Entorno 11. Sinónimos en la traducción: un argumento a favor del uso de textos auténticos

Sinónimos en la traducción: un argumento a favor del uso de textos auténticos para diferenciar entre significados similares

SYNONYMS IN UNDERSTAND COMPREHEND

translation:

An argument in favor of using authentic texts to differentiate between similar meanings

TAKE IN GRASP SEIZE

RESUMEN:

El estudio presentado en este artículo explora el uso de análisis de corpus como herramienta para la traducción de sinónimos. Primero, el autor explica por qué la traducción a través de diccionarios no siempre es suficiente para la desambiguación de significados de dos palabras similares. Segundo, el autor investiga el uso de las palabras ‘understand’ y ‘comprehend’ con un análisis cualitativo y cuantitativo usando el British National Corpus para demostrar que esta herramienta podría ser muy útil para un traductor con el propósito de llevar a cabo la diferenciación de significados de dos palabras aparentemente sinónimos.

PALABRAS CLAVE

Traducción · análisis de corpus · desambiguación de palabras

ABSTRACT:

The present study explores the use of corpus analysis for the translation of synonyms. The article first discusses how dictionary translations do not always sufficiently differentiate between the meaning of two similar words and goes on to investigate the meaning of the near-synonyms ‘understand’ and ‘comprehend’ both from a qualitative and quantitative perspective using the British National Corpus. The article concludes with the suggestion

that corpus analysis can be a useful tool for translators to disambiguate seemingly similar meanings.

KEY WORDS:

Translation ·corpus analysis · word disambiguation

INTRODUCTION

Using a large collection of authentic, computer-readable texts to extract patterns of actual language use is a rather recent development in applied linguistics. As such, it constitutes a new method to approach language, referred to as corpus linguistics, rather than a theory of language (McEnery and Hardie). The analysis of a corpus represents one important «source of evidence for improving descriptions of the structure and use of languages» (Kennedy 1) since, as Biber, Conrad, and Reppen aptly point out, our analyses of linguistic usage patterns «cannot rely on intuitions or anecdotal evidence [alone since,] in many cases, humans tend to notice unusual occurrences more than typical occurrences» (3). With this article, I am suggesting that translators refer to a corpus «to compare the use of apparent translation equivalents» ( Hunston 13) when faced with the decision of how to transfer meaning from one language to another instead of solely relying on a dictionary.

To support my argument, the present study investigates the meanings of the synonyms ‘understand’ and ‘comprehend’ in order to demonstrate how a corpus analysis can successfully distinguish between the similar meanings of these verbs. First, both verbs are studied using three online resources, the Merriam-Webster dictionary, WordReference, and the Collins dictionary, and the online translator Google Translate. Next, I provide a brief overview of corpus linguistics explaining the method I used to distinguish between the usage patterns of the two verbs before discussing the qualitative and quantitative corpus analyses in greater detail in section 4.

Por: Brita Banitz

FIGURE OUT

KNOW

UNDERSTAND

BEGREIFEN

Synonyms in translation

As a non-native speaker of English translating into German, my mother tongue, I have always wondered when to use ‘understand’ and when to use ‘comprehend’ since these two verbs are often treated as synonyms. According to Fromkin, Rodman, and Hyams, «synonyms are words or expressions that have the same meaning in some or all contexts» (196). However, perfect synonyms, two words with the exact same meaning, do not exist since «it would be inefficient for a language to have two words […] with absolutely identical meanings» (O’Grady, Archibald, Aronoff, and Rees-Miller 202). In order to differentiate between the two, I was taught at some point that ‘comprehend’ refers to some kind of a deeper, more profound understanding of something. However, that definition is, in my mind, too vague to give me any specific guidelines as to how to decide which verb to use in which context. However, as the following discussion illustrates, consulting a dictionary may not always provide a clear solution either. As a first step to distinguish between the verbs ‘comprehend’ and ‘understand’, I consult-

ed the online Merriam-Webster dictionary1, a well-known reference of the English language. The dictionary entries for both ‘comprehend’ and ‘understand’ indicate that the verbs are often used interchangeably but that while ‘comprehend’ stresses the « process of coming to grips with something intellectually » (Merriam-Webster.com), ‘understand’ is more often used to refer to the « fact of having attained a firm mental grasp of something » (Merriam-Webster.com). In addition, according to Merriam-Webster, ‘comprehend’ is used to express a complete understanding of the nature or meaning of something, indicating a deeper, or stronger, sense of understanding. The above explanation, nevertheless, does not, in my opinion, provide a clear enough distinction between the verbs’ meanings other than that ‘comprehend’ more often refers to the process and ‘understand’ the outcome of understanding. The picture remains murky when I looked up the senses and synonyms of the two verbs in WordReference2, an easily accessible tool for a translator, presented in Table 1 below:

