A key to understanding why a text is difficult to process. Lexical uniqueness of academic English texts

Akey to understanding why atext is difficult to process Lexical uniqueness of academic English texts Natalia Borza ABSTRACT: While the register of English language tertiary textbooks has been investigated substantially, moderately little is explored about the register analytical features of secondary textbooks. The purpose of the present pedagogically-driven study is to analyse the register of biology textbooks for secondary students from the point of view of English as asecond language (ESL) teaching by describing the lexical uniqueness of the register of the biology corpus (BIOCOR) 10th-grade students need to process during their studies at abilingual secondary school. The BIOCOR (consisting of 7,021 words) was compared to areference corpus (REFCOR) of general English texts at aCEFR B2 level (comprising 7,098 words) by exploring its high-value positive and negative keyness lexical items. The results of the investigation disclose that the lack of specialised uniqueness is prevalent in the BIOCOR with regard to academic English and specific biology terminology. The lexical plainness of the biology textbook can be regarded as one of the linguistic features revealing the non-academic but popularizing nature of the secondary textbook register. KEYWORDS: academic English, English for specific purposes (ESP), keyness, lexical uniqueness, register analysis 1. RATIONALE AND THE RESEARCH QUESTION Students at an English-Hungarian bilingual secondary school in Budapest tend to face an academically challenging situation in the second year of their studies when they start to master what is required in the 10th grade nationwide. The current pedagogically driven research to investigate one of the possible linguistic sources of the problem is motivated by my experience as apracticing English language teacher having observed the regular reappearance of the same hardships among the 10thgraders. The present study analyses the written register of English-language biology textbooks for secondary students from the viewpoint of English as asecond language (ESL) teaching by describing the register of the biology corpus students need to process during their studies. The register analysis is expected to result in apool of data relevant for gaining pedagogical insights applicable by teachers instructing in the intensive English language preparatory year of the bilingual secondary school as to what extent the language foci of the preparatory year enable students to handle the language use of the biology texts 10th-graders are assigned to process. Besides gaining adeeper understanding of the 10th-grade bilingual students’ needs in terms of English language and thus supporting my own and my colleagues’ professional develOPEN ACCESS 10 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 opment as general English teachers, this exploratory and descriptive corpus-based study can provide insights for future biology English for specific purposes (ESP) teachers, once biology ESP has been included in the ‘zero year’ language programme of the secondary bilingual school. Although the present research launches aclose investigation into describing the language use of two types of texts at aparticular bilingual secondary school in Hungary, the results of the enquiry are not restricted to the secondary school at hand, they can be meaningfully transferred and applied by educators working in any English-language international school where some of the students are non-native students. Keeping the 10th-graders’ difficulty of tackling academic subjects in English in the foreground, the present pedagogically motivated study aims to investigate the following problem from alinguistic point of view: To what extent do the general English reading texts (referred to in the study in its acronym form as the REFCOR for short) assigned in the intensive language preparatory course in the 9th grade at an EnglishHungarian bilingual secondary school enable students to handle the biology texts used in the subsequent term (hereafter referred to as the BIOCOR)? Since the most outstanding linguistic features along which registers differ from one another is considered to be vocabulary (Atkins, Clear & Ostler, 1992; Biber, 1989, 1993; Sinclair, 1991), the present research reports on the characteristic linguistic features of the BIOCOR with regard to its lexical uniqueness. Accordingly, the paper attempts to answer the research question: What lexical uniqueness is characteristic of the BIOCOR in comparison with that of the REFCOR? 2. REVIEW OF THE LITERATURE To ensure that the current analysis yields data practically valuable for ESL and ESP teachers, the register analytical approach rather than the genre analytical one was adopted in the present research (for the underlying reasons see Borza, 2015). The use of computer technology renders register analysis more reliable (Biber & Conrad, 2009), thus computerized corpus-linguistical methods were applied in the current project. Within the frame of corpus linguistics, Biber (1988) introduced acomprehensive methodological approach to describing patterns in register variations, the computerized method of multidimensional analysis (MDA). This method aims at finding underlying linguistic parameters, or dimensions, as well as specifying linguistic parallels and dissimilarities among registers along the dimensions identified. MDA relies on multivariate statistical techniques, especially factor analysis, to investigate the co-occurrences of linguistic features when discovering systematic patterns of variation among registers. As is characteristic in the register approach, the complexity of linguistic features is emphasised in the process of obtaining adequate descriptions of registers. In line with the early recognition of the importance of linguistic co-occurrences (Brown & Fraser, 1979), MDA follows Biber’s (1988) observation that statistically significant linguistic features tend to cluster in texts as they share communicative functions. Consequently, the method finds it misleading to focus on specific, isolated linguistic features and does not investigate single parameOPEN ACCESS NATALIA BOrZA 11 ters individually.1 The MDA perspective aims at finding groups of linguistic features