We do not hesitate, in poetry, to yield ourselves to the unreal - Wallace Stevens

Home About Automatic poetics Directed poetics Poems on file Source code The Prosthetic Imagination Stats To students The author Contact

You can find detailed directions below. Topic:  Next, pick a stanza form: Generic Open Form Stanza Hinge Stanza Paratactic Stanza Standard Stanza Now pick a grammar: Common DuPlessis Fragments hinge clause lyric Mimetic nominatives O'Hara Plath questions Snyder Subordinate Clauses titles Now pick one or more word "pools": topic context topic only topic synonyms topic antonyms alliteration Pick a preferred tense: present past future present perfect past perfect future perfect Pick a preferred subject: I we you he she it they Pick a preferred object: her me us you him it them Finally enter a minimum and maximum number of lines for this stanza and click "Add stanza." When you are satisfied, click "Compose poem." Minimum lines:      Maximum lines (max 5):  Stanzas so far (max 6):  0       Context source:  Complete Poems of Emily Dickinson Joseph Conrad's Heart of Darkness Topic Etc3 assigns a topic to each stanza. She takes the text you enter for the stanza's topic and parses it to discover words it recognizes and can reasonably use to construct poetic utterances. It ignores words it doesn’t recognize. If you asked it to write a poem about sub-prime mortgages, it would just ignore your topic, since neither of those words is in its semantic model—it would, however, write something on a topic of its choosing. Enter a topic. You can change the topic for every stanza. A topic can be any lexical string, a "real topic," a list of words, or nothing at all. If you don't change it between stanzas, Etc3 will just keep using the first one. Actually, there's a pretty good chance that she'll just ignore your topics altogether. Stanzas Generic open form stanza simply breaks utterances into short lines, indenting another tab stop at each break. Hinge stanza also ignores the number of lines options and writes a single line using the grammar that you select. It does not add an extra line break at its conclusion, in effect attaching itself to the next stanza as its first line. Its original purpose was to a way to have an oppositional kind of structure within a poem. It works best with the Hinge grammar. Paratactic stanza selects one syntactic structure from the grammar you select and then uses that for each of its lines. Standard stanza is just like the generic stanza but inserts no line breaks into an utterance. Word pools A word pool is a “bag of words” that directs Etc3’s morphological decisions. There are different kinds of word pools. You can select as many as you want. Etc3 will create a pool that is an amalgam of your selections. Topic only used by itself restricts the words in the pool to only the ones Etc3 recognizes from the stanza’s topic, resulting in the most limited vocabulary with repetition almost guaranteed. Used with other pools, it makes it more likely that words from your topic will work their way into the final piece. Topic synonyms is a collection of words other than topic words but with approximate semantic meanings. Topic antonyms is a collection of words other than topic words but with approximate opposite semantic meanings. Topic context collects words that appear near the topic words in Etc3’s source corpora. Etc3 often surprises with its context words. I recently asked it to write about “justice and mercy.” It found “sentence” as related to justice and used that as the basis for a meditation on writing. This gives the most expansive vocabulary. Alliteration simply gets words with the same initial letter, using a word from the topic as a starting point. It’s a bit silly, but also a bit fun. There are lots of different things you can do with these word pools. You can asked for a two-stanza poem, where the first stanza uses synonyms and the second antonyms to get a nice little hinged poem. You can throw in a one-line stanza using alliteration as a device. Lots of possibilities. Grammars A grammar is a set of syntactic templates from which Etc3 can construct meaningful utterances. Most simply, they are groups of TAG trees that follow some model. There are many more trees in the database than are exposed in the grammars. And the grammars are quite likely to use them for things like adjoining trees, performing lexical replacements, and a host of other actions. The exposed trees are higher level abstractions grouped around some organizational principle. Mimetic is a conflation of several of the grammars, with the less satisfying structures erased. In a sense it models what we have learned about describing poetic speech. And its the one most likely to be used in the automatic writing utility. Common Just structures that find there way into mulitple grammars, they are the results of studying blooded writers. (Though bloodless, we think re retain more of what makes for poetry than such writers currently do.) Fragments consists of various phrases that are not sentences. It works well with the Paratactic Stanza. Hinge clause has one structure that suggests a turn. Intended to be used with the Hinge Stanza but use it any way you want. It won’t break. Lyric contains some structures that I lifted from places like Best American Poetry. It had some use in developing test cases, but seems to me to have limited value in production. Nominatives contains a couple of subject-copulative verb-complement structures, again primarily for testing. O’Hara consists of structures lifted from Frank O’Hara’s poems. Snyder consists of structures lifted from Gary Snyder’s poems. Plath consists of structures lifted from Sylvia Plath's poems. This is not to suggest that we like the late Mrs. Hugh's work. We don't. But it really does contribute to the polemic. Titles consists of structures Etc3 uses in creating titles, but use it like you would fragments. Preferred tense Etc3 will try to use the tense you select, but some TAG trees will override it. You can use this option variably in the stanza of a poem to suggest temporal shifts. E.g.: put the first stanza in the past, the second in the future. Preferred subject If Etc3 selects a tree that has a personal pronoun subject, it will try to use the value you select. Used creatively with preferred object (and with the right grammar), you can approximate an MFA poem. Preferred object If Etc3 selects a tree that has a personal pronoun object complement, it will try to use the value you select. Used well and in conjunction with preferred subject, you can get things like “I hated you.”