CC BY-ND
51
https://doi.org/10.17323/jle.2022.12985
Semantic Prosody of Research Verbs: A Corpus-Informed Study
Ghuzayyil Mohammed Al-Otaibi ®
King Saud University, Riyadh, Saudi Arabia
ABSTRACT
Background. Synonymous words behave differently, and language users should be aware of the fact that though near-synonyms share similar denotational meanings, they require different collocates. Further, with specific collocates, they provoke a special affective meaning called semantic prosody. To give an example of this problematic area, researchers use a lot of reporting verbs that merely describe an opinion such as argue, claim, believe, etc. or state facts such as find, confirm, cite, etc. Such verbs cannot be used interchangeably as some novice researchers usually do when they discuss their findings or compare their results with others'. Purpose. This study aimed at examining the semantic prosody of 24 research verbs commonly used by researchers. For this purpose, collocational behavior of nearly synonymous verbs was examined. Compared to previous studies, this study considered only adverbs co-occurring with such research verbs.
Methods. The researcher used the Directory Open Access Journals (DOAJ), which is of 2.6 billion words and 659,132 texts, and focused on predicational adverbs that end in -ly. For the purpose of the study, adverbs with positive semantic prosody are those proving a stronger attitude towards the proposition, improving the quality, quantity, manner of a piece of information or its the relation to the topic or those suggesting a higher level of certainty.
Results. Investigating 24 research verbs related to hypothesizing, reporting, and summarizing, the researcher found that such verbs have different sets of collocates and thus distinct semantic prosodies. Results showed that 12 of the research verbs were positive (i.e., quantify, argue, claim, suggest, state, mention, indicate, outline, summarize, encapsulate, recapitulate, and reveal), whereas 12 verbs (i.e., hypothesize, review, conclude, presume, posit, assume, theorize, speculate, note, report, find, and postulate) were neutral.
Implications. The study has its own implications for writing instructors and researchers. Novice researchers should not use some research verbs interchangeably as they require different collocates of adverbs. Further, future research should address the relationship between word's etymology and semantic prosody as the present study showed that verbs derived from Latin (e.g., conclude, hypothesize, postulate, etc.) are neutral compared to those that are originally French.
Citation: Al-Otaibi, G. M. (2022). Semantic prosody of research verbs: A corpus-informed study. Journal of Language and Education, 8(2), 48-65. https://doi. org/10.17323/jle.2022.12985
Correspondence:
Ghuzayyil Mohammed Al-Otaibi, ghyzayel@gmail.com
Received: September 10, 2021 Accepted: June 21, 2022 Published: June 30, 2022
KEYWORDS:
collocation, corpus linguistics, research verbs, semantic prosody, synonymy
@ ®
INTRODUCTION
Synonymous words behave differently, and language users should be aware of the fact that though near-synonyms share similar denotational meaning (Xiao & McEnery, 2006), they require different collocates. Further, with specific collocates, they provoke a special affective meaning called semantic prosody (Stubbs, 2002). Research in this area has been popularized by corpus linguists who emphasize the precedence of use and frequency over meaning of individu-
al words. With frequent use of collocates with node words, the latter will eventually adopt such collocates as their closest friends. Thus, only corpora, not pedago-gists relying on intuition, can inform us about what is acceptable in terms of use (Xiao & McEnery, 2006); that is, which words should be used with which.
Corpus linguists found that some words are used with specific collocates by native speakers to give negative or positive meanings. Hence, language learners should be aware of such differences be-
sides dictionary definitions. As noted by Scarcella (1984), objectivity and balanced argumentation mark native speakers' writing as opposed to nonnative learners of the language who tend to generalize and show subjectivity in their writing. Therefore, enough attention should be drawn to the use of words (mainly verbs) especially in scientific writing or academic prose as they tell a lot about writers' attitude. Jaroongkhongdach (2015) noted that reporting verbs (e.g., report, investigate, note, argue) are problematic for novice researchers and graduate students. Such researchers tend to use them in a factive manner to show their supportive attitude of others' conclusions. However, Jaroongkhongdach (2015) added that expert researchers viewed such verbs as factive or non-factive where in the latter no signal of approval or disapproval is given. In general, such verbs cannot be used interchangeably as each may require a different set of collocates and convey negative, positive, or neutral prosody (Bloch, 2010).
Semantic prosody has been investigated by Sinclair (1991), Louw (1993, 2000), Stubbs (1995, 1996, 2001a, 2001b), Bub-litz (1996), Partington (1998), Hunston (2002), Schmitt and Carter (2004), Unaldi (2013), Begagic (2013), and Xiao and McEnery (2006). Nevertheless, a few research papers (e.g., Hunston, 1995; Bloch, 2010; Ilchenko & Kramar, 2022) have examined semantic prosody of research verbs. However, such studies have focused on exploring co-texts or concordances of research verbs (e.g., Hunston, 1995; Bloch, 2010; Ilchenko & Kramar, 2022), co-occurring subjects and objects (Hunston, 1995), lexico-grammatical structures, or grammatical context (Hunston, 1995; Ilchenko & Kramar, 2022). None has investigated semantic prosody of research verbs in relation to co-occurring predicational adverbs though adverbs modify verbs and may denote writers' attitude towards a certain claim. Examining which adverbs collocate with which research verbs might help novice researchers in using the accurate adverb with the right research verb to communicate a specific attitude.
Research Objective
This study aimed at examining the semantic prosody of 24 research verbs commonly used by researchers. For this purpose, collocational behavior of nearly synonymous verbs was examined.In addition, focus was on conducting a corpus analysis of research verbs and their co-occurring adverbs. More specifically, lemmas (basic forms) of research verbs and their variants (the past form) in academic prose were investigated using three different corpora.
LITERATURE REVIEW Collocation
Firth (1957) proposed the term collocation to refer to habitual co-occurrence of words. Thus, food collocates with fast but not quick though fast and quick express the same mean-
ing. The same goes for coffee that collocates with black but not brown. A collocation is of two parts: the node word and the collocate. The node word is the one whose collocability (tendency to come with certain words) is being examined (Stubbs, 1996).In the previous examples, food and coffee are node words, whereas the rest are collocates.
Semantic Prosody and Near-Synonymy
Stubbs (2002) observes that "there are always semantic relations between node and collocates, and among the collocates themselves" (p. 225). The type of collocational meaning that results from the interaction between a node and one of its typical collocates is called semantic prosody (Xiao & McEnery, 2006). Xiao and McEnery (2006) emphasized that the semantic prosody of a certain word is acquired through its typical collocates (i.e., nouns including subjects and objects) which can be one or more words and not only one adjacent word. The typical use of words determined by their collocates establishes to a greater extent their meaning. Researchers exploit such typicality to implicate some meaning. This exploitation is only possible if the associated semantic prosody is strong (Hunston, 1995).
Xiao and McEnery (2006) noted that a word may acquire a positive or negative meaning because of its typical collocates. Semantic prosody serves to express speakers' or writers' attitude (Louw, 2000). Louw (1993) argued that semantic prosodies associated with the majority of node words are negative. Additionally, Xiao and McEnery (2006) argued that with atypical collocates the word will continue to give its typical prosody. For example, though cause in cause happiness is used with happiness, it is still negative. It gives the impression that such happiness is impossible or undesirable since corpus investigation shows that cause is frequently used with death, problem, damage, pain, etc. (Xiao & McEnery, 2006).
Semantic prosody and collocational behavior are used to distinguish between words that are traditionally considered to be near synonyms (Xiao & McEnery, 2006). Near synonyms are words with similar denotational meaning (Partington, 1998). However, as reported by Tognini-Bonelli (2001), synonyms are not "collocationally interchangeable" (Xiao & McEnery, 2006, p. 108). Therefore, powerful cannot be used in place of strong to describe tea (Tognini-Bonelli, 2001). According to Stubbs (1996), semantic prosodies can be positive (pleasant meaning), neutral, or negative (unpleasant meaning) corresponding to Partington's (2004) favourable, neutral, and unfavorable prosodies.
Research Verbs
Citing others' work is one way that helps authors promote their work (Hewings et al., 2010). Reporting verbs are important elements in any research (Jaroongkhongdach, 2015). They function to "to give credit to other researchers and to use their work in the cumulative construction of knowledge"
(Charles, 2006, p. 320). In addition, they are utilized to report authors' claims or as indicators of their attitudes towards others' words (Hyland, 1999; Thompson & Ye, 1991). However, using reporting verbs effectively is problematic for novice researchers especially non-native English students such as Iranians (Yeganeh & Boghayeri, 2015), Malaysians (Manan & Noor, 2014), and Thais (Jogthong, 2001; Jirapanakorn, 2012). Further, Bloch (2010) and Pecorari (2008) reported that not using reporting verbs properly may result in readers' misinterpretation of writers' arguments. Hyland (2008) explained that such writers feel compelled to take "definite and self-assured positions" (p. 70). Additionally, Thompson and Ye (1991) and Bloch (2010) stated that non-native English students use fewer reporting verbs and of less variety compared to expert researchers.As noted by Pickard (1993), they use mainly say to introduce quotations. This denotes, as argued by Hyland (2002), students' inability to present sound discussions and convincing arguments. Other novice writers tend to cite others' work for the sake of citing them. On the other hand, expert researchers cite to synthesize relevant work, justify their claims, and give support to their argument (Mansourizadeh & Ahmad, 2011). To cite effectively, authors should use reporting verbs properly (Yeganeh & Boghayeri, 2015).
Types of Research Verbs
Previous research called verbs used in research as reporting verbs (Jaroongkhongdach, 2012; Hyland, 1999; Hyland, 2000a). However, arguments by (Thompson & Ye, 1991; Ziman, 1968) suggested that they should not be treated in a similar fashion because sometimes they refer to facts or simply opinions. Thompson and Ye (1991) differentiated between verbs that express the position of the writer (i.e., introducing 'averrals' or propositions given by the writer) or those that denote that of the author whose claims are being reported (i.e., attributions) or those accompanying the writer's interpretation of the issue. Hunston (2000) reported that young researchers should understand the difference between such verbs as this aids in establishing the importance of any claim to the research goal. Some citations or claims are consistent with the author's and some are inconsistent, flawed, or irrelevant (Ziman, 1968). Thus, authors should decide on the credibility of every claim and their attitude towards each (Hunston, 2000). Attitudes are either favorable or unfavorable (Bloch, 2010). Further, Hunston (1995) reported that attributions imply some evaluation by the attributor (i.e., the text's writer) because of the different interpretations an attribution can give. As for reporting verbs, they can be also factive or non-factive. Factive reporting verbs represent others' claims as facts, and non-factive ones give no clue of any attitude towards others' argument (Jaroongkhongdach, 2015).
