Analyzing and extracting features from requirement specifications is an indispensable activity to support Software Product Line Engineering. However, performing features extraction is a time-consuming and inefficient task, since massive textual requirements need to be analyzed and classified. Most of the current approaches exhibited limitations: hindered applicability with requirements in Japanese; the support tools proposed were not made available publicly and thus making it hard for practitioners' adoption. This paper proposes a feature extraction approach from requirement specifications in Japanese using natural language processing techniques. Also, we propose a ranking method for extracted features to reduce efforts reviewing feature candidates. A case study was conducted to evaluate the performance of the proposed approach. Initial results show that 90.7% features were extracted correctly, and the top 40% features extracted contained 79.1% true features.