On the Language Typology mailing list this morning, a note from Grev Corbett:
A sobering question is “In ten years time, how many people in this linguistics class are going to care about the definition of phoneme, clitic or right node raising?” If the proportion is small, then a linguistics class can be invaluable in getting over messages which will matter in ten years time, such as:
– beware of arguments from authority
– respect the data
– don’t guess when you can measure
– beyond what we think we know there’s a seething mass of uncertainly and ignorance out there
– when we hit the ‘in-between’ cases, we don’t throw our toys out of the pram, but we try to understand the apparently clear cases better
– “… the intensity of the conviction that a hypothesis is true has no bearing on whether it is true or not.” (Peter Medawar: Advice to a Young Scientist 1979 p. 39)
(You will note that these pieces of advice have a wider applicability than to linguistics.)