Hyland (2000a) proposed a categorization of reporting verbs according to the type of activity they indicate. Such activities can be embodied by research acts (i.e., actions carried out in
the research, e.g., discover), discourse acts (i.e., verbs verbally expressing cognitive or research activities, e.g., discuss), or cognition acts (i.e., verbs denoting mental processes, e.g., assume). More importantly, authentic texts of available corpora can inform researchers of which verb is used for which claim and for what stance (Bloch, 2010). As noted by Hyland (2000a), verbs referring to discourse acts are more common than others and especially in soft fields of knowledge that require speculation and interpretation and personal input. Hence, verbs of discourse acts such as suggest and argue followed by research verbs (e.g., find) are more common than others.
Thompson and Ye (1991), on the other hand, divided verbs according to the process they perform. The first group of verbs are known as textual verbs and they are used for verbal expressions (e.g., state, write). The second category includes mental verbs, and they are of mental processes (e.g., think, believe). The third group, however, are known as research verbs, and they refer to research activities (e.g., find, demonstrate). Hyland's (2002) discourse and cognition verbs correspond to Thompson and Ye's textual and mental verb categories. Thompson and Ye (1991) also stated that there is a relationship between reporting verbs and evaluation or the position of the reporting writer towards the claims of a specific author.Thus, verbs can be factive (e.g., establish), counter-factive (e.g., overlook), or non-factive (e.g., find).Similarly, Thompson (1994) differentiated between verbs where the writer believes strongly in what the author states (e.g., acknowledge, admit, point out, etc.) and those that show some disagreement with the author (e.g., claim, purport, misinform).
By the same token, Francis et al. (1996) distinguished between argue verbs (i.e., concerned with writing and different forms of communication, e.g., argue, suggest, assert, point out), think verbs (i.e., describing the process of thinking, believing; knowing, understanding, hoping, fearing, e.g., think, assume, feel), show verbs (i.e., used to indicate a fact or a situation, e.g., show, demonstrate, reveal), and find verbs (i.e., concerned with coming to know, e.g., find, observe, discover, establish). According to Hyland (2002), ARGUE verbs are used more commonly in social science by native speakers, whereas FIND/SHOW verbs prevail in natural science. Compared to previous research, this paper sheds light on three groups of frequent research verbs related to hypothesizing, reporting, and summarizing results. This classification roughly corresponds to some of the categorizing schemes mentioned above.
Previous Studies on Research Verbs and Semantic Prosody
Some studies (e.g., Sinclair, 1991; Louw, 1993, 2000; Stubbs, 1995, 1996, 2001a, 2001 b;Bublitz, 1996; Partington, 1998; Hunston, 2002, Schmitt and Carter, 2004; Unaldi, 2013; Be-gagic, 2013;and Xiao & McEnery, 2006) have examined semantic prosody and considered mainly verbs such as make
sense, provide, cause, happen, set in, occur, come about, take place, etc. Previous research indicated that the majority of verbs explored in the literature were negative (e.g., Unaldi, 2013; Partington, 2004; Sinclair; 1991; Stubbs, 2001) and that semantic prosody of words is closely linked to genre (Begagic, 2013). As for research verbs, a few research papers (e.g., Hunston, 1995; Bloch, 2010; Ilchenko & Kramar, 2022) have explored semantic prosody of such verbs.
Hunston (1995), for example, used the Bank of English to explore four verbs of attribution (i.e., verbs attributing statements to the writer of the text or to another author, e.g., acknowledge, insist, claim, and argue). As noted by Hunston (1995), attribution is used for hedging in research, introducing information that reflects that of the author's, highlighting a gap in research, transferring the responsibility of a claim, etc. The researcher considered word's co-text (i.e., the words surrounding a particular word) and found that acknowledge is used mainly in reluctant acceptance of an opponent's point of view. On the other hand, insist is typically used to contradict a previous statement. However, claim indicated neither acceptance nor rejection of a statement. As for argue, it is mainly associated with conflicts between the author and the writer. More importantly, with the first mention, argue was positively evaluated, but upon the second mention, the verb was negatively evaluated. Even if it is sometimes associated with positivity, the verb may carry some negativity. Hunston (1995) emphasized that her observation was not collocational and that she did not employ any statistical measure.
In another research, Bloch (2010) identified 27 research verbs (i.e., argue, assume, believe, claim, conclude, consider, demonstrate, describe, discuss, examine, explain, find, indicate, imply, mention, note, point out, predict, propose, prove, report, reveal, show, state, suggest, think) used commonly in research. The researcher used a corpus of research articles derived from Science, a group of journals that publish articles in different types of sciences. Search was not limited to the lemma (e.g., THINK) but covers wordforms such as thinks and thought.
Bloch (2010) has identified six categories of reporting verbs based on Swales' (1990) distinction. Two categories were relevant to the present study. For example, the fourth category examined in Bloch's (2010) was concerned with how writers use special verbs to denote their rhetorical attitude towards a claim. According to Hyland (2000b), an expression of attitude may include hedging, bolstering, or toning up or toning down a claim. The assessment of the writer's attitude towards the truth of a claim is what Radden and Dirven (2007) called epistemic modality. The verbs conclude, describe, examine, note, point out, report, and reveal are always positive. Others are typically positive such as consider, demonstrate, discuss, find, indicate, imply, predict, prove, show, and suggest. As for mention, it can be equally positive or negative. It is used more commonly in learner corpora than in Science (Bloch, 2010). Re-
garding explain, state, and claim, 25-30% of their meaning is negative. State in specific is used to present a claim that will be criticized in subsequent sentences. This finding was in line with Latour's (1987) argument who claimed that the purpose of referring to previous research is to enhance the validity of a claim through citing literature that is consistent with the writer's claim. If there is some negativity, negativity stems out from negating the verb itself not from the verb (Bloch, 2010) itself. As for argue, 40% of its instances were positive.
Another similar category is about the strength of writer's attitude towards a claim or what Hyland (1998) termed hedging. Non-native speakers tend to state stronger claims compared to native ones (Hyland & Milton, 1997). Maintaining a moderate position seems to be very difficult for such writers (Bloch, 2010).Bloch (2010) found that 61% of the verbs were strong, whereas only 2% were weak, and 37% were considered moderate. The verbs demonstrate, discuss, examine, report, reveal, show were always strong. As for conclude, consider, describe, explain, find, indicate, note, point out, predict, and state, they were often strong. However, argue, claim, imply, propose, suggest were mainly weak. According to Hyland (2002a), suggest was used to reduce the writer's responsibility towards the certainty of a claim. However, this lowered level of responsibility can be still boosted by using a number of rhetorical devices. Since suggest and indicate are considered to be moderate, a positive expression of the claim can be made stronger through the use of adverbs (e.g., strongly for suggest).
In a recent study, Ilchenko and Kramar (2022) examined three reporting verbs (i.e., argue, claim, believe) in 40 journals of linguistics. They focused on their rhetorical and discursive functions. They found that argue and claim were more associated with Others, whereas believe was used more frequently with Self and was very common in concluding sections, research limitations, and suggestions for future research. Compared to argue, claim was followed by refutations or simply reservations. Thus, argue and claim cannot be used interchangeably in contexts.
As shown above, previous research on semantic prosody of research verbs focused on a small number of research verbs (e.g., Hunston, 1995; Ilchenko & Kramar, 2022), used a smaller corpus (Ilchenko & Kramar, 2022), explored co-texts or concordances of research verbs (e.g., Hunston, 1995; Bloch, 2010; Ilchenko & Kramar, 2022), considered co-occurring subjects and objects (Hunston, 1995), or examined grammatical context (Hunston, 1995; Ilchenko & Kramar, 2022). Hence, what distinguishes the present study from previous ones is that the researcher made use of a bigger corpus, investigated 24 research verbs, and utilized the T-score and the MI to reach a conclusion about research verbs' collocational behavior. More importantly, the emphasis of the current study was on collocating predicational adverbs that could be used to express one's attitude towards a claim.
METHODOS
Background
This paper explored the semantic prosodies of nearly synonymous verbs that are used typically to hypothesize, report, and summarize results. Such verbs are employed frequently in scientific writing. To examine verbs' collocates, researchers such as Greenbaum (1974) and Hoey (1991) argued that the term collocation should strictly describe statistically significant co-occurrence of words. Hence, Firth (1957) believed that our investigation of collocation should be quantitative. Therefore, many linguists have adopted the statistical approach to the study of collocation using different corpora (Halliday, 1966; Greenbaum, 1974; Sinclair, 1991; Hoey, 1991; Stubbs, 1995; Partington, 1998; McEnery & Wilson, 2001; Hunston, 2002). Thus, statistically significant co-occurrence of the collocational components (the node and the collocate) justifies collocates' inclusion as typical ones.
Data Collection Tools
In this study, the researcher used two different English corpora to decide on frequent research verbs: Sketch Engine for Language Learning (SkELL) and Corpus of Contemporary American English (COCA). SkELL consisted of one million words, whereas COCA consists of 250 million words collected between 1990 and 2015. SkELL is a recent project derived from Sketch Engine (an online corpus tool) and designed for language learners. Where COCA describes the American Variety, SkELL is mainly of British English. Data cannot be driven from one variety as the researcher is trying to extract generalizable data that can describe English in general. Further, the use of more than one corpus was necessary to verify the results and report only statistically significant findings that each corpus depicts. SkELL can provide concordances (i.e., alphabetical lists of key words used in a specific text) and information on typical collocates and synonyms (Thesaurus1; Baisa & Suchomel, 20142). COCA can yield frequency information and help with identifying typical collocates, contexts of words, and concordances.
Since semantic prosody can affect texts and not merely phrases, investigation of relevant texts or genres is important. The researcher focused on verbs used frequently in scientific writing. Research verbs are typically used in academic journal articles. Choosing academic corpora such as the British Academic Written English Corpus (BAWE) did not yield significant results in terms of typical collocates. Hence the third corpus utilized was the Directory Open Access Journals (DOAJ) which is available in Sketch Engine, a corpus tool that includes about 500 different corpora. DOAJ consists of
journal articles in various areas of knowledge such as medicine, science, technology, humanities, and social science. At the time of data collection, it includes 2.6 billion words and 659.132 documents.
