Software requirements are ambiguous due to natural language and the natural language is generally ambiguous. The ambiguous requirements lead errors in software development. Therefore, there have been many approaches or techniques applied for detecting ambiguity in software requirements. This paper has selected three supervised machine learning algorithms with Bag of Words (BOW) features, which are Support Vector Machine (SVM), Random Forest and K-Nearest Neighbor (K-NN) for detecting syntactic ambiguities in software requirements. The summary of results showed that SVM achieved accuracy (80%), Random Forest attained (86.66%) and K-NN obtained (50%) respectively.