Never mind...
I decided to use
  test="generate-id(.) = generate-id(ancestor::llcd:word//llcd:morpheme[1])"
which of course tells whether the current morpheme is the first in the
current word, instead of the last, which is just as useful in this situation.
And I think it's more efficient than what I had before.  It's certainly
shorter and seems clearer.  (Other ideas are still welcome though.)

Lars


> My current test is:
> "following::llcd:morpheme[1][generate-id(ancestor::llcd:word) =
> generate-id(current()/ancestor::llcd:word[1])]"


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list