Almost forty years ago, Neil Smith published a diary study focused on the phonological development of his son Amahl (Smith Reference Smith1973). In addition to supplying the full list of the pronunciations of all identifiable words that Amahl used during the study, recording each new pronunciation as it appeared (a database that pre-dated, but is now available electronically through, CHILDES), Smith provided extensive analyses, giving realization rules to account for how the child's pronunciations differed from those of adults, and described how those pronunciations and rules changed during development. A few of the reported phenomena held particular interest (for example, the puzzle–puddle phenomenon; and the use of a default syllable /ɹiː/ in place of target initial unstressed syllables), especially since they have turned out to be of relatively low frequency across children. Smith also discussed the ramifications of the data for theories of language acquisition and for theories of phonology. The book had an immense impact, and is still heavily cited today. The database has frequently been mined by other researchers.

Smith has now published a sequel: a study of the phonological development of Amahl's son Zachary (Zak), over a similar age range, with similar methodology, and with similar analysis and discussion. As before, there is an appendix providing the child's pronunciation of all identifiable words, at different ages, along with an analysis. To facilitate comparison between the two children, the differences between the child and adult pronunciations are presented using rules in the style of the early 1970s. The theoretical discussion addresses changes in theory over the past forty years, and ramifications of the new (and the old) data for more recent theories.

The major attraction of this book is the cross-generation aspect: we see detailed data about the phonological development of two male children in succeeding generations. This is not a common thing. Most studies involving biologically related individuals involve two siblings (either separated by a few years or twins). Felsenfeld, McGue & Broen (Reference Felsenfeld, McGue and Broen1995) presented a cross-generational study on language disorders. They contacted children who had been identified as having protracted phonological development in a large study begun in the 1950s, and showed that the children of these children almost invariably had protracted phonological development (whereas the typically-developing children from the previous study almost invariably had typically-developing children). However, Felsenfeld, McGue & Brown's study did not compare the particular phonological patterns of each parent–child dyad; problems with phonological development occurred in both generations, but we do not know whether the particular difficulties were the same.

The cross-generational aspect of Smith's study is an attention-getter, but how much can it really tell us? Studies have shown that there can be large differences in the phonological patterns of siblings who share 25% of genes (see, for example, Dodd & McEvoy Reference Dodd and McEvoy1994); is it to be expected that sharing 50% of genes (as between father and son) will be that different? To evaluate that possibility, we would need (a) many more parent–child dyads, to show statistically that they are closer than expected by chance, or (b) a detailed comparison of all the phonological patterns of Amahl and Zak with the patterns discussed in the literature, to establish whether there were more processes in common than by chance, or (c) the presence of the same extremely uncommon patterns in both Amahl and Zak. The latter in particular might be considered compelling evidence. For example, word-initial unstressed syllables are subject to many different patterns of reduction across children and languages (including weak syllable deletion, initial consonant deletion, reduplication, and extra reduction of clusters or features as compared to initial stressed syllables), but there have been only two reports of the use of a default syllable to replace most adult target initial unstressed syllables: Amahl in Smith (Reference Smith1973), and the child in Gnanadesikan (Reference Gnanadesikan, Kager, Pater and Zonneveld2004). If such default syllables had occurred in Zak's speech, it is unlikely that it could be a coincidence. But such was not the case. Smith rarely comments on whether a given pattern is common or uncommon across children, and does not attempt an argument that Amahl and Zak have more-than-chance-related developmental patterns. While adding to the appeal of the book, the cross-generational aspect does not add much to our knowledge of phonological development.

