by the use of corpus evidence in hypothetically addressing their questions. I shall also discuss how in the course of addressing the teachers' questions, analyses of corpus data may lead to insights about linguistic patterns and meanings which apparently have not been given much attention in reference grammars, dictionaries, or in common practice.
5.1 Synonymous lexical items
One of the most frequently asked questions is whether there is any difference between words that are commonly taken as synonymous. There are cases in which teachers were not aware of any difference in meaning and usage, such as big and large, lastly and finally. But there are some in which they were aware of a difference in usage but could not quite articulate what the difference was, for example, tall and high.
Tall vs. high
The following is a message sent by a teacher who said that she knew that tall and high have different usage, but she could not quite explain the difference to her students:
The words "tall" and "high" have similar meaning but different usage. I have no problems in using the words myself. However, I find it difficult to explain the difference between these two words to my students. Is there any suggestions of teaching these two words?
The difference in usage between tall and high is particularly difficult for Brazilian learners to grasp because there is no such distinction in Portuguese. Both words will be translated as the same word in written Portuguese. Therefore explaining the words in Portuguese does not really help.
The Longman Dictionary of Contemporary English states that high is used for measurement of most things but not people, especially when we are thinking only of distance above the ground, such as a high shelf, a high building and a high mountain, whereas tall is used for people and ships, for example, a tall man and a tall ship. It further adds that tall is used for things that are high and narrow (e.g. a tall/high building and a tall/high tree).
In response to the teacher's question, I have searched the Freiburg-Brown Corpus of American English (FROWN)2 with roughly one million words, and examined the nouns that were modified by high and tall. What emerged from the search was that there was a tendency for high to be used in a metaphorical sense with more abstract nouns whereas tall tended to be used more frequently with concrete nouns such as people, trees and buildings. The following concordance lines elicit this difference in use.
These concordance lines are a useful start to get the teachers to think about a word not in isolation but in terms of its "semantic preference" (Sinclair, 1991). A further analysis of the FROWN corpus revealed that there are 605 instances of high whereas there are only 54 instances of tall. Except for 9 instances in which tall is used idiomatically, such as a tall order, walk tall, the rest are used in the context of talking about the height of people, buildings, and vegetation (see the above concordance lines for tall), with the highest frequency of tall co-occurring with people (about 50%) followed by buildings and structures (about 35%). In other words, the semantic preference of tall is quite restricted. By contrast, the contexts in which high is found is much more wideranging, including amount, intensity, quality and relative quantity. Taking 5 as the cut-off point for frequency yielded the following nouns that co-occur with high in the corpus (see Table 1).
5.2 Grammar rules and conflicting evidence
Teachers are often troubled by the fact the grammar rules that they have been taught as students do not accord with the authentic linguistic examples that they encounter (see also Tognini-Bonelli, 2001). Indeed, in the Orkut posts, one type of most frequently asked questions has to do with teachers who try to apply some usage rules but are confronted with conflicting evidence. The most frequently asked questions pertain to subject-verb agreement and the use of definite articles.
The following are some of the messages posted by teachers
We can see from the response given by Teacher C that she was trying to apply the rule of subject-verb agreement to the example provided by Teacher A. Teacher B intuitively felt that are can be used in both sentences but she was looking for some rules.
We could respond to the teachers' questions by pointing out that usually the singular form of be is used when the first noun that follows is singular and the plural form of be is used when the noun group after it is plural (see also the Collings Cobuild English Grammar, p. 416). However, a search through the corpus does show an instance of the following:
According to PACE, suspects can only be detained at designated police stations where there are a custody and a reviewing officer.
In other words, while what is stated in the Cobuild English Grammar is correct, teachers may benefit from knowing that occasionally the plural form of be can be used even when the noun following is singular. Therefore, the question is not about possibility but about probability of usage.
What is interesting is that the investigation of there's is in a corpus of academic spoken English reveals that it often appears before a plural noun. Although Quirk et al. (1985) have also made this point, it is much more convincing to provide teachers with corpus evidence. A search on the Michigan Corpus of Academic Spoken English (MICASE) showed that there are 3,785 instances of there's. The following concordance lines could be provided to teachers
By contrast there are only 30 instances of there're followed by a plural noun
The discussion on subject-verb agreement led to further questions of a similar nature from other teachers regarding whether the singular or the plural verb should be used in the context of one of the + plural noun.
