Previous Article in Journal
Use of Social Media by Health Science Degree Students in the Field of Organ Donation and Transplantation
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Journalism and Media
Volume 6
Issue 3
10.3390/journalmedia6030114
journalmedia-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Politics of Framing Water Infrastructure: A Topic Model Analysis of Media Coverage of India’s Ken-Betwa River Link

by
Harman Singh
1,*,
Matthew Hansen
2 and
Trevor Birkenholtz
1
1
Department of Geography, The Pennsylvania State University, University Park, PA 16801, USA
2
Institute for Computational and Data Sciences, The Pennsylvania State University, University Park, PA 16801, USA
*
Author to whom correspondence should be addressed.
Journal. Media 2025, 6(3), 114; https://doi.org/10.3390/journalmedia6030114
Submission received: 3 June 2025 / Revised: 29 June 2025 / Accepted: 9 July 2025 / Published: 22 July 2025
Browse Figures
Versions Notes

Abstract

The framing of water infrastructure in the news influences how the public perceives future infrastructure development and associated social-environmental risks. This study examines English-language newspaper coverage of the Ken-Betwa river link, the first component of India’s National River Linking Program (INRLP) to receive approval. Data for this analysis comprised 316 newspaper articles, collected via a keyword search in LexisNexis API, from seven Indian English-language newspapers (Free Press Journal (India), Hindustan Times, Indian Express, The Economic Times, The Hindu, The Times of India (TOI), and Times of India (Electronic Edition)) published between 2004 and 2022. By applying LDA topic modeling, a type of generative probabilistic model, to this dataset, this study examines how evolving media narratives frame water infrastructure in India. Our results identify 23 distinct topics and three dominant frames: (1) a government policy frame, (2) INRLP comparative frame, and (3) environmental conservation frame. We find that these frames evolve, with early coverage emphasizing feasibility and government-led negotiations, and later articles highlighting environmental risks. Our analysis shows how media discourse reflects institutional logic and infrastructure milestones. This study demonstrates the value of computational methods for longitudinal media analysis, has the potential to reveal shifts in public discourse, and highlights power dynamics in environmental reporting.

1. Introduction

The politics of framing water infrastructure in news media significantly influences how the public perceives infrastructure development and associated socio-environmental risks. Projected global investments in the renewal, repair, maintenance, and replacement of existing water infrastructure are estimated by UNESCO to reach USD 0.9–1.5 trillion annually worldwide (UNESCO, 2023). These large-scale projects often provoke intense public debate due to their costs and environmental and social impacts. In India, the Ken-Betwa river link is the first component of the National River Linking Program (INRLP) to gain approval. This large-scale water infrastructure project aims to redistribute surface water from areas of water surplus to areas of water deficit. The Ken-Betwa river link is expected to provide water for irrigation, drinking, and hydropower (Rai, 2018), but it is controversial due to the anticipated submergence of parts of the Panna Tiger Reserve and the displacement of human settlements (Parveen & Ilyas, 2021). This project is criticized by human rights and environmental advocates but favored by water resource planners. Hence, understanding how media contextualizes the evolving politics of large-scale water infrastructure projects is crucial, as public perception of such projects is shaped by media narratives (Crow-Miller et al., 2017).
News media offers a rich source of data to understand public narratives around the longitudinal development of water projects and has been recognized as a valid proxy to track societal values (Quesnel & Ajami, 2017). Historically, the sheer volume of newspapers has resulted in their underutilization (Yang et al., 2011). However, recent digitization efforts have increased their research potential (Popik, 2004). As news media repositories are curated and made available worldwide, scholars have begun to synthesize news frames in water infrastructure coverage. Previous studies focused on aging urban water infrastructure in the U.S. found that water infrastructure news is episodic and often linked to crises (Vedachalam et al., 2016). Other research has explored the intersection of water consumption trends and extreme climate events in California, revealing finer spatial and temporal relationships beyond issue-attention cycles (Quesnel & Ajami, 2017), and the framing of water reuse as a solution to water scarcity in London (Goodwin et al., 2018). In the South Asian context, Jiang et al.’s (2016) study analyzing the domestic and international politics of hydropower development on the Brahmaputra River has identified news framed centered on biophysical pressures, political campaigning, and water management reform.
Despite these valuable contributions, existing scholarship on water infrastructure framing in the Global South faces methodological challenges. Many studies rely on labor-intensive qualitative analysis, which can introduce bias. MacRobert (2020), for instance, analyzed news coverage over ten years (2008 to 2017) in two international newspapers to examine the social and environmental impacts of infrastructural development focused on South Africa. While the research demonstrated the value of investigating newspaper coverage to understand public concerns about the impacts of new infrastructure, the project relied on labor-intensive manual retrieval and the topical coding of relevant articles by student volunteers. Therefore, the analysis was limited both in its scope and scale (i.e., to two newspapers and ten years of coverage) and did not yield an automated method to transfer the technique to other topics, timeframes, or places (see also Silva et al., 2021). Consequently, there is a lack of studies using computational text analysis methods, such as topic modeling, to examine evolving media narratives around water and infrastructure in Asian countries. One of the few studies that used computational text analysis to examine news frames around the hydropower development of the Brahmaputra River found 30 topics and was able to summarize these topics and evolving media narratives by dividing them into four frames: hydropower development; divergent political perspectives; river water agreements shared between India, China, and Bangladesh; and the international media’s focus on environmental and energy issues (Jiang et al., 2017). This methodological gap and the absence of a readily available frame repository for India underscores the need for new approaches to explore the complexity and narratives around water problems highlighted in media coverage (Kallis et al., 2006; Weder et al., 2019).
This study addresses these gaps by leveraging the LexisNexis API, a research database with news archives, and topic modeling, a quantitative research design, to analyze 316 English-language news articles from seven major Indian news sources published between 2004 and 2022, focusing on the Ken-Betwa river link. This approach, when paired with framing theory, allows us to examine how complex infrastructure projects are contextualized and narrated over time while also overcoming the limitations of manual coding. Specifically, this study explores the following questions: (1) Which news topics related to river linking can be identified from news media coverage? (2) How prevalent are these topics between 2004 and 2022, and how do they vary temporally and relate to each other? (3) What media framings of water infrastructure emerge from the topic model analysis? In doing so, this study contributes to a growing body of work on topic modeling, framing theory, and public debates on large-scale water infrastructure development in India and beyond. It provides critical insights into the media’s role in shaping public perceptions of infrastructure projects and highlights power dynamics in environmental reporting that could be beneficial for policymakers, researchers, and environmental stakeholders.

