“Climate change” vs. “global warming”: A corpus-assisted discourse analysis of two popular terms in The New York Times

3.1 Data collection

This study focuses on The New York Times, a newspaper that is known as a “newspaper of record” (Cotter 2001: 416) in the US and is important in influencing media agenda in other countries. Two large corpora have been built by collecting the news texts with CC or GW in their titles in the NYT from the electronic database Lexis-Nexis. The time-span set for data collection is from January 1, 2000 to December 31, 2019. Previous studies on CC have found that climate issues have received increasing attention since the mid-2000s in most developed and developing countries (Song et al. 2021). All the collected news texts are further examined manually to make sure that they are closely related to GW or CC. The GW corpus consists of 363 news texts (279,839 tokens), while the CC corpus consists of 1,088 news texts (1,285,183 tokens). Therefore, CC was used much more frequently than GW during the examined period. Figure 1 shows the number of news texts collected each year, and Figure 2 shows the raw frequencies of the two terms in the two corpora.

Figure 1:

The number of news reports collected for each year.

Figure 2:

The frequencies of two terms in each year.

As Figure 1 shows, NYT features fewer uses in CC before 2005 (Bohr 2020; Feldman et al. 2015; Schmid-Petri et al. 2017). Before 2008, GW was more frequently used than CC in NYT, but this trend has been reversed since 2008. Compared with the growing coverage of CC, the coverage of GW was relatively stable and comparatively few after 2008. The first surge of attention to CC during 2007–2010 can be attributed to several activities, such as the release of the Fourth Assessment Report from the Intergovernmental Panel on Climate Change (IPCC), the release of an award-winning documentary by the former Vice President Al Gore, and the “Climategate” scandal (Bohr 2020). The coverage of CC reached its first peak in 2009, which can be attributed to COP15 in Copenhagen (2009). The coverage of CC has shown a dramatic increase since 2014 and reached its second peak in 2015, which can be attributed to COP21 in Paris (2015). Since then, the coverage of CC has been doubled due to the former US President Donald Trump who questioned the real existence of CC and made it a controversial issue.

3.2 Analytic methods and procedure

This study gives a corpus-assisted discourse analysis of the representations of CC and GW in NYT in terms of three aspects: (1) the topics/themes associated with CC and GW; (2) the particular ways of framing CC and GW; and (3) the perspectivization strategy taken towards CC and GW. Perspectivization strategy is a method to position writers’ viewpoints to express involvement and distance (Reisigl and Wodak 2016). It is usually reflected by the use of deictics, direct, indirect, free indirect speech, etc. We start with the identification of the topics/themes associated with the coverage of CC and GW. The corpus-analytic tool Wmatrix 3.0 is used for the analysis of the most prominent and preferential topics/themes (Liu 2017). Wmatrix 3.0 distinguishes itself from other corpus-analytic tools for the UCREL Semantic Analysis System (USAS) it incorporates. USAS divides the English vocabulary into 21 semantic domains which are further divided into 232 semantic categories (SMCs). Wmatrix 3.0 can do automatic semantic analysis and corpus comparison. It can produce a key SMC list by comparing one corpus with a specialized or general reference corpus. The key SMC list can rank these statistically significant key SMCs in terms of their log-likelihood values. The more their key values are, the more statistically significant these key SMCs are. In order to identify the most prominent themes/topics of the two corpora, this study first compares the two corpora respectively with the British National Corpus (BNC) Sampler written corpus in Wmatrix and produces two key SMC lists. Although it has to be acknowledged that an American sampler Written corpus is preferred here, the choice of American and British does not seriously affect the results because the semantic analysis categorizes these words in terms of meanings rather than spellings in Wmatrix. Then we focus on the top 20 key SMCs in each list to identify the shared topics/themes of the two corpora. After that, the two corpora are compared with each other to identify their preferential key SMCs. For the convenience of analysis, we still focus on the top 20 key SMCs in each corpus. The specific functions of each key SMC can be further examined through a close examination of the most frequently used tokens in each key SMC (L’Hôte 2010; Rayson 2008).

It is followed by a close examination of the particular ways of framing CC and GW in the two corpora (see Grundmann 2021). According to Entman (1993), to frame is to “select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (Entman 1993: 52, italics in original). The particular ways of framing CC and GW can be examined through their strong collocates (Baker et al. 2008). Different statistic measures can be used to measure the strength of collocation (Baker 2006), and the log-likelihood (LL) values are used in this study to rank the collocates of CC and GW. With the corpus-analytic tool Wordsmith 7.0, this study focuses on the top 100 collocates of CC and GW in their respective corpora to identify which collocates contribute to the four aspects of framing the issue: (1) problem definition; (2) cause; (3) consequence; and (4) solution/action.

