The mass media significantly shapes public perception through language. This study focuses on news discourse surrounding the Grand Ethiopian Renaissance Dam (GERD), a highly controversial project on the Blue Nile, initiated in 2013. The GERD's construction has fueled intense conflict among Nile basin countries, with media playing a crucial role in shaping public opinion. While existing research examines media coverage of the GERD, there's a gap in the linguistic analysis of the specific language and ideological underpinnings used. This study addresses this gap by employing critical discourse analysis (CDA) and corpus linguistics (CL) to compare the discourse in Egyptian and Ethiopian online English news corpora, totaling over a million words across 2655 articles. The study utilizes Fairclough's three-dimensional framework (text, discourse practice, and social practice) and van Dijk's ideological square (self-positive and other-negative representation) to analyze the data. The primary objectives are to compare lexical choices, identify statistically salient keywords and their collocates, and uncover ideological implications in the news reporting. The study seeks to answer: 1) What are the lexical items collocating with the highest statistically salient keywords related to the GERD in Egyptian and Ethiopian newspapers? 2) What does the collocation analysis of keywords reveal about the different stances adopted by the two conflicting countries regarding the GERD issue?

Existing research on the GERD conflict largely focuses on media framing, using content analysis on limited data, often lacking a detailed linguistic analysis. Studies like Bealy (2014), El-Tawil (2018), El Damanhoury (2023), and Sayed El Ahl et al. (2023) identify dominant frames (development, conflict, crisis, national interest), but lack in-depth linguistic analysis. A few corpus-based CDA studies exist, such as Elieba (2022) and Elsoufy and Ibrahim (2022), which compare lexico-syntactic features or identify recurring themes but lack the detailed collocation analysis undertaken in this study. Other linguistic studies examine press releases or speeches (El Shazly and El Falaky, 2023; Siraw, 2023), but do not analyze word choices in news discourse. This study differentiates itself through a critical perspective, combining CDA and CL, and employing a large dataset spanning 2013-2020, thus filling a significant gap in the literature by focusing on lexical choices and their ideological implications in a large, comparative corpus.

The study employed a corpus-based approach, compiling two comparable corpora from Egyptian and Ethiopian online English news sources (2655 articles, over 500,000 words each). Data were collected from three newspapers in each country – a state-owned and two private – using the Factiva database and the search term "Grand Ethiopian Renaissance Dam." Hard news articles were selected, excluding opinions and editorials. Wordsmith Tools software was used for keyword extraction (using Dunning's log-likelihood, BIC, and Hardie's Log Ratio), and Sketch Engine analyzed collocations (using LogDice). Keyness was assessed using three metrics: Dunning's log-likelihood (LL), Bayes Information Criterion (BIC), and Hardie's Log Ratio. Top 50 keywords were compiled from each corpus. A shortlist of five statistically salient keywords, most relevant to the study's focus, was chosen for in-depth collocational analysis. This analysis considered the grammatical categories of collocates (colligations), semantic preferences, and semantic prosodies. Data analysis followed Fairclough's three-dimensional CDA model (description, interpretation, explanation) and van Dijk's ideological square model. The analysis involved extracting keywords, analyzing their collocations, categorizing collocates grammatically, identifying semantic preferences and prosodies, identifying ideological strategies, and considering historical and social contexts.

The keyword analysis revealed significant differences between the Egyptian and Ethiopian corpora, except for the keyword "negotiations." In the Egyptian corpus, keywords like "crisis," "fears," "affect," and "interests" reflected a negative portrayal of the GERD. Collocational analysis of "crisis" revealed a strong negative prosody, with modifiers like "economic," "water," and "political," and verbs like "escalate" and "exacerbate." The keyword "fears" was frequently modified by terms such as "overpopulation," further highlighting anxieties about water scarcity. "Affect" appeared in collocations suggesting negative impacts on Egypt's water share and security. The keyword "negotiations" was used in both positive and negative contexts. In the Ethiopian corpus, keywords like "power," "growth," "nation," and "utilization" highlighted the positive aspects of the GERD, emphasizing development and economic progress. "Power" was frequently linked to "hydroelectric" and "generation." "Growth" strongly collocated with "economic" growth, and "utilization" appeared frequently with terms like "equitable" and "sustainable." The keyword "negotiations" was used in both positive and negative contexts. Both corpora used "negotiations" to refer to diplomatic efforts, but with different perspectives on success or blame for failure. The analysis revealed contrasting ideological strategies: Egypt's media employed victimization, dramatization, and attribution of agency, highlighting Ethiopia's negative actions, while Ethiopian media emphasized national self-glorification and positive self-representation while casting a negative light on Egypt's actions and motivations.

The findings directly address the research questions, demonstrating significant differences in lexical choices and framing between Egyptian and Ethiopian news coverage of the GERD. The identified collocations and their prosodies strongly suggest underlying biases reflecting national interests and political stances. The contrasting use of keywords and their collocates reveals how each country strategically constructs its narrative, portraying itself positively and the other negatively. This confirms the existence of media bias and ideological polarization, supporting existing research on media framing in international conflicts. The study's strength lies in its corpus-based approach, which overcomes the limitations of previous qualitative studies by providing quantitative evidence of lexical patterns and their ideological implications.

This study provides valuable insights into the linguistic construction of media bias in the context of the GERD conflict. The contrasting lexical choices and framing strategies employed by Egyptian and Ethiopian news outlets reveal the significant role of language in shaping public opinion and fueling political tensions. The study's contribution is its in-depth analysis of lexical collocations, demonstrating how subtle linguistic choices can reinforce pre-existing biases and contribute to the polarization of narratives. Future research could extend this analysis to other languages, media types, and time periods to further explore the dynamics of media representation in international conflicts. Investigating the influence of specific journalistic practices on news values and objectivity would be particularly valuable.

This study focuses on English-language news reports, limiting its scope. A wider linguistic analysis including Arabic, Amharic, and other languages used in the region would be beneficial. The study is limited to hard news reports, excluding opinion pieces and editorials, which could offer additional perspectives. The study also focuses on a specific time frame, and extending the analysis to cover a longer period could reveal shifts in media framing and discourse. The corpus might not fully represent the diversity of views within each country. Future research could consider analyzing other media forms such as television and social media to obtain a fuller picture of the media coverage.