1.1. Framing Theory and Evolution

News framing theory explores how the media shapes public opinion by selectively highlighting specific aspects of an issue. News frames are socially constructed devices that help make sense of complex phenomena reported by the media and serve to reduce the complexity of our everyday world (Goffman, 1974). Gamson and Modigliani (1987, p. 143) approached news framing by defining “packages” as “a central organizing idea or storyline that provides meaning to an unfolding strip of events.” In relation to packages, information is defined as facts or beliefs stated in the news. A frame exists between packages and information and ties these two together based on culture-based meanings, norms, and values (Gamson & Modigliani, 1987). To identify a news frame among textual information, other authors (Cappella & Jamieson, 1997) have suggested that news frames should be distinct and identifiable conceptually, recognizable by other scholars, and commonly observed in journalistic practice. For example, Jiang et al. (2017) examined the news coverage of hydropower development on the Brahmaputra River in China, India, and Bangladesh, and found that this project was framed primarily via a ‘political perspective’ rather than a water resource or energy issue (p. 13).
The theorization of frame evolution is often based on the “issue-attention cycle” which is rooted in the interactions between communication media, domestic problems, and the public (Downs, 1998). Frame evolution is a socially constructed process that reflects the shifting priorities and perspectives of media producers, journalists, and society. Given the cyclic nature of news, the theorization of frame evolution seeks to examine how frames in news coverage develop over time (Kee et al., 2013). However, Downs’s (1998) linear theoretical approach to frame evolution has been criticized, since frame evolution is not entirely dependent on linear stages of issue development (Chen et al., 2022). The issue-attention cycle model does not explain the cycles of media attention (Mccomas & Shanahan, 1999) and the complexities of different timelines of issue development stages, especially for ongoing issues (i.e., climate change) (Chen et al., 2022).
To address the limitations of the issue-attention cycle, researchers have emphasized the role of framing theory and its roots in the theory of social constructionism, which first appeared in Goffman’s frame analysis (i.e., Goffman, 1974), as well as the early works of Bateson (Bateson, 1972; Brossard et al., 2004; Vreese, 2005). In the case of news media, social constructionist approaches emphasize the role of cultural and social circumstances in shaping frames around an issue (Best, 2013; Chen et al., 2022). Additionally, exploring an ongoing issue, such as large-scale water infrastructure development, under a social constructionism lens can capture the dynamic interplay between news frames and shifting sociopolitical events (Hopp et al., 2020) and highlight newsroom priorities, as parallels are drawn to broader social and cultural contexts that transform everyday events into newsworthy stories (Ofori-Birikorang, 2019). In this study, we explore and discuss inductively identified news frames reinforced by social, cultural, and power structures.

1.2. Historical and Critical Contexts of Water Infrastructure Discourse in India

The framing of water infrastructure in Indian media reflects the historical evolution of environmental discourse. Media coverage of environmental narratives is determined by cultural, social, political, and economic factors. Since the 1990s, neoliberal policies in India have led to economic reforms that have strained natural ecosystems (Mishra, 2020). Foundational work by Ramachandra Guha (2006) has argued that neoliberalism has led to an “anti-green backlash,” and that Indian environmentalism differs from Western environmentalism, as it is grounded in livelihood and displacement. Environmental journalism in India is rooted in post-emergency civic activism, grassroots movements, and growing contestations over large-scale development (Mohanty, 2018). These early traditions of “Indian environmentalism” in the 1970s–80s foregrounded critical reporting on state-led infrastructure projects, highlighting their social and ecological impacts (Ravi Rajan, 2014). However, emerging patterns within recent Indian news media suggest that development stories often take precedence over environmental ones, often relying on “authority-bias” sources over marginalized ones (Mishra, 2020). Scholars such as Gadgil and Guha (2013) have documented the recent ecological and social costs of large-scale water infrastructure such as dams, while Baviskar’s (2007) book Waterscapes: The Cultural Politics of a Natural Resource also highlights the social and environmental conflicts arising around Narmada River’s dams, revealing the deepening power asymmetries between environmental initiatives and marginalized communities. This evolving historical and critical context on water infrastructure helps position our analysis of the Ken-Betwa river link within broader debates in environmental journalism and development.
Indian media often frames large-scale water infrastructure projects as national symbols of development and modernity (Bhattacharyya, 2023). Within Indian environmental reporting, English-language news sources such as The Hindu’s Survey of the Environment section, Frontline, Times of India, and Down to Earth, to name a few, have played a vital role in documenting the long-term environmental and social impacts of state-led infrastructure projects. At the same time, scholars have noted distinct patterns in the media framing of environmental issues in vernacular and English news outlets. Regional and vernacular media outlets tend to highlight place-based local perspectives, such as socially embedded accounts of water issues, including everyday grievances over access, and the short-term impacts of environmental events (Ghosh, 2019; Pal, 2025; Priya, 2024). On the other hand, Indian English-language environmental news reporting tends to highlight political frames focusing on national and international perspectives (Nirmala & Arul, 2017). While this study focuses on Indian English-language newspapers, given their wider circulation, influence in shaping national narratives, and longitudinal availability through digital repositories, future research could consider comparing the framing of water infrastructure across multilingual and regionally diverse news media.
Consequently, water resource management is inherently political, grounded in social processes that represent the interests of individuals or groups (Mollinga, 2008). Media narratives are influenced by ownership structures that can prioritize certain political or legal agendas over others and either support or challenge nationalist discourses around water infrastructure (Obertreis et al., 2016). In India, the consolidation of media ownership, due to political and corporate pressures, has led to limited critical reporting and commercialization (Chadha, 2017). These dynamics can play a significant role in shaping how water infrastructure is framed in the media. While our study focuses on the textual outputs and frames within Indian English-language newspapers, rather than the frame-building process itself, we recognize that such political-economic influences form the basis of media production.

1.3. Contentious New Water Infrastructure: The Indian National River Linking Project (INRLP) and the Ken-Betwa Water Transfer Link

In India, the concept of river linking dates to the colonial period and was formalized by the Ministry of Water Resources, now known as the Ministry of Jal Shakti, as the National Perspective Plan (NPP) in 1980. The project’s goal is to redistribute surface water from so-called water surplus areas beset by flooding into deficit areas prone to drought. The National Water Development Agency (NWDA) was established in 1982 to oversee interbasin water transfers to achieve this plan, which then became known as the Interlinking of Rivers Project (ILRP). Subsequently, in 2002, the Supreme Court of India ordered the Ministry of Water Resources to undertake a series of studies that considered the interbasin transfer of India’s rivers via interlinking (Iyer, 2012). These proposals were divided into two regional components: a Peninsular component with 16 links and a Himalayan component with 14 links (Bandyopadhyay & Perveen, 2008; Iyer, 2012). Figure 1 shows a map and a list of these water transfer links. Many feasibility studies have been completed, and the project is proceeding in phases.
The first project to gain approval and commence was the Ken-Betwa river link, which falls under one of the sixteen links for the peninsular region and connects two tributaries of the Yamuna River in the central Indian states of Madhya Pradesh and Uttar Pradesh. The NWDA presented the first plan to link these tributaries in 1995 (Rai, 2018). In 2005, central and state governments signed a detailed project report (DPR) to investigate inter-state cooperation. Later, in 2008, this project was granted National Project status, which made it eligible for central government funding. The project’s first component is expected to take eight years to complete (Bagla, 2006). This link transfers water from the Ken River basin to the Betwa River basin through a 231 km canal (Rai, 2018). The project will provide water for irrigation purposes in Bundelkhand, a drought-affected area straddling the states of Uttar Pradesh and Madhya Pradesh (Tripathy, 2021). The Ministry of Jal Shakti states that besides water for irrigation, the project is expected to supply drinking water to a population of about 6.2 million (62 lakhs) people and to generate 103 MW of hydropower and 27 MW of solar power (Rai, 2018). However, the project will submerge large areas of the ecologically sensitive Panna Tiger Reserve and displace an unknown number of human settlements. A study of sensitive flora and fauna in the Panna Tiger Reserve revealed that the interlinking of Ken-Betwa will result in a loss of unique biodiversity and adversely affect the sustainability of the sanctuary (Parveen & Ilyas, 2021). Water resource planners favor the project, while environmentalists and human rights advocates are vocal critics.
This study builds on these previous analyses of the Ken-Betwa link and a study that examined a limited number of news reports of the Brahmaputra River’s hydropower development through structural topic modeling (STM) (Jiang et al., 2017). In the next section, we identify and discuss topics that capture the evolution of dominant news frames around the Ken-Betwa river link.

