As time goes on in our age of increasing reliance on computerized kanji input, this question may become increasingly irrelevant, but when I'm writing a sentence with (gasp!) pen and paper, I have always struggled to remember where the dividing line between kanji and okurigana lies with each character. Is there any way (aside from brute-force memorization, which I'm attempting at the moment) to learn and reliably recall okurigana? Are there any patterns I can rely on?
Other than brute-force memorization (棒暗記), the only thing I can suggest is material regarding the Kanji-Kentei (漢検), because I know some of the (lower?) levels focus on 送り仮名. Some materials I have are books of tests from previous years (問題集), and a Nintendo DS 漢検 game. However, I got all of this in Japan, so I don't know how accessible this kind of stuff would be for you (Amazon.co.jp, have friends from Japan send you stuff, etc.) There's always Google to maybe search for some online 送り仮名 quizzes.
For い-adjective, of course the い is always 送り仮名. There are also the 〜やか／〜らか type adjectives, where that suffix is always 送り仮名 as well.
Learning verbs by their dictionary form will also help. Note however, that there are often times where multiple forms of 送り仮名 are accepted. For example, the verb とらわれる can be written both as 捕らわれる and 捕われる. This can throw a kink into some verbs, but I think the ratio of multiple-accepted-送り仮名 verbs to one-accepted-送り仮名 is probably pretty low.
Since the kanji part of both verbs and adjective is the part conveying the root meaning of this particular adjective form, the okurigana (i.e. everything that follows the part written in kanji) is rarely composed of anything else except for derivation and conjugation suffixes.
Conjugation suffixes are easy enough, and I doubt they pose any problem for you. You know that the
たかった in 見たかった is the part that changes by conjugation, it definitely has to be part of the okurigana. All other forms of the verb 見る wouldn't give you any problem either.
I guess your problem really begins when you get to verbs like 変える and 帰る. Both look essentially the same in their Rentaikei (base) forms, and in both only the last syllable (る) changes - so why is the え in 変える spelled out in the okurigana, while it is considered part of the Kanji in 帰る?
The answer for that is that the え here is part of the derivational suffix. If we look at the verb's history, then it's original stem was actually
kap-, from which both
kaperu (written かへる in old spelling and かえる today) and
kaparu (written かはる in old spelling and かわる today) are derived. As you can see, two different derivational suffixes have been attached to this verbal root to create the transitive (
kaperu) and intransitive (
kaparu) versions of the verb.
These origin of these derivational suffixes may be a bit obscure nowadays, but it's still easy to recognize the difference between 変える and 帰る by just comparing them to their counterpart verbs written with the same kanji: 変わる and 帰す. You can see that in 変わる the two last syllables change and in 帰す only the last syllable changes. That's why writing down the derivational suffixes in okurigana is still relevant even today: because these parts of the verb still change.
Now, it should be noted here (and it was already noted by istrasci, in fact) that some verbs have more than one acceptable okurigana. Looking at older texts, it seems to me that the standardization of okurigana is actually quite recent, so we naturally have quite a few exceptions, whether they are officially (or semi-officialy) accepted or not. But these exceptions to the rule shouldn't bother you most of the time, since all it really means is that you have an extra form, for which the derivational suffixes aren't fully spelled out in okurigana (e.g. 捕われる, which derives from 捕る) - the other form (with fully spelled out derivational suffixes) is still accepted, and usually it's also the more common one.
As for い-"adjectives", they are a bit easier (like istrasci has noted), since they are less often derived, but you should pay attention to the cases where such adjectives are derived. They're usually easy to detect because of the ending: adjectives ending with ない (such as 少ない and もったいない/勿体ない, but not 汚い), たい (重たい) and しい (悲しい、親しい).
The gist: There is a kind of computerized kanji input system that puts you back in charge of which letter to write in Kanji and which one in Kana. Namely, SKK. With SKK, you can actively learn okurigana rules even when writing text on the computer.
Basically, SKK converts words one by one (single-word conversion), without analysing syntax or grammar. Instead, users specify the border between Kanjis and Kanas. This means you can write dialects, written or spoken words, ancient words as well as standard words in the same manner. (From SKK Features)
The disclaimer: While I have been using SKK for 8 years now, I haven't intentionally used it for the purpose of learning okurigana. So I can't testify that SKK will improve your okurigana skills, nor do I have any study data on that matter.
The why: That said, SKK gives you the environment to actively and constantly engage in the act of deciding the okurigana, which can help you get better with it. This is just like pen-and-paper writing on the computer, but with instant feedback from the system.
a → あ
a, → あ、
a,O → あ、▽お
a,Oku → あ、▽おく
a,OkuR → あ、▽おく*r
a,OkuRi → あ、▼送り
a,OkuRiG → あ、送り▽g
a,OkuRiGa → あ、送り▽が
a,OkuRiGana → あ、送り▽がな
a,OkuRiGanaSpace → あ、送り▼仮名
a,OkuRiGanaSpaceEnter → あ、送り仮名
AquaSKK's "making the switch" article (ja) also has a good fleshed out example.
The caveat: SKK's okurigana dictionary lists extra forms alongside with standardized forms. If you want to restrict the acceptable okuriganas to standardized forms only, you need to find or create a dictionary file with extra forms weeded out.
Where to get it: There are various versions of SKK for several operating systems. You can check out the Japanese Wikipedia article for a comprehensive list.