Smith is aware that the choice of doing a diary study is a controversial one. In addition to making this study comparable to the previous book, his choice exploits the many strengths of diary studies. As his grandfather, Smith had frequent access to Zak, and was able to observe Zak's speech in 154 sessions over a period of 33 months. This is a level of data density that surpasses almost all studies involving recordings, giving much finer temporal information about changes in the system. (It is, however, less dense than the diary study of Amahl, where Smith as the father interacted with the child daily, and could potentially notice each new change on the day when it began.) Since a diarist perforce has to keep up with changes as they occur, Smith was uniquely able to focus on words that were in the process of change when collecting data. On the negative side, his data are more qualitative than quantitative, and no acoustic analyses were done. As Smith notes, any subtle, covert contrasts would not have been detectable. Some researchers focus on covert contrasts as the most important phenomenon of phonological development, suggesting that it renders all less-detailed analyses of dubious value. Other researchers suggest that covert contrasts provide more finely detailed information, but do not contradict analyses based on coarser, perception-based phonetic analysis. Smith did tape-record Zak's speech to some degree, and carried out useful perceptual tasks, showing that Zak, when listening to recordings of his own pronunciations, could not unambiguously identify which adult word was intended (contra a claim that has occasionally been made). Nevertheless, anyone who has already decided that the methodology used in Smith (Reference Smith1973) renders the data questionable will not find this book any more attractive. Those who feel that the strengths of the diary study methodology outweigh the limitations will find this book to be as useful as Smith's previous study of Amahl.

Chapter 4, ‘Z and his background’, is in many ways the heart of the book, along with the database in the appendix. The chapter presents a full analysis of Zak's phonological patterns, with changes over time. There are fourteen stages covered in just under fifty pages, with each stage presented in a comparable format. In addition to information on phonological development, Smith presents some information on cognitive and syntactic development, which is designed to show that Zak is typical for children of his age in these other areas. Unlike Smith (Reference Smith1973), only a few examples are presented in the analysis chapter, and those are usually mixed in with the text rather than put on separate lines. The intent seems to be for the reader to flip back and forth between the text and the appendix, to find illustrations that clarify the nature of the child's patterns. This requires a high degree of commitment on the part of the reader, which will undoubtedly be off-putting to some readers. However, many of the phenomena are quite interesting, and the committed reader will find a lot of absorbing material. Chapter 4 is thus a challenging but rewarding read.

Smith's presentation of Zak's phonological patterns is mostly clear, but the decision to use rules in the style of the 1970s causes problems on a few occasions. With one exception (the analysis of glottal stop as placeless), Smith makes minimal use of nonlinear phonology. When discussing the neutralization of /tr/ and /tʃ/ as affricates, which can be analyzed as incorporation of [+continuant] from the [r] into the /t/ to create a contour ([–continuant, +continuant]) on the [continuant] tier, Smith simply leaves the rule unformalized. In other cases, Smith does not adequately bring out why certain patterns are problematic. For example, it has been noted in the literature (see Bernhardt & Stemberger Reference Bernhardt and Stemberger1998) that the realization of /f v/ as [t d] is odd because they lose both labiality and continuance. In contrast, the realization of /f v/ as [p b] is faithful to labiality and hence expected to be more likely (and is in fact more likely across children). Smith's use of early-1970s-style rules simply allows the child's phonological reduction to be stated. He does not remark on any oddness when this pronunciation first appears; only when the transition from Stage 5 to Stage 6 leads to a striking change in the statement of the rule does Smith characterize the relation between the Stage-5 and Stage-6 versions of the rule as ‘bizarre’ (77), and notes that the simplest statement of the rule would incorrectly predict that /p b/ should also have changed to [t d]. Use of the archaic framework means that the meaning of some phenomena is not extracted sufficiently.

Occasionally, details of phonetics need to be addressed in more detail. In many dialects, the stops in /tr dr/ clusters are lightly affricated, a fact which Read (Reference Read1975) uses to explain why young children often spell these clusters with the affricate letters 〈ch〉 and 〈j〉, and which Bernhardt & Stemberger (Reference Bernhardt and Stemberger1998) use to explain common child pronunciations, such as [iː] or [iː] for tree. Do Zak's affrication patterns derive from mild affrication in adult speech or is it a developmental phenomenon? Similarly, Smith does not transcribe glottal stops before vowels in word-initial position, and remarks on ‘the absence from the adult phonological system of [ʔ]’ (95); he treats the presence of a glottal stop in place of non-initial /k/ and other phonemes as far more foreign to the adult system than others do (who view it as extending English post-pausal glottal stop in ‘vowel-initial’ words into new positions in the word, much as when otherwise-only-syllable-initial [h] substitutes for syllable-final /s/ in words like mouse).