Teacher E (responded as follows):
A search on the FROWN corpus showed that there are 71 instances of "one of the … (be) …" and both the plural and singular be forms are used. There is a higher frequency of occurrence of the plural form but the singular form is used often enough to be regarded as an acceptable alternative.
Some further examples of questions of similar nature are whether a plural or singular verb should be used after the structure "none of the …" and "more than one…". For example:
It is clear from the teachers' questions that they were puzzled by the lack of agreement between the subject and the verb. It could be pointed out to them that technically, none means literally not one, and thus it seems to be more logical to use a singular verb. However, because none of functions as a quantifier, it is often followed by a plural noun, and therefore a plural verb is used. A search on the FROWN corpus revealed both singular and plural verbs being used. For example:
The above discussions sparked off a series of questions from teachers posing related questions asking whether one should say there are no students in the room or there is no student in the room; I have no friends or I have no friend. The use of corpus evidence is particularly helpful because the question is not about possibility but probability. Moreover, teachers came up with so many variations of the subject-verb agreement structure that it would not be possible to provide some kind of comprehensive guiding principle. The best solution, I should think, would be to invite them to look for corpus evidence themselves.
The presence or omission of the definite article is another problematic area for teachers. They have difficulties finding some kind of consistency in the rules for using the definite article. For example, they have been told that the definite article should be used if there is only one of a kind being referred to, such as the sun, the moon, and the earth, the name of a country, and before a position, such as the Chairman and the Secretary. However they have also come across cases where the definite article was missing. For example, He was elected Chairman of the Association, She was appointed secretary of the committee.
Sometimes, the questions asked by the teachers can be very specific and it would not be possible to answer them without consulting a corpus. Take for example the following message posted by Teacher D:
Hello! Although I have been an English teacher for about 4 years, I still sometimes have difficulty in using articles. I would be very grateful if someone can help me in the following problem:
In relation to teacher's question it would be possible to point out that the definite article the is used when referring to systems of communication or mass media, such as the radio, the telephone and the mail. It could be observed that the use of the is a bit variable with television since a search through the FROWN corpus shows instances of television, with and without the. For example:
The question about "CD player" is a bit more problematic. There are only eight instance of "CD player" in the FROWN corpus. My intuition led me to conclude that since a "CD player" is in fact a name which has been generalized to refer to any normal size or small portable player for CDs rather than a system of communication, the tendency would be to use it as an ordinary countable noun, and therefore the indefinite article "a" or possessive pronoun would be used (the two examples in the corpus containing an article confirmed my intuition). I then conducted a further search on the BNC-online (http://sara.natcorp.ox.ac.uk/lookup.html) and found 56 occurrences of CD player which further confirmed my introspection (although there were two hits where "the" was used preceding "CD player"). Here are a few examples
5.3 Rationalization of collocations
The third type of frequently asked question has to do with rationalization of collocations. Teachers often try to look for rules governing which words can go with certain words and why. For example, the following is a message from a teacher asking whether one can say well-experienced.
A search conducted on the FROWN corpus on the adjective experienced revealed only fourteen instances. Of these fourteen occurrences there were only two where experienced was modified by the superlative most. It became obvious that a corpus of 1 million words would not suffice if we were to give a sound explanation for this teacher. I then used the BNC-online search (100 million words) to investigate what types of intensifiers (e.g. vastly, highly, very) and comparatives (e.g. more, less) could modify the adjective experienced. A search for well experienced yielded the following eight instances:
To investigate whether there is any difference in the behavior between experienced and other adjectives that take well as the modifier, a search was carried out on well in the BNC and it yielded the following compound adjective: well qualified, well educated, well organized, well equipped and well-known. To see whether the rare occurrence of experienced being pre-modified by the adverb well has to do with the semantics of experienced, a search was conducted on the modifying adverbs, highly, very, poorly and badly. The following are the results of the search on the BNC-online (see Table 2).
