This week’s Nature has a feature article on how a new breed of computational linguists are attempting to understand the evolution of language by using high powered computer models. The traditionalists are not impressed, and accuse the new school of reducing language to numbers and oversimplifying to the point of meaninglessness.
It’s an old debate in the human sciences, and relates to whether aspects of human experience can be meaningfully quantified.
Some psychologists, for example, completely reject any scientific approach to thought and behaviour because they say it strips human experience of exactly what it means to be human – the lived subjective experience of life.
It wonders how we can explain the fact that our conscious experience – which we understand subjectively, can arise from the biological function of the brain – which we understand empirically and objectively.
While not all problems are quite so intractable, many issues in human science bump up against the maxim “not everything that can be measured is meaningful, and not everything that is meaningful can be measured”.
Whether a particular method gets the balance right is a constant source of arguments.
The Nature article notes that traditional linguists tend to use their interpretation of word meaning combined with historical records to track how language has developed over time, while newer methods code rough assumptions into numerical models and then compute likely patterns.
It is putting it mildly to say that many historical linguists find the evolutionary biologists working on language histories to be bungling interlopers who have no idea how to handle linguistic data. It is also an understatement to say that some of these interlopers feel that their critics are hidebound traditionalists working on a hopelessly unverifiable system of hunches, received wisdom and personal taste. And that’s just the mood between the historical linguists and the newcomers. Lots of the newcomers don’t like each other either. ‚ÄúWhy get excited about it when it is still so preliminary?‚Äù says Johanna Nichols, a historical linguist at the University of California, Berkeley. ‚ÄúWe are not impressed by a computational or mathematical paper per se. We have to see that it blends well with what is known by historical linguistics and really adds to our knowledge. Then we will be excited.‚Äù
Link to Nature article ‘The language barrier’.