Courtesy of Mulboyne, a story at Asahi.com entitled ""Samu!", "Sugo!" spread rapidly in recent years: Japanese usage opinion poll (「寒っ」「すごっ」数年で急速に広がる　国語世論調査). (Similar article at Mainichi.) Seems that the 2010 Agency for Cultural Affairs opinion poll on Japanese usage found rapidly increasing acceptance of adjectives used in stem-only form (or rather, stem + /Q/, like samu! rather than samui, mijika! rather than mijikai.

The interesting thing is that different words are accepted at different rates. 85% of people either use samu! or are not bothered when others do, and this is what you'd expect since it's been appearing in books for more than a century; sugo! and urusa! and so on have lower acceptance rates, and I suppose the implication is that this is because they didn't appear in Edo literature (or didn't appear as often, or whatever). The logical conclusion is that stemifying adjectives is currently a relatively limited process but is gradually becoming productive for more and more people. (I am one of the relatively youthful offenders who sees nothing wrong with this pattern at all.)

Also uncovered in the survey:

• About half of Japanese people were not aware that endangered languages/dialects could be found in Japan (which is kind of baffling); having been made aware of this fact, 49.6% of those surveyed agreed that measures should be taken to prevent their disappearance, 17.8% thought it couldn't be helped and nothing should be done, while 29.5% declined to decide either way.
• 31.9% of respondents think that official government documents should use the Japanese-style comma "、" even in horizontally written text, while 30.0% are OK with using the "," (which is the current official style).
• 23.3% of respondents do not think it a good thing that English is a global lingua franca, but don't see what is to be done about it.
• When speaking to recently arrived non-Japanese folk, 61.2% of respondents try to speak slowly; 57.8% choose easy words; and 48.8% use gestures and diagrams. 6.8% mix in words from their interlocutor's native language. Only 3.1% speak exactly as they usually would.
• Fully three quarters of respondents feel that romanization should represent the difference between long and short vowels (so that, for example, 大野 and 小野 don't both end up as "Ono"). This is up from 70.8% ten years ago. Similarly, 56.9% are in favor of writing "Kōbe" and 53.0% in favor of "Ōsaka", up from 40.1% and 38.0% ten years ago. Long-vowel consciousness is rising in Japan. Interestingly, 10.8% of people are in favor of "Kobe" while 19.6% are in favor of "Osaka"; the reason for this difference appears to be because 11.6% are in favor of "Koube" (which is not an option for Osaka because its long vowel is two おs, while Kobe's is おう). Only 1.5% are in favor of Koobe, while 6.7% are in favor of Oosaka. This is actually a really fascinating topic that I would like to read more research about.
• Ranuki is proceeding apace, with the younger generation leading the way. Interestingly, though, kangaerenai is 80% rejected even by the yoof, which makes it much rarer than taberenai, korenai, etc. (which have clear majorities of the under-30 segment). Perhaps ranuki is somehow blocked if the last mora in the stem has no consonant? (This had never occurred to me before, even though I too would say taberenai and korenai but never kangaerenai.)

Thanks Mulboyne!