ACCORDING TO FROMKIN, RODMAN, AND HYAMS, «SYNONYMS ARE WORDS OR EXPRESSIONS THAT HAVE THE SAME MEANING IN SOME OR ALL CONTEXTS» (196).

Term Understand Comprehend sense to comprehend to understand synonyms comprehend grasp take in discern grasp perceive figure out know seize understand

Table 1. WordReference meanings.

As becomes evident in Table 1 shows, which lists the first five synonyms for each verb, neither the sense explanation nor the listed synonyms are helpful when trying to distinguish between the two verbs since ‘understand’ means and is synonymous with ‘comprehend’ and ‘comprehend’ means and is synonymous with ‘understand’.

Finally, I looked up the definitions of both verbs in the online Collins dictionary 3, another reputable source a translator may resort to.

Table 2 below lists the definitions, in British English, provided by the dictionary. The bolded definitions are the ones that result in synonymous senses. When compared, it becomes clear, again, that the two meanings cannot be easily distinguished from one another by consulting a dictionary only.

When focusing only on the first definition for each verb provided by the Collins dictionary, it is evident that the two verbs are synonymous. In addition, when I looked up the translation of

both verbs into German, the target language I am translating into, Collins suggested that both verbs be translated as ‘verstehen’, which, yet again, is not helpful since the proposed translation for both ‘comprehend’ and ‘understand’ is exactly the same.

To further investigate the possible translations of both verbs, I then consulted Google Translate, an online statistical machine translation system often used by translators. Table 3 below lists the top five German translations of ‘understand’ and ‘comprehend’

As Table 3 above shows, among the top five translation solutions for both verbs are ‘begreifen’ and ‘erfassen’. A translator translating one of these German verbs into English would be left wondering whether to translate these as ‘understand’ or ‘comprehend’.

What becomes clear from the discussion above is that neither the dictionary definitions nor the translations of ‘comprehend’ and ‘understand’ provide a clear picture as to the dif-

Term Definition understand

to know and comprehend the nature or meaning of to realize or grasp (something) to assume, infer, or believe to know how to translate or read to accept as a condition or proviso to be sympathetic to or compatible with comprehend to perceive or understand to comprise or embrace; include Table 2. Collins dictionary definitions.

Sample vs monitor Synchronic vs diachronic

Term Understand Comprehend

translations (listed from most to least frequent)

Table 3. Google Translate translations.

verstehen begreifen begreifen erfassen

erkennen nachvollziehen

erfassen umfassen fassen enthalten

ferent meanings of the synonyms. A translator who stops here will be left wondering which verb to use in which context. Consequently, I believe that only a corpus-based analysis of the verbs can shed light on their usage patterns. As Hunston argues, it is the semantic prosody of the words’ collocations, further explained in sections 3 and 4 below, that distinguishes between synonyms because they convey certain attitudinal meanings (Lin and Cung) that are complimentary rather than «collocationally interchangeable» (Hu 118). Yet, before discussing the meanings and usage patterns of both verbs as evidenced in the British National Corpus, the corpus I used for my subsequent analysis, I provide a brief overview of corpus linguistics in the next section.

Corpus linguistics

A corpus-based analysis, the method I propose the translator faced with synonyms employ in addition to consulting a dictionary, represents the following characteristics:

• it is empirical, analyzing the actual patterns of use in natural texts;

• it utilizes a large and principled collection of natural texts, known as a ‘corpus’, as the basis for analysis;

• it makes extensive use of computers for analysis […];

• it depends on both quantitative and qualitative analytical techniques.

