A Corpus-based Study of Semantic Prosody across a Native Corpus

Semantic prosody is the most abstract and elusive concept among Sinclair’s classification of the extended lexical units. As a novel linguistic topic in corpus linguistic, semantic prosody is defined here as the collocational meaning arising from the interaction between a given node word and its collocates. For instance, the verb ‘main’ is associated with unpleasant things—problems, issues, and the like. In this paper, we tried to compare two groups of near synonyms (CAUSE vs. BRING ABOUT & RESULT IN vs LEAD TO) in a native corpus (BNC) in terms of their semantic prosody. Anthony (2017) software was applied to extract the prosodic and collocational patterns of near synonyms in the corpus. The results confirmed previous research, indicating that native speakers treat near synonyms differently in terms of their corresponding semantic prosodies. Hopefully, the findings will create a cross-linguistic awareness of semantic prosody, contributing mostly to vocabulary teaching and learning.


Introduction
Focusing on semantic relations between words (vocabulary depth) researchers have recognized that full power for meaning is mostly displayed in discourse, that is, in the company of other words (Ahmadian & Darabi, 2011).For instance, we cannot predict whether the sindividual word strong describes a physical or a psychological quality (compare strong coffee with strong personality) (for more information see Almela, 2007, p. 23).One instance in which meaning is represented in the company of other words is collocation.Widdowson (2007, p.79) asserts that "there is a mutual attraction that draws the collocates as well as the nodes and the collocates together".He provides promising examples as the word unforeseen will attract the word circumstances, foregone will attract conclusion, crying will attract shame and so on (p.79).One example of collocational relationship that has been recently the interested subject of inquiry among researchers is semantic prosody.This concept was later being associated with different appellations such as semantic harmony (Lewandowska-Tomaszczyk, 1996), discourse or pragmatic Available online at http://jgu.garmian.edu.krdprosody (Stubbs, 2001), semantic associations (Hoey, 2003;Nelson, 2006), and evaluative meaning (Morley & Partington, 2009).This term has been widely used by post Firthian corpuslinguists such as Sinclair (1991), Louw (1993), Stubbs (1995), Partington (1998), and Hunston (2002).Stubbs (1995Stubbs ( , 2001)), as an example, provided and tested out some instances (cause vs. bring about, completely vs. perfectly, among many others) to show that each single word has a different connotation compared with its near synonym.Based on this study, cause and completely have been shown to reveal negative semantic prosodies, while bring about and perfectly, as their near synonyms, have shown positive semantic prosodies, (cause death vs. bring about happiness, completely failed vs. perfectly amiable).A working definition of semantic prosody is presented by Louw (1993) as following: semantic prosody refers to a form of meaning which is established through the proximity of consistent series of collocates often charactrizable as positive or negative and whose primary function is the expression of the attitude of its speaker or writer toward some pragmatic situation (p.8).
Although literature is replete with research on different corpus-based studies on semantic prosody, monolingual research on Native corpora is scant.In their study, Wang and Wang (2005) examined the semantic prosody of cause.The study showed that great differences exist in the semantic prosody of CAUSE between Chinese learners of English and English native speakers.Chinese learners of English underused the typical negative semantic prosody and at the same time overused the atypical positive semantic prosody.However, the study is confined to the semantic prosody of CAUSE without adequate attention to its collocation patterns.

