The current study investigates emotional attitudes of temperature terms in English via corpus analysis and sentiment AI analysis. First, based on the collocation patterns that are extracted from corpora in English (Corpus of Contemporary American English and iWEB), we attempt to specify a potential emotive feature for each temperature term. Second, we compare the results of corpus-based analysis in English with what a Sentiment Artificial Intelligence (Sentiment AI) model predicts. These results reveal how the meaning of temperature terms can be multidimensional: (a) a literal or figurative meaning in the semantic descriptive dimension; and (b) a speaker’s positive or negative emotional attitude in the evaluative dimension. Theoretical implications of the current study include the following: For one thing, we identify temperature terms as another clear case of sentiment-encoded words in English. Further, the current corpus analysis suggests adding to sentiment dictionary the newfound positive or negative connotational differences of temperature terms as a potential feature. Finally, our findings further support the notion of multidimensionality in meaning.
2. The landscape of temperature terms in English
3. Corpus analysis of sentiment in temperature terms
4. Sentiment AI analysis of temperature terms