(Biber et al., 4)

Types of corpora Characteristics

sample vs. monitor

synchronic vs. diachronic

finite vs. open size (new texts are being added on a continuous basis)

limited to one point in time vs. over a period of time

general (or reference) vs. specialized general language use vs. language for specific purposes

monolingual vs. bilingual or multilingual single language vs. two or more languages

written vs. spoken

mixed or multi-modal

limited to written texts vs. recordings of spoken texts both spoken and written language or including different modes such as images and sound

annotated vs. non-annotated includes linguistic labels and contextual information vs. plain texts without a previous linguistic analysis and labels

Table 4. Types of corpora.

Hunston defines a corpus as a «collection of naturally occurring examples of language, consisting of anything from a few sentences to a set of written texts or tape recordings, which have been collected for linguistic study» (2). Following a predetermined set of criteria, texts are selected and «stored in such a way that [the corpus] can be studied […] both quantitatively and qualitatively» (Hunston 2). According to Laviosa (80-81), the following types of corpora exist, summarized in Table 4.

A particular software 4 is then used to access the corpus. Apart from counting the total number of words present in the corpus, the researcher can investigate how many times a particular search term, called ‘node’, appears. Another option is to display the node together with a specific number of words to its left and right sides generally referred to as kwic (key word in context) concordances (Laviosa). These lines are subsequently studied to investigate «how the linguistic feature is systematically associated» (Biber et al., 6) either with particular words, its lexical associations, or the grammatical features of the surrounding text, its grammatical associations. In addition, statistical measures can be used to find the node’s collocations, the words that tend to co-occur with the search term (Hunston). It is these association patterns that are of particular interest for translators because, as Hunston points out, «if a word has several senses, each sense will tend to be associated most frequently with a different set of patterns» (139). In section 4 below, I will demonstrate that it is exactly these association patterns that ultimately distinguish successfully between the two synonyms my study is concerned with.

The corpus study

The corpus I used for my analysis is the British National Corpus (bnc), published in 1995. Considering the categories listed in Table 4, the corpus is considered a general, monolingual, diachronic, annotated monitor corpus consisting

of approximately 100 million words with 90% representing written and 10% spoken British English. Since the corpus includes texts from a wide variety of domains and speakers, it is considered representative of modern British English.

To find out whether the synonyms ‘understand’ and ‘comprehend’ do indeed prefer different contexts, I first investigated the distribution of both verbs in the entire bnc by searching for the lemma of each verb, defined by Stubbs as «the dictionary head-word» (2). For the lemma understand 22,946 hits were returned with a frequency of 233.4 instances per one million words with a relatively equal distribution between spoken and written texts. The lemma comprehend, however, occurred only 464 times in total resulting in a frequency of 4.72 instances per one million words when combining both spoken and written texts. We can see from these numbers that understand is almost 50 times more common than comprehend. In addition, comprehend is almost never used in spoken discourse with only 0.77 instances per one million words whereas it occurs 5.19 times per one million words in written texts.

In total, there are 20,510 occurrences of understand and 456 occurrences of comprehend in written texts. And since comprehend is rarely used in spoken discourse, I limited the next step of my investigation to written discourse. In addition, as evident from the distributions above, comprehend is most frequently used in academic prose (158 occurrences). I therefore decided to focus on the usage patterns of both lemmas within written academic discourse. Consequently, one important limitation of this decision is that my findings are valid for this particular text type only. Section 4.1 below discusses the results of the qualitative analysis.

Qualitative analysis

Within written academic discourse, 3,953 hits

4 In the case of a commercially available corpus, the software is provided together with the corpus. If the corpus was designed by the researcher himself, he can use free online software such as AntConc (available at http:// www.laurenceanthony.net/software/antconc/) to study the corpus.

WHEN FOCUSING ONLY ON THE FIRST DEFINITION FOR EACH VERB PROVIDED BY THE COLLINS DICTIONARY, IT IS EVIDENT THAT THE TWO VERBS ARE SYNONYMOUS.

understand understand+SP comprehend

comprehend+CP

WRITTEN ACADEMIC DISCOURSE, 3,953 HITS WERE

WITH A FREQUENCY OF 272.77 INSTANCES PER ONE MILLION WORDS FOR THE LEMMA UNDERSTAND.

were returned with a frequency of 272.77 instances per one million words for the lemma understand. More specifically, the following word forms were found, presented in Table 5.