To decide on the collocability of associated words, the researcher considered besides frequency, mutual information (MI), and t-score. Following Xiao and McEnery's (2006) suggestion, the minimum co-occurrence frequency of a collocate to a specific node word was set at three (for comparable corpora of one million words) and 20 for COCA. Mollin (2014), however, noted that for bigger corpora a word is considered a typical collocate if it occurs 50 times or more. Further, COCA and Sketch Engine support MI, a test that measures collocational strength (between a node word and a collocate) depending upon the occurrence of a collocate with a given word in various contexts (Xiao & McEnery, 2006).The researcher accepted a minimum MI score of 3 for a collocate to be considered so for a given node word. Besides MI, t-score, a measure of certainty or confidence, was useful to emphasize that there is an association. The t-score (significance threshold=2.57), as opposed to MI, is a more reliable measure since it accounts for frequencies of collocates. Thus, frequent collocates score very high on t-score. It is important to note that some combinations (e.g., proper names, technical terms, etc.) score high on MI but not on t-score. Thus, both measures were essential for the purpose of the study (Bartsch, 2004).In COCA and Sketch Engine, the span of a co-occurrence pattern was set four to the left and four to the right of the node word, a common setting to look for collocates. As for SkELL, the word-sketch feature (see figure 1) is very helpful in identifying typical collocates for a word based on their frequency and MI score (Baisa & Suchomel, 2014).
The Sampling Procedure of Research Verbs
Frequency and text category are important criteria for choosing potential research verbs. Hypothesizing, reporting, and summarizing verbs that are used very frequently in scientific writing were the focus of the present study. The researcher examined a number of research articles to collect a sufficient number of verbs. An online thesaurus (Thesaurus. com, 2017) and Oxford Online Dictionary3 were used to identify the synonyms of each group of research verbs. Results were verified using SkELL.
It is important to note that exploring semantic prosody was not limited to lemmas of research verbs (the unmarked form, HYPOTHESIZE), but it included one more form which is the past (hypothesized). The basic form of research verbs and the past (denoting the present perfect tense and the
1 Thesaurus. Com. (2017). Retrieved from http://www.thesaurus.com/
2 Baisa, V., & Suchomel, V. (2014). SkELL: Web interface for English language learning. Retrieved from https://www.sketchengine.co.uk/ wp-content/uploads/2015/05/SkELL-Web-Interface-for-English-Language-Learning.pdf
3 Oxford Online Dictinary. (2017). Retrieved from https://en.oxforddictionaries.com/
Figure 1
The word sketch of the verb assume as suggested by SkELL
5kE|L
Examples Word sketch Similar words More features
assume (S3 Effffi^»«»!
subject of assume
model theory player scenario calculation scholar analysis estimate everyone argument method approach author hypothesis people object of assume
responsibility command role control office leadership duty position presidency identity liability form power throne title adjectives with assume constant most many more modifiers of assume
implicitly mistakenly safely incorrectly wrongly automatically generally erroneously tacitly falsely reasonably naturally naively widely correctly
past) are two frequent forms used in research. For example, the past and the present perfect are used for reviewing relevant literature, describing the selected procedure (involving past events), and reporting results. The present tense, however, is used commonly to present conclusions and discuss implications4. The Word Sketch Feature in Sketch Engine was used to search for research verbs. The basic form of the verb (i.e., the lemma, e.g., REVIEW) yielded results related to the past form (i.e., reviewed) and the present one (i.e., reviews, review, reviewing). Further, variation in spelling was considered. Thus, hypothesize and hypothesise, theorize and theorise, and summarize and summarise were included in research since researchers are allowed to use either form.
Data Collection
Xiao and McEnery (2006) noted that prosodic meaning is closely determined by word forms, contexts, and genres. Thus, this study examined research verbs with reference to scientific writing and two forms of verbs (the basic form and the past). Since the investigated items were verbs, attention was drawn to co-occurring modifiers of verbs (adverbs). Adverbs are used to answer questions starting with How/ When/Where/How much, and To what extent. Hence, objects (i.e., nouns) occurring with such verbs were not considered though each object may contribute a different meaning with each verb. More importantly, the evaluative meaning obtained from the collocating adverb is only partial since mainly adverbs, not subjects or objects, were examined in this study.
As mentioned above, the focus of the study was on pred-icational adverbs that end in -ly. Predicational adverbs (i.e., associated with gradable predicates) are divided into manner or event adverbials (i.e., how the event takes place) including locative or temporal adverbs, subject-oriented adverbs (i.e., identify properties of the noun in the subject
position, e.g., accidentally, deliberately, etc.) which includes mental-attitude (e.g., reluctantly, willingly, etc.) andagent-ori-ented adverbs (e.g., wisely, rudely, etc.), and speaker-oriented adverbials which includes evaluative adverbs (e.g., expressing the attitude of the speaker towards the proposition, e.g., surprisingly, unfortunately, etc.), speech act adverbs (i.e., expressing an involved speech act, e.g., frankly, briefly, etc.), and epistemic adverbs (i.e., involving gradable modal adverbs, e.g., certainly, clearly, etc.; Ernst, 2002). Table 1 below is illustrative.
Such adverbs are used to present writer-oriented interaction with readers or what Hyland termed stance. Stance includes writer's attitudes towards a piece of information or how certain they are about its accuracy. As noted by Hyland (1999), stance can be represented by hedges (e.g., probably, generally), boosters (e.g., definitely), attitude markers (e.g., certainly), and self-mentions. Hedges are softening words that are used to illustrate writer's awareness of the cooperative principle maxims. Boosters, on the other hand, are used to express certainty in what writers say. However, attitude markers illustrate one's affective attitude to the stated proposition. Affect includes surprise, agreement, importance, frustration, etc. Hyland (1999) reported that in research papers hedges are more common than other categories and such interactional markers or reporting structures are more frequent in 'soft' fields of knowledge such as philology, applied linguistics, sociology, etc. than in science papers because such fields are considered to be interpretative.
Hyland (2002) stated that reporting verbs should be used with care as they imply one's evaluation of the source author's claim. They can either approve reported authors' conclusions through factive verbs such as show, solve, etc. or use counter-factive verbs (e.g., fail, ignore) to show disapproval. Experienced researchers, however, prefer to use verbs with
4 APA manual (publication manual of the American Psychological Association). (2017). American Psychological Association.
Table 1
Types of Predicational Adverbs with Examples
Types
Sub-Types
Definitions with Examples
denoting how the action is performed, e.g., loudly
event locative denoting where the event takes place, e.g., externally
temporal suggesting when an event takes place, e.g., recently, etc.
Subject-oriented mental-attitude denoting «a state of mind experienced by the referent of the subject of the verb, e.g.,
reluctantly, calmly, willingly, anxiously, eagerly, frantically, absent-mindedly, gladly, sadly» (Ernst, 2002, p. 63)
agent-oriented showing that «an event is such as to judge its agent as an adjunct [i.e., adverbials, ad-
jectivals, and relative clauses modifying a noun] with respect to the event, e.g., cleverly, stupidly, wisely, etc. (Ernst, 2002, p. 54)
Speaker-oriented evaluative expressing the attitude of the speaker towards the proposition, e.g., surprisingly, un-
fortunately, etc.
speech act expressing an involved speech act, e.g., frankly, briefly, etc.
epistemic involving gradable modal adverbs, e.g., certainly, clearly, etc.
no attitudinal hint (e.g., find, identify). For the purpose of the study, verbs with a positive semantic prosody are those proving a stronger attitude towards the proposition, improving the quality, quantity, manner of a piece of information or its relation to the topic or suggesting a higher level of certainty (Bloch, 2010). This was initially determined by the frequency of verbs' collocates of adverbs and then by checking concordances for accompanying subjects.
RESULTS
In the present study, the researcher examined the semantic prosody of research verbs used frequently in research papers. Though previous research (Jaroongkhongdach, 2012; Bloch, 2010; Thompson & Ye, 1991; Hyland, 2002) treated all research verbs as reporting verbs, for the purpose of the study, research verbs can be defined as those used to hypothesize, report, and summarize results.
Results of the Sampling Procedure of Research Verbs
As mentioned above, the researcher divided research verbs into three categories (see Table 1) corresponding to the three main sections (where they are commonly found) in a typical research paper and as they are used by research writers. This division was meant to avoid any overlap in categories. Hence, hypothesizing verbs are used in the introductory part of research papers, whereas summarizing ones are found mainly in the concluding part. Reporting verbs, on the other hand, occur more frequently in the review of literature section and the discussion where researchers report results of others or compare theirs to others' findings. Thus, such verbs cannot be used interchangeably. They correspond
roughly to Faber and Mairal's (1999) verbs of COGNITION (e.g., hypothesize, speculate, assume), SPEECH (e.g., note, mention, argue), and PERCEPTION (e.g., find), Thompson and Ye's (1991) mental (e.g., assume, speculate), textual (e.g., argue, claim), and research verbs (e.g., find, reveal, indicate), and Francis et al.'s (1996) ARGUE (e.g., argue, suggest), THINK (e.g., assume, posit), SHOW (e.g., reveal, indicate), and FIND verbs (e.g., find).
As stated above, the researcher used an online thesaurus (Thesaurus.com, 2017), Oxford Online Dictionary, and SkELL to collect hypothesizing, reporting, and summarizing verbs after checking a few research articles.For example, for the hypothesizing group, theorize, speculate, presume, presuppose, suppose, posit, postulate, deduce, ascertain, deduce, pinpoint, quantify, ponder, consider, etc. were established as typical synonyms. To minimize the number of verbs investigated, COCA with respect to typical text categories was checked.Table 4, 5, and 6 in the Appendix show the frequency score of each verb in five text categories (academic writing, spoken, fiction, magazine, and newspaper) along with its normalized frequency score (occurrence per million [opm]) because the size of each sub-corpus is not equal (i.e., academic writing [81 million words], spoken [85 million words], fiction [81 million words], magazine [86 million words], and newspaper [81 million words]) at the time of data collection.
