Ok, this would be done by JavaScript, through the VB:JavaScript linkage. VB is a better choice but it appears that it might be possible to do this with C#, not sure.
To find positions of text you use doc.getPageNthWord and doc.getPageNthWordQuads. This gives the position of each word and NOTHING ELSE. To look for consecutive words you would need to analyse the results. Text in a PDF is not just strings with newlines, but a complicated collection of graphics; getPageNthWord assembles the graphics and uses GUESSWORK to pick word boundaries.
I believe you can also use JavaScript to add highlight annotations.
You might use the OLE API to open, save and close the framing PDF, assuming it isn't already open in Acrobat.