2. Data and Methods

2.1. Data

We collected news coverage of the INRLP by employing the LexisNexis Web Services API (https://www.lexisnexis.com/en-us/products/lexis-api.page, accessed on 5 May 2023). The LexisNexis API is an academic research database established in 1970 that contains over 17,000 news, legal, business, and reference sources. We utilized the LexisNexis API for our study, as it houses one of the most extensive historical archives, provides advanced search functions, and delivers high-precision results. The API interface allows researchers to access the programming interface to conduct advanced queries and download data in batches, which is not possible with the LexisNexis Uni web browser interface. For this study, we chose to query all seven Indian English-language newspapers cataloged by LexisNexis: Free Press Journal (India), Hindustan Times, Indian Express, The Economic Times, The Hindu, The Times of India (TOI), and Times of India (Electronic Edition) for their entire period of availability between 2004 and 2022. We chose these sources because they were reputable, influential, and circulated throughout India. We examined more than one or two news sources to minimize reporting bias in our dataset (Sen et al., 2022). Therefore, these newspapers are likely to reflect both public opinion and its influence. After a process of trial and error, we used the following keywords to capture the breadth of data available on this topic: “India” AND “river” AND “link.” The search yielded more than 12,000 newspaper articles. We deleted duplicates and manually screened for irrelevant articles. This yielded 4437 pertinent news articles. We then searched these for the keywords “Ken” and “Betwa”. The final corpus used for the present paper contained 316 news articles. The distribution of articles by newspaper sources is shown in Table 1.

2.2. Method and Research Design

This study employs a quantitative content analysis research design using topic modeling to identify evolving media narratives. Topic modeling is an increasingly prevalent method for determining the topic and thematic characteristics of text data points in a document. Topic modeling originated in the 1980s under generative probabilistic modeling techniques (Liu et al., 2016). Since then, research has further branched into the classification and summarization of various data types, such as social science, geospatial, and image data. Topic modeling is defined in the data analytics community as a statistical tool that can extract latent variables (i.e., variables that are not directly observed) from large datasets and group them based on topical patterns (Blei et al., 2003; Liu et al., 2016). It can also be categorized as a text-mining approach that can derive patterns of recurring words (Alghamdi & Alfalqi, 2015). In this type of analysis, each “word” is treated as a fundamental unit, a “document” is made up of text strings of N fundamental units, a “corpus” represents the entire dataset comprising all available documents, and a “topic” found within a corpus is the probability distribution of a given vocabulary, i.e., distinct words (Vayansky & Kumar, 2020). Hence, topic modeling refers to a particular algorithm or method that identifies random but complex distributions of words in a corpus, words across documents, and patterns across a sequence of documents (Barde & Bainwad, 2017).
To compensate for humans’ limited ability to process large datasets, topic modeling has been applied to explore social science datasets that are otherwise too large and time-consuming to manage (Blei et al., 2003; Sukhija et al., 2016). However, Ramage et al. (2009) argued that topic models have limited value in discovering trends or unexpected patterns without informed human judgments. Hence, topic modeling of textual data creates a platform for interactive exploration, which can optimize the data model but is not a substitute for human assessment on its own. Compared with traditional content analysis methods, topic modeling minimizes the effects of researcher subjectivity on topic identification and reduces the chances of human error (Downe-Wamboldt, 1992).
For our research design, we selected Latent Dirichlet Allocation (LDA), a popular topic modeling algorithm, which assumes that a document contains several topics rather than a single dominant one (Grimmer & Stewart, 2013). LDA also works as a generative probabilistic model that allows the algorithm to generalize its approach to any data point to deliver a topic assignment (Vayansky & Kumar, 2020). This feature made LDA suitable for our dataset, which includes a mix of English and transliterated Hindi vocabulary. LDA has been used previously to examine topics of public debate in India, such as media coverage of gender and sexual violence after the 2012 Delhi gang rape (Shah, 2019) and caste-related violence (Fonseca et al., 2019). In both cases, this method helped overcome bias stemming from qualitative coding and led to a better understanding of the impact of news media on democratic debates.

2.2.1. Data Pre-Processing

We pre-processed the corpus dataset using the functions available in the RStudio 2024.12 stringr package (Wickham & Wickham, 2019) and hunspell (Ooms, 2018) R packages. Parallel processing tasks were performed using foreach and doParallel R packages (Weston, 2020). First, we converted instances of uppercase text to lowercase text. Any HTML tags remaining after the data were downloaded were removed. All dashes in the text were replaced by white spaces. All double white spaces were removed, as well as any white space at the start or end of the text. In line with standard LDA pre-processing practices, we replaced numeric values with characters and removed punctuation marks. Word contractions were also replaced to improve machine readability within topics. For example, “don’t” was changed to “do not.” Finally, after some trial and error, common words that were unique to our dataset, such as incomplete words and names of people, were removed. After tokenizing the text and creating a document-term matrix (DTM), additional steps were taken to remove stop words such as “the” or “a.”
Furthermore, stem words were reduced to their primary forms to minimize redundancy. We manually created a stemming function using the hunspell package in RStudio, which is a wrapper commonly used to find spelling errors in a text document (Ooms, 2018). Subsequently, words that appeared in fewer than 5% of documents and common words that appeared in more than 90% of documents were removed.

2.2.2. Identifying the Number of Topics (k)

We identified the optimal number of topics by determining the average probabilistic coherence for several potential topics. Topic coherence measures evaluate the quality of topics from a human-like perspective and define how interpretable a topic is based on the degree of relevance between the words within the topic itself (Blair et al., 2020). We used the tm and topicmodels packages to maintain a corpus structure for unsupervised classification (Grün & Hornik, 2011). The doParallel and foreach packages were employed to create a custom function for parallel processing (Weston, 2020). After comparing models ranging from 10 to 100 topics, we chose a model with k = 23 topics. Our chosen number of topics indicated high internal coherence and that they were exclusive to each other.

