Linguists Enthusiastically Catalog Symbols, Interpreting Carefully Occurred Nuances.
Lecsicon is a collection of over 24,000 acrostics generated by OpenAI's language model GPT. Following a specific prompt, GPT made a sentence with a series of words whose first letters sequentially spelled out the given word. At a large scale, this process gives rise to intriguing behavioral patterns in the model. The resulting acrostics also elicit gripping reflections on the nature of this literary device and its role in meaning making.
In response to my prompt, the model created acrostics in the form of sentences to each of the words provided. The accuracy of GPT's responses, as usual, was not guaranteed. At times, one or two extra words slipped into the sentence. In extreme cases, the model rambled on and lost track of the initial instructions. Out of tens of thousands of attempts, with GPT 3.5, 30% of the responses strictly followed my rules; and with GPT 4, 60% of them. When one visits the Lecsicon web page, the successful word-sentence pairs are typed out letter-by-letter by a program.
In Lecsicon, the resulting sentences offer a broader context for the original words. Many of the sentences perform similarly to Chomsky's “colorless green ideas sleep furiously,” but others, such as “Music: Many unexpected sounds indicate creativity” and “Date: Dinner and theater experience,” seem to flesh out how the units of meaning (letter, word, sentence, paragraph) feed into each other and meaning bleeds through the edges–the fractal character of the English language. Large Language model capture words based on their interrelations through computation processes, like a ghost hunting through a latent vector space. In such a space, words only exist in their associations with other words, sentences emerging from context.
The generated content was then passed through a series of text analysis. The dataset contains 24,393 sentences/200,049 words and among which, these are GPT's favorite words in our acrostics game:
Invented by Martin Wattenberg and Fernanda Viégas in 2007, the Word Tree visualization technique is a visual search tool for unstructured text. It lets you pick a word or phrase and shows you all the different contexts in which it appears. The contexts are arranged in a tree-like branching structure to reveal recurrent themes and phrases.
Below, you can explore how the integrity of 24,393 entries in Lecsicon unfolds in a tree-structure.
Acrostics involve the strategic placement of initial letters to a sentence to convey a message or word that is external to the sentence. Poets from Chaucer to Boccaccio, and from Edgar Allan Poe to Lewis Carroll have used this literary device to add depth, creativity, and intrigue to their work. This technique is particularly effective in shorter forms, such as sonnets or haikus, where each letter carries significant weight. Acrostics enhance the aesthetic appeal of poems, providing an additional layer of complexity through the arrangement of letters. Beyond their puzzle-like nature, acrostics serve as a unique form of self-expression for poets, allowing them to play with language and engage readers in an interactive exploration of the written word.
But the use of acrostics is not limited to poetry. They are often used in the creation of mnemonic devices that help retain large lists of information. Examples include the popular acrostic “My Very Excellent Mom Just Served Us Noodles” used to recall the names of the planets (including Neptune) and “Every Good Boy Deserves Fruit (or Favour in some versions)” used to recall the names of the notes placed on the lines of the treble clef staff. Acrostics are effective as mnemonic devices because they help fix in our minds what psychologist George A. Miller called “the magic number,” the average amount of discrete items the human brain is normally capable of retaining in short-term memory.
Linguistic devices like acrostics captivate me because there is a hidden layer behind putting words together following certain mechanisms. In Lecsicon, the resulting sentences provide a greater context for the original words and reveal new perspectives. It is not a coincidence that GPT makes sentences that relate to the original word's meaning. Instead, it is a deliberate calculation and balancing process to capture words based on their relations to each other. Any word in such a space is nothing but the linguistic associations around it, based on statistical computation of the patterns of the text humans have produced. And perhaps that's why some entries are intriguing and exciting, like this one: "snob: Surely, no one believes".
¡wénrán zhào!
theunthoughts.com
Last Update: Jan 2025
function meetRules(word: string, tokens: string[]): boolean {
if (!word || !tokens.length) {
return false;
}
// filter punctuations
const filteredTokens = tokens.filter(token => !RiTa.isPunct(token));
// first, check if the lengths match after filtering
if (filteredTokens.length !== word.length) {
return false;
}
// compare each letter in the word with
// the initals of each word in the sentence
return word.split('').every((char, index) =>
char === filteredTokens[index][0]
);
}