Only frequent verbs in academic prose compared to other text categories were chosen. Hence, hypothesizing verbs were minimized to eight verbs: posit, assume, hypothesize, theorize, speculate, postulate, presume and quantify. They all contribute the meaning suggested by the dictionary "to assume by hypothesis" (Oxford Online Dictionary, 2017). As illustrated by Table 4 in the Appendix, some verbs were more
common in the past such as hypothesize, theorize, posit, presume, and postulate.
The same procedure (see Table 5) was used to limit the number of reporting verbs, and hence the explored ones are 10: note, argue, claim, report, suggest, state, indicate, reveal, find, and mention. Some verbs were more common in other text types such as report, state, find, and mention which were more frequent in spoken English, newspaper, magazine and fiction, respectively. However, all potential verbs carry the meaning of "giving an account of something observed" (Oxford Online Dictionary, 2017).
As for summarizing verbs, Dictionary.com (2017)5 stated that a summarizing verb expresses something in a concise manner. Along with corpus investigation, dictionaries and thesauri suggested six verbs: outline, summarize, encapsulate, recapitulate, conclude, and review. Table 6 in the Appendix is illustrative. Thus, the total number of investigated verbs is 24. Table 2 below lists the research verbs that were explored in the study.
Frequency of Research Verbs
Using the DOAJ as a corpus of journal articles, the researcher found that reporting verbs were more frequent than hypothesizing and summarizing ones (see Table 3 below). Moreover, of reporting verbs, report (i.e., of 1.894.821 occurrences), suggest (i.e., occurring 1.887.652 times, indicate (i.e., used 2.114.053 times), and find (i.e., of 2.998.375 hits) were the most frequently used by researchers. As for hypothesizing verbs, assume (i.e., of 654.082 hits), quantify (i.e., occurring 268.657 times), and hypothesize (i.e., of 127.631 occurrences) were more frequent than the rest. Regarding the summarizing group, the verbs summarize (i.e., used 228.208 times [203.439 times as summarize and 24.769 times as summarise]) and outline (i.e., of 83.316 hits) occurred more than any other summarizing verb.
Collocating Adverbs of Research Verbs
Hypothesizing Verbs
As for corpus analysis of hypothesizing verbs in relation to their accompanying adverbs, results showed that they were in general neutral. Hypothesize (freq=116.881) was mainly associated with previously (freq=725, MI=8.57, t-score=26.8),
initially (freq=296, MI=9.72, t-score=17.18), originally (freq=191, MI=10.15, t-score=13.8), recently (freq=153, MI=7.5, t-score=12.3), and specifically (freq=53, MI=6.6, t-score=7.2). The same applied to hypothesise (freq=10.750) that was associated with previously. The verbs posit and theorize occurring 4.680 times (or theorise of 10,032 occurrences) were not associated with typical adverbs.
Regarding speculate, postulate, and presume, they suggest neither positive nor negative prosody. Speculate habitually occurred with reasonably (freq=66, MI=10.20, t-score=8.11) and previously (freq=174, MI=7.23, t-score=13.10), but presume with generally (freq=92, MI=8.60, t-score=9.5).Moreover, postulate collocated frequently with originally (freq=89, MI=10.3, t-score=9.4), initially (freq=79, MI=9.12, t-score=8.8), previously (freq=350, MI=8.8, t-score=18.6), and recently (freq=148, MI=8.8, t-score=12.3).
As for assume, it typically occurred with more than 40 adverbs and mainly with implicitly (freq=2230, MI=12.6, t-score=47.2), generally (freq=2963, MI=8.7, t-score=47.21), safely (freq=482, MI=10.7, t-score=21.9), reasonably (freq=597, MI=9.8, t-score=24.4),simply (freq=877, MI=8.4, t-score=29.5), commonly (freq=1101, MI=8.04, t-score=33.05), initially (freq=639, MI=8.01, t-score=25.18), and typically.Quantify is the only verb in the hypothesizing group that suggested a positive meaning (see Figure 2 below). It was habitually used with accurately (freq=1604, MI=11.08, t-score=40.03), objectively (freq=335, MI=11.8, t-score=18.2), reliably (freq=407, MI=10.8, t-score=20.16), precisely (freq=550, MI=10.16, t-score=23.4), rigorously (freq=89, MI=9.8, t-score=10.5), etc.In addition, quantify was used commonly with domain adverbs such as spectro-photometrically, densitometrically, colorimetrically.
As for accompanying subjects, assume, quantify, presume, and posit occurred more commonly with inanimate subjects (e.g., theory, calculation, approach, method, hypothesis) as opposed to hypothesize, speculate, postulate, and theorize which were used more frequently with investigator, researcher, author, theorist, scholar, colleague, subjects, economist, feminist, etc. This suggests that the latter set of verbs are typical attribution verbs. Examples of concordance lines are the following: (1) Considering that the large-scale PEV data is not presently available, the paper reasonably assumes a residential community with one ten-thousandth of Texas
Table 2
Hypothesizing, Reporting and Summarizing Verbs Used in Research Papers
Typical Sections Category Examples
Introduction Hypothesizing posit, assume, hypothesize, theorize, speculate, postulate, presume, quantify
Review of literature Reporting note, argue, claim, report, suggest, state, indicate, reveal, find, mention Discussion
Conclusion Summarizing outline, summarize, encapsulate, recapitulate, conclude, review
Figure 2
Concordance lines of the verb quantify.
I~1 > PLoS ONE ve abundance of IncPI plasmids can only be precisely quantified
2 Q > PLoS ONE ie concept documented in this study we cannot directly quantify
3 □ The Cryosphere ... d surface temperatures (GSTs) must be systematically quantified
4 Q > Journal of Mari... :hallenge for any reliability assessment is to accurately quantify
5 □ BMC Genomics Although yeast two-hybrid interactions can not be easily quantified
6 Q International J... ts. </s><s> We are currently carrying out further tests to quantify
7 □ Atmospheric Mea... ot be eliminated by our filter handling protocol has to be quantified
8 Q Atmospheric Mea... ing. </s><s> It can therefore be difficult to conclusively quantify
9 □ Indian Journal... expressed that the behavior of irregular solution can be quantified
10 Q] PLoS Computatio... inuously from the finger bases to their leader cell tip, as quantified
O PLoS Computatio... inuously from the finger bases to their leader cell tip, as quantified
by quantitative realtime PCR. </s><s> However, the rec< the individual oxygen contribution of each organism belc as an important background for understanding meso-sci or estimate the underlying failure rates. </s><s> There a , the very high level of p-Galactosidase expressed by th< more precisely the level of improvement that is achieval as accurately as possible and subtracted from the carbc charring by optical methods, especially on highly loaded more precisely using EHSA. </s><s> The procedure ca previously [22] (Figure 3 C). </s><s> A very similar trenc previously [22] (Figure 3 C). </s><s> A very similar tren
population, and all houses in this community are equipped with the smart meters (Mathematical Problems in Engineering); (2) This doctrine simply posits that a contract cannot confer enforceable rights or impose obligations on persons who were not parties to it and have not furnished considerations (Australian Journal of Business and Management Research); (3) Relating biome productivity to the mean an-
Table 3
Frequency of Hypothesizing, Reporting, and Summarizing Verbs Used in Research Papers
Research Verb
Hypothesizing Verbs assume quantify
hypothesize (-ise)
speculate
postulate
presume
theorize (-ise)
posit
Reporting Verbs
find
indicate
report
suggest
reveal
note
mention
state
argue
claim
Summarizing Verbs conclude summarize (-ise) review outline encapsulate recapitulate
Frequency
654.082
268.657
127.631
56.243
39.057
19.841
14.712
10.911
2.998.375 2.114.053 1.894.821 1.887.652 777.019 564.024 337.777 193.562 148.594 67.732
314.762 228.208 209.035 83.316 27.667 12.329
nual temperature, this model implicitly presumes a certain correlation between the climatic conditions of the growing season and those of the whole year (Carbon Balance and Management); (4) There are several theoretical perspectives that explicitly posit that human perceptual processes are supported by active prediction (PLoS ONE).
Reporting Verbs
Regarding reporting verbs, note was associated with more than 30 predicational adverbs including previously (freq=6354, MI=8.08, t-score=79.4), explicitly (freq=303, MI=7.1, t-score=17.2), finally (freq=237, MI=6.12, t-score=15.17), specifically (freq=334, MI=5.6, t-score=17.9), recently (freq=391, MI=5.2, t-score=19.2), etc. On the other hand, argue was mainly associated with positive predicational adverbs (see Figure 3 below) such as convincingly (freq=389, MI=14.2, t-score=19.7), persuasively (freq=144, MI=16.2, t-score=11.9), rightly (freq=84, MI=12.3, t-score=9.1), strongly (freq=1360, MI=9.7, t-score=36.8),-forcefully (freq=64, MI=13.4, t-score=7.9), plausibly (freq=57, MI=11.9, t-score=7.5), etc.
Similarly, claim was positive because it occurred frequently with rightfully (freq=31, MI=14.1, t-score=5.5), confidently (freq=39, MI=11.6, t-score=6.2), legitimately (freq=28, MI=13.1, t-score=5.2), justly (freq=22, MI=13.9, t-score=4.6), rightly (freq=27, MI=11.6, t-score=5.1), etc. Report habitually occurred with more than 40 predicational adverbs such as previously (freq=99546 , MI=10.13, t-score=315.2), recently (freq=21183, MI=9.11, t-score=145.28), commonly (freq=5530, MI=7.6, t-score=73.9), widely (freq=3031, MI=6.60, t-score=54.4), consistently (freq=2135, MI=7.6, t-score=45.9), originally (freq=1092, MI=7.1, t-score=32.8), etc., and hence it was coded neutral.
Suggest, which is also a frequent research verb, was positive occurring more frequently with strongly (freq=24351, MI=10.6, t-score=155.94), previously (freq=6338, MI=7.4, t-score=79.14), clearly (freq=2813, MI=7.5, t-score=52.7), etc. By the same token, state was of favorable pros-
Figure 3
Concordance lines of the verb argue.
□ Details Left context KWIC Right context
□ ©PLoSONE ¡nt modulation of somatic impedance. </s><s> It has been argued previously that active hair bundle movements may under!
