An Arabic Financial Sentiment Dictionary: Bilingual Corporate Disclosures and Market Reaction on the Saudi Stock Exchange

Alromeh, Saleh Hamad (2025) An Arabic Financial Sentiment Dictionary: Bilingual Corporate Disclosures and Market Reaction on the Saudi Stock Exchange. Doctoral thesis, University of East Anglia.

Preview

PDF Download (10MB) | Preview

Abstract

Arabic sentiment analysis is an emerging field compared to sentiment analysis in the English language. In the finance and accounting literature, the dictionary-based approach has been a widely recognized method for measuring textual sentiment. Additionally, finance-oriented sentiment dictionaries have been shown to be more accurate in capturing the textual sentiment compared to general-purpose dictionaries. As a result, Loughran and McDonald’s (2011) finance-oriented dictionary has become a predominant tool for capturing textual sentiment across various English financial settings and has been mapped to other languages. The Arabic language, however, is regarded as low-resourced and morphologically complex, calling for specialized sentiment analysis tools suited to business communication. To date, no finance-oriented Arabic dictionary addresses textual sentiment in finance and business communication.

The research introduces an Arabic Financial Sentiment Dictionary, a robust Arabic sentiment dictionary designed for financial applications. The Arabic financial sentiment dictionary is constructed by translating the Loughran and McDonald (2011) dictionary into Arabic. The dictionary consists of Positive, Negative, and Uncertainty categories and is designed to address textual sentiment classification in financial narratives. The research evaluated the Arabic financial sentiment dictionary in cross-language settings, using 728 Arabic annual reports, 274 CEO letters, and their English counterparts from Saudi public firms. Additionally, the study evaluates the proposed dictionary against an Arabic general dictionary in capturing the sentiment in financial disclosures. The findings, consistent with those in the finance and accounting literature across languages, indicate that the Arabic finance-oriented dictionary is more accurate in capturing the tone of financial narratives, whereas general purpose dictionaries often misclassify words in financial applications and include words that lack sentiment value.

The study further investigates the impact of machine translation on Arabic-English financial text, using a sample of annual reports and CEO letters from Saudi public firms. A growing body of literature employs machine translation to overcome the scarcity of available natural language pre-processing resources in low-resource languages, leveraging more suitable tools in other languages. However, the language pair in translation and the text’s domain pose challenges to retaining the tone of the original text. To assess whether sentiment is preserved across Arabic and English financial disclosures, the study conducts a bidirectional translation experiment. The findings indicate that sentiment is largely preserved and that machine translation maintains relative intensity of sentiment, although positive language shows a degree of discrepancies in the Arabic-to-English direction.

Lastly, using a sample of 686 firm-years of annual reports in Arabic and their corresponding English versions, the study examines whether tone (positive, negative, and uncertainty) in bilingual disclosures on the Saudi Stock Exchange is associated with returns differently following annual report releases. By applying an event study methodology, the findings reveal that tone in annual reports is associated with abnormal returns following their publication. In particular, negative tone across both languages is statistically significantly associated with lower cumulative abnormal returns, although the market’s reaction is delayed. In contrast, positive and uncertain sentiments are neither informative nor statistically associated with returns. More importantly, the results indicate that the Saudi Stock Exchange responds similarly to the negative tone conveyed in the Arabic and English annual reports.

Item Type:	Thesis (Doctoral)
Faculty \ School:	Faculty of Social Sciences > Norwich Business School
Depositing User:	Bethany Turner
Date Deposited:	17 Sep 2026 14:28
Last Modified:	17 Sep 2026 14:28
URI:	https://ueaeprints.uea.ac.uk/id/eprint/104579
DOI:

Downloads

Actions (login required)

View Item