that co-occur in registers. To map registers onto the groups of linguistic markers or dimensions, texts in the corpora are automatically analysed for linguistic features representing numerous major grammatical and functional characteristics. After carrying out the quantitative, numerical analyses, the frequent (positive) and rare (negative) features in the dimensions detected through factor analysis are interpreted in terms of communicative functions. The qualitative analysis specifies how the language features with statistically significant values are well-suited to the communicative purposes of the text. Using the numerical and functionally interpretive method of MDA developed in Biber’s (1988) seminal work, numerous registers have been explored, among them are letters (Biber & Finegan, 1989), medical academic prose (Atkinson, 1992), 18th-century authors across different registers (Biber & Finegan, 1994b), spoken and written registers in avariety of languages (Biber & Finegan, 1994a; Biber, 1995), research articles and textbooks (Conrad, 1996), internet-based and computer-mediated communication (Herring, 1996), newspapers (Biber & Finegan, 1997), scientific prose (Atkinson, 1999), newspapers, magazine articles and medical writing (Vilha, 1999), elementary school writing (Reppen, 2001), disciplinary texts (Conrad, 2001), historical and contemporary registers (Conrad & Biber, 2001), speech and writing at university (Biber et al., 2002; Biber, 2006), radio and TV sports commentary (Reaser, 2003), biology research articles (Biber & Jones, 2005), university classroom talk (Csomay, 2005), biochemistry research articles (2007), blogs (Grieve et al., 2011), academic registers and sub-registers (Nesi & Gardner, 2012), movie language (Forchini, 2012). Applying Biber’s (1988) rather complex MDA for capturing register specific features has been challenged by Tribble’s (1999) proposition claiming that the application of the keyword function of WordSmith (Scott, 2008) could reveal similar patterns as MDA. Xia and McEnery (2005) investigated whether this assertion proves to be correct. Contrary to the most straightforward implication of the term ‘key words,’ they are not the most frequently used words in the register, neither are they the ones that carry the most important propositions in the text; however, key words make the text characteristically different compared to alarge reference or benchmark corpus. Key words can be identified through statistical comparison carried out by the keyness function of keyword programs. The test of keyness is predicated on alog-likelihood test, Dunning’s procedure (1993) most typically, which is not based on the presupposition that data have anormal distribution in the text (McEnery et al., 2006). Showing the lexical uniqueness of atext, keyword lists reveal register specificity by containing words that are either significantly frequent or on the other end of the spectrum, significantly infrequent in the collection of texts. In the first case the list allows to investigate positive keyness, that is, words and structures that make the target corpus different from alarger reference corpus, while the second list provides information about negative keyness, about the words, expressions and structures that are dramatically missing from the corpus under scrutiny compared to abenchmark corpus. 1 Biber’s (1988) work was ground-breaking in examining 67 linguistic features in 481 texts and 23 registers. OPEN ACCESS 12 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 Through investigating the effectiveness of the keyword function, Xia and McEnery (2005) endeavoured to find alabour-effective method that could substitute the rather complex and “extremely time-consuming” (McEnery et al., 2006, p.308) MDA procedure, which resists any simple characterisation. Although MDA is apowerful tool in register analysis, which has been used to uncover various registers as demonstrated above, it is undoubtedly demanding to carry out. The reason for its laborious nature is the fact that it requires the sophisticated statistical analysis of alarge number of linguistic features to identify the groups of features that co-occur in the text with high frequency. To show that MDA fails to be irreplaceable with aless arduous tool for register analysis, Xiao and McEnery (2005) undertook akeyword analysis to compare three registers (conversation, speech, and academic prose) by producing wordlists of corpus files extracted from large American corpora (the Santa Barbara Corpus of Spoken American English, the Corpus of Professional Spoken American English, and the Freiburg-Brown corpus of American English), which were compared to areference corpus, the British National Corpus, to detect and compile those words whose frequency differed from the reference corpus either by being unusually high (positive keywords) or extremely low (negative keywords). The results of their study confirmed that applying the keyword approach is capable of producing comparable results to the MDA approach and can identify important register patterns, despite creating aless nuanced comparative contrast of registers, which is “not as likely to work for finer distinctions among texts” (Conrad, 2015, p.318). 3. METHODS In order to make the study replicable and the results transferable, the following Methods Section comprises four sections. First, the steps of the keyness analysis are explicated (Section 3.1.1), then the methods of compiling the biology corpus under investigation are accounted (Section 3.1.2), which is followed by those of the reference corpus (Section 3.1.3). Next, it is explained why amini-corpus was adopted in the research (Section 3.1.4). 