To get an idea of the fics' distinguishing words, I started off by calculating the term frequency-inverse document frequency for each of them. This required that I collect some other fanfics to use as a background for comparison, so I grabbed 31 T-raged gen 'fics of >10000 words off AO3 to add to the mix.

Term frequency-inverse document frequency tries to pick out the most informative words in the document, those that appear with high frequency in a given document and are rarely found in other documents. This means that proper names tend to pop out like crazy, and I did some of these clouds early in the morning, so you'll note that my filtering of them was, in places, a bit spotty. Mostly I didn't bother with any that crept in at the smallest text sizes anyway. Pokémon species were difficult for me to call; at times they double as proper names, after all, but I thought it was to some extent interesting to see which species you featured that other stories tend not to. I tried to only remove species that were either on one of the MC's teams or plot-important, which showed up with higher frequency just because characters were referring to them a lot, and their prevolutions. Which, let me tell you, was still plenty. Did you realize your 'fic has loads and loads of characters?

I liked the TFIDF representation a lot better than the raw frequencies, but it did bug me that it pretty much selected for relatively rare words above all else. It would tell you words you used a lot that basically nobody else did, but it wouldn't tell you about any common-ish terms that most people use, but you happen to use much more frequently than the norm. To try and get at that I computed a crude "enrichment score," which is just the frequency of a word divided by the total number of words in the document, then divided by the per-word frequency of all other documents in the dataset. This didn't make for quite such nice word clouds, since the numbers ended up having a lot less variance even after some rescaling, but I think you can see some interesting things by putting the two representations side by side.

To begin, the clouds for The Quest for the Legends:

I think you'll agree that this representation shows off more interesting words, as well as giving a much better idea of the story's themes. TFIDF representation is on the left, enrichment on the right.

You can definitely see Scyther's influence on the TFIDF cloud, even though his actual name was removed. Many of the other terms are either battle-related or have to do with places/things/concepts unique to Ouen. Also, you kind of like dragons I guess?

I was surprised "said" popped up like that in the enrichment cloud, but then I guess it's a word you don't really notice, even when the author's using it a lot. Here, the Ouen influence is a lot less, so you get some interesting concept words coming to the fore instead, like "pain," which doesn't appear in the TFIDF cloud but does show a fair amount of enrichment (it's just to the right of "attack."

Note that in both cases I had to remove "pokémon" from the word list because it otherwise completely swamped everything else out. I don't really know how to explain that, since "pokémon" is obviously a fairly common word in the other stories as well, so I'm not sure how you're managing to use it that much more often than them.

Then Morphic:

TFIDF on the left again. "Dully" is one of the more interesting words out of the overrepresented ones to me; lots of people feeling dead inside, eh? I also like the abundance of words indicating a more "mundane" setting that you don't normally see in pokéfic, like cellphone, sidewalk, bus, etc. Also, this might be the only story in the sample to give slugma the time of day. Meanwhile, the enrichment plot does suggest that you use "looked" a lot more than the average. I'm somewhat intrigued by your apparent fascination with doors, personally.

As you can see, something about the way é was encoded in the version of Morphic on your site caused the character to get stripped out during the preprocessing step, hence the majestic "pokmorph," "pokmon," and so on. I think taking a look at the differences between their representation in the two clouds really emphasizes the differences between what the two measures can tell you. This is probably the only story in the corpus where that error occurred, so the words' importance gets boosted a ton in the TFIDF representation. They're more subdued in the enrichment cloud because they don't actually come up all that often in the text. Also telling, I think, of how different this is from a "standard" pokéfic.

And I definitely made clouds for my own fanfic, too. As for cursing, there's nowhere near what I think you'd expect.

I mean, yes, there's that giant "fuck" right at the heart of everything in the TFIDF version, but I was expecting a whole bunch of collateral with things like "shit," "asshole," and so on making up the majority of the other top hits, but it turns out they aren't really that prevalent. I'm a little disquieted by the way "tentacles" shows up prominently in the TFIDF cloud; I guess it's clear that I filtered out the explicit 'fic, then. And then I have absolutely no idea how "can" is coming through so strongly in the enrichment cloud; a can-do sort of attitude really does not pervade the story, and I can't think of how I'm managing to hit it so often. "You're" is also interesting to me as the only "you" variant that shows up prominently and also a word I'd expect to come up fairly often in dialogue in other stories. I'm wondering whether it's high mainly because it's acting as a state-of-being verb ("you're thinking...") or because it's associated with characters yelling at each other ("you're so X! You're so Y!").

Unfortunately I think these visualizations are a bit less useful for this story because it's written in present tense, and evidently the stemmer doesn't handle collapsing different verb forms into their root. The fact that I'm getting so many verbs coming out in these clouds, whereas you're much heavier on the nouns and adverbs, would be interesting, except that I think it's probably just an artifact of my story having, say, "leaps," where others would have "leaped," and those getting counted as two different words. All in all I think this makes my clouds a bit less descriptive; the ones for your stories give more of an indication of what they're about (since stories tend to be about "things"), while mine are more about the kinds of things that get done.

I do like the tiny little "lies" that shows up near the middle of the TFIDF cloud, though. Very pertinent.

And I have a confession to make. As with "pokéball" for TQftL, there was one word that I had to prune out of both these clouds...