Online marketplaces are growing internationally with sellers coming from around the world. Product description texts often stand as the first piece of information soliciting buyers' attention, where the seller’s culture and linguistic backgrounds drive great variations in description style and appeal. Product descriptions, on top of being a self-selected disclosure of product information, offer first-impression signals to infer a seller's credibility, competency, experience, and more for a buyer’s decision-making. Sellers therefore make strategic decisions in composing product descriptions to woo potential buyers. In this research, we studied composition quality of product descriptions, a type of marketer-generated content, in an online marketplace. We collected online product description data and quantified their properties using data mining methods to define perceived description quality and characterize quality differentiating dimensions. Our analysis identified ten defining characteristics for differentiating native versus non-native speaker drafted descriptions. By combining business text analytics methods with modern linguistic theories and tools, this study is the first research to systematically quantify and characterize the concept of composition quality within the context of consumer-perceived nativeness. Our findings further complete and advance the literature on marketer-generated content (MGC), consumer decision-making, and seller operations decisions.