3.1 THE PrOCESS OF THE KEYNESS ANALYSIS OF THE COrPUS 3.1.1 THE STEPS OF KEYNESS ANALYSIS Although Biber’s (1988) multidimensional analysis (MDA) has along record of reliably uncovering linguistic patterns of registers, the present research follows amore recent analytical method which is considered to be areplacement of MDA (Tribble, 1999). The reason for choosing the keyword application of the WordSmith program (Scott, 2008) instead of carrying out aMD analysis on the BIOCOR is not simply due to the novelty of the software. In pragmatic terms, the decision was based on considering Xia and McEnery’s (2005) empirical research results. Their study proves that revealing keyness with the WordSmith program is amethod that provides comparable results to MDA since the new application can identify similar linguistic patOPEN ACCESS NATALIA BOrZA 13 terns among registers. In theoretical terms, the models of already identified dimensions to explore the characteristic linguistic features of texts by the application of MDA (Biber, 2001;2 Biber et al., 2014;3 Staples et al., 20184) fail to appear to be utterly relevant considering the focus of the present research. Neither the ESL teachers instructing 9th-grade students in the bilingual programme, nor ESP teachers in general would benefit directly from the linguistic data of these dimensions in their teaching practice. The reason why the above-mentioned dimensions do not fully address the key aspects relevant in the present educational setting might lie in the fact that these dimensions were identified with the aim of finding generalizable parameters of linguistic variations for adifferent discourse domain: 1) general parameters of variations among spoken and written registers in English and 2) patterns of grammaticality and lexico-grammatical characteristics in university student writing/speaking tasks, that is, productive skills were in the focus, which are greatly different from the receptive skill of processing reading tasks. Thirdly, the multivariate statistical technique on which MDA is based is factor analysis, which is asophisticated method that can be applied effectively to large corpora. In practical terms, factor analysis does not work effectively on amini-corpus, thus the current corpus of 7,000 running words cannot be investigated fruitfully along the Biberian lines of factor analysis. Keyness describes the distinguishing lexical characteristics of aregister by comparing its language use to that of another register (Xia & McEnery, 2005). The keyword application of WordSmith version 5 (Scott, 2008) was used in the present research to extract lexical items that are present in the BIOCOR, ones which are, however, not typically used in the REFCOR. That is, keyness results show the lexical uniqueness of the BIOCOR by compiling lexical items that make the register markedly different from the REFCOR. Inversely, the keyness application was also applied to collect lexical items that are underrepresented in the BIOCOR compared to the REFCOR, which set of lexical tokens is labelled as displaying negative keyness values. 2 Seven dimensions of the Biberian multidimensional analysis (2001) 1) Involved versus informational 2) Narrative versus nonnarrative 3) Elaborated reference versus situation dependent reference 4) Overt expression of argumentation 5) Abstract style versus nonabstract style 6) Online informational elaboration marking stance 7) Academic hedging 3 Four dimensions of the multidimensional analysis by Biber et al. (2014) 1) Literate versus oral response 2) Information source: Text versus personal experience 3) Abstract opinion versus concrete description/summary 4) Personal narration 4 Four dimensions of the multidimensional analysis by Staples et al. (2018) 1) Compressed procedural information versus stance toward the work of others 2) Personal stance 3) Possible versus completed events 4) Informational density OPEN ACCESS 14 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 It needs to be underlined that keyness, either positive or negative, does not reveal lexical items that frequently or infrequently occur in the BIOCOR but ones which are characteristically different with respect to their frequencies when the register is compared to the REFCOR. Using this method ensured that lexical items which are not register specific, ones which occur with similar frequencies in both corpora, such as the, of, or and, are not compiled. Keyness is determined by statistical comparison carried out by keyword programs. Aword is considered to be key if its frequency in the corpus when compared with its frequency in the reference corpus is such that the statistical probability as computed by the appropriate procedures described below is smaller than or equal to ap value of 1E–6 (Scott, 2008).5 To compute the keyness of an item, WordSmith version 5 (Scott, 2008) calculates four values, which are consequently cross-tabulated. The four values include the raw frequency of the item in the corpus, the number of running words in the corpus, the raw frequency of the item in the reference corpus, and the number of running words in the reference corpus. The statistical procedure of finding key words includes the chi-square test of significance with Yates’ correction for continuity to reduce the error in approximation. The test of keyness in the case of the WordSmith program (Scott, 2008) relies on alog-likelihood test, Dunning’s procedure (1993). The fact that Dunning’s procedure is not based on the presupposition that data have anormal distribution in the text (McEnery et al., 2006) increases the instrument’s reliability. The application of alog-likelihood test, disfavouring normal distribution, was especially important in the present research environment, where the REFCOR does not contain considerably more running words than the target corpus but was compiled to be approximately of the same size as the BIOCOR. WordSmith version 5 (Scott, 2008) treats words which are not represented in the REFCOR as if they occurred 5.0E–324 times (that is 5.0×10–324) in the baseline corpus. To apply akeyword program that assigns such asmall value to non-represented lexical items in the corpora was adecisive factor in the choice of the software. Without this slight modification, uncovering stark contrasts between the two registers would have been impossible since cross-tabulation with values of zero does not produce any meaningful result. An infinitesimally small number, however, allows for the handling of lexical items that do not occur in either of the two corpora, and due to the number’s close-to-zero value, it does not affect the calculation materially. To ensure reliability, WordSmith version 5 (Scott, 2008) defines those items as key whose p value is smaller than or equal to 1E–6, that is 0.000001. The p value shows the danger of being ungrounded when claiming relationships. Consequently, an extremely low p-value threshold increases reliability. In the present case the chance of