Online word of mouth through which consumers share their opinions and experiences on online platform has been given new significance in academia due to the popularity of the proxies, the valence and the volume of numerical ratings. In this paper, we underline multi-dimensional attributes of textual reviews, marginalized proxies of online word of mouth due to the difficulty of the measurement, and empirically show whether and how different the multidimensional attributes of textual reviews affect the sale of the product. Furthermore, we also show the impacts can be moderated by the status and the reputation of the hotels, and the impacts are different depending on whether the similar criteria of multi-ratings, new features of numerical ratings, are overlapped with the attributes extracted. To show the interesting causalities, we crawl hotel information including all the details of the hotels, prices, numerical and textual reviews, from hotels.com. We limit our analysis to New York-located hotels to exclude any possible endogeneity issues from strong seasonality or locational preference. In the analysis, we first extract frequently commented topics and match the topics with sentiments. Then, we build up an econometric model to analyze our hypotheses. As expected, the multi-dimensional attributes play different role on the sales. We split the sample by the status and the reputation, and found the impacts are significantly different.