1 Overview

In her 2019 book, Phonological templates in development, Marilyn May Vihman lays out a career's worth of work on the role of templates in children's phonological and lexical development. The book provides rich historical context and compiles a huge wealth of data in one place. It will be of interest to any student of phonological development, especially those who are interested in a concrete framework for defining and measuring a child's phonological patterns from a larger lexical perspective. At the same time, with its references to exemplar theory (e.g. Jusczyk Reference Jusczyk, Ferguson, Menn and Stoel-Gammon1992) and Dynamic Systems Theory (Thelen & Smith Reference Thelen and Smith1994), the book is also relevant for those more broadly interested in cognitive development. Data from children learning a variety of languages, along with detailed appendices of children's productions and adult templatic forms, will be tantalising for linguists who may want to do further analyses.

Vihman's main proposal is that, after an early stage of relatively accurate productions, children expand their vocabulary and phonetic repertoire by taking a templatic approach, selecting words that fit those templates and adapting those that do not. These templates reflect universal truths about motor and cognitive development, but do not otherwise rely on any sort of linguistic universal; in fact, the templates are in large part idiosyncratic from one child to the next. Vihman outlines how templates are defined, calculated and measured, illustrating these steps with data from a variety of languages along the way. She situates her theory within the broader Dynamic Systems Theory, focusing on the role of the child's own actions and perceptions in shaping their phonologies, and on the complex interplay between ever-changing forces such as motor skills, vocabulary size and external language input. She rejects the notion that there must be continuity between the child grammar and the adult grammar, instead noting that both continuities and discontinuities are expected. The book's chapters can be generally divided into three sections: background information (Chapters 1 and 2), detailed characterisations of templatic analyses (Chapters 3–5) and implications of the theory, especially as it relates to fully developed phonologies (Chapters 6–9).

2 Background information

The first two chapters of the book offer an overview of the theoretical advances that have led to the templates proposed in later chapters, and are especially valuable to any student of the field who wants to understand the study of phonological development from a historical perspective. In Chapter 1, ‘Perspectives on phonological development’, Vihman sets the stage by defining template (‘a fixed pattern or schema to which output must conform – regardless of the prosodic or segmental make-up of the input or of the processes that derive output from input’; p. 2) and providing a broad review of the foundational research on which her theory rests. Vihman situates her work within Dynamic Systems Theory, a theory of development that may be less familiar to linguists than to psychologists. Dynamic Systems Theory posits that child development is the emergent result of change within a complex system. Thelen & Smith originally applied this theory to motor development, but it has been extended to many domains, including linguistic development (e.g. Rvachew & Bernhardt Reference Rvachew and Bernhardt2010, Samuelson & Faubel Reference Samuelson, Faubel, Schöner and Spencer2016, Samuelson et al. Reference Samuelson, Kucker and Spencer2017), where it is argued that the child's own actions and perceptions allow her to construct a model of language, rather than rely on an innate mechanism. Through the lens of Dynamic Systems Theory, Vihman summarises other theories and findings that are key to her central proposal, including exemplar models, prosodic templates, the relationship between babbling, perception and later productions, and lexical selection and organisation within the developing lexicon. The chapter places each topic in its historical context, and draws clear links between topics.

Chapter 2, ‘Whole-word phonology: historical and theoretical overview’, is a detailed review of the use of words (rather than phonemes or features) as fundamental linguistic units. Vihman traces this analytical approach from its inception (Firth Reference Firth1948) to its recent role in usage-based models of phonology (e.g. Vihman & Croft Reference Vihman and Croft2007, Menn et al. Reference Menn, Schmidt, Nicholas, Vihman and Keren-Portnoy2013). The research summarised in this chapter has in common the assumption that a child's phonological development is dependent on the child's lexical development, rather than the other way around. Only by looking at the level of the word, these authors argue, can we understand the structure of a child's phonology. It is striking to see that, as phonological theory has evolved and advances in experimental methodology have progressed, the idea that words are the most important units in developing phonologies has returned again and again.

3 Templatic analyses

The middle section of the book lays out Vihman's proposal in detail. In sum, in Chapters 3–5, Vihman argues that children's early productions can best be analysed through templates, rather than sets of phonological processes. She illustrates her approach with case studies from six languages, with some cases showing the progression of templates as more words are learned, and others demonstrating the influence of language environment on templatic structure. Though the data discussed in these chapters are largely available in other venues, this book is the first to bring all of the data together. Any reader who hopes to do a templatic analysis like Vihman's will want to read these chapters carefully, as they include an extraordinary level of detail and sample analyses.

Chapter 3, ‘Building the evidence: from item learning to templates’, draws on data from six children learning American English to illustrate templates at two stages of development: first words and the 30–50 word stage. Vihman presents a brief analysis of the children's earliest words, and then links those to their later patterns of production. She describes the iterative procedure by which she identifies prosodic structures and templates; tables of data illustrate the breakdown into prosodic structures (different depending on the child), separated by selected and adapted words. This distinction is one that is important throughout the book: selected words are those that the child attempts precisely because they fit a template, while adapted words are those that the child modifies to fit a template. Vihman notes that a child who uses templates will either overselect words that match the templatic patterns or adapt forms to fit the pattern, and that the use of specific segments or classes of sounds within the templates provides strong evidence of their existence in a grammar (p. 79). In detailed analyses of each of the six American English-learning children, Vihman describes consonant inventories and the relevant templates. It is interesting to note that, although all six children are learning the same language, their templates are not the same; some overlap (especially in the avoidance of multiple consonantal places of articulation within the same word), but others are idiosyncratic. This idiosyncrasy, presumably due to each child's own experiences, is used as evidence that an analysis relying on universals (like constraints) would be untenable.