The most common word form of understand is "understand" accounting for almost half of the occurrences. For the lemma comprehend , 141 hits were returned with a frequency of 9.73 instances per one million words. Table 6 summarizes the frequency of the verb’s word forms.

For comprehend, the most frequent word form was "comprehend" accounting for more than half of the occurrences. Next, both lemmas were studied in context in order to investigate the collocational patterns of both verbs.

‘Collocations’ are the different usages of a node (Kennedy) that are extracted from the search term’s concordance lines. They are of interest to translators because words tend to occur in «semi-preconstructed phrases that constitute single choices» (Sinclair 110) for the language user rather than appearing at random. This phenomenon has been referred to by Sinclair as the ‘idiom principle’. The different collocational patterns of the node can be fur-

ther summarized into its lexico-grammatical associations, called ‘colligations’ (Hunston). According to Biber et al., colligations allow us to distinguish between the different senses and usage patterns of synonyms. Using a corpus, it can be established that these patterns follow certain sequence preferences in which a particular lemma tends to co-occur with words from a semantically related set (Begagić), referred to as ‘phraseology’ (Hunston) or ‘semantic preference’ (Begagić). These preference patterns, on the other hand, represent a «unit of meaning» (Hunston 257) which carries a particular discourse function referred to as ‘semantic prosody’ (Hunston). Hunston claims that «if the phraseology changes, the semantic prosody is also different» (258) indicating that «semantic prosody is a discourse function of a sequence rather than a property of a word» (Hunston 258).

Semantic prosody is therefore not just a simple collocation but rather «arises from the pragmatic level» (Fuqua 77). And even though native speakers may be able to list some examples of a word’s co-occurrence patterns, «they certainly cannot document collocations with

any thoroughness, and they cannot give accurate estimates of the frequency and distribution of different collocations» (Stubbs 1). Corpus-based studies are thus important for the field of translation because they shed light on the semantic prosody of different words (Fuqua) compared to «the traditional dictio -

nary [which] cheerfully represents words as often having several discrete meanings, but gives no help whatsoever as to how in practice the language user distinguishes among them» (Sinclair 7).

In what follows, the collocations of both verbs are discussed in greater detail. After gen-

that this is

comprehend and interpret COMPREHEND + CONJ + VP

comprehend how this has come about COMPREHEND + CP 5%

comprehend in terms of COMPREHEND + SP 4% comprehend the ways COMPREHEND + NP

Table 9. Right-hand collocations and colligations of COMPREHEND.

can comprehend easier

erating the kwic concordances for each lemma, I sorted the lines one word to the right of each node in alphabetic order and used the thinning command to obtain 100 random but reproducible samples. This was done to obtain a sample of all possible words occurring immediately to the right of each node assuming that 100 instances are sufficient to provide me with enough data to find any patterns. This decision implies, however, that my results are based on a small sample only and are to be considered hypothetical until tested with a bigger sample. Next, I analyzed the right-hand collocations for their lexico-grammatical patterns and summarized them as the node’s colligations. These steps where repeated to analyze the left-hand context of each node. Table 7 lists the most frequent right-hand collocations and colligations of understand

As Table 7 indicates, the most common pattern is understand followed by a noun phrase

(understand the … mechanism) which accounts for more than one third of the data. Since it appears that this is an important pattern, it will be compared to comprehend below to examine whether any differences in semantic prosody can be detected. When ordered one word to the left, the following colligations emerge, summarized in Table 8.

Of the pattern to + understand, 6% were preceded by a noun (for example, ability to understand), and 3% by an adjective (easy to understand). Since this pattern accounted for more than one third of the data, it will also be compared to comprehend below.

Next, I investigated the collocations of the lemma comprehend7. Table 9 is a summary of the most common colligations found one word to the right of the node.