Literature review Studies on Semantic Prosody
Further to corpus-based studies on semantic prosody, experimental studies on this issue have also captured the attention of researchers.For example, Ahmadian, Yazdani , and Darabi (2011) constructed, validated and used a test of semantic prosody to experimentally measure EFL learners' knowledge of semantic prosody.The results of their study showed that EFL learners have little or no knowledge of prosodic features.In the same line, the results of Ahmadian and Darabi's (2011) study on the relationship between receptive and productive semantic prosody showed that EFL learners have problems both in terms of their receptive and productive prosodic knowledge.The implication of these studies was that information on semantic prosody should be included in the curriculum so that EFL learners be aware of semantic prosody and take benefit of this concept.
Fang (2010) and Zhang (2010) compared the data from a learner and native corpora to investigate semantic prosody.The findings of their study revealed some similarities between the two corpora in terms of semantic prosody.Zhang (2010), in her study, used the main corpus CLEC and the reference corpus BROWN to explore the semantic prosody of COMMIT in Chinese EFL.The result indicated that Chinese EFL learners exhibit similar semantic prosody as compared with those of native speakers.Fang's similar study made a contrastive analysis of the collocational features of cause and lead to in SWECCL (Spoken and Written English Corpus of Chinese Learners) and BNC by using the collocational study methods of corpus linguistics.The data showed that English-major learners demonstrated similar semantic preferences with the native speakers, but that there were still great differences in their underlying collocational patterns Therefore, within the realm of semantic prosody of lexical pairs monolingual studies carried out to investigate semantic prosody across a native corpus are few and far between.The present study tried to fill this void by carrying out a study which investigates the extent to which native speakers treat lexical pairs differently in terms of semantic prosody.

Hence, the following research question was raised:
To what extent do native speakers treat near synonyms differently in terms of their semantic prosody?

METHOD:
The data for the present study were extracted from a native corpus known as British National Corpus (BNC).The British National Corpus (BNC), built between 1991 and 1994, is a 100-million-word collection of samples of written and spoken language from a wide range of sources, designed to represent a wide crosssection of British English from the later part of the 20th century.The spoken component of the BNC constitutes approximately 10 percent (10 million words) of the total and the written component 90 percent (90 million words).There are nine written domains in the corpus: applied science, arts, belief and thought, commerce and finance, imaginative, leisure, natural and pure science, social science, and world affairs.For the present study, social science sub-corpus was used.

Procedure
The processes of search from BNC occurred through two stages.In the first stage, the prosodic features of the first group of near synonyms (CAUSE & BRING ABOUT) were analyzed.The result of a search in the corpus was displayed as a list of up to 50 randomly selected instances headed by a note of the total frequency of the search string.Then to fulfill the requirement of data triangulation, we analyzed the second group of near synonyms (RESULT IN & LEAD TO) in BNC.The second investigation went through the same process as the first one.In searching process, Anthony (2017) (version 5.3.0) was applied for the analyses.Capital letters are deliberately used here to refer to a lemma, which stands for all the word-forms of the verb.That means, CAUSE stands for cause, caused, causes, causing.

Results and discussion
The analysis of semantic prosody on the near synonyms of CAUSE vs. BRING ABOUT and RESULT IN vs. LEAD TO was carried out.First, a comparison of the distribution of semantic prosody of CAUSE vs. BRING ABOUT was made in the native (BNC) corpus.Second, the second group of near synonyms, namely RESULT IN vs. LEAD TO was compared in BNC corpus in terms of their semantic prosody.Each of these processes is elaborated upon in the following section.

Semantic Prosody of CAUSE and BRING ABOUT
In BNC corpus, among the random selection of 50 instances of the word cause, 36 uses were shown to have negative semantic prosody, 9 were neutral, and 5 were positive.The negative collocates associated with cause, randomly selected from BNC corpus, are revealing in Table 1.
The unfavorable collocates which accompany cause typically contribute to its negative affective meaning, hence confirming what Stubbs (1995) in his study found that cause has a strong negative semantic prosody.In the same line, Xiao and McEnery (2006) also stated: "the negative semantic prosody of cause has been widely observed" (p.114) Table 1

. Semantic Prosody of CAUSE in BNC
Looking back at the data drawn from BNC, neutral collocates of this item are also observed in academic or technical texts (see Table 2).

Table 2. Neutral Semantic Prosody of CAUSE in BNC
Therefore, based on the above typical examples we can argue that 'text type' has a dramatic effect on the choice and condition of semantic prosody.It means that semantic prosody is context-specific.This finding supports what Hunston (2007) found in his study in which he concluded that the attitudinal meaning ascribed to some words is not mostly fixed across different contexts.Rather, they vary from context to context.
For Hunston (2007, p. 266) the 'transfer' of attitudinal meaning from one context to another is a kind of "genuine dilemma".One line of argument which is mainly strong is that meaning does not exist except in context (Teubert 2003).Thus, it seems to Hunston that it is not logical to say that a word or phrase can carry its meaning across from one context to another (see Whitsitt, 2005 for further arguments in support of this view).Therefore, all we need to say is that CAUSE often occurs in the context of undesirable situations, but not always (Hunston, 2007, p. 266).
Contrary to well-established negative prosody of CAUSE in this study, there are some instances of positive semantic prosody which are represented in the following examples, randomly taken from BNC corpus: (a) the cause of this extraordinary behaviour.