2 □ ( Interdiscipllna... n prove immensely valuable. </s><s> Brian Lancaster has argued convincingly that the study of Jewish, Taoist and Euddhis
3 Q 0 PLoS Medicine ous by their absence. </sxs> Orentlicher and Hehir have argued compellingly that If advertisements for luxury goods were i □ O PLoS Medicine evelsofSP resistance, these important uncertainties also argue strongly for the evaluation of alternatives. </s><s> Proguai
6 PLoS Medicine levels of SP resistance, these important uncertainties also argue strongly for the evaluation of alternatives. </sxs> Progui G Potchefstroom E... atlon of the Convention </sxs> Mendellvich33 correctly argues that determined practical measures should be used to sup
7 Q] Potchefstroom E... en years are actually not in school.91 Smolln92 rightfully argues that for the large majority of children not in school It is diffic
0 Potchefstroom E.. by ruling out full-time employment.93 Smolin94 rightfully argues that the "desire of a child labour movement to support com
s □ PLoSONE ;onsumptioh of the respective infected host cells strongly arguing for oxygen-dependent HIF-1 activation mechanisms. </s><
PLoS ONE irmeable cell culture dishes, see Fig. 2, 3, 7] and strongly arguing for a role of PHD-2 and against transcriptional HIF-1 Indue
□ BMC Public Heal... :ing that PLWH may not be as marginalized as previously argued [ 3,14,15] and that the relationship between HIV and un 2 Q] PLoS Genetics I to normal by 10 wk of age. </s><s> These data strongly argue against any structural defect in the components of the sorr
13 □ PLoS Genetics e of conserved motifs outside the homeodomain strongly argues for orthology with ShxC, as does overall protein sequence
14 □ 0 Mediators of In IL-17 producing proinflammatory Tti17 cells, theoretically arguing against a generally Immunosuppressive role of 5-Aza as A
ody (see Figure 5 below) since it occurred habitually with explicitly (freq=2772, MI=11.2, t-score=52.6), clearly (freq=3163, MI=9.2, t-score=56.1), previously (freq=4256, MI=8.4, t-score=65.04), specifically (freq=536, MI=7.2, t-score=22.9), simply (freq=568, MI=7.9, t-score=23.7), precisely (freq=212, MI=7.6, t-score=14.4), correctly (freq=164, MI=6.8, t-score=12.6), etc.
Similarly, indicate was of positive associative meaning since it commonly occurred with clearly (freq=17253, MI=9.8, t-score=131.2), strongly (freq=4047, MI=7.7, t-score=63.3), possibly (freq=1790, MI=7.5, t-score=42.08), previously (freq=1775, MI=5.2, t-score=41.04), etc.Further, reveal was also of positive meaning (see Figure 6 below) used habitually with clearly (freq=2874, MI=8.5, t-score=53.4), consistently (freq=350, MI=7.3, t-score=18.5), recently (freq=646, MI=6.3, t-score=25.10), potentially (freq=273, MI=5.6, t-score=16.20), previously (freq=537, MI=4.8, t-score=22.3), significantly (freq=840, MI=4.6, t-score=27.8), etc.
Nevertheless, find (freq=2.998.375) was used frequently with more than 40 predicational adverbs including commonly (freq=9979, MI=8.5, t-score=99.6), previously (freq=8824, MI=6.6, t-score=92.9), recently (freq=4484, MI=6.8, t-score=66.3), exclusively (freq=2825, MI=8.3, t-score=52.9), typically (freq=3600, MI=7.1, t-score=59.5), mainly (freq=4630, MI=6.4, t-score=67.2), consistently (freq=2517, MI=7.9, t- score=49.9), etc. Also, mention collocated more with previously (freq=22684, MI=9.4, t-score=150.4), briefly (freq=1215, MI=9.2, t-score=34.7), explicitly (freq=1287, MI=8.8, t-score=35.7), specifically (freq=956, MI=6.7, t-score=30.6), commonly (freq=363, MI=5.2, t-score=18.5), etc., and apparently mention was of positive meaning. The above-advanced analysis showed that seven of the ten reporting verbs were positive (argue, clam, suggest, state, in-
dicate, reveal and mention), whereas the rest were neutral (note, report, find).
In general, reporting verbs such as claim, argue, find, mention, report, and note were used more frequently with animate subjects such as author, scholar, researcher, respondent, informant, customer, interviewee, participant, patient, woman, investigator as opposed to indicate, suggest, reveal, and state which were utilized more commonly with result, line, bar, datum, finding, study, analysis, evidence, etc. The following are examples of concordance lines with reporting verbs and their common subjects and adverbs: (1) These results clearly indicate significant variability between regions, with sites in the Middle East and India (triangles and hexagram) having much lower lidar ratios and higher refractive indices than sites in the Sahel (squares), which in turn have lower lidar ratios and higher refractive indices than the sole site in the Sahara desert (Atmospheric Chemistry and Physics Discussions); (2) All these experiments had clearly revealed that the A-M system is uninvolved mechanically in tension generation. (International Journal of Molecular Sciences); (3) The paper briefly reviews main types of Virtual Learning Environments and analyses the use of VLEs in Lithuania (Informatics in Education).
Summarizing Verbs
As for summarizing verbs, starting with conclude (occurring 314.762 times), it collocated with safely (freq=286, MI=11.4, t-score=16.9), finally (freq=435, MI=9.2, t-score=20.8), easily (freq=427, MI=7.7, t-score=17.9), recently (freq=323, MI=7.2, t-score=17.8), generally (freq=231, MI=6.5, t-score=15.03), etc. However, outline was positive collocating with briefly (freq=1156, MI=12.13, t-score=33.9), clearly (freq=514, MI=8.2, t-score=22.5), previously (freq=977, MI=7.9,
Figure 4
Concordance lines of the verb report.
□ Details Left context KWIC Right context
1 □ © PLoS ONE ases of their differentiation into DC, </s><s> We recently reported a detailed analysis of AEC conditioning of DC using an in
2 □ © Evidence Based conferences (B6.8%). </s><s> Librarians also frequently reported flexible schedules, support and funding for professional dt
3 □ © PLoS ONE ■ne of date palms is much less extensive than previously reported </sxs> We are not suggesting that intra-cultivar heterof
4 □ © PLoS ONE artial or complete degradative pathways were previously reported to be localized on plasmids belonging to the lncP-1, IncP-
5 □ © PLoS ONE imidswith high similarity to pWWO. that were previously reported to carry degradative genes (Fig. 4) [21,22]. </s><s>The p
6 □ © PLoS ONE nd biodégradation processes in sediments as previously reported already for mercury-contaminated sediments in KazachsU
7 □ © PLoS ONE 3nd biodégradation processes in sediments as previously reported already for mercury-contaminated sediments in Kazachst,
8 □ © PLoS ONE Reliability and validity of this scale have been previously reported [17], </s><s> Motivation related to the time trial was meas
9 □ ) Journal of Biorn fpe of microorganisms [50]. </s><s> We have previously reported the efficiency of a crude Ulva extract against the hemibiotr
10 □ © Disease Markers diabetic nephropathy, as determined through previously reported mlcroarray analyses [31]. </s><s> However, in this subgrc
11 □ © Disease Markers npared with valsartan/hydrochlorothiazide. as previously reported in the ADDM study [9]. </s><s> However, the alterations ir
12 |~) Atmospheric Che.. of reactive iodine species (IO, OIO, 12 ) has thus far been reported primarily from coastal sites. <lsxs> An overview of the t
13 □ Atmospheric Che.. detection of molecular iodine at Mace Head was recently reported by Saiz-Lopez and Plane (2004), who found highly elevatï
14 □ © Atmospheric Che.. . >per limits for BrO in the MBL of 1.5-2 ppt. as previously reported from several coastal sites (Sander et al.. 2003. and référé
,—, _____ ..... . . . . _ ... ............
Figure 5
Concordance lines of the verb state.
□ Details Left context KWIC Right context
1 □ © Evidence Based ... 3050 years old (although this percentage is not explicitly stated In the paper except in a table). </s><s> Nearly half of the i
2 □ © Theoretical and.. lber one objective is SlretBàràgan Channel - as recently stated the Romanian minister of agriculture, Mr. Valeriu Tab^rS.«
3 n Journal of Cont. Dn the judgement of citizens. </s><s> Tsakatika correctly states that the EU system of governance shows important limitat
4 □ © Atmospheric Che... ! mlneraloglcal composition of dust, one can equivalents state that the regional variability of the lidar ratio is caused by r<
S Q © International J . ow methods are evaluated and so it is reckless to overtly state that which methods indeed have the lowest error rates </:
6 O International J... » AbCDe ABCDe aBCde Diploidy and Dominance clearly state that double information in genotype is reduced by half in it
7 □ Informatics in... aphical illustration to present data. <ls><s> Although they stated explicitly that the suitable abstract data type for their prot
8 □ ©PLoS ONE tied [51]. <1 sxs> In addition, our consent form explicitly stated that all information Is confidential and would not be sharec
9 □ © Linguistik Onli... ■s Aderopo of plotting against him. </s><s> This is clearly stated In his next non-proverbial statement: If you think like a tort
10 □ © Entropy lore and more thermodynamics researchers and teachers stating explicitly that entropy Is not dlsorder[22], </s><s> Howev
11 □ © BMC Neuroscienc. . id variance. </s><s> The reason for this can be precisely stated in elementary mathematical terms. </s><s> Moreover, it is
12 □ ©PLoS ONE lotions and EC numbers. </s><s> It has been previously stated that folds and primary EC numbers are unrelated [64,65]
13 □ © Finance Chall.. ce the bank liquidity, contrary to the hypothesis originally stated . </s><s> If this ratio increase with 1% then the banks' liqL
14 "2 Journal of Elec.. ■—i jbjectlves are broadly defined objectives, which explicitly state the organization's objectives in terms of the results it want
t-score=31.12), etc. Likewise, summarize was used habitually with briefly (freq=2773, MI=12.8, t-score=52.6), succinctly (freq=88, MI=12.8, t-score=9.3), concisely (freq=71, MI=13.19, t-score=8.4), shortly (freq=128, MI=9.6, t-score=11.2), etc., and thus it was positive.
