How can morph dictionary be built using postgresql?

Question

I need to create special morphological vocabulary for a natural language. Each word should contain a set of characteristics. Does PostgreSQL 9.* help in such situation? I mean: should I create table from the scratch or there are some predefined means?

Answer 1

You are looking for things well beyond what typical solutions on PostgreSQL are designed to build. Your best bet is to build your own custom table.

Now, I am just an interested reader of linguistic topics, but for morphological analysis, I suspect you are going to run into problems with different languages having fundamentally different morphological systems. For example, the morphological concepts in Austronesian, Indo-European, and Athabascan (all of which I know just enough to be dangerous) do not strike me as very conducive to a single relational model. For example, we might have distinct morphologies for verb tenses in IE, but not the other two. Reduplication in Austronesian languages adds a bit of a curve ball (particularly when dealing with prefix/suffix/infix combinations), and Athabascan has slots that don't fit the other two language's expectations of parts of speech.

So I don't think you are likely to find many general-purpose morph database schemas around. Things are likely to be language-specific and purpose-specific. For example, building a database to look for morphological changes between Middle and Modern English is going to be quite different than something designed merely to analyse current morphology in, say, tweets.