erroneously listing words with similar frequency in the two corpora as key words is 0.00001%. In order to arrive at data which are practically useable for ESL teachers instructing in the bilingual programme of the school and for biology ESP teachers alike, words of the same root were lemmatized by the keyword program before running the keyword application. That is, keyness values were determined for word families rather than 5 1E–6 is astandard scientific notation for the value of one times 10 to the power of –6, which equals one over 1 million, or 0.000001. OPEN ACCESS NATALIA BOrZA 15 for individual word forms. Lemmatization was treated as fundamentally essential since the investigation of word families produces more useful data for ESP teachers than that of conjugated verb forms and various word formations in the process of working out the lexical dimension of ESP syllabi.6 The same argument supports the practical reason why word lists for learners of English also tend to group words into families (West, 1953; Xue & Nation, 1984). Besides, compiling words in word families instead of listing isolated elements of different word forms was chosen for theoretical reasons too, namely, word families form aunit in the mental lexicon (Bauer & Nation, 1993; Nagy et al., 1989). Lemmatization rendered the following different word forms as one group: — singular and plural forms, e.g., cell— cells, parasite— parasites, segment— segments; — nominative and genitive forms, e.g., mosquito— mosquito’s; — regular inflections of the verb (verbs in different tenses), e.g., cause— caused, reproduce— reproduces; — verbs and gerunds, e.g., spread— spreading; — base, comparative and superlative adjectives, e.g., small— smaller— smallest; — derivations of the word: amoeba— amoebic, blood— bleeding, chemicals— chemically, class — classify, contract— contractile— contraction, dead— death— die, digestive— digestion, granules— granular, saliva— salivary, slime— slimy. Yet compound words were not joined in one batch, thus flat and flatworm, stream and streamlined for instance were computer-counted separately. The reason for not lemmatizing compound words lies in the strong possibility that the parts of the compounds cover relatively distant meanings, for instance cow and cowslip or Mary and marigold. After running the appropriate statistical procedures of the keyness software, the key words of the BIOCOR were listed by the software in rank order. The computercounted keyness values of the lemmatized items on the list reveal to what extent the frequency of the particular item is different when compared to that in the REFCOR. Subsequently, the key words were manually correlated to the most frequently occurring lexical items in the BIOCOR (for the most frequently occurring lexical items in the BIOCOR see Borza, 2014). Such acorrelation was considered to be important in order to find out more in depth about the nature of the biology register. The most prevalent words in the BIOCOR were recorded in rank order, and arranged in frequency bands. Band 1 contains the most pervasive, most frequent words in the BIOCOR, the ones which are used no fewer than 30 times, while Band 10 comprises more rare items, word families which appear four times. Table 1 shows the frequency of items in particular bands, expressed both in the number of their raw occurrences and in percentages. 6 For insights regarding the working out of the grammar dimension of the ESP syllabi, e.g., tenses, modal auxiliaries, active-passive voice, sentence complexity, see other research such as Borza 2013, 2016. OPEN ACCESS 16 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 Rank order Raw frequency of lemmas Frequency of lemmas Band 1 30 or more 0.42% or more Band 2 20–29 0.28%– 0.41% Band 3 15–19 0.21%– 0.27% Band 4 12–14 0.17%– 0.20% Band 5 10–11 0.14%– 0.15% Band 6 8–9 0.12%– 0.13% Band 7 7 0.10% Band 8 6 0.08% Band 9 5 0.07% Band 10 4 0.06% Table 1:The frequency bands in the BIOCOR. Individual lexical items and lemmatized tokens which occur fewer than four times in the BIOCOR were not compiled in this investigation. The reason for disregarding low-frequency lexical items is the assumption that in an informational, educational register, such as that of the biology textbook for secondary school students, essential lexical items appear repeatedly in order to fulfil the textbook’s instructional function. Next, in order to gain abetter understanding of the degree of the use of specific lexis in the register, the key words were classified into three categories: biology terms, academic English and general English. The category of biology terms contains lexical items which have aspecific meaning within the context of biology, ameaning or ashade of meaning which differs from the everyday use of the word. Acurrent dictionary of biology (Thain & Hickman, 2004) served as areference point in determining if alexical item is to be categorized as abiology term or if the biology related word falls into the category of general English. Thain & Hickman’s biology dictionary (2004) was applied as the baseline of categorization since its entries, according to the dictionary’s editors, venture to explain the most indispensable notions in biology for teachers and students alike, that is, its selection of information is perfectly relevant in the present educational setting. Words and expressions which appeared as separate entries in the biology dictionary were grouped as biology terms. Every single member of alemmatized word family was checked in the dictionary in order to ensure that word classes did not affect the labelling of biology terms. For instance, the noun reproduction appears as an entry in the biology dictionary; however, the verb reproduce does not. In this case the lemmatized word family including the items reproduce, reproduction, reproductive was labelled as abiology term. Yet multi-word dictionary entries, where alexical item was the head of the entry in conjunction with other words, were not grouped as biology terms unless they were present in the BIOCOR with the exact same word combinations. For example, the word body is not aseparate entry in the biology dictionary, while the lexical item carotid body is. Accordingly, the word body was not categorized as abiology term in the present research unless it was used in the BIOCOR in