As for review (used 209.035 times), it occurred frequently with more than 30 adverbs including briefly (freq=3759, MI=11.9, t-score=61.2), retrospectively (freq=2160, MI=13.01, t-score=46.4), critically (freq=1953, MI=11.4, t-score=44.1), etc. On the other hand, recapitulate was of positive meaning since it is used with faithfully (freq=282, MI=16.2, t-score=16.7), accurately (freq=175, MI=10.9, t-score=13.2),-
fully (freq=295, MI=10.3, t-score=17.16), briefly (freq=63, MI=10.5, t-score=7.9), etc. Similarly, encapsulate was of pleasant meaning used mostly with efficiently (freq=94, MI=10.3, t-score=9.6), successfully (freq=138, MI=10.1, t-score=11.7), completely (freq=111, MI=9.30, t-score=10.5), fully (freq=91, MI=8.9, t-score=9.5), etc. The above detailed account of summarizing verbs showed that summarizing verbs were essentially positive (outline, summarize, encapsulate, and recapitulate). However, conclude and review were neutral.
Regarding co-occurring subjects, all summarizing verbs occurred more frequently with inanimate subjects (e.g., table, diagram, chart, article, literature, paper, section, chapter,
Figure 6
Concordance lines of the verb reveal.
□ Details Left context KWIC Right context
1 □ © PLoS ONE analyze the oxygen profiles above the irregular interface, revealed strongly reduced net photosynthesis and dark respiratior
2 □ 0 International J... o the final year in the institution </sxs> This will surely reveal more interesting patterns. </sxs> With all these observa
3 □ CD international J... )iverse features on mobility and relay range deliberately reveals that the. Relay Station (RS) can be grouped up into three
r □ Q ISRN Neurology males. </s><s> Planned comparisons between quartiles revealed significantly lower 5-HIAA and HVA levels of patients in
5 □ 0 epj web ofcont ],[35]). </s><s> The magnetic nature of CVs is generally revealed by the detection of coherent pulsations in the optical and
6 □ 0 BioMed Research.. . :planation for this perceived heterogeneity was probably revealed by a recent study in which bone marrow biopsy samples ;
7 □ 0 PLoS ONE The association of PLC-c with the H202 stress has been revealed previously . as there is a direct evidence to support the <
8 □ 0 Genetika 3th coeffi- </s> </p> <p> <s> cient analysis undoubtedly revealed more importance of leaf water content, especially in the p
9 □ 0 Journal of Libr... ? Daily life must be regarded as the social totality vividly revealed from the inside. </$><s> Hence, we can find that the faga
10 □ 0 PLoS ONE jnically-loaded joint tissues [46], the current work clearly reveals that T-VSCC plays a direct role in cellular events associa
11 □ 0 Clinical and De... )llows. </s><s> (1) In the BP group, DIF most frequently reveals linear deposits of IgG and C3 at the demnoepidermal junc
12 □ 0 PLoS ONE correct number of built-in communities is systematically revealed , in all instances, by a sudden drop of some of the persist
13 □ 0 Gastroenterolog... sa. </s><s> These two methods are even able to clearly reveal the localization, appearance, degree of mesenteric infiltrs
14 □ 0 PLoS Genetics ing a silver staining procedure: active caspase 3 staining revealed intensely stained neuronal profiles (Figure 6C. 6D), </s>
enhancer) except for conclude that was employed equally with author, investigator, researcher, colleague, expert, etc. as well as with summary, survey, literature, etc. This finding suggests that summarizing verbs were not commonly used for attribution or evaluation. The following lines are illustrative: (1) This paper briefly reviewed the botanical, traditional, phytochemical, pharmacological and conservation related aspects of this plant (Journal of Costal Life Medicine); (2) As a result, it is difficult for current models to accurately recapitulate all of the pathological features associated with IMR (PLoS ONE)
DISCUSSION
This study surveyed the collocational and prosodic behaviors of 24 research verbs (hypothesizing, reporting, and summarizing verbs) besides their frequency information. Results showed that reporting verbs were more frequent than others and that find, suggest, report, indicate were more commonly used in scientific writing. This finding has been emphasized earlier by Hyland (2000a) who noted that verbs referring to discourse acts (e.g., suggest) and those known as research verbs (e.g., find) are more common than others. More importantly, results revealed that 12 of the research verbs are positive (quantify, argue, claim, suggest, state, indicate, outline, summarize, encapsulate, recapitulate, mention, and reveal), 12 (review, conclude, find, report, note, posit, assume, presume, hypothesize, theorize, speculate and postulate) are neutral. As opposed to Louw's (1993) argument, semantic prosody is not essentially negative since the researcher found that none of the explored node verbs are negative. This conclusion also reflected Hyland's (2002) who noted that when writers adopt a position towards authors' claims, the position is either neutral or positive. Hyland (1998) explained that if a specific writer wishes to express a negative evaluation of author's viewpoint, this is accomplished not through reporting verbs as this is consid-
ered a face-threatening act in academic writing and may result in paper's rejection by publishers.
As emphasized by previous studies, words that are considered near-synonyms do not sometimes convey the same semantic prosody as they frequently occur with different collocates. Such differences in semantic preferences illustrate the fact that such words should not be used substitut-ably by researchers. As reported by Louw (2000), research verbs tell a lot about writers' attitudes. For example, argue was mainly positive and proves a stronger point discussed, whereas note is of neutral meaning suggesting that the writer's argument needs some further investigation. In addition, hypothesize and assume are near synonyms (can be used interchangeably) since they express the same affective meaning, but not summarize and conclude. Thus, the new differentiating criterion for near synonyms is semantic prosody. Another dimension of prosody (besides formality) should be considered to distinguish between words.
Such findings confirmed Bloch's (2010) who noted that the verbs reveal, indicate, suggest, mention were used to show that the writer's attitude is positive towards a claim. Further, as mention was used frequently by learners more than researchers (Bloch, 2010), the present paper found its frequency in fiction. In terms of the strength in attitude expressed by writers, the study reflected results of Bloch (2010). Thus, reveal was strong since it collocates with clearly and significantly. Similarly, indicate was associated with strong claims because it was used commonly with clearly and strongly. Additionally, state was of favorable meaning because it was associated with clearly, specifically, explicitly, etc. As for argue, claim, and suggest, Bloch (2010) explained that such verbs are generally weaker than others. Further, Hunston (1995) reported that argue and claim in specific are mainly associated with conflicts and disagreements. Hence, argue, claim, and suggest need to be associated with hedges of adverbs that suggest strength in attitude. The same finding
has been emphasized by Ilchenko and Kramar (2022) who noted that argue was frequently used with hedges. Thus, results showed that argue was commonly found with convincingly, strongly, persuasively, plausibly, etc. and claim with confidently, justly, rightfully, etc.As for suggest, the researcher found that its typical collocate is strongly (freq=24,351). As reported by Hyland (2002a), suggest was used to reduce the writer's responsibility towards the certainty of a claim. However, this lowered level of responsibility can be still boosted by using a number of rhetorical devices. For Hyland (2002a), suggest and indicate were considered to be moderate. A positive expression of the claim can be made stronger through the use of adverbs (e.g., strongly for suggest).
Compared to show, prove, and demonstrate, which were used to show agreement with a previous statement, indicate and suggest are hedges that allow for some evaluative space (Thompson & Ye, 1991) where the writer becomes obliged to present a contrasting view. The two verbs have almost the same collocates of adverbs (i.e., clearly, strongly, possibly, previously, collectively, consistently, indirectly, actually). As reported by Hyland (2002), suggest is commonly found in philosophy, sociology, applied linguistics, marketing, and similar fields of interpretation and speculation.
Hyland (2002) argued that find is a non-factive verb. Hence, the researcher found no preferred collocates that signaled a specific attitude. Thus, according to Hyland (2002), find is typically used by researchers in physics, applied linguistics, and marketing.Therefore, typical collocates were frequently, commonly, previously, recently, typically, etc. Such non-factive verbs are commonly found in science and engineering as they indicate impartiality that characterizes hard knowledge fields where writers acknowledge previous research without corrupting it with their personal stand (Hyland, 2002).
As a cognitive verb, speculate is a verb that embodies a tentative view towards a reported claim. Thus, findings illustrated that speculate collocates with tentatively, cautiously, and initially. In addition, the researcher found that postulate and hypothesize are discourse verbs that indicate a tentative attitude. Thus, they were of neutral prosody collocating with initially, tentatively, and originally. Though suggest belongs to the same category, it shows more of positive semantic prosody since it collocates with strongly and clearly.
Findings also showed that report is neutral, and this confirmed Hyland's (2002) statement that report is a discourse verb used frequently by researchers to pass information without interpretation besides state and summarize. Hyland (2002) mentioned that report is used more frequently in biology and mechanical engineering. As for note, argue, and claim, they are positive except for note. Hyland (2002) called such verbs assurance discourse verbs used by researchers to bolster their argument. Hence, results indicated that claim collocates with rightfully, rightly, legitimately, falsely, confidently and argue with convincingly, persuasively, rightly, strongly, forcefully, and plausibly. More importantly, as found
by Hyland (2002), argue and claim are more commonly used in philosophy and argue in specific is frequent in sociology, applied linguistics, and marketing. Nevertheless, note is more frequent in sociology. Hyland (2002) argued that writers' evaluative stance is more evident in soft domains of knowledge which is supported by using adverbs (e.g., correctly, rightly) with such reporting verbs. In this way, writers firmly align themselves with a specific position. Writers are ought to present existing literature in a convincing manner for their readers. Thus, research verbs commonly used in soft domains of knowledge are positive (e.g., argue, claim, reveal, indicate, suggest) and occur more frequently with rightly, strongly, etc.
Moreover, this corpus investigation showed that previously is a common collocate for almost every research verb (state, suggest, indicate, reveal, find, mention, conclude, outline, summarize, and review). Previously is a temporal discourse marker that can function globally (i.e., connecting the writer's argument to others') and locally (i.e., linking the writer's argument to what has been mentioned earlier in the same article; Sarda et al., 2014). Because of its dual function, it is the most typical adverb for many research verbs.
Implications
Pedagogically, language teachers including L2 writing instructors should draw students' attention to the fact that verbs may tell a lot about one's attitude, and hence find and report are neutral in general, whereas argue and claim are of favorable prosody. Thus, they cannot be taught as synonyms in class. Moreover, mention is positive in academic writing, and further research should explore its prosody in fiction as it is more frequent in the latter genre. Additionally, researchers will benefit a lot from the present study since it informs them about which verbs are emotionally loaded with positivity (e.g., argue, claim), and hence they should be avoided at all costs in hard science in which one should show an impartial attitude typical of scientific writing as opposed to soft science.
