"Hastily diagrams sentence"

« previous post | next post »

Thoreau, "I don’t cry for yesterday; there’s an ordinary world", Unqualified Offerings 10/22/2015:

Ah!  Clinton vs. Bush with an insane billionaire in the mix and the latest Whitewater/Vince Foster sequels already brewing!  It’s 1992 all over again.  Except this time I’m not single and trying to work up the courage to ask a girl out.  Things are much better for me.

Comment by mds:

Except this time I’m not single and trying to work up the courage to ask a girl out


… I say go for it.

The online demo of the Stanford parser makes the same "mistake":

As does the online demo of the Berkeley parser, with the added puzzle of construing "ask a" as a verb:

Exercise for the reader: What is the Penn Treebank representation of the analysis that the writer intended? (Both of those parsers are aiming for the Penn Treebank style, more or less…)

Extra credit (lots, because it requires poking into the code): What's with that business of analyzing "ask a" as a verb???


  1. Christopher said,

    October 25, 2015 @ 10:19 am

    Surely the meaning here is "not (X and Y)", which is logically distinct from "(not X) and Y"?

    [(myl) Exactly.]

  2. Y said,

    October 25, 2015 @ 12:33 pm

    OT, when I saw the phrase Berkeley parser, I first read it as parsley barker. But I'm fine now.

  3. Lingshan said,

    October 25, 2015 @ 2:01 pm

    Christofer, I thought of that too.

    But why "ask" is not a verb? Sorry if it is silly to ask it (pun intended), I might add that I am no native English speaker.

  4. Jerry Friedman said,

    October 25, 2015 @ 2:36 pm

    Lingshan: The question is why "ask a" (both words together) is listed as a verb in that tree. Normally we'd say the verb is just "ask".

  5. Rubrick said,

    October 25, 2015 @ 6:14 pm

    Parsely Barker would be a really, really challenging job.

  6. Mark Johnson said,

    October 26, 2015 @ 5:38 am

    The Charniak-Johnson parser gets the scope of negation right, but I don't know about the analysis of "work up". I don't know how to paste in a graphic, so here's the parser's PTB representation:

    (S1 (S (PP (IN Except) (NP (DT this) (NN time))) (NP (PRP I)) (VP (VBP 'm) (RB not) (VP (JJ single) (CC and) (VBG trying) (S (VP (TO to) (VP (VB work) (PRT (RP up)) (NP (DT the) (NN courage)) (S (VP (TO to) (VP (VB ask) (NP (DT a) (NN girl)) (PRT (RP out)))))))))) (. .)))

    [(myl) Here you go:

    The up in work up (the idiomatic version that means "To intensify gradually to some state" or "To develop or produce") should be intransitive, right?

    Also, shouldn't "trying to etc." have its own phrasal node?]

  7. Akito said,

    October 26, 2015 @ 7:56 am

    I am only an amateur, so the result is of no consequence, but I have just noticed that the Stanford parser yields "Not that I" as a constituent for the sentence "Not that I know of."

  8. David L said,

    October 26, 2015 @ 11:12 am

    Parsley Barker would be a really, really challenging job.

    Hard for a person or a cat, easy for a dog.

  9. bratschegirl said,

    October 26, 2015 @ 12:07 pm

    "Get yer frrrreeeeesh parsley right here! Parsley!"

    Won't somebody think of the poor sage, rosemary, and thyme?

  10. Old Gobbo said,

    October 26, 2015 @ 12:15 pm

    "Parsley Barker would be a really, really challenging job"

    Sounds to me like a street trader in greengrocery.

  11. Kaleberg said,

    October 26, 2015 @ 9:17 pm

    It's good to see he's no longer single and trying.

RSS feed for comments on this post