The figures in Table 2 show that there are several ways in which experienced behaves differently from the other five adjectives. First, despite the fact that well experienced is found in the BNC, its occurrence is much more restrict than the rest. Second, while there is a large number of instances of experienced taking the intensifier very, there are very few or no instances of the other five adjectives co-occurring with very. Third, these five adjectives, however, take the intensifier very when they combine with well to form compound adjectives. Fourth, while educated, organized, equipped and known can be modified by poorly and/or badly, experienced cannot. These four characteristics suggest that it is likely that experienced denotes a positive quality which renders the modification by well superfluous and the contradictory modification by poorly and badly unacceptable. By contrast, except for qualified, the other adjectives can be modified by adverbs denoting negative qualities, suggesting that they can be used neutrally, though they commonly denote positive qualities.
6. Implications for language teacher education
In the above discussion, we have seen that teachers often look for generalizations about grammar rules so that they can provide some guidelines to their students. This is perfectly legitimate especially in second/foreign language learning situations where learners do not have the same amount of exposure to the language as in first language learning situations. The problem is whether the rules and generalizations indeed capture how language is actually used rather than how language is perceived to be used, and whether they reflect the dominant patterns of use. The easy accessibility of corpora allows teachers to check prescribed rules and generalizations against linguistic data, it encourages them to be sensitive to patterns that emerge from the data and to make their own interpretations and generalizations of these patterns (see also Hunston, 1995).
Indeed, the constant use of corpus evidence in addressing such questions may help teachers to reflect on their knowledge of the language as well as critically examine grammar rules and patterns that they have always taken for granted. They may begin to look at corpus evidence for answers, instead of just relying on dictionaries and reference grammars. For example, messages like the following have emerged in the post quite frequently:
The process of using corpus evidence to investigate questions like the above may be beneficial for the teachers because in this process they will probably notice linguistic patterns and pragmatic loads carried by linguistic items that they might not be aware of, some of which may not be found in typical reference materials. The following is just one example, among many, of how a question from a teacher can lead to interesting discoveries of linguistic facts.
imply and Infer
These two words cause some confusion because some people use the word infer to mean imply, as observed by the Collins Cobuild English Dictionary (p. 862, entry infer). For example,
The police inferred, though they didn't exactly say it, that
A teacher posted a question regarding these two words in the English as a Second Language list:
Can anyone please provide some examples for me to
Thank you very much!
Another teacher responded by saying that to imply means "to suggest something indirectly" whereas to infer means "to guess something is the case or to conclude".
The explanation provided by the above teacher is very much in agreement with that provided by the Collins Cobuild English Dictionary for infer and imply, which is given below.
on the basis of information that you already have.
I infer from what she said that you have riot been well.
1. If you imply that something is the case, you say something
2. If an event or situation implies that something is the
By carefully looking at the semantics of infer and imply in the FROWN corpus I would like to suggest that if the teachers in question had access to corpus data they could not only clarify their doubts in relation to how these items are used (by relying on several examples authentically contextualized), but also note that there are a number of instances where imply is modalized, and where the writer makes it clear that what has been said does not imply what the reader has inferred. In fact, out of a total of 61 instances of imply found in the FROWN corpus, there are 27 instances of imply being modified by either modal verbs, may, might, would, could or lexicalized modality, such as seem to, appear to, tend to. For example:
There are 18 instances of negative forms of imply, such as does not imply, should not be taken to imply, need not to imply and so on. For example,
A search on the FROWN corpus for instances of infer showed that there is a similar tendency for infer to be modalized. There are 12 instances of infer, out of which nine co-occur with the modals can, would, may and lexicalized modality, such as tempting and tend to (two are in negative forms).
The common characteristics shared by imply and infer is that both pertain to what is not explicitly stated. Therefore, people tend to hedge statements about implications and inferences with modals or lexicalized modality. The much higher proportion of the negative form of imply as compared to that of infer suggests that there is a difference in the semantics of the two lexical items. The negative form of imply serves to pre-empt possible misinterpretations of what is not directly said. This kind of evidence may help teachers to better understand the difference in meaning between these two items spelled out in the Collins Cobuild English Dictionary. These characteristics would not have been easily detected without the help of the corpus and the concordancer showing the environments in which these two words occur.