(b) the cause of competent management, (c) a cause that would be electorally popular found it in anti-communism.
In (a) the expression this extraordinary behavior is apparently associated with pleasant or positive semantic prosody of CAUSE, and this nominal collocate is the object of the sentence in which 'what' is supposed to be the doer of the action.However, a closer look at the context in which this expression is represented shows that this nominal collocate may be intentionally used to express an ironical effect.In other words, we can say that by using 'irony' the addressor violates the condition of semantic prosody to process the discourse and it is the responsibility of addressee to process the meaning.In this regard we are safe to say that semantic prosody is closely related to discourse.
In (c), the positive semantic prosody of 'cause' is represented through the adjectival phrase electorally popular.In the above example 'anti-communism is associated with the possible CAUSE of the popular election.This typical occurrence of 'cause' shows that sometimes, though in rare cases, 'cause' has a positive semantic prosody.
This finding again corroborates the claim made by Stubbs (1995).He argues that "although such negative prosodies are probably more common, positive prosodies also exist (1995, p. 25).He provides the example causing work which usually means bad news, where as providing work is usually a good thing: (1) when you over draft your account, you cause extra work for the bank staff.
(2) this will provide work; it will raise the standard of living.
By and large, we can say that based on random selection of 50 instances of CAUSE this word has a negative semantic prosody, and if it associates with positive affective meaning, it is mostly ironical.
Although bring about and cause are near synonyms, they don't have similar semantic prosodies.This is what we can observe in the present study.Looking at the 50, randomly selected examples of bring about in BNC, it can be seen that about 25 instances were associated with positive SP.While most of the uses for this item were favorable, only 5 instances were reported to have negative SP: his destruction, quick destruction of England, a loss of associability, the extinction of endemic species, a reduction.This finding considered, literature also documents that bring about has been recurrently assigned with a positive feature (e.g, Stubbs, 1995;Xiao and McEnery, 2006).To our surprise, about 20 uses of bring about were neutral, indicating that this phrasal verb is used more neutrally than cause, while it was reported that cause is more associate with negative prosodies than bring about.Table 3

. Semantic Prosody of BRING ABOUT in BNC
This may largely be attributed to the variety of BNC data sources and the limit of writing topics of the learner corpus, showing that learners rely heavily on a limited choices of words with which they are familiar (Fan, 2010).

Semantic prosody of LEAD TO and RESULT IN
Among the random selection of 50 instances of the phrase lead to, about 34 uses were shown to have negative semantic prosody, 10 instances were positive, and 6 were neutral.This finding supports Xiao and McEnery's (2006) study on the near synonyms in which they found about 49% of the distribution of this phrase is associated with negative semantic prosody.Some negative collocates associated with lead to are revealing in Table 4. Table 4. Semantic Prosody of LEAD TO in BNC About 16 instances of use associated with lead to in BNC were shown to have positive or neutral semantic prosody.It is also observed that these typical examples are mainly represented in technical or specialized language.This finding supports the arguments made before that SP is sometimes contextspecific and the prosodic condition attributed to some words or phrases is not fixed (Zethsen, 2007).Rather, it changes from context to context.Put it simply, we can say that SP for a word or phrase is, in some cases or contexts, associated with negative prosody, while in some others it may be rendered as positive or neutral.To help clarify the point, we can refer to some uses of lead to in BNC in which case the conditions of SP is fluctuating.For example, in: -How might this lead to the differential treatment of whites and blacks?
We see that lead to here has a negative semantic prosody, while in -About wet conditions that may be dangerous but give rise to regeneration, and scorched ones that appear daunting but may lead to growth.We observe that lead to is represented as having positive SP.That said, it can be argued that lead to has a strong negative SP when it is used in general contexts.However, when it is used in scientific or specialized contexts, it may be characterized as having positive or neutral prosodies.
A close look at the randomly selected examples of result in extracted from BNC shows that about 27(%54) instances of use of this phrase are displayed as negative which is somehow a big number.Again like lead to the negative semantic prosody of this phrase does not run counter to Xiao and McHonery's (2006) finding in which case about 47% of the instances of SP were negative.
It was also observed that 14 instances of result in are shown to have positive SP, while only 9 uses are neutral.Some of the negative instances of result in extracted from BNC are reported in Table…… Table 5

