Contract conditions are crucial as they outline an agreement between different parties. The semantic terms in contract conditions need to be precisely designated. Where these conditions contain vague meanings, the interpretation of the conditions will vary, especially since the parties of the contract will be differently motivated to pursue their different expectations from it. The vague terms in contract conditions may thus cause a dispute and conflict among the parties that can jeopardize the eventual success of a construction project. The conventional practice of identifying vagueness in construction contract conditions is done manually, which is prone to error, time-consuming, and requires expert involvement. This study develops a methodology to automate the identification of vague terms in construction contract conditions with the sequential application of natural language processing (NLP) and machine learning (ML) techniques. Morphological and lexical analysis procedures are used to evaluate the corpus data obtained from a widely used typical construction contract published by International Federation of Consulting Engineers (FIDIC). Classifications of contract conditions in the corpus data are searched using several supervised ML techniques to determine the best performing classifier. The results show that the developed methodology reduces time spent on contract review, is reliable with a high level of accuracy in predicting the presence of vagueness, and removes dependence on expert participation in the contract review processes.