Two of the best places to start would be D. A. Carson's Exegetical Fallacies (Baker, 1984), especially chapter 1 on "Word-Study Fallacies" (pp. 25-66) and also Peter Cotterell & Max Turner's Linguistics and Biblical Interpretation (IVP, 1989), especially chapter 4, "The Use and Abuse of Word Studies in Theology" (pp. 106-128). Both of these deal with the problems of older lexical works, though Cotterell & Turner goes into more detail. The chief problem of many older works is a fascination (some would say preoccupation) with etymologizing, often known popularly as "root fallacy," the notion that you can derive the current meaning of a word by breaking it down into its component parts. While this exercise may have historical interest in cases where the actual derivation of the word is known, it amounts to little more than speculation in cases where the derivation is not known. Some common English examples would be breaking down the word "broadcast" into "broad" and "cast," or the word "Watergate" into "water" and "gate," and then thinking that these components have some bearing on the contemporary meaning of the (combined) word. Another major problem with the older reference works is that they simply do not deal with all the discoveries of ancient manuscripts that illustrate first-century word usage in Hellenistic Koine Greek, like the discoveries at Oxyrhynchus in Egypt.

Hall Harris