. Semantic Prosody of RESULT IN in BNC
The most significant collocates associated with result in are: loss, death, problems, disease and rubbery for negative prosody and growth for positive prosody.The neutral collocates are used in a more scientific meaning: regulation, shifting, test, non-volatile product and presence.Something worthy of note here is that despite the near synonyms cause and bring about which were associated differently in terms of their semantic prosody, here lead to & result in are more similar with regard to their prosodic features.Therefore, it would be fair to say that both are almost associated with negative semantic prosody.

Conclusion and implications
The results showed that native speakers exhibit knowledge of semantic prosody in their writing samples.A great many of their use of semantic prosody was appropriate.However, a few of their prosodic manifestations were inappropriate and were different from those proven by research.This indicates that appropriate knowledge of semantic prosody is sometimes counter-intuitive.Based on these findings, it can be concluded that although native speakers may show the prosodic knowledge of near synonyms in their writing samples, they may sometimes have only superficial knowledge of semantic prosodies and do not realize the underlying collocational patterns.Therefore, awareness of semantic prosody may be even necessary for native speakers.The findings of this study can have some implications too.Based on this study, it is implied that awareness of this concept is crucial for language teachers in that they can benefit by using concordance lines taken from the analysis of computer corpora, or by instructing words with close meanings but with different prosodies.Thus, semantic prosody should be integrated into ESL/EFL vocabulary teaching to help develop language learners' communicative competence (Zhang, 2009).In this study, it was shown that native speakers have insufficient knowledge of semantic prosody, let alone EFL learners with no satisfactory knowledge of collocational patterns.Therefore, more emphasis should be put on the teaching of collocational behavior instead of teaching separate words without context.Information on semantic prosody is also useful for translators.This information will help them know that equivalent words or near synonyms may not have the same semantic prosodies in two languages, thus affecting their text evaluation or interpretation.Discourse analysts can also benefit a lot from information on semantic prosody not only in using irony but in avoiding using mixed messages in a flow of discourse.Finally, being aware of different conditions of semantic prosody may be crucial for lexicographers in compiling their bilingual or monolingual dictionaries, a condition taken into account by Co-Build dictionary.However, with all above-mentioned understanding, awareness, recognition, and information related to semantic prosody, little work has been done to deeply explore this concept or to comprehensively apply it in ESL/EFL pedagogy (Zhang, 2009).Thus, further research is urgent to practically and empirically apply semantic prosody in language pedagogy.
Corpus-based studies carried out in the literature to investigate semantic prosody are well articulated (e.g., Dodd, 2000; McEnery & Xiao, 2006; Sardinha, 2000; Sinclair, 1991; Stubbs,1995; Tognini-Bonelli, 2001; Wang & Wang, 2005; Zhang, 2009; Zhang & Ooi 2008, to name but a few).For example, in a cross-linguistic, semantic study, Zhang and Ooi (2008) compared the concept emotion/feeling with its Chinese equivalent quing.Applying Sinclair's lexical model, they used two monolingual corpora (Chinese Internet Corpus of 280 million words and the Bank of English comprising 450 million words) for the analysis of instances of use.The authors, then, concluded that the Chinese quing terms ganqing/gingan differ from their English near-equivalents feeling/emotion in terms of colligation, collocation, semantic preference and semantic prosody.The study also shows that the nuances of meaning are influenced by specific cultural difference, which ultimately affects semantic prosody.