Similar to understand, the pattern comprehend + noun phrase was the most frequent. In fact, for comprehend, that pattern represent-

10. Presents the most common colligations one word to the left of the node.

able key(s) reverence easy to comprehend

the + N mechanism actions distinction behavior foundations cause issue complexities story (differentiated) nature meaning direction (and scope) language (diverse) system part (elusive) data pathophysiology emergence

UNDERSTAND + the + N COMPREHEND + the + N

suggestion (indian’s) reverence motives nature attitudes message resistance ways

attempt ability desire faculties effort failure key(s) inability

ed half of the occurrences. On the other hand, the pattern comprehend + adverb was not found in the data. For comprehend + complement phrase, no instances of ‘that’ as the complement were encountered while this pattern occurred four times with understand (for example, understand that this is the way the counsellee’s life is predicted). Contrary to understand, the pattern noun + comprehend did not occur and personal pronoun + comprehend occurred only once. Similar to understand, the most common pattern found was to + comprehend preceded by either an adjective (for example, difficult to comprehend ) or a noun ( failure to comprehend ). And while nouns preceded both lemmas equally often, adjectives preceded comprehend more

than twice as often (8%) than understand (3%). What is also interesting is that the nouns preceding comprehend are difficulty, problems, and trouble, indicating a semantic set we might label evaluative and negative whereas the two nouns preceding understand are bureaucracy and importance which I consider ‘neutral’ since they do not carry an inherent positive or negative meaning. This shows a tendency of comprehend to co-occur with evaluative-negative collocations, to be investigated further next. In order to gain a better understanding of the kinds of words that occur with each node, I had a closer look at the collocates, specifically nouns and adjectives occurring within a span of six words, three to the left and three to the right, of the nodes since «collocations can be state import

THE WORDS THAT OCCUR WITH UNDERSTAND APPEAR TO CARRY A NEUTRAL OR EVEN POSITIVE MEANING WHEREAS THE WORDS COLLOCATING WITH COMPREHEND SEEM TO FOCUS ON EVALUATING SOMETHING.

UNDERSTAND WELL DIFFICULT

COMPREHEND

DIFFERENT STATISTICAL MEASURES CAN BE USED TO CALCULATE THE DEGREE TO WHICH A NODE TENDS TO CO-OCCUR WITH OTHER WORDS WITHIN A PREDETERMINED SPAN OF CONTEXT.

used to obtain a profile of the semantic field of a word» (Hunston 78). Table 11 provides a list of the nouns occurring up to three words to the right of the node.

When comparing both lists, it appears that understand collocates with words that I would consider neutral whereas comprehend tends to prefer words that carry a complex or evaluative meaning. A similar picture emerges when studying the collocates to the left of the nodes, presented in Table 12.

Again, the words that occur with understand appear to carry a neutral or even positive meaning whereas the words collocating with comprehend seem to focus on evaluating something. And while some of these collocates carry a positive meaning, there are also a number of negative meanings not present with understand . In sum, the qualitative analysis revealed that there seems to be a pattern of complimentary semantic preferences with understand collocating with neutral-positive words and comprehend co-occurring with evaluative-negative words. Whether these patterns actually imply an opposite semantic prosody which would clearly distinguish between the two verbs is investigated in the quantitative step of the analysis below.

Quantitative analysis

Different statistical measures can be used to calculate the degree to which a node tends to co-occur with other words within a predetermined span of context. If the result turns out to be statistically significant, «the probability of its chance occurrence is sufficiently low» (Barnbrook 94). Hunston also argues that using

a statistical measure is a more reliable method than informal observation to assess the «significance of each co-occurrence» (70). Yet, since the question of which statistical measure to use is an ongoing debate, Lijffijt, Nevalainen, and Säily et al. suggest that various methods should be used.

The default option in BNCweb is the log-likelihood (LL) test. This statistic is based on the assumption that all the words present in the corpus occur so independently, i.e., there is no tendency for any one of them to co-occur with another word (Lijffijt et al.). If the LL-value surpasses 10.83, the critical value to indicate that the result did not occur by chance, the association between the node and its collocate is statistically significant and we can conclude that they actually do form a pattern (McEnery, Xiao, and Tono). However, this measure has been criticized because, as Sinclair points out, words do not have an equal chance of occurring in a text, resulting in potentially erroneous conclusions (Lijffijt).

The mutual information (MI) score, on the other hand, «indicates the strength of a collocation» (Hunston 71) by measuring to what degree the co-occurrence pattern is not accidental or random. According to Hunston, an «MI-score of 3 or higher can be taken to be significant» (71) indicating that the node and its collocate are indeed associated with one another.

