This is not intended to be a complete list, but I need to put these somewhere that I’m unlikely to lose them. Here are some things I believe, have no evidence for, but would like to work toward in my research:
- language change in general and sound change in particular are not (merely) destructive. languages are not in any kind of decline and there was no golden age.
- the same is true for writing and punctuation.
- coarticulation is not ``a problem’’. It enriches the information content available to listeners in a speech signal and is key to understanding articulation.
performance' is every bit as interesting and informative ascompetence’.
- humans are exceptionally good at seeing patterns in data and making useful generalizations based on those patterns. you might call this statistical learning and you’d probably be right.
- sonority, markedness, and extrametricality are nonsense concepts, generally defined circularly, and are to be referred to only with the most extreme care.
- killing people is wrong. violating people’s privacy is wrong. I need to come to terms with what it means to do research (e.g. machine translation, text summarization, etc.) which can be used directly (and is typically funded by) people who are interested in killing and eavesdropping.
- `computational linguistics’ needn’t be just about language technology. the same mathematical tools and empirical approach can and should be applied to the rest of theoretical linguistics.