Word Search Cipher

Question

I have come up with a cipher idea that probably isn't original but I can't find anywhere, and I wanted to get an idea of just how hard it is by seeing if anyone here can crack it. I would especially like to know whether there's any realistic chance of breaking it with a pen and paper as I would like to use it for a codes and ciphers club I run, but if you can't, please do try to crack it with whatever program you can find.

Here's the concept:

1. It uses a basic substituted alphabet, which in this case will be shown as the Hymmnos script. You could try to refer to the official version of that but it won't help as the symbols don't correspond to the same letters.

2. The sender builds a word search using a pool of words that the receiver knows and expects, and hides a secret message between them. Not every word can be fit but the words are ordered in some objective way. They are thematically related so solving one word will make it possible to solve the rest in the pool.

3. To decipher it, the receiver goes through the pool of words completing the word search until the remaining letters spell a coherent message

First method: Look for sequences that are in both messages (including backwards, vertically, and diagonally)... with enough words, you may be able to guess what letters the symbols represent.

Second method: I made a shameful mistake with the second message... I moved the letters in the wrong direction through the alphabet before changing the font. Which means you can now compare the wrong and right versions of the second message to the official Hymmnos script in order to work out exactly what each letter is.

Massive Hint:

The list of words are Pokemon ordered by national dex no.

Hint 2:

Of Weedle and its evolutions, one didn't fit

First message:

Second message:

Second message with each letter shifted the wrong way through the alphabet relative to the official Hymmnos script:

Answer 1

Partial answer:

Here's the boards with each symbol being given a letter (so it's easier to write and differentiate), the number of times it appears, and the boards written using the assigned letters. Now that we have a second board, the frequencies change (frustratingly) much - note that this is because this translations was done before the mis-shifting of the second message was discovered:

The boards as text:

ABCDEFFGAHIACCD
JJKLMDFAKINJFMO
JFFKKPQNGNQFMFR
CPEGDDHSDIBGTEC
RFQAUFGGKIDDCIF
EPFVCITPCBTWJAK
MPJFPINDNFVFGJG
IFKPINQIMIIGIFK
IGKHQPMDQVQTXGF
KHNFCGGCVMPOFUC
LFYFUQTQBFPIUFK
QGNPTPQMDGQTDMM
THIDBGIPFAMDPTI
DQMJCFNCTMFCWIO
MEFUQXHANISFDGR

AKWPLCKPWAYASFO
YBOBAPKQKYGIPHO
UAQLYZAWZVEOBPW
OFCAWYZAKPXAKLS
HFEFFAKOKKSNSWA
GKAOOUUFOFHEAPW
FPFYOBBKCZOHWLL
ZOLBPOFPAAKELCS
FFOVZFSKAZFBSIZ
HCOAOOAFCSYHBEK
WPWSPCLKOFKAAOO
OOKNKOFFEHKSSOM
OUOPGKOFAYAAZWP
JOWESAEKCWSEMFZ
BBPKLOOHOSAKUOF

We might take the frequency of the first board and assign each symbol a letter based on the frequency in English (where the order is ETAOINSRHDLUCMFYWGPBVKXQJZ), but there are too many collisions for this to be a usable solution strategy (you can check the edits to find a visual board if you want to try anyway)

---

Now that we know (almost) what the word search contains, I had another go with a different technique:

Given that the words are Pokemons sorted by national dex number, the grids are likely to contain both "WEEDLE" (#13) and "BEEDRILL" (#15). As these share a lot of letters, and are somewhat easy to find thanks to their double letters, they should make it possible to find the first symbols that correspond to at least "E", "D" and "L". Unfortunately there is no solution where the two words share symbols.