The study investigates the relationship between the similarity of word contexts and the degree of synonymy between words. It hypothesizes that the proportion of words common to the contexts of two words is a function of their semantic similarity. The study tested this hypothesis using various context definitions and found that similarity of context is a reliable criterion only for detecting pairs of words that are very similar in meaning. The study compared human judgments of synonymy with contextual overlap measures. Sixty-five pairs of words, ranging from highly synonymous to semantically unrelated, were analyzed. Contexts were defined in four ways: all words in a sentence, content words within a frequency range, words closest grammatically to the theme word, and words judged as closely associated. Overlap was measured using two methods: type overlap (M_y) and token overlap (M_k). The results showed a positive relationship between semantic similarity and contextual overlap, but the relationship was strongest for highly synonymous pairs. For intermediate values, overlap was relatively constant, suggesting that subjects did not perceive the scale as consistently. The study also found that grammatically defined contexts and contexts based on association had higher discrimination power in detecting synonymy. The study concluded that while there is a positive relationship between synonymy and contextual similarity, the relationship is not strong enough to reliably infer synonymy from contextual overlap for all pairs of words. The findings suggest that the relationship may be influenced by the specific language materials and procedures used in the study. Valid generalizations depend on further research into factors such as corpus size, content homogeneity, and information packaging.The study investigates the relationship between the similarity of word contexts and the degree of synonymy between words. It hypothesizes that the proportion of words common to the contexts of two words is a function of their semantic similarity. The study tested this hypothesis using various context definitions and found that similarity of context is a reliable criterion only for detecting pairs of words that are very similar in meaning. The study compared human judgments of synonymy with contextual overlap measures. Sixty-five pairs of words, ranging from highly synonymous to semantically unrelated, were analyzed. Contexts were defined in four ways: all words in a sentence, content words within a frequency range, words closest grammatically to the theme word, and words judged as closely associated. Overlap was measured using two methods: type overlap (M_y) and token overlap (M_k). The results showed a positive relationship between semantic similarity and contextual overlap, but the relationship was strongest for highly synonymous pairs. For intermediate values, overlap was relatively constant, suggesting that subjects did not perceive the scale as consistently. The study also found that grammatically defined contexts and contexts based on association had higher discrimination power in detecting synonymy. The study concluded that while there is a positive relationship between synonymy and contextual similarity, the relationship is not strong enough to reliably infer synonymy from contextual overlap for all pairs of words. The findings suggest that the relationship may be influenced by the specific language materials and procedures used in the study. Valid generalizations depend on further research into factors such as corpus size, content homogeneity, and information packaging.