The MI-score, however, is not an indication of whether the relationship between the two words is meaningful (Hunston). As Barnbrook suggests, considering the frequency of any word co-occurring with the node is misleading because function words such as articles and

prepositions will always collocate with the node since they are so frequent in language in general. In order to be certain that the encountered association is «the result of more than the vagaries of a particular corpus» (Hunston 72), we can use the t-score since it takes the size of the corpus into account. If the t-score is two or above, the discovered co-occurrence pattern is considered statistically significant (Hunston).

In sum, the MI-score is a measure of strength of the collocation based on the frequencies of both words in the corpus while the t-score is a measure of certainty that the two words actually do tend to co-occur (Hunston). Both indicators are useful because the MI-score provides information about lexical behavior whereas the t-score gives an indication as to the grammatical behavior of a word (Hunston).

As a final step in my investigation, I searched for the collocates of understand and comprehend within a span of three words to the left and three words to the right of each node. I expanded the span to three words in each direction since the qualitative analysis above revealed that some of the adjectives and nouns co-oc-

curring with the node were up to three words removed from it. Once the collocations were obtained, the LL, MI, and t-score statistics were recorded. Table 13 lists the collocates that were statistically significant.

Table 13 shows that all three statistical measures indicate a significant association between well + understand and difficult + comprehend once the function words and the modal verb can, which collocates with both verbs indicating a partial overlap in collocational patterns, were eliminated. On the other hand, there were zero occurrences of difficult + understand and well + comprehend. Suggesting that well carries a neutral-positive meaning and difficult an evaluative-negative one, I can confirm that in agreement with the qualitative analysis discussed above, the semantic prosody of understand is ‘neutral-positive’ whereas it is ‘evaluative-negative’ for comprehend. This clearly shows that not only do both verbs prefer separate semantic fields but that these are also complimentary which, in turn, corroborates Hu’s claim above that synonyms are indeed not «collocationally interchangeable» (118).

Lexical categories Examples

noun (N) Harry, boy, wheat, policy, moisture pronoun (PN) he, herself, their verb (V) arrive, discuss, melt, hear adjective (Adj) good, tall, old, intelligent, beautiful preposition (Prep) to, in, near, at adverb (Adv) slowly, quietly, now, always

Nonlexical categories

Examples determiner (Det) the, a, this, these, no (as in no books)

auxiliary verb (Aux) modal auxiliary (MAux) nonmodal auxiliary (NMAux) conjunction (Conj) and, or, but

will, can, should be, have

degree word (Deg) too, so, very, more, quite

Syntactic complements

Examples complement (Comp) that, why, whether, how, if phrase (P) The team will win. complement phrase (CP) (He knew) that the team would win. set phrases (SP) in terms of, as + NP, by + NP (passive voice)

Phrases

Example

noun phrase (NP) the dog verb phrase (VP) (Paul) is reading slowly.

CONCLUSIONS

As the detailed discussion above illustrated, a translator may have to resort to a corpus analysis in order to accurately distinguish between the different usage patterns of synonyms. Since neither the dictionaries consulted in the present study nor Google Translate were clearly discriminating between the meanings of the verbs ‘comprehend’ and ‘understand’, the British National Corpus was used to study the collocational contexts of both. A first analysis revealed that the verb understand occurs almost equally frequently in written and spoken texts and is significantly more common overall than comprehend. comprehend, on the other hand, is very rare in spoken texts and occurs more frequently in written academic discourse compared to any other written texts. The subsequent steps of my investigation were therefore limited to written academic discourse only.

The analysis of both verbs’ colligations revealed that for both the most prominent lexico-grammatical patterns were ‘VERB + the + noun’ and ‘adjective/noun + to + VERB’. A closer look at the nouns and adjectives collocating with each verb showed that understand tends to co-occur with words that carry a neutral-positive meaning whereas comprehend occurs together with more evaluative-negative words. When subjecting the collocations of both verbs to a statistical analysis, all three measures used (LL, MI, and t-score), indicated that the verb understand collocated with well whereas the verb comprehend preferred difficult . Since I suggest that well is neutral-positive and difficult is evaluative-negative, I argue that the statistical analysis confirmed that, indeed, understand occurs within neutral-positive contexts whereas comprehend collocates with evaluative-negative contexts. My analysis therefore affirms that

