Talking out of the box IV
By esj on Friday 11 March 2011, 01:29 - Permalink
In the previous blog post, I showed an applied example of working with non-speakable text. This example highlighted some of the issues pertaining to name transformation. In this post, we shall briefly discuss a way at looking at non-speakable text as a precursor to using these concepts in a practical environment.
The core concept is something I call "landmarks". Everything has landmarks. At first glance, this paragraph has characters, words, sentences, paragraphs, all are landmarks. It also has quoted text landmarks as in the word inside quotes in the first sentence. There are other landmarks possible in natural language text. Code, because of it's structure is full of landmarks. Taking a code example from an earlier post, let's highlight its landmarks.
Landmarks are what we use to grab a hold of some piece of text without specifying that text. "Select first method" would highlight the region occupied by the word "rename". "Rename second instance" would select the second instance name,"fn", apply a reverse transform to generate the plaintext name and pop up the speakable dialog box for editing/changing said names.
There are lots of other landmarks visible in code as well as in other structured non-speakable text environments. Proper use of landmarks in a natural language grammar enables effective efficient navigation.
Next post, pulling name generation and landmarks together.