It's not all that easy to compare though -- I'd say when you are looking at a small computer, just how small and light it is is very important. The Asus is 20% heavier than the macbook air (3.7lbs vs 3.0lbs).

Huh, and here was me thinking that it is the fourth most visited site on the web. But if you personally don't interact with those visitors, I guess they aren't important.

Surely a scone is a bread, albeit leavened with soda rather than a yeast...

It shouldn't matter what the name is, but who the person is and where they intend to do the work (i.e., what university they are at) are very important as the person needs to prove that they can achieve what they propose. So the reviewers can't be blinded from these facts. In this respect, a grant is nothing like a peer-reviewed article.

Speaking as a statistics PhD, I can say that the correlation he finds between distance and phoneme diversity is pretty robust -- real data from complex systems is always going to be noisy. The model used could probably be improved (eg using a definition of distance that better reflected migration routes) but that would require making a lot more assumptions.

Cherry-picking extreme cases does not prove that a trend does not exist. But, as argued in the paper, the Austronesian languages could be expected to see a very strong founder effect because of the island-hopping style of migration. This has been studied extensively and there is genetic and cultural evidence to suggest founder effects.

This is evidence that supports the hypothesis. It doesn't prove it, by any means, but it is consistent with the hypothesis.

You may well be right that it is not a strict founder effect in action here in the sense that there weren't some phonemes that go left off the boat, as it were. But it is quite possible that there was some other cultural force in effect as new colony established themselves causing the same bottle-necking that is seen with a strict founder effect.

I think linguists often get lost in the complexity of the individual systems they study (often to a very impressive level) and are unable to step back with a study like this and see a broader trend. Indeed, a very broad and quite weak trend like this probably doesn't really tell a linguist much if they are interested in relationships between languages of the past few thousand years because it has very little predictive power. But I see it as quite a nice use of statistical reasoning in that it ignores a lot of the details, simplifying the data down to a point where it is easy to work with but still contains some signal that can reasonably be interpreted.

There is no argument by Atkinson that all language change is monotonic or predictable. But the observation that, on average, phoneme diversity increases with speaker population size and decreases with distance from Africa both seem true. These observations support the hypothesis that language spread out from a single origin in Africa , going through multiple bottlenecks that left this specific pattern of change. Of course, this is only one small part of the difference between languages and the signal is not without noise. It also fits in with broader theories that language was one of the things that allowed early humans to greatly expand their range. Are you arguing that people left Africa without language and it developed independently in multiple locations? By the way, your Hawaiian example is a classic of the founder effect seen throughout the Pacific.

Founder effects can occur with large groups, too. Think of two towns that speak the same language but have some noticeable variation between the two, for example, in one town they use the voiceless dental plosive, the other they use both the interdental fricative and the voiceless dental plosive. If these towns become isolated from each other evolving distinct languages, one will have fewer phonemes than the other. This is the founder effect.

I'm not sure sure that your theory is even correct there. It is quite hard to make a lot of different sounds -- other great apes have far less vocal ability than we do -- whereas it is easy to convey lots of different meaning by rearranging just a few sounds.

It's not my theory, it is the theory they use in the paper. Whether this is the cause or not it is hard to say. But the trend is apparent and this seems a reasonable explanation. Also, it is just a trend -- the graph in the linked article shows there is huge variation around the trend so counter-examples of the type you point out are many.

The number of phonemes in a language has nothing to do with intelligence. In theory, the more modern languages have fewer phonemes because of the "founder effect". If you think about this in terms of vocabulary, it is obvious -- no-one knows all the words in any language, so if a small group set off to start their own colony, the language of that colony won't have the words that none of the founders knew. New words may be invented to substitute for the missing words but they will be different. It is the same with sounds (and genetic diversity, where this was first observed). Since new sound formation is a very slow process, the signal remains for a long time.

This is crucial --- cable would have nothing if it weren't for sports.

It was conceived as a "porous wall". You can still read it as a casual user without bypassing anything.

Uh, the value they are charging for is the reporting they do with a staff of about 2000 journalists. A lot of people value that and quite rightly so.

It has little to do with road repair -- damage to roads is primarily done by heavy vehicles. A small car does virtually no damage to a well-made road.