Words with multiple meanings (e.g. bark of the tree/dog) have provided important insights into several key topics within psycholinguistics. Experiments that use ambiguous words require stimuli to be carefully controlled for the relative frequency (dominance) of their different meanings, as this property has pervasive effects on numerous tasks. Dominance scores are often calculated from word association responses: by measuring the proportion of participants who respond to the word 'bark' with dog-related (e.g. "woof") or tree-related (e.g. "branch") responses, researchers can estimate people's relative preferences for these meanings. We collated data from a number of recent experiments and pre-tests to construct a dataset of 29,542 valid responses for 243 spoken ambiguous words from participants from the United Kingdom. We provide summary dominance data for the 182 ambiguous words that have a minimum of 100 responses, and a tool for automatically coding new word association responses based on responses in our coded set, which allows additional data to be more easily scored and added to this database. All files can be found at: https://osf.io/uy47w/.