conjunction with the word carotid. The label of academic vocabulary was given to those lexical items that appeared on Coxhead’s (2000) extensive list of academic vocabulary comprising 570 word families. OPEN ACCESS NATALIA BOrZA 17 Coxhead’s academic word list (AWL) was selected to be used in the present research since it is asystematic collection of academic English, aset of wide-ranging lexis typically used in the register of academic English. Furthermore, the list is applied with ahigh rate of validity in the present research environment as the collection of lexical items was particularly compiled for pedagogical purposes. The AWL was gathered in order to provide insights for English teachers preparing students for their tertiary studies in English as Coxhead aimed at showing what specific lexis was prevalent in academic text. Thus, the AWL accords well with the educational context of the current research as it is concerned with the teaching applications to improve second language students’ success in an academic environment when studying disciplines in English. The AWL has been proven to pinpoint the collection of lexical items that makes academic registers markedly different from other registers (Coxhead, 2000), thus it is areliable instrument to find academic vocabulary in texts in English. The corpus in which the frequency of words was run by Coxhead (2000) embraces four sub-corpora of the following faculty sections: arts, commerce, law, and science. Each of these faculty sections are further divided into seven subject areas. Biology is one of the subject areas of the science sub-corpus, which allows its use as abaseline in the present research environment with ahigh rate of construct validity. The AWL contains word families that appeared in over half of the twenty-eight subject areas. Words that occurred in fewer than fifteen of the subject areas were labelled as narrow range words and were excluded. This principle ensured that the list could be used for any academic subject area, its coverage is not restricted to specific subjects. In the development of the list, frequency played akey role, word families that were used more than 100 times in the 3,500,000-word-long corpus were shortlisted. Basic vocabulary, words that are among the first 2,000 most frequently occurring words of English as compiled by West in his General Service List (1953), were not involved in the short list, since academic reading presupposes the learner’s familiarity with basic vocabulary at tertiary level. From this respect, AWL is advantageous to be used in the current research environment since 10th-grade students are also expected to be familiar with the most widely used words in general English. This similarity ensures ahigh rate of criterion related validity for the present research. Besides basic lexis, proper nouns, for example names of places and people, as well as Latin forms, such as etc., i.e., were also removed from the AWL short list. Finally, the list was organized into ten sublists based on the frequency of the particular word family. The sublists were numbered consecutively, where sublist one contains the most common academic words in the corpus, while sublist ten comprises less frequent academic lexis. The present research uses Coxhead’s (2000) findings in order to see whether the biology texts assigned to 10th-grade students in the bilingual secondary school are difficult to read due to the fact that they contain alarge number of academic lexical items among their key words. Thirdly, lexical items which failed to fit either in the category of biology terms or in the group of academic vocabulary were assigned the label general English. Highkeyness lexical items within the general English category were collected and listed in order to help general English teachers and biology ESP teachers gain knowledge about the nature of the general English lexis used in the biology textbook for secondary school students. OPEN ACCESS 24 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 into aseries of segments). The lemma also shows the possibility of being combined in aprepositional phrase (e.g., in each segment). The lemma genus, with the third highest keyness value (k=31), shows arather scarce variety of collocations. It combines only in noun phrases and verb phrases (see Table 7). There is one single noun with which it goes together in the BIOCOR (name). In asimilar fashion, neither is the number of verbs it collocates with more numerous, since it is used in no more than one verb collocation, with the verb belong. In anoun phrase genus name name of the genus Verb it collocates with belong to agenus Table 7:Lexical environment of the biology term genus in the BIOCOR. The token intestine, with an outstandingly high keyness value (k=29), forms word combinations within anarrow range (see Table 8). It appears in noun phrases which refer either to its type, large intestine or small intestine, or to its structure, wall of the intestine. The number of verbs it combines with is even less manifold; the lemma appears only within one verb phrase, live in the intestine. In anoun phrase the host’s digested food the host’s digestive juices the host’s faeces intermediate host Verb it collocates with an intermediate host carries it Table 5:Lexical environment of the biology term host in the BIOCOR. In anoun phrase gut segments mature segments new segments the youngest segment Verb it collocates with to pass asegment produce segments rings divide the body up into segments segments drop off segments mate segments reach the rear end of the worm With averb in the passive voice the body is divided up into aseries of segments Prepositional phrase in each segment Table 6:Lexical environment of the biology term segment in the BIOCOR. OPEN ACCESS NATALIA BOrZA 25 In anoun phrase large intestine small intestine wall of the intestine Verb and aprepositional phrase live in the intestine Table 8:Lexical environment of the biology term intestine in the BIOCOR. The next significantly high keyness value item (k=28), drug, is applied in the BIOCOR in aslightly more versatile way (see Table 9). It forms verb combinations both in the active voice (drugs save lives) and in the passive voice (drugs are taken and be treated with certain drugs). Also, the lemma is capable of forming an adjective phrase with