Studies of applications of corpus linguistics to second/foreign language teaching and learning have emphasized the importance of adopting a data-driven approach to language learning so that learners go through a process of self-discovery (see for example Johns, 1991). The discussion in this paper attempts to shows that it might be equally important for teachers to go through this process of self-discovery and to experience formulating generalizations about linguistic patterns that they have observed so that they try to grasp the grammar as much as linguistic researchers do.
1 Orkut is defined as "an online community which connects people through a web of reliable friends". To make sure that abusive content (e.g. pornography, racism, pedophilia, etc…) does not circulate among the different communities the idealizers have created standards which are the shared values of the orkut.com community. According to the idealizers "the standards are a living document and will change based upon the needs of the broad community and the available tools. The Community Standards will be upheld through a combination of human and automated moderation. As you may have noticed, an automated abuse-detection system is already at work. The system temporarily suspends the accounts of individuals who are abusing the orkut.com community." (quoted from www.orkut.com)
2 For further information access http://khnt.hit.uib.no/icame/manuals/frown/INDEX.HTM.
Berry, Roger (1994). Using concordance printouts for language awareness training. In C. S. Li, D. Mahoney, & J. Richards (Eds.), Exploring Second Language Teacher Development (pp. 195-208). Hong Kong: City University Press. [ Links ]
Biber, Douglas (1988). Variation across Speech and Writing. Cambridge: Cambridge University Press. [ Links ]
Biber, Douglas; Conrad, Susan; Reppen, Rendi. (1994). Corpus-based approaches to issues in applied linguistics. Applied Linguistics, 15(2), 169-189. [ Links ]
Bygate, Martin; Tonkyn, Alan; Williams, Eddie (Eds.). (1994). Grammar and the Language Teacher. New York: Prentice Hall. [ Links ]
Carter, Ronald; McCarthy, Michael (1997). Exploring Spoken English. Cambridge: Cambridge University Press. [ Links ]
Carter, Ronald; McCarthy, Michael (2001). Size isn't everything: Spoken English corpus, and the classroom. TESOL Quarterly, 35(2), 337-340. [ Links ]
Collins Cobuild English Grammar (1990). London: Harper Collins. [ Links ]
Hawkins, Eric (1999). Foreign language study and language awareness. Language Awareness, 8(3/4), 124-142. [ Links ]
Holmes, Janet (1988). Doubt and certainty in ESL textbooks. Applied Linguistics, 9, 21-44. [ Links ]
Hunston, Susan (1995). Grammar in teacher education: The role of a corpus. Language Awareness, 4(1), 15-31. [ Links ]
Johns, Tim (1991). Should you be persuaded: Two examples of data-driven learning. In T. Johns & P. King (Eds.), Classroom Concordancing. ELR Journal 4 (pp. 1-16). Birmingham: CELS University of Birmingham. [ Links ]
Kennedy, Graeme (1998). An Introduction to Corpus Linguistics. London: Longman. [ Links ]
Kjellmer, Goran (1994). A Dictionary of English Collocations Based on the Brown Corpus, 3 Vols. Oxford: Clarendon Press. [ Links ]
Ljung, Magnus (1991). Swedish TEFL meets reality. In S. Johansson & A,-B. Stenstrom (Eds.), English Computer Corpora (pp. 245-256). Berlin: Mouton de Gruyter. [ Links ]
Mindt, Dieter (2000). An Empirical Grammar of the English Verb System. Berlin: Cornelsen. [ Links ]
Partington, Alan (1998). Patterns and Meanings. Using Corpora for English Language Research and Teaching [Studies in Corpus Linguistics 2]. Amsterdam: John Benjamins. [ Links ]
Quirk, Randolph; Greenbaum, Sydney; Leech, Geofrey; Svartvik, Jan (1985). A Comprehensive Grammar of the English Language. London: Longman. [ Links ]
Sinclair, John (1991). Corpus, Concordance, Collocation. Oxford, UK: Oxford University Press. [ Links ]
Tognini-Bonelli, Elena (2001). Corpus Linguistics at Work [Studies in Corpus Linguistics 6]. Amsterdam: John Benjamins. [ Links ]