2.2.3. Testing the Validity of the Topic Model

Finally, we tested the validity of our LDA model by running it through 100 random seeds (Levy & Franklin, 2014) with k values ranging from 20 to 40. This step ensured the model’s robustness and reliability across all parameters (Maier et al., 2018). Furthermore, to ensure the validity of topics and topic labels, one of our authors, who is an expert in this topic, qualitatively examined the model’s results and corresponding topic labels to ensure that they were coherent. Qualitatively inspecting the results of a topic model is standard practice and helps determine the number of topics (Jiang et al., 2016).

3. Results

3.1. What Topics Can Be Identified?

Table 2 summarizes the five most frequent terms for each of the 23 topics generated through the LDA topic model. In Table 2, the “Labels” column contains the most frequently occurring term in each topic, and the “Perspective Labels” column contains labels based on our qualitative assessment of the 30 most common words in each topic, which is accepted practice (Keller et al., 2020). The topics in Table 2 are sorted according to their prevalence (% proportion of documents devoted to a given topic), from the highest to the lowest percentage score. The most dominant topic was Topic 12, which had a prevalence of 8.32%, while the rarest topic was Topic 13, which had a prevalence of 1.78%. The top ten topics by prevalence were perceptively labeled as Topics 12 (Irrigation and river sharing agreement), 17 (Environmental impact assessment and clearances), 14 (Water sharing demands and needs), 16 (Feasibility for peninsular components), 4 (Supreme court government clearance process), 19 (Bundelkhand politics), 5 (Interbasin transfer and water quantity), 10 (Narmada River interlinking), 6 (Wildlife conservation and habitat), and 9 (Water nationalism) with prevalence scores ranging from 8.32% to 4.23% and in total accounting for a prevalence of 57.58% in our corpus. Overall, the topics in the corpus include discussions of other INRLP projects, environmental conservation, natural hazards, government/politics, and funding schemes.

3.2. How Prevalent Are These Topics Between 2004 and 2022?

We examined how the coverage of the 23 topics evolved to understand temporal shifts in media discourse. Figure 2 illustrates distinct topics and their prevalence in news coverage between 2004 and 2022. The availability of digitized news sources varies, as only the Hindustan Times, Times of India (Electronic Edition), The Times of India (TOI), and the Free Press Journal (India) were available before 2014. Topics 13 (Parliament legislation and amendments), 14 (Water sharing demands and needs), 15 (Land submergence), 16 (Feasibility for peninsular components), 17 (Environmental impact assessment and clearances), 18 (Environment and land), 19 (Bundelkhand politics), 20 (Climate change policy), and 21 (Central government schemes) dominated the articles before 2012. The empty column for 2012 suggests that our data extraction and pre-processing did not capture any articles related to the Ken-Betwa link in that year. After 2012, there was a mix of topics that dominated the news, such as Topics 11 (Agricultural funding and budget) and 12 (Irrigation and river sharing agreement), which were also discussed more recently in 2021 and 2022. The re-emergence of these topics in 2021–2022 reflects the formal approval of the Ken-Betwa project, as well as renewed attention on funding it and on its impact on river-sharing agreements.

3.3. How Are These Topics Related to Each Other?

We plotted a cluster dendrogram (Figure 3) to illustrate the similarities among the topics. We utilized Ward’s method to minimize the variance between clusters (Murtagh & Legendre, 2014; Ward, 1963). In Figure 3, each branch collection is considered a group. For example, Topics 12, 19, 11, 5, and 8 form a group, and Topics 5 and 8 form a subset within that group. The groups logically make sense since Topic 8 (Peninsular interlinking) and Topic 5 (Interbasin transfer and water quantity) both have similar themes. Topics 12 and 19 fall into the same cluster as Topics 5, 8, and 11 but are not as closely related to each other due to overlaps with other themes such as politics and funding schemes.
We used a statistical method called t-distributed stochastic neighbor embedding (t-SNE) to further visualize high-dimensional data. This technique reduces the tendency to crowd points together in the center of the plot and gives each data point a location that can reveal structures at different scales (Van Der Maaten & Hinton, 2008). The t-SNE algorithm is nonlinear and adapts to the underlying data. We used the “Labels” column as topic headers to visualize clusters in the t-SNE model. We paid particular attention to the “perplexity” parameter within the t-SNE algorithm to balance attention between local and global aspects of our data. Van Der Maaten and Hinton (2008) suggested that perplexity values between 5 and 50 tend to provide robust performance. Since we are examining 23 topics, we chose the lowest value of perplexity suggested, which is five, since the perplexity value should be smaller than the number of points. The Epsilon or learning rate was set to 5000 iterations until a stable configuration was reached. In Figure 4, clusters between the 23 topics in our data can be visualized in a two-dimensional space. Topics with similar content were clustered together in the t-SNE plot, and the closer the topic labels to one another, the more similar the topics (and their subset words). For example, documents that focused predominantly on the topics “inter” and “component” were graphed relatively close to each other. Similarly, topics related to government policies such as “committee”, “board”, “chief”, “clearance”, and “share” clustered together.

4. Discussion: The Politics of Framing Water

In this section, we discuss how news framings on water infrastructure are distilled from the topics identified through LDA topic modeling, how the Ken-Betwa river link is framed in English-language news media, and how these frames vary over time. Three dominant frames were identified via the results of the topic modeling analysis: (1) government policy frame, (2) INRLP comparative frame, and (3) environmental conservation frame. These frames not only organize how the project is presented but also reflect deeper institutional patterns and sociopolitical functions that structure public discourse around water infrastructure.

4.1. Government Policy Frame

The government policy frame reflects the government’s efforts to justify and promote the INRLP as a solution to water scarcity, inter-state cooperation, and regional development. Topic 16 (Feasibility for peninsular components), which was a dominant discussion topic from 2006 to 2010, and Topic 21 (Central government schemes), which was most prominent from 2004 to 2011, reveal early narratives on feasibility studies and government schemes, illustrating how technical assessments shape public discourse. These topics also reveal the government’s role in managing interstate tensions between Uttar Pradesh and Madhya Pradesh over the Ken-Betwa link by positioning interlinking as a national imperative. These narratives reflect Gamson and Modigliani’s (1987) ideas that news frames act as interpretive ‘packages’, specifically as institutional logics frame large-scale development.
Topics 11 (Agricultural funding and budget) and 12 (Irrigation and river sharing agreement) emerged as dominant topics after the government approval of the Ken-Betwa link in 2020. These topics demonstrate a shift in narratives toward irrigation, agricultural productivity, and project financing. The prevalence of the term “lakh” as a unit equal to 100,000 INR and “budget” reveals how economic rationality demonstrates a higher operational priority for the approval and directions of the INRLP. Our findings align with Jiang et al.’s (2017) study, which also found that national water infrastructure projects are framed through the lens of state interests. Overall, news discourse emphasizes project milestones reaffirming institutional rhythms.

