Word count for inline equations

[manu-mannattil]

Currently, texprlcount assumes that 1 inline equation = 1 word. However, the PRL guidelines for calculating the length of an article doesn't mention the word equivalent for inline equations. I contacted APS about this and this is what they had to say about inline equations:

Counting inline math terms is perhaps the most difficult feature to pin down, according to the length estimators APS has contacted. Years ago, in borderline cases, the inline math would have been counted manually, at great time and expense. Now we provide estimates only and try to arrive at a reasonable total.

However, on experimenting with the APS submission system and comparing its estimates with texprlcount's, it seems that inline equations are given a higher word count compared to other words in the text. For instance, the following phrase

one two three $a=b=c$

is counted as having 8 words, whereas texprlcount would count it as 4 words. I think symbols like =, ≠, ~, etc., are assumed to be word boundaries by APS's system. I don't know of a good solution to this problem since APS's way of counting inline equations don't seem to be documented anywhere. A possibility is to use a higher weight for inline equations while calling texcount, e.g.,

my $texcount = `texcount $name.tex -opt=tcrules -utf8 -sum=1,1,1,0,0,2,0`;

This seems to reproduce APS's estimate reasonably well.