Skip to content

Dealing with Ambiguousness the Way Humans Do

Here I wish to outline (and sort a bit) two principles I have in mind in dealing with ambiguousness in linguistic input. My point is to suggest ways that might reflect upon the way the human mind deals with this issue. Ambiguity issues are solved by the human mind seamlessly and with probably very little intervention of higher logic and conscious thinking. This is shown by the fact that ambiguous input is not seen as such in most cases. Its ambiguous nature is only revealed after a second look and guided search. This suggests that the issue is dealt with low level simple methods. Continue reading ›

Tagged , , ,
Comments (0) Permalink

Nautral Language Representation

Natural language is both a human creation and used only by humans. This suggests a strong connection between the traits of the language itself and the inner works of the human brain. It seems to me like that the structure of the brain is well optimized for the task and at the same time the structure of natural languages is optimized for the human brain. This, of course, hints at Chomsky’s universal grammar idea. However, it can also mean that structures that are ‘natural’ to humans might underlie the logical structure of languages. Continue reading ›

Tagged , , ,
Comments (0) Permalink

Semi-hardwiring of Langauge Specific Attributes

Kids have an easy time getting a new language. This difference is not only in the ease of acquisition, but also in the higher quality of the result. It seems that a child’s brain is more susceptible and ready to acquire a new language. It’s known that the human brain is being massively hard wired during the first years of a child’s life. It can be hypothesized that certain attributes of the language being learnt at such early stages are hard-wired into the infant’s brain. Continue reading ›

Tagged , , ,
Comments (0) Permalink

Source Credibility in Language Acquisition

The issue of false inputs in language acquisition is a well known one and considered critical to NLP applications. For instance, computer automated acquisition of language through data on the Internet will require to sift through a large percentage of false data. Statistical models try to overcome this difficulty by assuming that most of the data is correct and therefore the ‘right’ knowledge will overcome the ‘wrong’ one and practically drown it in the statistical pool created. Continue reading ›

Tagged , , , ,
Comments (0) Permalink