4.2. INRLP Comparative Frame

This frame illustrates that news media has substantially reported on the various developments of the Ken-Betwa link in the past decade (2013–2020) and situated its development within the broader ensemble of pending national river linking strategies. For example, in Topic 10, two out of the top five terms were “Narmada” and “Tapi” Rivers; in Topic 3, two out of the top five terms were “Ganga” and “Yamuna” Rivers; in Topic 2, “Ganga” River is mentioned again; and in Topic 8, “Godavari” River is mentioned, demonstrating a comparative framing among techno-managerial discourse. By contextualizing the Ken-Betwa link through a comparison with pending projects with similar infrastructure and functional characteristics, such as the Par-Tapi-Narmada link, which is also envisioned to transport surplus water from the Western Ghats to drier regions of Gujarat (Mehmood et al., 2014), and the Godavari-Krishna Pennar-Cauvery link, which is aimed at resolving water scarcity and flooding, media discourse fosters a narrative of national water infrastructure ambition. This bundling of different components of the INRLP across news perhaps suggest that media sources’ prioritize building a coherent national story on infrastructural vision.
As Kallis et al. (2006) argue, a techno-managerial narrative can gloss over the nuanced socio-ecological complexity that accompanies large-scale hydrological initiatives. Such comparisons can overshadow localized risks and community-level concerns. For example, one of the motivations behind the Ken-Betwa link is to irrigate the drought-prone Bundelkhand region, where water access is an issue. In Topic 19, “Bundelkhand” is one of the top five terms and is one of the most prevalent topics in 2010, 2013, and 2020. Between 2010 and 2014, detailed project reports for Phase 1 and Phase 2 of the project were completed. In 2021, the Union Cabinet approved funding for the Ken-Betwa link after a lengthy interstate dispute regarding water sharing and funding. In the years 2021 and 2022, Topics 11 (Agricultural funding and budget) and 12 (Irrigation and river sharing agreement) dominate the news media narrative, alluding to the proposed reservoirs, canals, and dam projects that will support agricultural productivity. The emphasis on the technical components of this project in the media suggests an engineering-centered vision of hydrological development. Furthermore, the socio-economic impacts of the link are estimated to displace farmers and tribal people (Pathak, 2016), a topic not explicitly discussed within structuralist views on Ken-Betwa. Purely structural narratives on development are also critiqued by Crow-Miller et al. (2017), who suggest that infrastructure discourses in Asia lack community-centered concerns and are shaped by imaginaries of progress. This highlights the importance of exploring distributional impacts in future frame-building analyses for water infrastructure projects. Overall, the use of LDA topic modeling and temporal trends captures latent patterns in the news as frame development occurs alongside water infrastructure milestones. This connects back to Goffman’s (1974) suggestion that framing theory is grounded in social constructionism and also affirms Hopp et al.’s (2020) argument that frame development reflects socio-political interactions.

4.3. Environmental Conservation Frame

The third frame captures opposition to the Ken-Betwa project due to environmental conservation concerns, especially the potential partial flooding of the Panna Tiger Reserve (Topics 17–21). The Ken-Betwa link was proposed by the NWDA in 1995, and since then, the project has fallen short of acquiring environmental clearances. Topics 6, 9, 17, 18, and 20 featured terms such as “committee,” “clearance,” and “expert,” reflecting the procedural hurdles and environmental assessments tied to the project. Topic 6 (Wildlife conservation and habitat) and Topic 9 (Water nationalism) have remained prevalent in the last five years. Topic 20 (Climate change policy) has been part of the narrative since 2004, and as illustrated in Figure 2, it was a much more prevalent topic in the years before 2012. This timeline reflects the approval of numerous environmental clearances in 2017, and concerns about the flooding and ecological destruction of the Panna Tiger Reserve remain (Parveen & Ilyas, 2021). These topics highlight ongoing tensions between developmental ambitions and conservation mandates. Interestingly, Topic 23, Diamond mining, refers to Rio Tinto’s proposed project, which further complicates environmental governance in the area. Our findings show that environmental concerns consistently recurred across our dataset’s timeline, intersecting with socio-cultural complexities beyond news issue development cycles, as also discussed by Chen et al. (2022). The environmental conservation frame also affirms Goffman’s (1974) social constructionist ideas of frame development, specifically shaped by cultural and institutional practices in the case of the Ken-Betwa river link.
While empirical studies of the Ken-Betwa project are limited, Pathak (2016) argues that the Ken-Betwa project would submerge portions of the Panna Tiger Reserve and farmland, lead to a loss of timber trees, increase air pollution, and influence local employment. Additionally, public health risks such as high fluoride and silica concentrations in the Ken and Betwa rivers (Avtar et al., 2011) remain under-discussed in media framing but have significant implications for sustainable water management. This highlights the limits and gaps within news discourse in surfacing place-specific environmental knowledge. Overall, media coverage over time reveals a shift in discussions from environmental concerns to project completion and economic utility, which reflects the changing political economy of the project. The identification of frames through topic modeling allows us to study long-term media coverage, beyond episodic environmental frames, and discuss non-linear frame evolution (Mccomas & Shanahan, 1999) such as the emergence of and decline in patterns within environmental reporting.

5. Conclusions

In this paper, we advanced methods and theory in LDA topic modeling and scholarship on media framing of water infrastructure projects. We advanced LDA topic modeling by developing a technique to download, extract, and analyze a corpus of news media coverage using the LexisNexis API. We demonstrate that LDA topic modeling is a rigorous quantitative approach for identifying topics in news discourse over extended timeframes while overcoming biases common during manual coding. In doing so, this study contributes to methodological innovation within computational media analysis and offers new insights into how water infrastructure, such as the Ken-Betwa river link, is framed in evolving news coverage.
By applying LDA to news articles focused on the Ken-Betwa link, we identified 23 unique topics and three dominant frames: government policy, INRLP comparison, and environmental conservation. The Ken-Betwa link, approved in 2021, was associated with discussion on project funding and support for irrigation, farming, and drinking water, highlighting how news discourse reflects infrastructure milestones. Future work will examine whether these promises are realized.
Importantly, this study highlights how media framings of water infrastructure projects reflect and reproduce power dynamics. Our analysis revealed that earlier years of media discourse focused on feasibility and government-led negotiations, illustrating the prominence of institutional logic in shaping public discourse. News articles published later in the dataset focused more on environmental risks associated with the Ken-Betwa link. These patterns were further clarified using visualizations like dendrograms and t-SNE clustering. By making visible the sociopolitical functions behind these frames, we show how topic modeling can be leveraged to identify and critically discuss media frames.
This work also has important implications for infrastructure studies by providing a replicable computational method for analysis of infrastructure project framings in the news and an empirical case rooted in South Asia. Moreover, as newspaper media are increasingly being ‘born digital,’ not just in advanced economies like the US but also throughout the Global South, approaches like ours will be essential to mapping discourse longitudinally. Overall, this study provides a critical lens for interpreting news media representations of infrastructure development, environmental risk, and contested water futures, offering a baseline to assess public discourse and understand the politics of water infrastructure.
This study was constrained by the size of the corpus, the choice of newspapers, the choice of topic model and its performance metrics, and the requirement for a certain degree of human intervention and perspective. The primary limitation of this study was its focus on English-language newspaper coverage, which may not fully represent the broader media landscape such as regional language newspapers, radio, television broadcasts, or social media. Additionally, textual analysis offers insights into the outputs of media production and provides limited insight into the complex political, legal, and media ownership dynamics. Future research could build on this study and explore diverse media platforms and ownership to analyze how water infrastructure is framed and to offer a more comprehensive understanding of public discourse. Future comparative studies examining regional and English-language media sources could also reveal socio-cultural and political divergences or convergences on how large-scale water infrastructure projects are perceived. Additionally, future work will expand this analysis to the full INRLP corpus. In doing so, we hope to further innovate automated techniques for topic identification and labeling.

