10

I'm writing my masters thesis and my supervisor has picked up my sentence structure, in that they are too long. She commented that Microsoft Word has a feature to investigate 'large' sentences. How can I do this with TeX? I'm using a mac and TexShop.

Improve this question
  • 2
    Not a very TeXy way: copy and paste the text from the generated pdf file to Word. – Caramdir Aug 25 '10 at 17:49
  • 2
    Or if you don't have Word: there are tools to count the number of words in a tex file. To count sentences, we'll just count the number of periods (unless you use a lot of commands/math with the character, you shouldn't need to strip the tex file) cat file.tex | sed 's/[^.]//g' | wc -c should give you a number that is one more than the total number of period marks used. – Willie Wong Aug 25 '10 at 18:34
  • Willie Wong: comment should be an answer! – Matthew Leingang Nov 30 '10 at 16:06
8

I think this is the easiest way. First, make sure you have pdftotext and diction installed. These should be available via MacPorts.

  1. Render your document to a PDF. Let's assume it's called paper.pdf.
  2. Grab the plain text from the PDF using pdftotext. At the terminal, run this: pdftotext paper.pdf paper.txt
  3. Now run style -l N paper.txt, where you should replace N with a number. This will print out all lines of your document that are longer than N words.

Alternatively, you can do it all as a one-liner:

$ pdflatex paper.tex && pdftotext paper.pdf - | style -l 20

style is extremely powerful and has many other features. For a good overview, see here.

Improve this answer
  • One could alternatively use DeTeX instead of pdftotext to get a plain text version to send to style. I find, however, that pdftotext works a lot better, especially if one has a lot of markup in his or her document that DeTeX doesn't understand (e.g., TikZ). pdftotext also works a lot better if one splits one's original document up into multiple .tex files that one \include{}'s. – ESultanik Aug 26 '10 at 12:38
  • using style without any parameters will give you a summary that will tell you the % of long ("at least 27 word") sentences. – Geoff Aug 26 '10 at 13:35
  • @ESultanik, do you know how to tell diction/style that abbreviations are not (usually) ends of sentences? (Dr., Fig., etc.) – Geoff Aug 26 '10 at 13:42
  • @Geoff This should work for most cases (except for 'etc.', but it's easy to extend the regex for those special cases): pdftotext paper.pdf - | sed 's/([A-Z][a-zA-Z]*)\./\1/g' | ... – ESultanik Aug 26 '10 at 14:37
  • @ESultanik. Just drop the period, okay. It seems to me that style should support this. – Geoff Aug 26 '10 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.