The other chapters in the book address theoretical issues, providing a challenging read. The typical presentation is one where an issue or proposal is brought up, and a source is cited, without detailed elucidation in the text. Most readers are likely to find one or more sections in these chapters to be ‘cryptic’ and not self-contained. The book would have been more readable, and the arguments more compelling, had the text been less compact.

The issues that receive the greatest attention are: competence vs. performance, representations, the question of whether the child has his own phonological system that differs from that of adults, and the psychological reality of the child's output patterns. In the discussion, Smith makes no attempt to define the meaning of terms or argue for particular definitions. He concludes that child phonological patterns are part of performance rather than competence, that they are not represented, that the child has no system of his own, and that child phonological output patterns have no psychological reality. Smith's ultimate preference is to assume that child phonological processes are created in a connectionist associative network, and are thus not symbolic. He suggests that this is particularly problematic for Optimality Theory, which requires comparison between an input representation and an output representation. He also raises the possibility that adults too make use of this network, so that adult output is also performance and not represented in the system.

As is often the case, this discussion of competence and performance is an oversimplification. Competence cannot occur by itself; it can be observed only as it is filtered through performance. The most crucial question is what constitutes the relationship between competence and performance. If the answer is that performance is competence plus some other things, it follows that performance rather than competence may underlie some of the patterns that we observe. For every universal, an explicit argument must be put forward, proving that it is part of competence and cannot be due to performance. Smith takes a position, common today, that competence for grammar can arise on the basis of perceptual input, and that production of sounds is performance. But is this compatible with the notion of generative grammar? Previous approaches to grammar (as, for example, under structuralism) were based on parsing; researchers would take naturally occurring sentences and develop procedures for parsing them into their constituent parts. Chomsky's introduction of generative grammar took the perspective that parsing is not enough, that a speaker must be able additionally to generate the sentences of his/her languages. However, if production (generation) is only performance, can anyone still ‘do’ generative grammar? The question demands more detailed discussion.

Smith presupposes that representations (by which he apparently means symbolic representations) are required for perception (though he provides no arguments), but suggests that they are not needed for production. Neural networks ‘store’ information in a form that is not strictly symbolic, but the general patterns can be described in similar high-level terms. If we cannot call these stored patterns ‘representations’, what are the arguments that representations are needed for anything? Smith assumes that if no representations are involved (but simply patterns of activity that map onto articulation), then the patterns in that network do not constitute a system and are not ‘psychologically real’. But many psychologists, working with psychological theories of processing and learning, posit exactly such networks with exactly such sub-symbolic patterns, and refer to them as a system. Smith implies that such psychologists work with things that have no psychological reality. It is not clear exactly what Smith means, making it difficult to evaluate the conclusions.

Smith also sends mixed signals about how changes occur over time. He assumes that much of the learning in phonological development comes about automatically from improvements in motor processing. This assumes a passive learner, who will get everything right simply by waiting and observing. Most theories today presuppose active learners, who note that their performance is less than perfect and actively learn to do better. Connectionist models incorporate error-driven learning as a key part of learning in neural networks, so Smith addresses how that could work. He notes that some cognitive psychologists have posited an output Response Buffer as part of short-term memory. The output patterns of language production are copied into the Response Buffer, where they are held temporarily in a state that allows for self-monitoring, including evaluation as to how well they correspond to the optimal output. On the basis of less-than-optimal output patterns, the child can alter weights in the neural network to improve subsequent performance. However, if these sub-symbolic output patterns are qualitatively different from the symbolic input patterns, it is not clear how the learner can evaluate their adequacy. Smith's discussion is interesting, but far too abbreviated. Perhaps the intent was to raise issues for discussion, rather than to provide any compelling answers, and if so, the discussion is successful.

Overall, this book is a dense read but well worth it. The database of Zak's phonological development is interesting and detailed, and will undoubtedly be mined by others. The discussion of theory is interesting but terse; terms are not defined, and ultimately it is difficult to decide whether Smith's arguments are compelling. Despite these weaknesses, the book has many strengths, and it is a worthwhile read. It will undoubtedly attract the same sort of attention as Smith's previous book has had over the past forty years.