Author Contributions

Conceptualization, H.S. and T.B.; methodology, H.S.; formal analysis, H.S.; data curation, M.H.; writing—original draft preparation, H.S.; writing—review and editing, T.B.; visualization, H.S.; supervision, T.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Restrictions apply to the availability of these data. Data were obtained from LexisNexis API through institutional access and are not publicly available.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Alghamdi, R., & Alfalqi, K. (2015). A survey of topic modeling in text mining. International Journal of Advanced Computer Science and Applications (IJACSA), 6(1). [Google Scholar] [CrossRef]
  2. Avtar, R., Kumar, P., Singh, C. K., & Mukherjee, S. (2011). A comparative study on hydrogeochemistry of Ken and Betwa rivers of Bundelkhand using statistical approach. Water Quality, Exposure and Health, 2(3–4), 169–179. [Google Scholar] [CrossRef]
  3. Bagla, P. (2006). Controversial rivers project aims to turn India’s fierce monsoon into a friend. Science, 313(5790), 1036–1037. [Google Scholar] [CrossRef] [PubMed]
  4. Bandyopadhyay, J., & Perveen, S. (2008). The interlinking of Indian rivers: Questions on the scientific, economic and environmental dimensions of the proposal. In Interlinking of rivers in India (pp. 75–98). CRC Press. [Google Scholar] [CrossRef]
  5. Barde, B. V., & Bainwad, A. M. (2017, June 15–16). An overview of topic modeling methods and tools. 2017 International Conference on Intelligent Computing and Control Systems (ICICCS) (pp. 745–750), Madurai, India. [Google Scholar] [CrossRef]
  6. Bateson, G. (1972). Steps to an ecology of the mind. University of Chicago Press. [Google Scholar]
  7. Baviskar, A. (2007). Waterscapes: The cultural politics of a natural resource. Permanent Black. [Google Scholar]
  8. Best, J. (2013). Constructionist social problems theory. Annals of the International Communication Association, 36(1), 237–269. [Google Scholar] [CrossRef]
  9. Bhattacharyya, D. (2023). India’s new water works. Current History, 122(843), 158–160. [Google Scholar] [CrossRef]
  10. Blair, S. J., Bi, Y., & Mulvenna, M. D. (2020). Aggregated topic models for increasing social media topic coherence. Applied Intelligence, 50(1), 138–156. [Google Scholar] [CrossRef]
  11. Blei, D., Ng, A., & Jordan, M. (2003). Latent dirichlet allocation. Journal of Machine Learning Research, 3, 993–1022. [Google Scholar]
  12. Brossard, D., Shanahan, J., & McComas, K. (2004). Are issue-cycles culturally constructed? A comparison of French and American coverage of global climate change. Mass Communication and Society, 7(3), 359–377. [Google Scholar] [CrossRef]
  13. Cappella, J. N., & Jamieson, K. H. (1997). Spiral of cynicism: The press and the public good. Oxford University Press. [Google Scholar]
  14. Chadha, K. (2017). The Indian news media industry: Structural trends and journalistic implications. Global Media and Communication, 13(2), 139–156. [Google Scholar] [CrossRef]
  15. Chen, Y., Thorson, K., & Lavaccare, J. (2022). Convergence and divergence: The evolution of climate change frames within and across public events. International Journal of Communication, 16, 107–129. [Google Scholar]
  16. Crow-Miller, B., Webber, M., & Molle, F. (2017). The (re)turn to infrastructure for water management? Available online: https://www.water-alternatives.org/index.php/alldoc/articles/vol10/v10issue2/351-a10-2-1/file (accessed on 13 May 2025).
  17. Downe-Wamboldt, B. (1992). Content analysis: Method, applications, and issues. Health Care for Women International, 13(3), 313–321. [Google Scholar] [CrossRef] [PubMed]
  18. Downs, A. (1998). Up and down with ecology—The “Issue-Attention Cycle”. In Political theory and public choice (pp. 100–112). Edward Elgar Publishing. [Google Scholar] [CrossRef]
  19. Fonseca, A. F., Bandyopadhyay, S., Louçã, J., & Manjaly, J. A. (2019). Caste in the news: A computational analysis of Indian newspapers. Social Media + Society, 5(4), 205630511989605. [Google Scholar] [CrossRef]
  20. Gadgil, M., & Guha, R. (2013). Ecology and equity: The use and abuse of nature in contemporary India. Routledge. [Google Scholar]
  21. Gamson, W. A., & Modigliani, A. (1987). The changing culture of affirmative action. In Research in political sociology (Vol. 3, pp. 137–177). JAI Press. [Google Scholar]
  22. Ghosh, B. (2019). Making water media in 21st-century South Asia. In Water histories of South Asia (pp. 226–242). Routledge India. [Google Scholar] [CrossRef]
  23. Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Harvard University Press. [Google Scholar]
  24. Goodwin, D., Raffin, M., Jeffrey, P., & Smith, H. M. (2018). Evaluating media framing and public reactions in the context of a water reuse proposal. International Journal of Water Resources Development, 34(6), 848–868. [Google Scholar] [CrossRef]
  25. Grimmer, J., & Stewart, B. M. (2013). Text as data: The promise and pitfalls of automatic content analysis methods for political texts. Political Analysis, 21(3), 267–297. [Google Scholar] [CrossRef]
  26. Grün, B., & Hornik, K. (2011). topicmodels: An R package for fitting topic models. Journal of Statistical Software, 40(13), 1–30. [Google Scholar] [CrossRef]
  27. Guha, R. (2006). How much should a person consume? Permanent Black. [Google Scholar]
  28. Hopp, F. R., Fisher, J. T., & Weber, R. (2020). Dynamic transactions between news frames and sociopolitical events: An integrative, hidden Markov model approach. Journal of Communication, 70(3), 335–355. [Google Scholar] [CrossRef]
  29. Iyer, R. (2012). River linking project: A disquieting judgment. Economic and Political Weekly, 14(14), 33–40. [Google Scholar]
  30. Jiang, H., Qiang, M., & Lin, P. (2016). Assessment of online public opinions on large infrastructure projects: A case study of the Three Gorges Project in China. Environmental Impact Assessment Review, 61, 38–51. [Google Scholar] [CrossRef]
  31. Jiang, H., Qiang, M., Lin, P., Wen, Q., Xia, B., & An, N. (2017). Framing the Brahmaputra River hydropower development: Different concerns in riparian and international media reporting. Water Policy, 19(3), 496–512. [Google Scholar] [CrossRef]
  32. Kallis, G., Kiparsky, M., Milman, A., & Ray, I. (2006). Glossing over the complexity of water. Science, 314(5804), 1387–1388. [Google Scholar] [CrossRef] [PubMed]
