The language question is actually a good one for Shadow World given that most races are actually bio-engineered races that, at one time, were slaves to the Althan and probably used a common language (probably not Iruaric since the language requires partial telepathic ability for complete fluency). But More than 100000 years of isolation can alter languages radically, and migrations create surprises in the linguisitic map (especially if you forget that languages, most of the time, are cultural constructs, not racial ones).
But barring that, there is evidence (from the Master Atlases) that Iylar, Loar and Erlin languages are very different from one another. Furthermore, there are differences between variants of those languages (Emeri-Elven and Muri-Elven are listed as examples, but I would assume that at least two, and possibly three other Erlin variants would exist on Jaiman alone - one for Urulan of they do not speak Muri-Elven, one for the Mah-Ilari elves of the north (and possibly another, now dead variant for the Jaimani Elves of Lu'Nak), and possibly one for the elves of Remiraith.
I would also assume that Shulur is widely different from the others because of the specific communication needs of the Shuluri. I would assume that most Shuluri speak another language when they want to communicate with strangers (and the language of the Kinsai would be a prime candidate given the proximity, with Erlin - being the closest the world has to a "common tongue" - being a distant second). Loariki is, for all purposes, a fabricated language, and it is probably specific to one Loar culture (I would suspect Namar-Tol). It would be odd if the language spoken by the emerian Dyari was close to Iylar, given the widely different geographical origins of the people.
And elven languages are the easy part, because elves are immortal and language changes would only occur very slowly. Mannish tongues, on the other hand, would diverge from any common root much quicker - and I seem to recall that one of the Master Atlases indicates that there are "at least seven major variants" of Rhaya, the "common mannish tongue" of Jaiman, which is probably a distant relative of the language the Shay spoke before they migrated to Jaiman to escape the yoke of the Masters of Emer in the Second Era. In Tanara alone, there are at least five different spoken tongues - not including Rhaya or Erlin.
The way I resolved that explosion of languages was, as Jdale suggested, to create a "language map" with each language being connected to the languages it was related to, with a number associated with the connection - this number indicated the number of ranks "lost" when trying to use a language as a basis for another. It was interesting to see that, for instance, learning Old Emer (a dead language) was likely efficient as it was connected to several languages (at least four different languages in Haestra, plus one in Reandor) - it was similar to european Latin in a way for most south european languages. It also showed, for instance, that the language spoken by the Zori of Jaiman was completely different from Old Emer (which is the root of all the languages spoken by the Laan in Emer) even though Zori and Laan are basically the same race... but with very different histories and cultural origins.
Overall, the map covered 66 languages and all of Jaiman, Emer and Kelestia, plus 10 magical languages. Of course, it is not canon in any way (my version of Kulthea having drifted from the official one during the twenty years I've been GMing it), and I did not map minor variants, but it helped (my players, mostly).
As a matter of fact, the idea of language map was borrowed from Danger International (HERO system in the modern world), which included one for real-world languages. There was also one in the "Time of the Dragon" supplement for the Dragonlance D&D setting (a setting I used for a RM campaign, incidentally).