Limitations
The present corpus investigation was limited to the old version of COCA which was of 250 million words collected between 1990 and 2015.Hence, the new genres (webpages, blogs, TV series) added in 2020 were not considered.Additionally, co-texts for each verb were not examined because Sketch Engine allows only 150 to 185 words of context for each verb. Additionally, some sentences were removed from context which makes it difficult for the researcher to decide whether the attributor was refuting or supporting an earlier argument.
More importantly, as noted above, the researcher focused on predicational adverbs that end in -ly and can be used to express one's attitude towards a claim. Selecting predicational adverbs implied that quantificational adverbs such as always and frequently were not examined besides domain adverbs
such as medically, mathematically, physically, etc. Additionally, focus adverbs (e.g., only), adverbials introducing new participants to discourse (e.g., For George, with a hammer, etc.) and almost were not considered in the present study.
CONCLUSION
The present study examined the semantic prosody of 24 synonymous research verbs (related to hypothesizing, reporting, and summarizing results) which are established in the dictionary as near synonyms since they share the same deno-tational meaning. However, the researcher found that such verbs have different sets of collocating adverbs, as suggested by the DOAJ, and thus distinct semantic prosodies. Indeed, research verbs convey either neutral or positive prosodies, but they have never become associated with negative meanings. Hence, results showed that 12 of the research verbs were positive (i.e., quantify, argue, claim, suggest, state, mention, indicate, outline, summarize, encapsulate, recapitulate and reveal), whereas 12 verbs (i.e., hypothesize, review, conclude, presume, posit, assume, theorize, speculate, note, report, find, and postulate) were neutral. This is mainly attributed to the fact that expert writers follow the norm in scientific writing and only boost their argument in fields that require interpretation and justification. However, the obtained results cannot be generalized to all research verbs used in all disciplines since the researcher did not examine verbs in terms of surrounding words and grammatical structure. Nonetheless, this study is of significant implications for researchers, language instructors, and learners.
More specifically, future research should address the semantic prosody of research verbs as they are found in
their extensive co-texts and used in scientific writing. Context may reveal some disagreement or refutations in which one can attribute such negativity to the research verb in question. Thus, more attention should be paid to verbs of attribution (e.g., argue, claim, etc.) since the researcher found that there are verbs that are commonly employed to introduce different types of propositions such as summarize, outline, mention, review, etc., but they are utilized more frequently with inanimate subjects.
Additionally, future research should address the semantic prosody of research verbs using learner corpora. Previous research emphasized that students of non-native speakers should be aware of how verbs are used conventionally by expert researchers to communicate their argument more effectively. Moreover, further research can explore which collocates are used in which disciplines as some studies emphasized that expert researchers in soft fields of knowledge used reporting verbs differently from those of hard knowledge domains.
Furthermore, future research should investigate the relationship between word's history and semantic prosody. The present study showed that verbs derived from Latin (e.g., conclude, posit, note, assume, hypothesize, postulate) are basically neutral compared to those taken from French (e.g., argue, claim, state, mention, reveal). Based on this, any further research should relate such areas to extensions of meaning and diachronic and synchronic studies of meaning.
DECLARATION OF COMPETING INTEREST
None declared. ■
REFERENCES
Bartsch, S. (2004). Structural and functional properties of collocations in English: A corpus study of lexical and pragmatic constraints on lexical co-occurrence. Gunter Narr Verlag.
Begagic, M. (2013). Semantic preference and semantic prosody of the collocation make sense.Jezikoslovlje, 74(2-3), 403-416.
Bloch, J. (2010). A concordance-based study of the use of reporting verbs as rhetorical devices in academic papers. Journal of Writing Research, 2(2), 219-244. https://doi.org/10.17239/jowr-2010.02.027
Bublitz, W. (1996). Semantic prosody and cohesive company. Leuvense Bijdragen, 85(1-2), 1-32.
Charles, M. (2006). Phraseological patterns in reporting clauses used in citation: A corpus-based study of theses in two disciplines. English for Specific Purposes, 25(3), 310-331. https://doi.org/10.1016Zj.esp.2005.05.003
Ernst, T. (2002). The syntax of adjuncts. Cambridge. https://doi.org/10.1017/CB09780511486258
Faber, P. B., & Uson, R. M. (2012). Constructing a lexicon of English verbs (vol. 23). Walter de Gruyter.
Firth, J. (1957). Papers in linguistics. Oxford University Press.
Francis, G., Hunston, S., & Manning, E. (1996). Grammar patterns 1: Verbs. HarperCollins Publication.
Greenbaum, S. (1974). Some verb-intensifier collocations in American and British English. American Speech, 49, 79-89. https:// doi.org/10.2307/3087920
Halliday, M. A. K. (1966). Lexis as a linguistic level. In C. Bazell, J. Catford, M. A. K. Halliday, & R. Robins (Eds), In memory of J.R. Firth (pp. 148-62). Longman.
Hewings, A., Lillis, T., & Vladimirou, T. (2010). Who's citing whose writings? A corpus-based study of citations as interpersonal resource in English medium national and English medium international journals. Journal of English for Academic Purposes, 9, 102-115. https://doi.org/10.1016/jjeap.2010.02.005
Hoey, M. (1991). Pattern of lexis in text. Oxford University Press.
Hunston, S. (1995). A corpus study of some English verbs of attribution. Functions of Language, 2(2), 133-158. https://doi. org/10.1075/fol.2.2.02hun
Hunston, S (2000). Evaluation and the planes of discourse: Status and value in persuasive texts. In S. Hunston & G. Thompson (Eds.), Evaluation in text: Authorial stance and the construction of discourse (pp. 176-207). Oxford University Press.
Hunston, S. (2002). Corpora in applied linguistics. Cambridge University Press. https://doi.org/10.1017/CB09781139524773
Hyland,K (1998). Hedging in scientific research articles. John Benjamins. https://doi.org/10.1075/pbns.54
Hyland, K. (1999a). Academic attribution: Citation and the construction of disciplinary knowledge. Applied Linguistics, 20, 341367. https://doi.org/10.1093/applin/20.3341
Hyland, K. (1999b). Disciplinary discourses: Writer stance in research articles. In C. Candlin & K. Hyland (Eds.), Writing: Texts, processes, and practices (pp. 99-121). Longman. https://doi.org/10.4324/9781315840390-6
Hyland, K. (2002a). Activity and evaluation: Reporting practices in academic writing. In J. Flowerdew (Ed.), Academic discourse (pp. 115-130). Longman.
Hyland, K. (2002b). Specificity revisited: How far should we go now? English for Specific Purposes, 21(4), 385-395. https://doi. org/10.1016/S0889-4906(01)00028-X
Hyland, K. (2008). Make your academic writing assertive and certain. In J. Reid (Ed.), Writing myths: Applying second language research to classroom writing (pp. 70-89). University of Michigan Press.
Hyland, K. & Milton, J. (1997). Qualification and certainty in L1 and L2 students' writing. Journal of Second Language Writing, 6, 183-205. https://doi.org/10.1016/S1060-3743(97)90033-3
Ilchenko, O., & Kramar, N. (2022). Reporting verbs argue, claim, and believe in linguistics research articles: A corpus-based study. Journal of Language and Linguistic Studies, 1S(Special Issue 1), 203-213.
Jaroongkhongdach, W. (2015). Idea sharing: Reporting verbs in research writing: Lessons from experts. PASAA: Journal of Language Teaching and Learning in Thailand, 50, 143-154.
Jirapanakorn, N. (2012). How doctors report: A corpus-based contrastive analysis of reporting verbs in research article introductions published in international and Thai medical journals. The Bangkok Medical Journal, 4, 39-46. https://doi.org/10.31524/bk-kmedj.2012.09.007
Jogthong, C. (2001). Research article introductions in Thai genre analysis of academic writing [Unpublished doctoral dissertation]. West Virginia University.
Johns, T. (1991). It is presented initially: Linear dislocation and interlanguage strategies in Brazilian academic abstracts in English and Portuguese [Unpublished doctoral dissertation]. University of Birmingham.
Latour, B. (1987). Science in action. Harvard University Press.
Louw, B. (1993). Irony in the text or insincerity in the writer? The diagnostic potential of semantic prosodies. In M. Baker, G. Francis, & E. Tognini-Bonelli (Eds), Text and technology (pp. 157-76). John Benjamins. https://doi.org/10.1075/z.64.11lou
Louw, B.(2000). Contextual prosodic theory: Bringing semantic prosodies to life. In C. Heffer, H. Sauntson, & G. Fox (Eds), Words in context: A tribute to John Sinclair on his retirement. University of Birmingham.
Manan, N. A., & Noor, N. M. (2014). Analysis of reporting verbs in master's theses. Procedia - Social and Behavioral Sciences, 134, 140-145. https://doi.org/10.1016/j.sbspro.2014.04.232
Mansourizadeh, K., & Ahmad, U. K. (2011). Citation practices among non-native expert and novice scientific writers. Journal of English for Academic Purposes, 10, 152-161. https://doi.org/10.1016/jjeap.2011.03.004
McEnery, A., & Wilson, A. (2001). Corpus linguistics (2nd ed). Edinburgh University Press.
Mollin, S. (2014). The (ir) reversibility of English binomials: Corpus, constraints, developments (vol. 64). John Benjamins Publishing Company. https://doi.org/10.1075/scl.64
Partington, A. (1998). Patterns and meanings. John Benjamins. https://doi.org/10.1075/scl.2
Partington, A.(2004). Utterly content in each other's company: Semantic prosody and semantic preference. International Journal of Corpus Linguistics, 9(1), 131-56. https://doi.org/10.1075/ijcl.9.1.07par
Pecorari, D. (2008). Academic writing and plagiarism: A linguistic analysis. Continuum.
Radden, G., & Dirven, R. (2007). Cognitive English grammar. John Benjamins. https://doi.org/10.1075/clip2
Pickard, V. (1993a). Citing previous writers: What can we say instead of «say». Paper presented at RELC Conference, 1993. Singapore.
Pickard, V. (1993b). Should we be teaching refutation? Concordanced evidence from the field of applied linguistics. In N. Bird, J. Harris, & M. Ingham (Eds.), Language and content. Institute of Language in Education.