Nevertheless, the analysis of framing in terms of their collocates may not present an accurate picture about the exact stance of the newspaper towards the (un)certainty of CC or GW in view of the journalistic norm of balanced reporting (Boykoff and Boykoff 2004). In order to examine to what extent the newspaper adheres to the journalistic norm of balanced reporting and (dis)aligns with the scientific consensus on the existence of CC or GW, this study gives a further analysis of the perspectivization strategy adopted towards the (un)certainty of CC and GW. Perspectivization refers to how the news reporters position themselves towards others’ viewpoints and express involvement or distance in discourse (Reisigl and Wodak 2016), and it can manifest at all levels of discourse (KhosraviNik 2010). The concordance lines of some collocates with a contradictory evaluation of the existence of CC and GW are further examined to reveal the specific perspectivization strategies taken towards these statements.

4.1 Analysis of topics/themes

Table 1 shows the top 20 key SMCs of both corpora. Among them, 17 can be identified in both corpora, including W4 (“Weather”), A2.1+ (“Change”), Y1 (“Science and technology in general”), W5 (“Green issues”), Z99 (“unmatched”), W3 (“Geographical terms”), O1.3 (“Substances and materials: Gas”), O4.6+ (“Temperature: Hot/on fire”), Q2.1 (“Speech: Communicative”), G1.1 (“Government”), W1 (“The universe”), O4.6 (“Temperature”), F4 (“Farming & Horticulture”), A2.2 (“Cause & effect/connection”), N5.1+++ (“Entire; maximum”), A13 (“Degree”), and X5.2+ (“Interested/excited/energetic”). They show that the representations of CC and GW share many similarities in the choice of topics/themes.

An examination of the most frequent tokens in these shared key SMCs reveals the presence of three themes: (1) environment; (2) politics; and (3) science. The majority of the key SMCs belong to the theme of “environment”, including W4 (e.g. climate, weather, drought), A2.1+ (e.g. change, changes, changing), W5 (e.g. environmental, environment, pollution), O1.3 (e.g. gas, carbon-dioxide, air, gases), O4.6+ (e.g. warming, heat, fire), W1 (e.g. world, planet, solar), W3 (e.g. global, earth, forests), O4.6 (e.g. temperatures, temperature, Fahrenheit), F4 (e.g. farmers, agriculture, agricultural), N5.1+++ (e.g. extreme, extremes), and X5.2+ (e.g. energy). The prominence of this theme is not surprising in view of the fact that both corpora are concerned with environmental issues. Therefore, both corpora underline the change in climate, weather, and temperature, the release of greenhouse gases, the impact on agriculture, and extreme weather. The theme “politics” can be perceived from G1.1, which consists of such tokens as president, government, state, country, nations, and officials. They underline the significance of government in the representations of the two issues. The key SMC Y1 suggests the presence of the theme “science”. Its most frequent tokens are scientists, science, scientific, technology, technologies, and science. They suggest that both corpora make frequent references to science in the representations of the two issues. However, Y1 ranks higher than G1.1 in both corpora, suggesting that “science” is more highly foregrounded than “politics”.

Nevertheless, a comparison of the two corpora with each other also reveals their preference for some themes and topics (see Table 2). CC features the preference for a variety of topics/themes, including “politics” (G1.2, G1.1, G2.1−), “economy” (I1 and I1.1), “food” (F1), “education” (P1), “drinks and alcohol” (F2), “work and employment” (I3.1), “flying and aircraft” (M5). The preference for “politics” suggests that the representations of CC tend to be more politicized than the representations of GW in NYT. Besides, the emphasis on the topics of “food”, “drinks and alcohol”, “economy”, and “work and environment” actually indicates that the representations of CC tend to highlight the economic and environmental impacts of CC. This can also be seen from other key SMCs with negative meanings, including E4.1− (e.g. suffer, suffering, suffered), and A1.1.2 (e.g. damage, devastating, destruction). By contrast, GW shows the preference for a limited number of topics/themes, including “science and technology” (Y1), “gas emissions” (O1.3 and “emissions” in A9−), “substances and materials” (O1), “temperature change” (O4.6), “energy” (X5.2+), and “media” (O4.3). It may be attributed to fewer uses of GW in the recent years. The preference for “science and technology” suggests that GW tends to put more emphasis on scientific consensus than CC. This will be further illustrated in Section 4.3. The preference for “gas emissions”, “substances and materials”, and “energy” also sheds light on its preference to highlight the cause and solution of GW. This can also be seen from its preference for other key SMCs which underline the action taken towards GW, including N5.1 (e.g. limits, and limit), N5− (reduce, reducing, reductions), N3.1 (e.g. measures, measure, measured), N3.8− (e.g. slow, slowdown, slowing), and A1.7+ (e.g. regulate, limit, curb). The preference for “media” (O4.3) can be attributed to the fact that GW tended to draw more media attention in the last two decades. This is in contrast to the use of CC and GW on Twitter, where Shi et al. (2020) find that GW tends to be more politicized and concerned about the general phenomenon, while CC prefers to foreground scientific perspective and refer to specific topics and issues.