  33. Kee, C. P., Ismail, N. S., Hashim, N., & Nie, K. S. (2013). Framing Bersih 3.0: Online versus traditional mass communication. Procedia Technology, 11, 161–168. [Google Scholar] [CrossRef]
  34. Keller, T. R., Hase, V., Thaker, J., Mahl, D., & Schäfer, M. S. (2020). News media coverage of climate change in India 1997–2016: Using automated content analysis to assess themes and topics. Environmental Communication, 14(2), 219–235. [Google Scholar] [CrossRef]
  35. Levy, K. E. C., & Franklin, M. (2014). Using topic models to examine political contention in the U.S. trucking industry. Social Science Computer Review, 32(2), 182–194. [Google Scholar] [CrossRef]
  36. Liu, L., Tang, L., Dong, W., Yao, S., & Zhou, W. (2016). An overview of topic modeling and its current applications in bioinformatics. SpringerPlus, 5(1), 1608. [Google Scholar] [CrossRef] [PubMed]
  37. MacRobert, C. J. (2020). A review of impacts associated with infrastructure news coverage in South Africa. South African Geographical Journal, 102(2), 236–248. [Google Scholar] [CrossRef]
  38. Maier, D., Waldherr, A., Miltner, P., Wiedemann, G., Niekler, A., Keinert, A., Pfetsch, B., Heyer, G., Reber, U., Häussler, T., Schmid-Petri, H., & Adam, S. (2018). Applying LDA topic modeling in communication research: Toward a valid and reliable methodology. Communication Methods and Measures, 12(2–3), 93–118. [Google Scholar] [CrossRef]
  39. Mccomas, K., & Shanahan, J. (1999). Telling stories about global climate change. Communication Research, 26(1), 30–57. [Google Scholar] [CrossRef]
  40. Mehmood, K., Patel, A., Singh, V., Prajapati, S., Kalubarme, M. H., Prakash, I., & Gupta, K. P. (2014). Submergence analysis using geo-informatics technology for proposed Dam reservoirs of Par-Tapi-Narmada River link project, Gujarat State, India. International Journal of Geosciences, 5(6), 622–633. [Google Scholar] [CrossRef]
  41. Mishra, M. (2020). Environmental journalism in India: Past, present, and future. In Routledge hand-book of environmental journalism (pp. 291–305). Routledge. [Google Scholar]
  42. Mohanty, R. (2018). Democratizing development: Struggles for rights and social justice in India. SAGE Publications, Inc. [Google Scholar]
  43. Mollinga, P. P. (2008). Water, politics and development: Framing a political sociology of water resources management. Water Alternatives, 1(1), 7–23. [Google Scholar]
  44. Murtagh, F., & Legendre, P. (2014). Ward’s hierarchical agglomerative clustering method: Which algorithms implement ward’s criterion? Journal of Classification, 31(3), 274–295. [Google Scholar] [CrossRef]
  45. Nirmala, T., & Arul, A. (2017). Framing of environment in English and Tamil newspapers in India. Journal of Media and Communication Studies, 9(1), 1–9. [Google Scholar] [CrossRef]
  46. Obertreis, J., Moss, T., Mollinga, P., & Bichsel, C. (2016). Water, infrastructure and political rule: Introduction to the special issue. Water Alternatives, 9(2), 168–181. [Google Scholar]
  47. Ofori-Birikorang, A. (2019). News construction on the pivots of framing and ideology: A theoretical perspective. Research on Humanities and Social Sciences, 9(2), 41–51. [Google Scholar] [CrossRef]
  48. Ooms, J. (2018). hunspell: High-performance stemmer, tokenizer, and spell checker. (version 3). R Archive Network. [Google Scholar]
  49. Pal, A. (2025). Representation of freshwater access, equity; Community empowerment in leading mainstream vernacular newspapers: A sustainable development perspective in Indian Sundarban Region’s Moushuni and Gobordhanpaur Islands. International Journal for Multidisciplinary Research, 7(3), 1–65. [Google Scholar] [CrossRef]
  50. Parveen, T., & Ilyas, O. (2021). The inter-linking of rivers and biodiversity conservation: A study of Panna Tiger Reserve, Madhya Pradesh, India. Current Science, 121(12), 1572. [Google Scholar] [CrossRef]
  51. Pathak, H. (2016). Socio-economical, environmental evaluation of Ken-Betwa River link project, India. Present Environment and Sustainable Development, 10(2), 35–42. [Google Scholar] [CrossRef]
  52. Popik, B. (2004). Digital historical newspapers. Journal of English Linguistics, 32(2), 114–123. [Google Scholar] [CrossRef]
  53. Priya, R. (2024). Reframing the narrative: An analysis of print media reporting on Bihar floods. Frontiers in Water, 6, 1039240. [Google Scholar] [CrossRef]
  54. Quesnel, K. J., & Ajami, N. K. (2017). Changes in water consumption linked to heavy news media coverage of extreme climatic events. Science Advances, 3(10), e1700784. [Google Scholar] [CrossRef] [PubMed]
  55. Rai, A. (2018). National river linking project significance & challenges. International Education and Research Journal, 4, 25–27. [Google Scholar]
  56. Ramage, D., Rosen, E., Chuang, J., Manning, C., & McFarland, D. (2009). Topic modeling for the social sciences. NIPS-Workshop on Applications for Topic Models: Text and Beyond, 5(27), 1–4. [Google Scholar]
  57. Ravi Rajan, S. (2014). A history of environmental justice in India. Environmental Justice, 7(5), 117–121. [Google Scholar] [CrossRef]
  58. Sen, A., Ghatak, D., Khanuja, G., Rekha, K., Gupta, M., Dhakate, S., Sharma, K., & Seth, A. (2022, June 29–July 1). Analysis of media bias in policy discourse in India. ACM SIGCAS/SIGCHI Conference on Computing and Sustainable Societies (COMPASS) (pp. 57–77), Seattle, WA, USA. [Google Scholar] [CrossRef]
  59. Shah, A. H. (2019). How episodic frames gave way to thematic frames over time: A topic modeling study of the Indian media’s reporting of rape post the 2012 Delhi gang-rape. Poetics, 72, 54–69. [Google Scholar] [CrossRef]
  60. Silva, G. D. P. d., Sherren, K., & Parkins, J. R. (2021). Using news coverage and community-based impact assessments to understand and track social effects using the perspectives of affected people and decisionmakers. Journal of Environmental Management, 298, 113467. [Google Scholar] [CrossRef] [PubMed]
  61. Sukhija, N., Tatineni, M., Brown, N., Van Moer, M., Rodriguez, P., & Callicott, S. (2016, July 18–21). Topic modeling and visualization for big data in social sciences. 2016 Intl IEEE Conferences on Ubiquitous Intelligence & Computing, Advanced and Trusted Computing, Scalable Computing and Communications, Cloud and Big Data Computing, Internet of People, and Smart World Congress (UIC/ATC/ScalCom/CBDCom/IoP/SmartWorld) (pp. 1198–1205), Toulouse, France. [Google Scholar] [CrossRef]