Sarda L., Carter-Thomas S., Charolles M. & Fagard, B. (2014) Adverbials: From predicative to discourse functions. In Sarda L., Carter-Thomas S., Charolles, M., & Fagard, B. (Eds.), Adverbials in use: From predicative to discourse functions (pp. 13-38). Presses universitaires de Louvain.
Scarcella, R. C. (1984). How writers orient their readers in expository essays: A comparative study of native and non-native English writers. TESOL Quarterly,18(4), 671. https://doi.org/10.2307/3586582
Sinclair, J. (1991). Corpus, concordance, collocation. Oxford University Press.
Sinclair, J. (1996). The search for units of meaning. Textus, IX, 75-106.
Schmitt, N., & Carter, R. (2004). Formulaic sequences in action: An introduction. In N. Schmitt (Ed.), Formulaic sequences (pp. 1-22). John Benjamins.https://doi.org/10.1075/lllt9
Stewart, D. (2010). Semantic prosody: A critical evaluation. Routledge. https://doi.org/10.4324/9780203870075
Stubbs, M. (1995). Collocations and semantic profiles: On the cause of the trouble with quantitative methods. Function of Language, 2(1), 1-33. https://doi.org/10.1075/fol.2.1.03stu
Stubbs, M. (1996). Text and corpus linguistics. Blackwell.
Stubbs, M. (2001a). On inference theories and code theories: Corpus evidence for semantic schemas. Text, 21(3), 437-65. https://doi.org/10.1515/text.2001.007
Stubbs, M. (2001b). Words and phrases. Blackwell.
Stubbs, M. (2002). Two quantitative methods of studying phraseology in English. International Journal of Corpus Linguistics, 7(2), 215-44. https://doi.org/10.1075/ijcl.72.04stu
Stubbs, M. (2006). Corpus analysis: The state of the art and three types of unanswered questions. In G. Thompson & S. Hunston (Eds.), System and corpus: Exploring connections (pp. 15-36). Equinox.
Swales, J. (1990). Genre analysis: English in academic and research settings. Cambridge University Press.
Thompson, G., & Ye, Y.Y. (1991). Evaluation in reporting verbs used in academic papers. Applied Linguistics, 12, 365-382. https://doi.org/10.1093/applin/12.4365
Thompson, G. (1994). Guide to reporting. Harpercollins.
Tognini-Bonelli, E. (2001). Corpus linguistics at work. John Benjamins. https://doi.org/10.1075/scl.6
Unaldi, i. (2013). Genre-specific semantic prosody: The case of pose. Mustafa Kemal Universitesi Sosyal Bilimler Enstitusu Dergisi, 10(21), 37-54.
Xiao, R., & McEnery, T. (2006). Collocation, semantic prosody, and near synonymy: A cross-linguistic perspective. Applied Linguistics, 27(1), 103-129. https://doi.org/10.1093/applin/ami045
Yeganeh, M. T., & Boghayeri, M. (2015). The frequency and function of reporting verbs in research articles written by native Persian and English speakers. Procedia-Social and Behavioral Sciences, 192, 582-586. https://doi.org/10.1016/j.sb-spro.2015.06.097
Ziman, J. M. (1968). Public knowledge: An essay concerning the social dimension of science. Cambridge University Press.
APPENDIX
Table 4
The Sampling Procedure for Hypothesizing Verbs
Verb / TextCategory Academic Writing Spoken Fiction Magazine Newspaper
Size 81 million 85 million Words 81 million Words 86 million Words 81 million Words
Words (opm) (opm) (opm) (opm) (opm)
hypothesize 371 18 16 54 8
hypothesized 2.594 19 27 88 17
hypothesizing 51 4 9 19 3
hypothesizes 72 0 8 55 3
Total 3,088 (38,1) 41 (0,5) 50 (0,6) 216 (2,5) 31 (0,3)
posit 418 38 21 80 23
posited 644 16 38 88 41
positing 183 6 12 25 8
posits 594 13 17 121 38
Total 1,839 (22,7) 73 (0,9) 88 (1,08) 314 (3,6) 110 (1,3)
ponder 373 (4,6) 142 (1,7) 352 (4,3) 493 (5,7) 471 (5,8)
consider 17.526 7990 4.730 14.508 10.943
considered 23,904 9,012 7,833 10,034 11,875
Total 41,430 (511,4) 17,002 (209,9) 12,563 (155,09) 24,542 (285,3) 22,818 (28,2)
assume 6123 4550 2509 3.799 2330
assumed 6,038 1,197 3,551 2,716 1,721
assuming 2,469 1,513 1,304 1,635 1,189
assumes 2,095 298 304 837 511
Total 16,725 (206,4) 7,558 (93,3) 7,668 (94,6) 8,987 (104,5) 5,751 (71)
conceptualize 472 (5,8) 21 (0,2) 14 (0,17) 46 (0,53) 13 (0,16)
ascertain 994 (12,2) 139 (1,6) 135 (1,6) 124 (1,4) 83 (1,02)
expect 7.013 (86,5) 12.719 (157,02) 7.241 (89,3) 10.788 (125,4) 11.035 (136,23)
ascertain 994 (12,2) 139 (1,6) 135 (1,6) 124 (1,4) 83 (1,02)
expect 7.013 (86,5) 12.719 (157,02) 7.241 (89,3) 10.788 (125,4) 11.035 (136,23)
elucidate 400 8 18 56 15
postulate 277 14 22 65 19
postulated 522 9 24 78 22
postulating 59 1 6 10 5
postulates 264 9 17 45 6
Total 1,122 (13,8) 31 (0,3) 69 (0,8) 198 (2,4) 52 (0,6)
think 26.677 (329,3) 413.902 (5,109,9) 111.874 (1,381,1) 67.017 (7779,2) 76.889 (949,2)
suppose 1.760 (21,7 3.652 (45,08) 8.871 (109,5) 1.882 (21,8) 938 (11,5)
theorize 220 41 35 136 65
theorized 366 47 64 187 85
theorizing 578 13 36 79 24
theorizes 62 13 4 123 49
Total 1,226 (15,1)) 114 (1,4) 139 (1,7) 525 (6,4) 223 (2,7)
presume 426 540 433 264 158
presumed 1,310 337 487 488 405
presuming 98 72 58 58 46
presumes 259 67 37 93 56
Total 2,093 (25,8) 1,016 (12,5) 1,015 (12,5) 903 (11,1) 665 (8,2)
predicate 134 (1,6) 75 (0,9) 20 (0,2) 14 (0,1) 10 (0,1)
presuppose 173 (2,1) 10 (0,1) 5 (0,06) 30 (0,3) 10 (0,1)
speculate 823 851 302 600 496
speculated 520 157 308 413 401
speculating 105 320 184 150 127
speculates 147 34 22 335 119
Total 1595 (19,6) 1362 (16,8) 816 (10,07) 1498 (17,4) 1143 (14,1)
Verb / TextCategory Academic Writing Spoken Fiction Magazine Newspaper
quantify 787 162 29 241 242
quantified 531 26 15 93 41
quantifying 288 7 5 46 25
quantifies 96 1 0 17 5
Total 1,702 (21,0) 196 (2,4) 49 (0,6) 397 (4,6) 276 (3,4)
conjecture 289 (3,5) 106 (1,3) 159 (1,9) 186 (2,1) 89 (1,09)
contemplate 510 (6,2) 273 (3,3) 474 (5,8) 515 (5,9) 340 (4,19)
Table 5
The Sampling Procedure for Reporting Verbs
Verb / Text Category Academic Writing Spoken Fiction Magazine Newspaper
Size 81 million 85 million 81 million Words 86 million Words 81 million Words
Words (opm) Words (opm) (opm) (opm) (opm)
note 148.888 (1,838,1) 6.604 (77,6) 8.550 (105,5) 9.518 (110,6) 7.888 (97,3)
argue 62.288 (768,9) 4.826 (56,7) 2.406 (29,703) 3.912 (45,4) 4.476 (55,2)
claim 84.616 (1,044) 5.563 (65,44) 3.457 (42,67) 7.312 (85,02) 6.660 (82,22)
report 204.224 (2,521,2) 208.024 (2,44) 46,712 (576,6) 122,960 (1,429,7) 193,568 (2,389,7)
suggest 148.000 (1,827) 5,158 (60,6) 2,364 (29,18) 7,405 (86,104) 4,827 (59,5)
state 704.144 (8,693,1) 493,504 (5,8) 102,608 (1,266,7) 394,160 (4,583,2) 959.064 (11,840,2)
find 222.480 (2,746,6) 420,104 (4,9) 407,224 (5,027,4) 442.952 (5,150,6) 297,864 (3,677,3)
indicate 106.464 (1,314) 2.010 (23,6) 1.046 (12,9) 3.319 (38,5) 2.054 (25,3)
reveal 40.895 (504,8) 1.976 (23,2) 3.779 (46,6) 4.373 (50,8) 2.045 (25,2)
mention 32.208 (397,6) 37,248 (438,2) 46.640 (5,758) 39,552 (459,9) 36,064 (445,2)
Table 6
The Sampling Procedure for Summarizing Verbs
Verb / Text Category Academic Writing Spoken Fiction Magazine Newspaper
Size 81 million 85 million Words 81 million Words 86 million Words 81 million Words
Words (opm) (opm) (opm) (opm) (opm)
sum up 1,288 (15,9) 1,323 (15,5) 469 (5,7) 959 (11,15) 686 (8,4)
outline 13,144 (162,2) 3,688 (43,3) 9,304 (114,8) 8,104 (94,23) 4,032 (49,7)
outlined 19,928 (246,02) 5,200 (61,17) 4,528 (55,90) 5,488 (63,81) 5,992 (73,9)
summarize 7,592 (93,7) 1,488 (17,5) 608 (7,50) 1,152 (13,3) 520 (6,4)
encapsulate 720 (8,8) 208 (2,4) 88 (1,08) 320 (3,7) 184 (2,2)
condense 824 (10,3) 264 (3,10) 416 (5,1) 1,056 (12,2) 232 (2,8)
recapitulate 440 (5,4) 56 (0,6) 56 (0.6) 96 (1,11) 56 (0,6)
conclude 25,904 (319,8) 7,192 (84,6) 3,080 (38,02) 8,624 (100,2) 7,352 (90,7)
review 193,112 (2,384,09) 37,656 (443,01) 20,456 (252,5) 49,216 (572,27) 67,168 (829,23)