4.2 Analysis of framing

An examination of the top 100 lexical collocates finds that they can be divided into four categories: (1) problem definition; (2) cause; (3) impact; and (4) solution/action (see Table 3). A number of collocates can be identified for both CC and GW, and they are marked in bold. These shared collocates suggest that CC and GW are represented in a much similar way in the NYT. As regards “problem definition”, both CC and GW are constructed as contentious issues, because these collocates communicate opposite meanings. Some underline the issue as real (i.e. real, reality), while others question the certainty of its existence (i.e. hoax, debate, deny, and skeptics). In view of “cause”, these shared collocates underline both human causes (i.e. human) and non-human causes (i.e. greenhouse, carbon, coal). With regard to “impact”, these collocates highlight the serious impacts and threats of CC and GW, including effects, threat, impact, concern, and poses. In terms of “solution/action”, these shared collocates highlight the fight against the issue (i.e. fight, fighting, combat, tackling, address, and addressing) and the mitigation of the issue (i.e. mitigate). In other words, both CC and GW are framed as contentious issues, caused by both human and non-human factors, with serious consequences and in need of addressing and mitigation actions.

Their preference for other collocates suggests their divergence in framing the two issues. In terms of “problem definition”, while CC shows a strong collocation with issue (124), GW shows a strong collocation with both problem (21) and issue (21). An issue refers to a situation that requires a discussion or a debate, whereas a problem involves a negative situation that requires an instant solution. Therefore, the nomination of CC or GW as an issue underlines that it is a topic in which different people might have different viewpoints, while the framing of GW as a problem suggests that it must be solved. The representations of GW thus tend to be more negative than the representations of CC. The strategic difference in the choice of the two words can be detected in the following example from NYT:

The difference can also be identified in the collocates related to “solution/action”. While solutions (16) and solving (5) occur as strong collocates of GW, they do not appear as strong collocates for CC. Besides, there is more consensus on the certainty of GW than on the certainty of CC. CC shows a strong collocation with hoax (40), deniers (28), and denial (26). They underline the uncertainty or non-existence of CC. However, GW shows a strong collocation with consensus (13) and unequivocal (4), which underlines the consensus on GW.

As regards the “causes” of CC and GW, while both CC and GW underline human causes, CC tends to put more emphasis on the anthropogenic causes than GW. This can also be seen from the predication strategy for CC and GW. Predication strategy is a strategy to attribute features to entities (Reisigl and Wodak 2016). It can be reflected in the form of adjectives, appositions, prepositional phrases, etc. CC tends to be pre-modified by such expressions as human-caused (51), human-induced (10), and human-driven (3). However, GW is seldom pre-modified by such expressions as human-caused (3), human-induced (2), and human-driven (2). As Taylor (2003: 455) suggests, ‘‘attributive adjectives tend to characterize a thing in terms of a stable, inherent property, whereas predicative adjectives tend to denote more temporary, circumstantial properties”. It suggests that NYT tends to underline human causes as an inherent property of CC, as in the following:

In terms of the “impact” of CC and GW, they show differences in both the nature and severity of the impacts. CC tends to be concerned with national security, as can be seen from the strong collocate security (42). However, security does not occur as a strong collocate of GW. Besides, CC tends to collocate with such tokens which dramatize the negative impact of CC, including serious (34), catastrophic (24). However, they do not appear as strong collocates with GW. Even though GW also shows a strong collocation with other tokens like dire (4) and doom (3). Their frequencies are comparatively low.

With regard to the “solution/action”, while NYT underlines the necessity to address and mitigate the issue, it also underlines the adaptation strategy to CC, as can be seen from the strong collocate adapt (34). Mitigation strategy refers to “activities to reduce or prevent carbon emissions”, while adaptation strategy concerns “activities to adjust economic and social systems to the effects of climate change” (Günay et al. 2021: 1). This is in contradiction to previous studies which claim that news reporting on CC rarely covers adaptation (Boykoff and Roberts 2007; Ford and King 2015; Liu et al. 2008; Moser 2014). However, adapt does not appear as a strong collocate of GW. Instead, as mentioned above, NYT also underlines the solution of the problem of GW, as can be seen from the strong collocates solutions (16) and solving (5). Therefore, while mitigation strategies have been emphasized for both CC and GW, adaption strategy has also been emphasized for CC, solution strategy for GW.

Figure 3 shows the percentages of the total frequencies of the collocates in the four categories when they are compared with the total frequencies of GW and CC in their respective corpus. It shows that CC shows the preference for “impact”, while GW shows the preference for “solution/action”, and “cause”. This further supports the findings in Section 3.1.