  62. Tripathy, L. (2021). Environmental geotechnical issues in river interlinking proposal (pp. 149–157). Springer. [Google Scholar] [CrossRef]
  63. UNESCO. (2023). Valuing hydraulic infrastructure. UNESCO. [Google Scholar]
  64. Van Der Maaten, L., & Hinton, G. (2008). Visualizing data using t-SNE. Journal of Machine Learning Research, 9, 2579–2605. [Google Scholar]
  65. Vayansky, I., & Kumar, S. A. P. (2020). A review of topic modeling methods. Information Systems, 94, 101582. [Google Scholar] [CrossRef]
  66. Vedachalam, S., Lewenstein, B. V., DeStefano, K. A., Polan, S. D., & Riha, S. J. (2016). Media discourse on ageing water infrastructure. Urban Water Journal, 13(8), 861–874. [Google Scholar] [CrossRef]
  67. Vreese, C. H. (2005). News framing. Information Design Journal, 13(1), 51–62. [Google Scholar] [CrossRef]
  68. Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association, 58(301), 236–244. [Google Scholar] [CrossRef]
  69. Weder, F., Voci, D., & Vogl, N. C. (2019). (Lack of) Problematization of water supply use and abuse of environmental discourses and natural resource related claims in German, Austrian, Slovenian and Italian media. Journal of Sustainable Development, 12(1), 39. [Google Scholar] [CrossRef]
  70. Weston, S. (2020). foreach: Provides foreach looping construct. (version. 1.5.1). R Archive Network. [Google Scholar]
  71. Wickham, H., & Wickham, M. (2019). Package ‘stringr’. Available online: https://github.com/tidyverse/stringr (accessed on 13 May 2025).
  72. Yang, T.-I., Torget, A., & Mihalcea, R. (2011, June 4). Topic modeling on historical newspapers. 5th ACL-HLT Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities (pp. 96–104), Portland, OR, USA. [Google Scholar]
Journalmedia 06 00114 g001
Figure 1. Map and list of water transfer links in India. Source: National Water Development Agency (https://nwda.gov.in).
Figure 1. Map and list of water transfer links in India. Source: National Water Development Agency (https://nwda.gov.in).
Journalmedia 06 00114 g001
Journalmedia 06 00114 g002
Figure 2. Temporal changes in the coverage of 23 topics between 2004 and 2022.
Figure 2. Temporal changes in the coverage of 23 topics between 2004 and 2022.
Journalmedia 06 00114 g002
Journalmedia 06 00114 g003
Figure 3. Cluster dendrogram of topics created using Ward’s method.
Figure 3. Cluster dendrogram of topics created using Ward’s method.
Journalmedia 06 00114 g003
Journalmedia 06 00114 g004
Figure 4. t-SNE visualization of topics.
Figure 4. t-SNE visualization of topics.
Journalmedia 06 00114 g004
Table 1. List of newspaper sources and the number of articles used in this study.
Table 1. List of newspaper sources and the number of articles used in this study.
Newspaper SourcesNo. of Articles
Hindustan Times100
The Times of India paper edition (TOI)77
The Times of India (Electronic Edition)42
Indian Express32
The Economic Times27
Free Press Journal (India)22
The Hindu16
Total316
Table 2. Summary of topics, perspective labels, and top 5 terms in each topic.
Table 2. Summary of topics, perspective labels, and top 5 terms in each topic.
TopicsLabelsPerspective LabelsPrevalenceTop 5 Terms
Topic 12Lakh/Project Cost (unit in the Indian numbering system equal to 100,000, in this case, referring to INR)Irrigation and river sharing agreement8.32%hectare, budget, agriculture, lakh, power
Topic 17CommitteeEnvironmental impact assessment and clearances7.36%committee, dam, clearance, expert, impact
Topic 14ShareWater sharing demands and needs5.79%share, house,
speak, bjp, amend
Topic 16ComponentFeasibility for peninsular components5.73%component, nwda, feasibility, peninsular, tail
Topic 4ClearanceSupreme court government clearance process5.67%clearance, district,
court, bharti, centre
Topic 19BundelkhandBundelkhand politics5.44%bundelkhand, region, people, add, congress
Topic 5BasinInterbasin transfer and water quantity5.24%basin, transfer, flow, country, drought
Topic 10IRL (Indian River Linking)Narmada River interlinking5.11%ilr, gujarat, maharashtra, narmada, tapi
Topic 6HabitatWildlife conservation and habitat4.69%habitat, conservation, specie, population, wild
Topic 9Jal (water in Hindi)Water nationalism4.23%country, jal, day, campaign, world
Topic 20EnvironmentClimate change policy4.13%environment, change, people, manage, time
Topic 11FarmAgricultural funding and budget4.04%farm, budget, agriculture, lakh, power
Topic 3InterINRLP implementation3.98%inter, component, yamuna, nepal, ganga
Topic 7ChiefProject authority and implementation3.86%chief, department, office, official, secretary
Topic 1BoardCentral government agencies3.80%board, director, field, nwda, chief
Topic 15Hectare (a metric unit of square measure)Land submergence3.53%hectare, km, land, basin, involve
Topic 22GangaGanga rejuvenation3.53%ganga, bharti, clean, pollute, rejuvenate
Topic 2CentreCentral government with national concerns3.34%centre, include, ganga, bihar, cost
Topic 18LandEnvironment and land3.30%land, hectare, commend, diverse, committee
Topic 21SchemeCentral government schemes2.82%scheme, flood, manage, central, complete
Topic 8GodavariPeninsular interlinking2.47%interlink, godavari, million, scheme, andhra
Topic 23MineEnvironmental challenges: mining and sugar1.85%mine, sugar, diamond, bhopal, white
Topic 13BillParliament legislation and amendments1.78%bill, house, speak, bjp, amend
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Singh, H.; Hansen, M.; Birkenholtz, T. The Politics of Framing Water Infrastructure: A Topic Model Analysis of Media Coverage of India’s Ken-Betwa River Link. Journal. Media 2025, 6, 114. https://doi.org/10.3390/journalmedia6030114

AMA Style

Singh H, Hansen M, Birkenholtz T. The Politics of Framing Water Infrastructure: A Topic Model Analysis of Media Coverage of India’s Ken-Betwa River Link. Journalism and Media. 2025; 6(3):114. https://doi.org/10.3390/journalmedia6030114

Chicago/Turabian Style

Singh, Harman, Matthew Hansen, and Trevor Birkenholtz. 2025. "The Politics of Framing Water Infrastructure: A Topic Model Analysis of Media Coverage of India’s Ken-Betwa River Link" Journalism and Media 6, no. 3: 114. https://doi.org/10.3390/journalmedia6030114

APA Style

Singh, H., Hansen, M., & Birkenholtz, T. (2025). The Politics of Framing Water Infrastructure: A Topic Model Analysis of Media Coverage of India’s Ken-Betwa River Link. Journalism and Media, 6(3), 114. https://doi.org/10.3390/journalmedia6030114

Article Metrics

Back to TopTop