Skip to content

[that godling year]

September 27, 2011
that 
godling year
         I fore
by 
looks
ware study
knapsacreached carried 
    moonlike
      metasting more shool
I fore
the study
    or aped 
        books
screenlit, 
   and 
trounded so 
         more
burn 
is flame the study
of myself 
with 
    dignified shore

I 
served
of the shove 
   bent ware
armlike lamblike 
leastingles an 
obliclike
      Now, 
trounded dopamined 
     not just begun the fasting of shool



         They 
    looking moonlike 
       looking burn 
within a pattern

        shadow sacreenlike
false tured brinklike
dissecrativity fromise, 
sometasting years 
the dirtuous groans
      just 
         burn 

they 
looks

screath 
the 
          looks

     God, I've thered his of evengeful 
    an obliclike 
looks
   so 
          the cured carried sacklike 
and at psyop 
the dictated brinklike 



They 
        looking 
burn 
    the wing
laughed
with 
         brinklike 
          the dissected 
metastingled to product 
  in a 
      panic
armlike Spiritus groans

 
Sept 26 2011, selections from character n-gram generation using charNG. Source text: unsupervised generation by davetolkacz from Infinite Monkeys on Borges’ Mirrors.

There’s been a bunch of discussion around here lately about mimesis or death of the author, and some of it I haven’t looked at long enough to figure out.

But I thought as a reference point, I’d take some word-n-grammed verse that davetolkacz generated, re-generate it through a character n-gram generator, and use a constrained editing method to produce a poem that was ego-driven and lyric but still open to multiple readings.

‘Cause the funny thing is, even with the unsupervised generation and constrained editing techniques, I still feel some thoughts and emotions of mine coming through. In the example above, it’s about my experiences in grad school and every school before that. I’m not 100% sure what every phrase means, but it still feels deeply personal.

Anyways, here’s the constrained editing method I used:

Method 4a7b54dd-8eea-40e9-9164-d8783c29565f (mining, arranging and pruning)

1. generate a number of lines

2. re-arrange the lines in any way

3. you may remove words from the beginning or end of a line (potentially removing the line entirely)

4. you may split a line into two new lines (but not join lines together)

5. you may add line breaks, to create stanzas for example

and here are the lines I generated:

Number of chars: 2400
N-gram length: 4
Type of chaining: markov
30% chance of inserting a newline after a word
50% chance of inserting 6 ± 5 initial spaces 

they 
        looking 
burn 
    the wing more study
on 
          the dissected 
that 
which inhabitern

    or aped 
ove 
        books
screenlit, 
   and 
by they looks
         Not 
ware study
  of 
knapsacreached carried to 
    moonlike
that 
godling year
         I fore
in the 
with 
    dignified shore
   in the shood
of 
        shadow sacreenlike convice are serted his 
false tured brinklike
that tear
I fore
trounded so 
         more
I 
served
within that makes 
of the shove 
   bent ware are armlike lamblike 
leastingles an 
obliclike
       troundi or aped chases
of 
shood
         of myself with for 
          aped his jing 
burn 
is flame the study
of myself 
withing of service act
      of 
the serve whose 
          number,
      that he stral reflexive, in is not watchs 
  in 
suffered

They 
looks
screath 
the 
          looks
        screen serve books
     screenlike 
       act
          of knapsacreate
seeks, to more
in 
is -ry pretended so them 
      as an oth for an 
obliclike 
        leastness
which in surroubler of shadow of 
evengeful homeone’s which in abnegation
did not just 
     but of spects 
organic
with fore
in to fastingles —
       
metastingled to product 
  in a 
      panic
with 
         brinklike 
armlike Spiritus groans
  
         but 
of evengeful 
hometishion
  that within that tear
      of spindle an evengeful hometasy

        They 
         looks
screen surreceipt 
of laughed
   
       They look out 
       of 
  thing use, 
sometasy
     
         in is 
          flameness.

   lifelihoot just 
but of with dissecrativity fromise, 
sometasting years 
   uses by 
not 
      just 
         burn 
the dirtuous groans

but but 
      of laughed 
his not but thered 
castings, noose the 
service armlike armlike 
preted sometasting 
    uses
of thefts
   
     God, I’ve thered his of evengeful 
      hometastingles 
    an obliclike 
looks
   so 
          the cured carried sacklike and at psyop the is of 
the dictated brinklike 
laughed

They look 
   that’s 
       use, in that ther 
Ruthe is jing use, in an evengeful hometishizings, organic
with bricks, 
not just 
burn 
its of catchs in a pattempt 
of 
all least, 
          an 
    obliclike 
armlike 
act
of evengeful 
home

Ever
    by ther 
sometasting moonlike Spiritus groans
   which in all 
the dissecrator, and that the lameness.