Figure 3:

Percentages of five semantic preferences in the corpus.

4.3 Analysis of perspectivization

This section gives a close examination of the perspectivization strategy adopted by the NYT in representing the (un)certainty of CC and GW. The concordance lines of the collocation of GW/CC with two collocates with contradictory meanings (i.e. real and hoax) are further examined. As Boykoff and Boykoff (2004) argue, the adherence to the norm of balanced reporting in US prestige press has resulted in biased coverage of “anthropogenic contributions to global warming and resultant action” (Boykoff and Boykoff 2004: 125). This is what they call “false balance”. Therefore, an examination of the perspectivization strategy in the representations of the (un)certainty of CC or GW can further expose how the contradictory viewpoints are coherently organized and whether “false balance” can still be identified in NYT. A distinction between attribution and averral can be made in examining these concordance lines (Partington 2012; Sinclair 1988). Attribution refers to the explicit attribution of what is being said to an external source, while averral refers to the statement of the viewpoints by the newspaper itself. As Example (3) shows, the statement that CC is real is attributed to hundreds of scientists rather than the news reporter. Therefore, it is an example of attribution rather than averral.

An examination of the collocation of real (70) and CC finds that 60 (86%) address whether CC is real or not. All of the 60 concordance lines underline CC as real, and all the statements are attributed to external sources rather than averred by the news reporters themselves. Five groups of external sources can be identified, including “science” (16), “common people” (15), “politics” (13), “business” (4), and “Religion” (1). Examples are as follows:

Therefore, “science”, “common people” and “politics” occur as the three dominant groups of sources. Their similar frequencies can be attributed to the journalistic norm of “balanced” reporting (Boykoff 2008; Boykoff and Boykoff 2004), which requires the relatively equal recognition to multiple sides of contentious issues (Bennett 1996). Therefore, NYT still features the balanced representations of the voices from politics, science, and common people in its representations of CC as real, even though Boykoff (2007) claims that this “false balance” is less noticeable after 2005.

While adhering to the journalistic norm of “balanced” reporting, the news reporters also indicate their stance towards these statements through reporting verbs. The most frequently used reporting verbs in these examples include say* (8), agree* (5), acknowledge* (4), convince* (3), believe* (5), know* (2), tell* (2), and understand* (2). Among them, acknowledge*, accept*, know*, agree*, believe*, understand*, and endorse* are mental process verbs (Halliday 1994), and some of them carry the implicature that the represented statement is true, such as acknowledge*, accept*, know*, and understand* (Haarman and Lombardo 2009).

NYT’s explicit stance can also be revealed through its use of hoax. An examination of the 49 concordance lines of the collocation of hoax and CC finds that 46 (94%) underline that CC is a hoax. However, all of them are attributed to the politicians in the US, the former US President Donald Trump in particular. Besides, NYT also shows disalignment with these statements in three ways. First, scare quotes are often used in these examples to show that these are the words used by the attributed sources and that the news reporter does not share the responsibility for these words (Fairclough 1992, 1995), as in the following:

Reporting verbs are also used to indicate the news reporter’s disalignment with these viewpoints. The most frequently used reporting verbs in these concordance lines are call* (15), think* (7), claim* (3), argue* (3), denounce* (1), mock* (1), and deride* (1). Most of them are verbal process verbs (Halliday 1994) and suggest that what is stated is personal and malicious, as in the following:

Finally, the newspaper’s dealignment with these statements can also be perceived through their overt negative evaluation of these statements, as in the following:

In these examples, the claim that CC is a “hoax” is often negatively evaluated by such expressions as falsely and absurd. They suggest that the represented statements are not true. In other words, the news reporter does not share the same view with the attributed sources.

The same is true for the collocation of GW and real (18). An examination of all the concordance lines finds that 13 out of 18 (72%) instances state whether GW is real or not. 12 of them underline GW as real, and only 1 occurrence underlines GW as not real. Examples are as follows:

Similar to the representations of CC, the sources attributed include “politics”, “science”, “common people”. These statements are often introduced by the reporting verbs and some expressions which also carry the implicature that what is stated is a matter of fact, as can be seen from the expressions like declare*, acknowledge*, the mounting evidence, and actually the mainstream view.

Hoax (4) does not appear as a frequent collocate of GW. This is also in contradiction to the statement that GW frequently collocates with hoax (Shi et al. 2020). Like CC, the external sources are also attributed to the politicians in the US.

Therefore, NYT still adheres to the journalistic norm of “balanced” reporting by citing contrarian viewpoints (Boykoff and Boykoff 2004). However, it also seeks to correct “false balance” through showing explicit and implicit alignment with the scientific consensus on the existence of CC and GW and disalignment with these viewpoints that challenge or question the existence of CC and GW (Brüggemann and Engesser 2014, 2017; Schmid-Petri et al. 2017).