resistant. Verb it collocates with drugs save lives Adjective it collocates with resistant to drugs With averb in the passive voice drugs are taken be treated with certain drugs Table 9:Lexical environment of the biology term drugs in the BIOCOR. The lemma gut, with ahigh keyness value (k=27), shows adiverse set of lexical collocations (see Table 10) in the BIOCOR. It appears in various noun phrases (e.g., human gut and gut wall) and verb phrases alike (the gut has aspecial region, or gut segments contain). Besides, the token is also used as areference to location in prepositional phrases, such as above the gut or beneath the gut. In anoun phrase animal’s gut human gut gut wall Verb it collocates with the gut has aspecial region gut segments contain Prepositional phrase above the gut beneath the gut in the gut Table 10:Lexical environment of the biology term gut in the BIOCOR. The last lemma with asignificantly high keyness value (k=24), agar, appears in relatively few combinations (see Table 11) in the BIOCOR. There is one single noun phrase it forms (agar jelly), and its verb collocations is no more miscellaneous, there being only one verb with which it collocates in the passive voice (the agar is put in petri dish). OPEN ACCESS 26 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 In anoun phrase agar jelly With averb in the passive voice the agar is put in petri dish Verb and aprepositional phrase grow bacteria on the agar put bacteria on the surface of the agar Table 11:Lexical environment of the biology term agar in the BIOCOR. Besides the above listed biology terms, the BIOCOR contains no other subject specific terms with significantly high keyness values. Among the general English items with high keyness value, however, there are two lemmas worthy of attention. The token figure (k=34) is notable from the point of view of the ESL and biology ESP teacher, since its meaning in the BIOCOR (data or number) is different to agreat degree from its similarly-formed Hungarian version (figura, which only conveys the meaning of bodily shape). The other conspicuous high-keyness word (k=26) is the proper noun John, which is hardly expected to be an item distinguishing the biology register from the register of general English reading tasks. The reason behind the high rate of appearance of the proper noun in the BIOCOR compared to its use in the REFCOR is the fact that the biology texts incline to use vivid sample situations instead of providing theoretical explanations for the teenage target readers. The exemplified imaginary person in these situations is called John, which makes the lemma’s frequency of occurrence in the BIOCOR extremely high. 4.3 NEGATIVE KEYNESS Lemmas with high negative keyness value reveal words which are systematically untypical in aparticular register compared to areference corpus. In the present study, lemmas with high negative keyness value show the set of lexical items which occur in the REFCOR but are significantly less often used in the BIOCOR. In other words, tokens with high negative keyness value shed light on aspecial group of words which 9th-grade students process during their general English studies: it is the collection of word families which are underrepresented (or not present at all) in the biology texts the students read the following term. The findings of running the keyword application of WordSmith version 5 (Scott, 2008) strikingly show that the BIOCOR contains no such item. Notably, there is not one single lemma in the BIOCOR with significantly high negative keyness value when compared to the REFCOR. 4.4 HIGH-FrEQUENCY LOW-KEYNESS WOrDS It is not insignificant to take note of the fact that the BIOCOR encompasses agreat many frequently occurring lemmas that do not appear among the word families with high-keyness value (for an extensive list of words frequently applied in the BIOCOR see Borza, 2014). This group of words, the set of high-frequency low-keyness items, show that the majority of the frequently used lexis of the BIOCOR is present in the REFCOR with asimilar rate of frequency. Table 12 displays the collection of all these words, shows each item’s frequency expressed in frequency bands, as well as the type OPEN ACCESS NATALIA BOrZA 27 of the lexical item (biology term, academic English or general English). It can clearly be seen that the group of high-frequency low-keyness words embraces nearly exclusively general English terms; only two instances of biology terms occur (growth and reproduce) and there are no academic terms at all. Key word Type Band growth biology term 1 animal general English 1 get general English 1 reproduce biology term 2 do general English 2 make general English 2 person general English 2 small general English 2 way general English 2 see general English 3 use general English 3 cause general English 3 contain general English 3 move general English 3 place general English 3 shape general English 3 water general English 3 Table 12:High-frequency low-keyness words in the BIOCOR. 5. CONCLUSION The present study investigated the lexical uniqueness of biology texts for secondary students used in abilingual secondary school in the 10th grade (BIOCOR) from the point of view of English language teaching. The pedagogically-driven research analysed the lexical characteristics of the BIOCOR by unveiling its high-keyness value lexical items. The aim of discovering and describing the key lexical features of the BIOCOR was twofold: i) to gain insights into the lexical difficulties which might pose obstacles to 10th-grade students in smoothly processing the corpus and ii) to collect information about the lexical uniqueness of the corpus which is applicable for ESL (English as asecond language) and ESP (English for specific purposes) teachers in the process of building the lexical dimension of the syllabus of an intensive language preparatory course and that of abiology ESP course, respectively. The keyness characteristics of the BIOCOR provide revealing information about the register of the biology textbook for secondary school students. 