While Chapter 3 focuses on American English as the developing language, Chapter 4 (‘First words and prosodic structures: a cross-linguistic perspective’) introduces data from six additional languages or varieties: Estonian, Finnish, French, Italian, Welsh and British English. (A minor criticism is that the child data in the book come primarily from European languages, though there are brief analyses of Arabic and Brazilian Portuguese in a later chapter, and Vihman also notes that work on child templates in Farsi, Pashto and Urdu is underway.) This chapter shows the impressive scope of Vihman's research career, as many of the data come from her own work (in collaboration with many others). By examining cross-linguistic data, Vihman investigates the roles of universals and the ambient language in templatic formation. She finds that several generalisations can be made across most or all of the languages in question, including a preference for CV structures, a reliance on mono- and disyllabic word shapes, and the prevalence of harmony processes. On the other hand, the ambient language also influences children's target selection; consonant inventory size is important, and children learning languages with (perceptually salient) geminates produce them at a higher rate than would be expected. Likewise, different prosodic structures occur at different rates, presumably because of the target language's lexical structure. The strongest argument for templates as an analytical tool, though, comes from Vihman's discussion of consonant variegation. Across the board, children had difficulties in producing consonants at multiple places of articulation within the same word, though their repairs differed across languages and across children. Vihman contends that children attempt more words with variegated consonants if they can fit them into a template, and that the use of templates thus allows children to grow their vocabularies.

Chapter 5, ‘Phonological templates in development’, is the last chapter that focuses on in-depth analyses of child data. Here Vihman introduces case studies of individual children from the six languages described in Chapter 4. With these longitudinal samples, she characterises templates at the segmental level where possible, and considers the child- or language-specific factors that lead to the construction of templates. For example, Vihman cites the example of Jude, a child learning British English, who produces a wide range of consonants, but a small set of prosodic structures. Vihman agrees with Szreder-Ptasinska (Reference Szreder-Ptasinska2012), who speculates that Jude's limited set of prosodic structures is exactly what allows him to attempt more articulatory gestures for different consonants. Some children use templates extensively, while others have very little evidence for templatic structure in their productions. And yet there are many cross-linguistic similarities among the children's deployment of templates. Vihman argues that those similarities are due to articulatory constraints in early speech and deficits in phonological working memory. Templates help circumvent both problems, by providing familiar and accessible articulatory routines and by limiting the amount of phonological information that must be stored in memory.

4 Theoretical implications

The final few chapters of the book discuss more details about templates as a theoretical construct. Chapter 6, ‘Issues around child templates: timing, fading, quantification, and function’, focuses on the logistics of templates: when do they emerge, when do they disappear and what is their developmental role? To answer the first two questions, Vihman provides longitudinal analyses of children learning Lebanese Arabic, Brazilian Portuguese and British English. She finds that the timing of template introduction varies from child to child, as does the timing of when the templates fade from use. With the goal of quantifying these individual differences, Vihman discusses the ‘Pscore’ measure, which quantifies each child's use of templates (Vihman et al. Reference Vihman, Keren-Portnoy, Whitaker, Bidgood and McGillion2013). The Pscore awards points for words that fit a templatic pattern, with more points awarded for words that have been adapted to fit that pattern than words that were selected because they already fit the pattern, and more points awarded for templates that include specific sound classes. Interestingly, though Vihman and colleagues hypothesised that high Pscores at early stages in late talkers would predict greater development at later stages, this hypothesis was not supported. Vihman concludes this chapter by noting again that templates allow a child to increase complexity in their productions, essentially by taking some of the articulatory and memory demands away.

In Chapters 7 and 8, Vihman relates child templates to adult templates. Chapter 7, ‘Relation of child to adult templates, I: Parallels in core grammar’, reviews the role of templates in adult phonology. McCarthy's (Reference McCarthy, van der Hulst and Smith1982) use of templates and prosodic morphology to account for languages like Arabic is one of the best-known examples, and it has been extended to account for child data as well, as in Wauquier (Reference Wauquier, Bendjaballah, Faust, Lahrouchi and Lampitelli2014). Vihman summarises a number of templatic analyses by McCarthy and others, for example for Arabic, Tiene, English, Cupeño, Hausa, Estonian and Czech. Though the theoretical link between the proposal of templates in adult languages and its application to child language is well described, I felt that there was a missed opportunity to explicitly discuss the impact of learning a templatic language on the development of child templates, a topic that is not addressed in the examination of children learning Arabic in the previous chapter. That is, does a child learning Arabic, Tiene or Hausa approach templates differently than one learning a language that does not rely on templatic structure? In these cases, we might not expect the templates to fade, but instead to become more adult-like over time. Though Vihman follows Dynamic Systems Theory in eschewing the continuity hypothesis, studies of children learning templatic languages might shed light on whether there is any continuity between child and adult templates.

At the end of Chapter 7, Vihman notes one of the biggest differences between child and adult templates: adults typically do not need to reduce their articulatory or memory demands to create new words, while children are constantly learning new words and fitting them into their developing lexicons. In Chapter 8, ‘Relation of child to adult templates, II: Language at play’, Vihman concentrates on language games and hypocoristics as a venue to see templates in action in adult languages. After considering truncated and hypocoristic forms in French, clippings and hypocoristics in Estonian and rhyming compounds in English, she notes five parallels between child templates and adult word formation (pp. 259–260). First, in both cases the application of a practised pattern allows for the use of more words. Second, both templates and newly created forms typically conform to a prosodic structure, and third, they often define a segmental specification. Fourth, neither child templates nor adult word formation rely on the initial consonant as the fixed element, possibly because the high salience of that position results in preservation of contrastive information. And finally, both templates and adult word formations are variable, such that the adaptations required to fit a word into its template are not always consistent.