CHILDES and graphing
Here are some notes from class, and also a (slightly editing and annotated) Jupyter notebook file representing what we did.
Notes relating somewhat to tasks from homework 4:
If you see an “index out of range” error it almost always means that you have tried to do something with an element of a list that turned out to be empty. You probably didn’t expect the list to be empty.
To demonstrate this, here’s an example where you might very reasonably
have not thought the list would be empty.
Suppose that you’ve done the following, created a little grammar,
and created a parser for that grammar, and parse the sentence
‘I left’ by using parser.parse(sentence).
import nltk
littlegrammar = nltk.CFG.fromstring("""
S -> NP VP
NP -> "I"
VP -> "left"
""")
sentence = ['I', 'left']
parser = nltk.RecursiveDescentParser(littlegrammar)
treegen = parser.parse(sentence)
If at this point you check to see that you have what you want, you might do something like…
list(treegen)
…which will, happily, display a list of trees resulting from parsing the sentence…
[Tree('S', [Tree('NP', ['I']), Tree('VP', ['left'])])]
“Perfect!” you think to yourself, and decide you’d like to give it a name so you can refer to it later…
trees = list(treegen)
…and then try to draw the tree…
trees[0].draw()
…you will be confronted with an unfriendly list index out of range error.
“But that’s nonsense!” you think to yourself. “I just looked at it, and
list(treegen) gave me a non-empty list. See?”
print(trees)
Which shows you
[]
See, it turns out that using the generator that parser.parse(sentence)
provided you with actually uses it up. You can only use it once.
And we burned through it just checking it out by displaying
list(trees). After that, there was no generator left, so doing
list(trees) a second time just produced an empty list.
It’s subtle, even perhaps darkly magical, but it is a way that you might wind up with an empty list you weren’t expecting. So just be aware that sometimes when you get a generator back, you may only be able to use it once.
As for the rest of what we did in class, I’ll include the Jupyter notebook file here, which has been lightly annotated and organized a bit from when it was originally typed.