That could work too. In both cases you get the word being formed in the spoken language, and then interfering on the spelling only afterwards. The difference is if defining the word syntactically (like I did) or phonologically (like your reasoning leads to).

[Kind of off-topic trivia, but for funzies] I've seen similar phenomena in other languages, like:

• Italian - "per questo" (thus, therefore; lit. "for this") vs. *perquesto
• Portuguese - "por que" (why; lit. "for what") vs. "porque" (because)

Both of our explanations would work fine for those two too, mind you; they both sound like unitary words and behave as such. (e.g. they repel syntactical intrusion).