1) The keyness results uncover that there is anearly absolute scarcity of academic words among the key lexis. That is, the lexis of the BIOCOR can hardly be distinguished from that of the REFCOR on account of the use of academic English terms. OPEN ACCESS 28 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 This finding goes contrary to the expectations of the biology teachers and the students of the bilingual programme alike, who expressed their certainty about the biology texts being abundant in academic vocabulary, which makes the register of biology texts starkly different from other registers in their perception (Cserép, 1997). Based on the findings, the difficulty of processing the BIOCOR cannot originate from the texts’ extensive use of academic English. 2) The great majority of biology key words appear with high frequency in the BIOCOR. This indicates that 10th-grade students are expected to read astring of texts which contains recurrently repeated biology key words. In other words, the students’ difficulty of processing the biology texts is hard to be accounted for by the students’ unfamiliarity with the biology lexis due to the sporadic appearance of the specific lexis. 3) The noun John is used so extensively in the BIOCOR that it appears among the key words. The abundance of the proper noun demonstrates that the register intends to convey and clarify its subject information more through practical examples than through highbrow, scholarly theoretical lines of thought. This tendency is in line with Shapiro’s (2012) findings highlighting the fact that the register of science textbooks written for secondary students is more popularizing than academic. As aresult, the language of pre-college textbooks for secondary students, who are non-experts in sciences, is less technical than that of tertiary textbooks, which are used in the discourse community of would-be scientists. 4) The BIOCOR contains no lemmas with high negative keyness value at all, in other words, there are no lexical items which occur significantly less often in the BIOCOR than in the REFCOR. That is, the register of the biology texts cannot be distinguished from the REFCOR in this respect, there are no significantly underrepresented general English lexical items. From the point of view of the ESL teacher, this similarity signifies that the vocabulary of the general English reading tasks assigned in the 9th grade cannot be characterized by asuperfluously expanded vocabulary in comparison with the lexis used in the biology texts. 5) Finally, the BIOCOR can be characterized by abounteous use of not register specific frequently occurring words, which are also frequently present in the REFCOR. This indicates that by the time students start pursuing their biology studies in the 10th grade they have already become familiar with agreat part of the lexis of the BIOCOR through reading the texts of the REFCOR in the 9th grade. Thus, processing the REFCOR in the ‘zero year’ provides afirm linguistic grounding for the students. Considering the lexical dimension of the language preparatory programme, reading the texts assigned in the 9th grade prepares bilingual students substantially for their academic studies in English the following year. To draw pedagogical implications, it is important to point out, however, that the level of difficulty of the general lexis in the BIOCOR does not go beyond the CEFR B2 level. Since the CEFR level of the lexis of the BIOCOR ranges from A1 to B2,8 language preparatory courses should not necessarily aim at more advanced levels. 8 The online software developed by the Lifelong Learning Programme of two departments of the University of Cambridge (Cambridge University Press and Cambridge English Language Assessment, http://vocabulary.englishprofile.org) was applied to define the CEFR levels of the particular lexical items. OPEN ACCESS NATALIA BOrZA 29 Considering all the aspects of the keyness results above, the lexis of the BIOCOR can hardly be described as challenging for 10th-grade bilingual students. The BIOCOR fails to show amore intriguing complexity in its key vocabulary than the REFCOR. The results of the research reveal that the lack of specialised uniqueness is prevalent in the BIOCOR with regard to academic English and specific biology terminology. The lexical plainness of the biology textbook can be regarded as one of the linguistic features typical of the register of non-academic but popularizing secondary textbooks. The prevailing lexically straightforward character of the BIOCOR, however, suggests that the perceived challenges 10th-grade bilingual students face during their studies in English are not explicable in terms of the lexis of their textbook; that is, they should stem from adifferent source. REFERENCES: Atkinson, D. (1992). The evolution of medical research writing from 1735 to 1985: The case of the Edinburgh Medical Journal. Applied Linguistics, 13, 337–374. Atkinson, D. (1999). The philosophical transactions of the Royal Society of London, 1675–1975: Asociohistorical discourse analysis. Language in Society, 25, 333–371. Atkins, S., Clear, J., & Ostler, N. (1992). Corpus design criteria. Literary and Linguistic Computing, 7, 1–16. Bauer, L., & Nation, I.S.P. (1993). Word families. International Journal of Lexicography, 6, 253–279. de Beaugrande, R. (2001). Large corpora, small corpora, and the learning of language. In M.Ghadessy (Ed.), Small Corpus Studies and ELT. Theory and Practice (pp. 3–28). Philadelphia, PS: John Benjamins. Biber, D. (1988). Variation across Speech and Writing. Cambridge: Cambridge University Press. Biber, D. (1989). Atypology of English texts. Linguistics, 27, 3–43. Biber, D. (1993). Representativeness in corpus design. Literary and Linguistic Computing, 8, 243–257. Biber, D. (1995). Dimension of register variation: Across-linguistic comparison. Cambridge: Cambridge University Press. Biber, D. (2001). Multi-dimensional methodology and the dimension of register variation in English. In S.Conrand & D.Biber (Eds.), Variations in English: Multi-dimensional Studies (pp. 13–42). London: Longman. Biber, D. (2006). University Language: ACorpusbased Study of Spoken and Written Registers. Amsterdam: John Benjamins. Biber, D., & Jones, J. (2005). Merging corpus linguistics and discourse analytic research goals: Discourse units in biology research articles. Corpus Linguistics and Linguistic Theory, 1, 151–182. Biber, D., & Conrad, S. (2009). Register, Genre, and Style. Cambridge: Cambridge University Press. Biber, D., & Finegan, E. (1989). Drift and the evolution of English style. Language, 65, 