the semantic prosodies of both verbs are complimentary and that the verbs are not used at random or interchangeably. However, one important limitation of the present study is that my results are only valid for written academic discourse since other types of texts were not included in the analysis. Another important point is that my findings are only true for the corpus I used and even though the BNC is considered representative of modern British English, the study would have to be expanded to include other corpo -

ra in order to obtain a more complete picture of the usage patterns of these verbs. The takeaway of this study is, however, that translators should go beyond dictionaries when deciding how to translate synonyms since only a corpus-based analysis is capable of revealing the complimentary semantic prosodies involved. And while this additional, and admittedly time consuming, step may not always be feasible, it does provide translators with yet another resource to improve the accuracy of their translations.

Brita Banitz

Dr. Brita Banitz is a Senior Associate Professor of Applied Linguistics and Chair of the Language Department at the Universidad de las Américas Puebla, Mexico. Her research interests include Humor Studies, Translation, Pragmatics, Language Testing, and Technology in Language Teaching. brita.banitz@udlap.mx

REFERENCIAS

• Barnbrook, Geoff. Language and Computers: A Practical Introduction to the Computer Analysis of Language. Edinburgh University Press, 1996.

• Begagić, Mirna. «Semantic Preference and Semantic Prosody of the Collocation ‘make sense’». Jezikoslovlje vol. 14, no. 2-3, 2013, pp. 403-416.

• Biber, Douglas, Susan Conrad, and Randi Reppen. Corpus Linguistics: Investigating Language Structure and Use Cambridge University Press, 1998.

• Fromkin, Victoria, Rodman, Robert, and Hyams, Nina. An introduction to language 9th ed., Thomson/Wadsworth, 2011.

• Fuqua, Jason. «Semantic Prosody: The Phenomenon of Prosody’ in Lexical Patterning.» The Journal of Language Teaching and Learning, vol. 2, 2014, pp. 76-83.

• Hu, H. C. Marcella. «A Semantic Prosody Analysis of Three Adjective Synonymous Pairs in COCA». Journal of Language and Linguistic Studies, vol. 11, no. 2, 2015, pp. 117-131.

• Hunston, Susan. Corpora in Applied Linguistics. Cambridge University Press, 2002.

• Hunston, Susan. «Semantic Prosody Revisited.» International Journal of Corpus Linguistics, vol. 12, no. 2, 2007, pp. 249-268.

• Kennedy, Graeme D. An Introduction to Corpus Linguistics Longman, 1998.

WELL CAN

• Laviosa, Sara. «Corpora». The handbook of translation studies, edited by Yves Gambier and Luc van Doorslaer, John Benjamins, 2011, pp. 80-86.

• Lijffijt, Jefrey, Terttu Nevalainen, Tanja Säily, Panagiotis Papapetrou, Kai Puolamäki, and Heikki Mannila. «Significance Testing of Word Frequencies in Corpora». 2015, https://users.ics.aalto.fi/lijffijt/articles/lijffijt2015a.pdf. Accessed 3 September 2019.

• Lin, Yen-Yu, and Siaw-Fong Chung. «A Corpus-based Study on the Semantic Prosody of Challenge». Taiwan Journal of TESOL, vol. 13, no. 2, 2016, pp. 99-146.

• McEnery, Tony, and Hardie, Andrew. Corpus Linguistics: Method, Theory and Practice. Cambridge University Press, 2012.

• McEnery, Tony, Richard Xiao, and Yukio Tono. Corpus-based Language Studies: An Advanced Resource Book Routledge, 2006.

• O’Grady, William Delaney, John, Archibald, Mark Aronoff, and Janie Rees-Miller. Contemporary Linguistics: An Introduction. Bedford/ St. Martin’s, 2005.

• Sinclair, John. Corpus, Concordance, Collocation. Oxford University Press, 1991.

• Stubbs, Michael. «Collocations and Semantic Profiles: On the Cause of the Trouble with Quantitative Studies.» Functions of Language vol. 2, no. 1, 1995, pp. 1-33. https://www.uni-trier.de/fileadmin/fb2/ANG/Linguistik/ Stubbs/stubbs-1995-cause-trouble.pdf. Accessed 3 September 2019.