In international travel, customers are willing to express their experiences and feelings about the trip by posting reviews on platforms in their native language. According to Customer-based Discrepancy Theory, customers with different language backgrounds will form different anchors when browsing or reading eWOM in the corresponding language due to the differences in expression forms, nouns, and other information in different languages, and even if they have consistent offline consumption experience, customers with different language backgrounds may have different satisfaction levels due to the differences in established anchors. Prior studies have been leveraging reviews to understand the information contained in eWOM in different languages, however, the analyses of multilingual reviews still face challenges. The current joint linguistic and statistical analysis methods suffer from information overload in terms of massive online data. In this paper, we address the above challenges by utilizing cross-linguistic deep learning and multiple linear regression model of attribute-level effects on customer satisfaction. The results of the data experiment in English, Spanish, French & Dutch based on online restaurant reviews from both Yelp and TripAdvisor platforms show that there are significant differences in restaurant satisfaction across customer groups with different language backgrounds. Furthermore, there are differences in the impact of attributes level satisfaction such as restaurant service between customer groups with different language backgrounds. Our findings contribute to the development of effective marketing strategies for corporate policies for international travel services by providing a more responsive experience for customers from different language backgrounds.