I should say that the three stars I am giving this book are well-deserved simply because of the author's meticulous description of Nahuatl grammar. It is evident that he knows the language well and that he has given a great deal of thought to how to best explain its structure. It is also interesting, from the point of view of a linguist, to see the way in which he has organized his grammar. In his own words, he seeks to examine Nahuatl as "a system operating from within its own structuring principles and rules".
There is something very appealing about that idea, as it is a common complaint that the native languages of the Americas are too often analyzed from a biased Western perspective. However, it strikes me as highly presumptuous of the author to assume that his terminology and organizing principles are the only ones that accurately reflect the structure of Nahuatl. In this case, as Andrews' grammar purports to be a learning grammar, indeed an "introduction" to the language, what would seem most important to me is that it give students a firm ability to correctly interpret Nahuatl texts -all that remain of the classical language.
This is precisely where Andrews' text fails, in large part due to his insistence on using his own ponderous terminology. Students will grow accustomed to terms like "verbal nuclear clause", but descriptions like "mainline specific projective causative-object pronoun" must frustrate even someone familiar with Andrews' terminology. The merit in using more standard linguistic terms is that they are readily understood by anyone with a background in linguistics, or anyone who has carefully studied the grammar of another language. I believe it is possible to use standard terms and still accurately describe Nahuatl; both Michel Launey and James Lockhart have written grammars that do just that (Lockhart's is very much an introductory grammar while Launey's is quite complete). This is corroborated by the fact that this second edition of Andrews' grammar, which as he says gives "even more attention to Nahuatl's individuality", is considered inferior to the first edition by every Nahuatl researcher I have spoken to.
Also annoying, especially from a linguist's perspective, is his tendency to want to divide and subdivide every morpheme; many of the divisions he makes are likely etymological and not pertinent to understanding the language, and others feel invented. For instance, he divides object pronouns like "nech" and "mitz" into "n+ech" and "m+itz", where both "ech" and "itz" are objective case markers. Indeed, "ech" and "itz" may be related, but he offers no explanation for why the two variants are used as they are, nor for why they are not expressed in the third person forms "c"/"qui" and "quim". This is roughly equivalent to saying that English "them" is in fact "the+m", where "m" is an objective case marker (also found on "who+m" and "hi+m"), which is not expressed in first and second person forms like "me" and "us". No linguist would actually propose such a rule for English, yet it is just what Andrews does for Nahuatl. Here, his attention to detail becomes overanalysis.
If you are considering buying this book because you would like to begin to study Classical Nahuatl, I would instead recommend Lockhart's "Nahuatl as Written", which is very easy to come by and is also much less expensive than this book, or Launey's grammar, which can be found in both French and Spanish editions but which is harder to find in the U.S. For the serious researcher of Nahuatl, Andrews' book is useful to have as a reference because of the author's thoroughness and attention to detail, but by all accounts the 1975 edition is the better of the two.