Second paragraph of third chapter:
There are two obvious reasons why minilectal approaches cannot explain or predict, why the fundamental concept of naturalness (see later) is so vague and shifting, and why linguistics faces something of a philosophical crisis. First, it is clear that explaining and predicting depend on the study of developments that produce structures: developments explain states, not vice versa. Of course, a given combination of existing structures is a conditional cause for what changes are possible, but the efficacious causes for linguistic changes must be sought elsewhere. This brings us to the second reason why static, or minilectal, approaches can never attain to the theoretical goals of explaining and predicting, namely, the fact that they are autonomous approaches: they prefer inert 'formal' explanations (if any) or economy metrics to developmental explanations – those based on how new languages get formed (from pidgins) and develop from then on – the sort of explanation made use of in cosmology and astrophysics. (Development is of course excluded from the purview of status approaches.)
I got this book because I knew Bailey briefly when I was much younger, and really admired his charm and his incomprehensible expertise in linguistics. He kindly taught me the rudiments of New Testament Greek. This book pulls together his own choice of his best academic pieces, which unfortunately means that they are pretty impenetrable to those not familiar with the internal arguments of linguists over the last forty years; it's clear though that he was a bit of an outlier (he's till alive, but long retired) and that his particular axe to grind is that linguists pay too little attention to language change as a dynamic, perpetual phenomenon, driven both by internal shifts and external stimuli.
The two most accessible chapters are fairly near the end. One sets out his argument that English should not be described as a Germanic language. He argues that all languages have more than one parent anyway, so the traditional "family tree" model is simply wrong. Much better, he says, to think of Middle English as a creole of Middle French and Anglo-Saxon. The vocabulary may be Germanic, but the grammar is much more Romance.
I find this a provocative and not completely convincing argument. I think there's a good reason why we instinctively tend to group languages together by vocabulary rather than grammatical structure, and it's not just oppressive Saussurian propaganda. I agree that there are indeed some intriguing underlying geographies – to take two examples that Bailey doesn't use, the definite article suffix which is shared by Albanian, Bulgarian, Macedonian and Romanian, but by none of their neighbours; and the English "-ing" and "do" constructions which appear to have come from Celtic. But Bulgarians and Macedonians still find it easier to communicate with article-less Serbs and Russians, while Romanians find it easier to communicate with article-before-noun Italians, and even English speakers will find more echoes of their own language in Dutch and German than in Irish or Welsh. So I'll happily concede that we need a more nuanced picture, but I think the traditional family structure does tell us something useful.
The next chapter, much more digestibly, argues simply that the Greek letter xi – uppercase Ξ, lowercase ξ – was sometimes pronounced "sh" /ʃ/ rather than "x" /ks/. Appreciating the evidence in depth would take more knowledge of ancient Greek than I have, but Bailey has me convinced at "Xerxes". The Persian ruler's name is spelt Ξέρξης in ancient Greek, but the Persian original was something like Xšaya-ṛšā, where that first X is a chi sound (as in "loch") rather than X as in "axe". The consonant after the "r" or "ṛ" is pronounced "sh" in Ancient Persian, also Modern Persian ش (both the second and second last letter in خشایارشا), also Hebrew שׁ (third and last in אֲחַשְׁוֵרוֹשׁ) and ς (usually "s" rather than "sh") in the Book of Esther (Ασουηρος) – not a hint of a "k" sound in any of them. NB also that in the Latin alphabet, "x" is routinely pronounced /ʃ/ in Basque, Catalan, Galician, Portuguese and Maltese. Given the weight of the evidence I was left wondering if this was really controversial at all.
Anyway, I'm glad I read this even if I didn't understand as much as I would have liked.