Large amount of textual, visual, and audio data are generating in social networking sites by the users nowadays. Social media users are generating these data in high increasing rate than any other time. Status updates/tweets, likes, comments, and shares/re-tweets are the basic features provided by the online social networking (OSN) sites. This paper utilizes the status updates of users to analyze and extract relevant natural language features to map them into predicting personality traits of those users. It is evident that using more features in a supervised learning system can predict more accurately. However, the linguistic features such as function words, character-level, word-level, structure-level features could be considered as relevant features for this case. While predicting the big five personality traits: openness-to-experience, conscientiousness, extraversion, agreeableness and neuroticism, the highly correlated features are determined applying feature selection algorithms. For experimentation, the research question is “What are the highly correlated features which are commonly found for all five personality traits?” In this paper, we have presented the experimental findings while determining the highly correlated features with the class and found that the percentage of “conjunction words” is always a common feature for each of the personality traits. The underlying (secret) relationship of this feature is analyzed in this paper.