This paper reports the results of a quantitative experimental study designed to reveal semantic changes involved in lexical variation in Turkish. Lexical variation under investigation has been the result of a linguistic policy, which has foreseen the replacement of Arabic and Persian loanwords with neologisms. It is hypothesised that lexical variation is not semantically free; semantic changes have occurred in the interaction process as each form is socially embedded. A questionnaire was conducted on more than one hundred subjects. Being a quantitative study, it makes use of sociodemographic factors such as sex, education, and age. Prototype theory was employed in the definitions of lexical items under investigation. It was found that semantic changes took place in the majority of the pairs tested