487–517. Biber, D., & Finegan, E. (1994a). Sociolinguistic perspectives on register. New York: Oxford University Press. Biber, D., & Finegan, E. (1994b). Multidimensional analyses of authors’ styles: Some case studies from the eighteenth century. In D.Ross & D.Brink (Eds.), Research in Humanities Computing (pp. 3–17). Oxford: Oxford University Press. Biber, D., & Finegan, E. (1997). Diachronic relations among speech-based and written registers in English. In T.Nevalainen & L.Kahlas-Tarkka (Eds.), To Explain the Present: Studies in the Changing English Language (pp. 253–275). Helsinki: Societe Neophilologique. OPEN ACCESS 30 STUDIE ZAPLIKOVANÉ LINGVISTIKY 1/2020 Biber, D., Gray, B., & Staples, S. (2014). Predicting patterns of grammatical complexity across language exam task types and proficiency levels. Applied Linguistics, 37(5), 1–31. Biber, D., Conrad, S., Reppen, R., Byrd, P., & Helt, M. (2002). Speaking and writing in the university: Amultidimensional comparison. TESOL Quarterly, 36, 9–48. Borza, N. (2013). Register analysis of biology texts: Acorpus-based exploratory study of grammar. Working Papers in Language Pedagogy, 7, 29–47. Borza, N. (2014). Does specific lexis make biology texts difficult? Acorpus-based lexical analysis of the register of biology texts. Practice and Theory in Systems of Education, 9(2), 181–196. Borza, N. (2015). Analysing ESP texts, but how? Practice and Theory in Systems of Education, 10(1), 1–15. Borza, N. (2016). The bewildering complexity of the biology register: An investigation into the syntactic complexity of secondary biology texts. Research in Corpus Linguistics, 4, 9–24. Brown, P., & Fraser, C. (1979). Speech as amarker of situation. In K.R.Scherer & H.Giles (Eds.), Social Markers in Speech (pp. 33–62). Cambridge: Cambridge University Press. CEFR: Council of Europe. Language Policy Unit, Modern Languages Division. (1996). Common European Framework of Reference for Languages: Learning, Teaching, Assessment. Strasbourg: Cambridge University Press. Conrad, S. (1996). Investigating academic texts with corpus-based techniques: an example from biology. Linguistics and Education, 8, 299–326. Conrad, S. (2001). Variation among disciplinary texts: Acomparison of textbooks and journal articles in biology and history. In S.Conrad & D.Biber (Eds.), Variations in English: Multidimensional Studies (pp. 94–107). London: Longman. Conrad, S. (2015). Register variation. In D.Biber & R.Reppen (Eds.), The Cambridge Handbook of English Corpus Linguistics. Cambridge: Cambridge University Press. Conrad, S., & Biber, D. (2001). Variations in English: Multi-dimensional Studies. London: Longman. Coxhead, A. (2000). Anew academic word list. TESOL Quarterly, 34(2), 213–238. Cserép, S. (1997). Technical terms in biology. An investigation into scientific English. Unpublished master’s thesis, Budapest: University of Economic Sciences. Csomay, E. (2005). Linguistic variation within university classroom talk: Acorpus-based perspective. Linguistics and Education, 15, 243–274. Dunning, T. (1993). Accurate methods for the statistics of surprise and coincidence. Computational Linguistics, 19, 61–74. Flowerdew, J. (Ed.). (2002). Academic Discourse. Harlow: Longman. Forchini, P. (2012). Movie Language Revisited: Evidence form Multi-dimensional Analysis and Corpora. Bern: Peter Lang. Grieve, J., Biber, D., Friginal, E., & Nekrasova, T. (2011). Variation among blogs: Amultidimensional analysis. In A.Mehler, S.Sharoff & M.Santini (Eds.), Genres on the Web (pp.303–322). Dordrecht: Springer. Herring, S.C. (1996). Computer-mediated Communication: Linguistics, Social and Cross-cultural Perspectives. Amsterdam: John Benjamins. Hoey, M. (2005). Lexical Priming: ANew Theory of Words and Language. London: Routledge. Kanokshilapatham, B. (2007). Rhetorical moves in biochemistry research articles. In D.Biber, U.Connor & T.Upton (Eds.), Discourse on the Move (pp. 73–103). Amsterdam: John Benjamins. McEnery, A., Xiao, R., & Tono, Y. (2006). Corpus-based Language Studies. London: Routledge. Ma, K.C. (1993). Small-corpora concordancing in ESL teaching and learning. Hong Kong Papers in Linguistics and Language Teaching, 16, 11–30. Nagy, W., Anderson, R., Schommer, M., Scott, J., & Stallman, A. (1989). Morphological families in the internal lexicon. Reading Research Quarterly, 24, 262–281. OPEN ACCESS NATALIA BOrZA 31 Nesi, H., & Basturkmen, H. (2006). Lexical bundles and discourse signalling in academic lectures. International Journal of Corpus Linguistics, 11(3), 283–304. O’Keffee, A., & McCarthy, M. (2010). The Routledge Handbook of Corpus Linguistics. London: Routledge. Prodromou, L. (1998). First Certificate Star. Oxford: Macmillan Publishers Limited. Reaser, J. (2003). Aquantitative approach to (sub)registers: The case of sports announcer talk. Discourse Studies, 5(3), 303–321. Reppen, R. (2001). Register variation in student and adult speech and writing. In S.Conrad & D.Biber (Eds.), Multi-dimensional Studies of Register Variation in English, 187–199. Harlow: Pearson Education. Roberts, M.B.V. (1981). Biology for Life. Surrey: Thomas Nelson and Sons. Scott, M. (2008). WordSmith Tools (Version5) [Computer software]. Liverpool: Lexical Analysis Software. Shapiro, A.R. (2012). Between training and popularization: Regulating science textbooks in secondary education. Isis, 103(1), 99–110. Sinclair, J. (2004). Trust the Text: Language, Corpus and Discourse. London: Routledge. Staples, S., Biber, D., & Reppen, R. (2018). Using corpus-based register analysis to explore the authenticity of high-stakes language exams: Aregister comparison of TOEFL iBT and disciplinary writing tasks. The Moderns Language Journal, 102(2), 310–332. Thain, M., & Hickman, M. (2004). The Penguin Dictionary of Biology. London: Penguin Books. Tribble, C. (1999). Writing Difficult Texts. Unpublished doctoral dissertation, Lancaster: Lancaster University. Vilha, M. (1999). Medical Writing: Modality in Focus. Amsterdam: Rodopi. West, M. (1953). AGeneral Service List of English Words. London: Longman. Xia, Z., & McEnery, A. (2005). Two approaches to genre analysis. Journal of English Linguistics, 33(1), 62–82. Xue, G., & Nation, I.S.P. (1984). Auniversity word list. Language Learning and Communication, 3, 215–229. Natalia Borza | Pázmány Péter Catholic University, Budapest, Hungary <[email protected]> OPEN ACCESS