Thoughts on analogy and some problems in interpreting phonological experiments

An inconclusive pilot study in Danish phonology gives rise to questions about testible differences between explanation by analogical algorithm versus rules. The literature on analogy shows its resistance to valid limitations on its operation in terms of markedness, similarity, or on the basis of purely conceptual or grammatical considerations. It is argued that being the less constrained mechanism, it is inferior as a working hypothesis to rules. It is suggested that the convincing instances of synchronic analogy are special cases where a speaker resorts to a more basic cognitive strategy as ill-defined and hence powerful as our ability to recognize similar aspects of nonidentical complexes.