This is a very simple Crash Blossoms Generator.

Play with the live project, and read more on Medium.

It creates ambiguous pseudo-headlines (in Title Case).

It uses some fun ambiguities that exist in natural languages - specifically one that is pervasive in English:

English is a weakly inflected language, meaning that generally words don't change according to their grammatical position. A lot of meaning is done with word order instead. Other languages change their verbs and nouns to expressing grammatical concepts, making such type of ambiguity less likely to occur.

Reading this article and doing some crazy drawings brought me to the conclusion that I'll try to make Crash Blossoms based on the following:

• focusing on noun-verb ambiguity
• additionally utilizing that both plural nouns and verbs in 3rd person singular end with an -s
• putting two pairs of (potential) composite nouns next to each other
• where the two middle ones should be noun-verb-ambiguous

That allows the reader to flip-flop between two possible interpretations, which is essentially the basic fun of all that.

Of course there's much more to the humor of Crash Blossoms, and this one-day-hack for learning purposes doesn't catch a lot of it.

But, if you're lucky and enter the right words, you can get sentences such as:

Gator Attacks Puzzle Experts

Which sseems to actually be a real headline and I think it's super fun! (at least if you're not a puzzle expert) 😱🐊

1. I checked out a few Crash Blossoms to look for some generalized pattern I'd be able to re-create.

2. I read a bit and drew a bit.

3. Then I distilled these half-actionable rules:

   • Capitalize words to increase possible ambiguity
   • Use nouns that pluralize with -s
   • The inner two need to be noun-verb ambiguous
   • The noun-verb associated with the composite noun needs to end with -s

4. I wrote some HTML and JS that takes word input and sticks it onto the page

5. Only when I saw that all this could work, I went to scrape for Noun-Verbs

   • and used BeautifulSoup to get a list of Noun-Verbs from the net

6. Then I puzzled it together (Constantly looking over my shoulder for that 🐊!)

7. And finally did some mini prettification and put it up on github

• Use Collocation info to make the results actually fun
• Use Sentiment Analysis to filter the happy words from that small list of words I'm using (okay... to practice Sentiment Analysis)
• Add more possibilities for creating different types of Crash Blossoms