Skip to content

little tender blue

June 14, 2010

won’t you do little raindrops
your little raindrops
your little raindrops
your little tender blue

you knew all you knew
count on little
count on little
won’t you can you knew

a little tender
a little girl
a tender blue
you can you just
little tender blue blue blue
you can you might
count on you

 
 

June 13-14, selected from type-based bigrams, source text: lyrics to “Little girl blue” as sung by Nina Simone, Generator: ePoGeeS

1) This is based on the lyrics to “Little girl blue” as sung by Nina Simone:

Sit there and count your fingers
What can you do, little girl you’re through
Sit there, count your little fingers
Unhappy little girl blue

Sit there and count the raindrops
Falling on you, it’s time you knew
All you can ever count on are the raindrops
That fall on little girl blue

Won’t you just sit there
Count the little raindrops falling on you
It’s time you knew all you can ever count on
Are the, the raindrops that fall on little girl blue

Ain’t no use old girl, ain’t no use old girl
You might as well surrender
‘Cause your hopes are getting slender
‘Cause your hopes are getting slender
Why won’t somebody send a tender blue, blue boy
Won’t somebody send a little tender blue boy to cheer
To cheer a little girl blue

2) Traditional poetic forms are finite. So, while a language model might in theory be able to generate infinite strings (like “you can you can you can …”) a traditional poetic form of the kind supported by gnoetry or epogees would not use such an infinite string. For small data sets, like the lyrics of the song being considered here, it might be possible to do an exhaustive listing of all possible combinations that the language model could generate.

So there’s a finite set of traditional-form poems that a language model could generate from a given single poem. The end-user’s job is to work with the poetry tool to select elements (i.e. individual poems) from this finite set.

3) N-gram poetry generators create novelty by combining pairs of words in interesting ways. What interesting types of combinations are there? Previously I mentioned combinations from different sources that are new yet familiar. The example above shows another: by subverting explicit textual meanings. The original text never explicitly says anything like “you can you might / count on you”; it superficially says all you can count on are raindrops and maybe little boys blue. (Interestingly Simone, a militant activist, sometimes explicitly sang that all you can count on is yourself, although it’s not in this text.)

4) so anyway, I wrote this poem by building a bigram model on the lyrics, and examining the language model for interesting lines (consistent with the bigram model) that were not in the original text.

5) why tf am I numbering my paragraphs?
5.1) oh, it must be cause I skimmed the Tractatus
5.2) lol. later!

Advertisements
No comments yet

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: