There are (at least) two types of data-oriented linguists: those who go for data across lots of languages and those who go for lots of data within a single language. David Fertig is clearly a linguist of the second type, as he has put together a masterful and painstakingly detailed study of verbal inflection in the German dialect of Nuremberg in the period between 1356 and 1619, thus based on data which Fertig characterizes as “drawn from a single local variety of a single language” (p. 1). The data for this study come from a collection of texts Fertig assembled consisting of letters, journals, diaries, reports, treatises, bookkeeping records, and protocols, and for all of the items included, a fairly accurate dating was possible, as was the identification of the author.

While Fertig gives a remarkably in-depth description of verbal inflection in his well-defined corpus, his goals are not (merely) descriptive in nature. In fact, as he states his aims, they are “to build a theoretical investigation of morphological change on a solid empirical foundation” (p. 1). Clearly, as the above account of the corpus indicates, the empirical foundation he works with is solid, and then some! Fertig is aware of the fact that one can sometimes drown, as it were, in too much data (see on this point Lass 1997, who, as Klein (1999:88–89) puts it, seems to believe that “despite our interest in taking into account as much data as possible in applying the comparative method, too much data can sometimes be a hindrance in that it may muddle the picture by making it harder to know what forms to take as input to the method.”). Yet, Fertig realizes, wisely I would say, that a rich database is perhaps the only way that the historical linguist can overcome the basic problem faced by those engaged in analyzing language history, stated by Labov (1972:100) as the need “to make the best of […] bad data—‘bad’ in the sense that it may be fragmentary, corrupted, or many times removed from the actual production of native speakers.” As a result, we are treated here to a study with an exhaustive basis—Fertig included in the database “every token of every verb that occurs in this collection of nearly one-half million words, about 86,000 tokens in all” (p. 1)—and with a foundation like this, it is fair to say that any theoretical conclusions Fertig reaches inspire confidence.

After two brief introductory chapters about the overall goals and the nature of the corpus, Fertig lays out in chapter 3 his views about language change and especially morphological change. This chapter is well thought out, and contains some provocative and downright iconoclastic ideas. Rejecting the semiotic principle of “one-form-to-one-meaning” as a viable principle of morphology, Fertig argues for the SEPARATION HYPOTHESIS and suggests that “indirect, conditional, non- one-to-one mapping between function (or meaning) and form is [to be] regarded as normal and expected in morphology” (p. 16). He goes on to take issue with the importance that some (see, for example, Bybee 1985) have placed on diagrammatic iconicity (involving the extent to which fusion of a stem with an inflectional marker reflects the relative semantic relevance of each piece to the lexical item's meaning, to dispute the rareness of exaptation (Lass 1990's term for the reuse of linguistic “detruitus” by speakers in novel yet rational ways), to reassess the relationship of analogy and rules, and to emend the definition of paradigm leveling to “the paradigm-internally motivated elimination of an allomorphic stem alternation” (p. 32). This last point is especially important since, as he notes (p. 31), “a very high proportion of the changes in verbal inflection observable in the Nuremberg texts involve [sic] what is traditionally referred to as analogical or paradigm(atic) leveling.” He also recognizes the importance of blends and hypercorrection (which he sees as “a kind of analogical development”, p. 37) for the data in his corpus, but is doubtful about traditional typologies of analogical change.

At this point in the book Fertig takes me to task for my lumping “all types of ‘change due to the influence of one form on another’ […] together under the heading ‘analogy’” (in Joseph 1998:362), saying that such “a practice […] has led to the frequent criticism of analogy as a ‘catchall’ term for processes that do not really have anything interesting in common” (p. 36). Let me offer, as a brief excursus, the following defense of my approach. My claim is that various changes, including such traditionally recognized phenomena as paradigm leveling, form-class (external) analogy, contamination (blending), re-compounding (renewal), reanalysis, and even folk etymology, do have much in common. For example, they are typically sporadic (as opposed to the regularity of sound change), typically show the involvement of some other form and a perceived relation to that other form (whereas sound change is impervious to such perceived relations), and typically are embedded in some grammatical subsystem of the language (whereas sound change is blind to grammatical involvement). We might add as well that these characteristics show these changes to have a psychological/cognitive grounding, whereas sound change can be seen as strictly phonetic and physiological in nature. This approach may indeed represent the defining of analogy “negatively as whatever is not sound change, semantic change, or borrowing,” a practice Fertig is clearly not impressed by, but so be it—to me, these characteristics represent valid properties of this otherwise disparate assortment of changes and show how they cohere as a group as opposed to other classes of changes with their own motivating factors (such as physiological ones, in the case of sound change, or social ones, as in the case of borrowing and diffusion more generally).

To return to Fertig's study, chapters 4 through 7 constitute the core of the presentation and analysis of the data. Chapter 4 offers first a detailed look at the inflectional endings, organized by person and number, with a special section on the 1st and 3rd plural forms of the present of sein ‘be’, whereas chapter 5 treats stem alternations. In this latter chapter, the organizing principle is essentially strong versus weak verbs, with a separate treatment of the modals, wissen ‘know’ and tun ‘do’, with discussion of stem-final consonant alternations as well (for instance, Verner's Law alternations). The theoretical point to be drawn from chapter 5 has to do with directionality in leveling, and Fertig finds that the notions of “local markedness and relative token frequency” (espoused, among others, by Tiersma 1982 and Bybee 1994) do “not appear to be applicable to the cases of leveling in verbal morphology” seen in the Nuremberg corpus (p. 107). Chapter 6 examines shifts that some verbs show between inflectional classes, a phenomenon that provides an interesting test—and to some extent confirmation—of the theory of inflectional class stability advocated by Wurzel (1984) wherein the importance of the stem vowel in determining inflectional class membership was stressed. Chapter 7 focuses on the ge- participial prefix, clarifying, among other things, some aspects of the absence of this prefix where it might otherwise be expected.

In his concluding chapter, Fertig makes an important nod in the direction of sociolinguistics and variation and how change is to be understood in this context. Given the acuity of his remarks throughout the book, and the general reasonableness of what he says in this chapter, with its reference to usage differences seen in “the opposition between the chancery and administrators, on the one hand, and the women, on the other” (p. 144), one can only regret that this chapter is so short, a mere four and a half pages! Healthy iconoclasm is found here too, as Fertig concludes, following Labov 1989, that while there is a relation between variation and change, “the earlier view equating variation with change in progress is now obsolete” (p. 147).

This work is clearly written and very cleanly produced, with only a few typographical errors at most. There is no index—a minor failing— but Fertig does include an extensive (17-page!) bibliography, and three appendices, one a ten-page verb frequency list giving the frequency of occurrence for every verb in the corpus (not surprisingly, sein ‘be’ and haben ‘have’ are the two most common, followed by werden ‘become’ wollen ‘want’, sollen ‘ought to’, lassen ‘let’, kommen ‘come’, tun ‘do’, and schreiben ‘write’, to round out the top ten), one a two-page sampling of lines from his data tables, and the last a two-page listing of sources for the texts in his collection. All in all, this is a most satisfying contribution to our understanding both of the development of German verbal inflection and of the nature of morphological change. Despite the appeal of broad cross-linguistic surveys, real progress in our field is made, I would say, with fine-grained exhaustive studies of the sort that this excellent and provocative work represents.