As the author points out, song structure biases the results, so that artists with structured and repeated lyrics end up with a lower "vocabulary". Wouldn't it m

As the author points out, song structure biases the results, so that artists with structured and repeated lyrics end up with a lower "vocabulary". Wouldn't it make sense to simply combine repeated lines before counting? So you'd be measuring vocabulary in first 35,000 non-structurally-repeating lyrics.