Now, armlikelike lensecration
     of shood
of the laughed 
      them as a pattempt of 
      evengeful 
hometasting 
but tears us.That the fantil the is nother Ruthe shore
in whose cult
        to makes by nothing, 
pretend by its of spindle 
      act
  of shood
of 
laughed from devources 
  —
     
      metasting more shool
the study
    on the 
          all the fantasy
      
They looks
      Now, 
armlike
trounded dopamined 
   his 
     not just begun the fasting of shool
          the 
with 
fore
      that 
to moved

         They 
    looking moonlike 
       looking burn its 
of its 
that within a pattern
      
or aped that g
Advertisements
5 Comments leave one →
  1. September 27, 2011 9:24 am

    Another interesting point to be made about ego-poems, is the issue of the divine analogy. The author:book :: god:the world, the “creator”/”originator”. This seems like the sense of authorship that is being contested/”killed” by Eric and authors. It’s interesting to see my poems now in the light of Eric’s analysis, especially “god is the polar coordinate plane” since it focuses on the “death of god” vis a vis Nietzsche. More to come, I’m strapped for time. BTW, the Borges’ Mirrors IM poem is really one of the best unsupervised examples from IM I’ve ever seen. Also, at least a few of the unsupervised examples have past the Turing Test. People (to my great shame and delight) thought I constructed them.

  2. September 27, 2011 11:48 pm

    BTW, the Borges’ Mirrors IM poem is really one of the best unsupervised examples from IM I’ve ever seen. Also, at least a few of the unsupervised examples have past the Turing Test. People (to my great shame and delight) thought I constructed them.

    Yeah, Crowning the Blood popped out without needing any editing, too.

    But evaluation is a notoriously hard problem… Manurung and Gervas and Knight and everyone else who does NLP approaches to poetry generation mentions it… if you think of a good answer, let me know. Respectfully, I’m not sure about the Turing Test; in my old Prosthetic Imagination post I talk a little about it. (Basically, you always have human intervention. For example, you programmed the generator, you decided the output was good enough to test, maybe you selected part of it and threw out other parts, etc. So we the humans did construct them, at least in part.) In evaluation, unless you look at adherence to form (rhyme, meter, etc.) you’re working with qualitative information, which will probably involve some kind of human ratings on a Likert scale… I’m thinking of two conditions, generated from ePoGeeS for example; in one of them you have end- and inner-rhyme, in the other you don’t. Provide Likert scales asking the test subjects (via Amazon Mechanical Turk maybe) whether they find the poem emotionally moving, likely to be generated by a human vs machine, etc, and see if there are statistically significant differences in which scores higher. One of these days I’ll run that experiment…

  3. September 28, 2011 4:20 pm

    Yes, the experiment was not scientifically rigorous.

    Anyway, this poem is interesting. With the context of knowing it’s about your school experience, and how it was built it really completes the experience of reading it. The question is, (and I am not advocating this necessarily) is the poem “weaker” because it requires that dependency, or does it even require that dependency. For me knowing the background completed the experience. I think this adds the complexity of the poem, as opposed to detracting from it.

  4. October 2, 2011 5:22 am

    Yes, the experiment was not scientifically rigorous.

    Sorry, I didn’t mean to imply you were claiming scientific rigor. I mentioned evaluation because it’s a difficult open question when dealing with aesthetic works in the language technology fields, and it’s also problematic to researchers in the digital humanities who suddenly find themselves arguing with stats-obsessed engineering grad students. Personally I suspect the best evaluative methodology for a field like poetry generation is the production of artistic artifacts.

    Anyway, this poem is interesting. With the context of knowing it’s about your school experience, and how it was built it really completes the experience of reading it. The question is, (and I am not advocating this necessarily) is the poem “weaker” because it requires that dependency, or does it even require that dependency. For me knowing the background completed the experience. I think this adds the complexity of the poem, as opposed to detracting from it.

    I guess this is one of those poems that tries to be open to multiple interpretations. The question is: does it provide the reader with the right affordances for at least one useful interpretation? To me it does: about education. For others it might be something else, or it might not work at all.

  5. October 2, 2011 11:07 am

    “Personally I suspect the best evaluative methodology for a field like poetry generation is the production of artistic artifacts.”

    I agree. I didn’t mean to imply that you were implying that I was implying scientific rigor. :)

    “I guess this is one of those poems that tries to be open to multiple interpretations. The question is: does it provide the reader with the right affordances for at least one useful interpretation?”

    I think it does too, looking over the poem, and the multiple references to school things, we can get some sense that you are talking about school. A lot of it is keeping us at a distance so looking for some grounding in the text without the context of school becomes harder, and requires more mental energy on behalf of the reader. But people will be interested in it just in terms of the process used.

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: