Lodge describes Fundamental concepts in phonology as ‘an attempt to bring together various strands of my work from over the past forty years’ (p. vi).Footnote * Whatever its title might suggest, the book is not a textbook. Part career retrospective and part monograph, it focuses on the concepts of sameness and difference, and aims to hold up to scrutiny a number of related issues in phonology, such as segmentation and biuniqueness. It also outlines Lodge's ‘declarative’, ‘polysystemic’, ‘abstract’ and ‘non-segmental’ phonological framework, with acknowledgement of the influence of (neo-)Firthian approaches to phonology (Firth Reference Firth1948, Palmer Reference Palmer1970, Kelly & Local Reference Kelly and Local1989, Ogden Reference Ogden1999a).
‘Declarative’ refers to the fact that like Coleman (Reference Coleman1998), Ogden (Reference Ogden1999a), Bye (Reference Bye, Toivonen and Nelson2007) and others, Lodge accounts for alternations using inviolable constraints only, and avoiding all structure-changing and destructive processes. ‘Polysystemic’ concerns the idea that a phonology consists of multiple subsystems which are to some extent interrelated, but can be formalised as coherent systems in their own right (see Ogden Reference Ogden1999a: 62–66, and Firth Reference Firth1957 for an early explanation). A polysystemic approach recognises that particular morphological or structural domains (for example verbs, or stressed syllables) may be associated with phonological constraints that have no function in other domains (for example nouns, or unstressed syllables), and explicitly restricts the application of constraints to their respective domains. Importantly in the context of Lodge's book, this includes the recognition that different systems of contrast operate in different positions in prosodic structure (for example, vowel systems in stressed vs. unstressed syllables). Lodge accepts the idea that such differences may have an impact on the representation of the contrastive units involved: a phoneme x that occurs in contexts y and z does not necessarily have an identical feature representation in both contexts. Finally, ‘abstract’ and ‘non-segmental’ refer to the fact that Lodge advocates a clear division of labour between phonological representation and phonetic description – of which more later – and that the phonetic segment has no primary phonological status in his framework (see Coleman Reference Coleman1992, Ogden Reference Ogden, Hulst and Ritter1999b, Local Reference Local2003).
A book focusing on sameness and difference in phonology is timely given the current interest in exemplar-based approaches (e.g. Coleman Reference Coleman, Durand and Laks2002, Silverman Reference Silverman2006, Johnson Reference Johnson, Solé, Beddor and Ohala2007), which bring to the fore the fundamental role of categorisation and abstraction from phonetic detail in the establishment of phonological representations. It is also timely given the renewal of interest in the nature of phonological representations and contrast, following a long period of focus on processes. For example, in recent years, Morén (Reference Morén2006) has proposed a new model of feature geometry that incorporates underspecification and language-specific phonetic interpretation of features, Mielke (Reference Mielke2008) has developed an ‘emergentist’ approach to distinctive feature theory, and Dresher (Reference Dresher2009) and papers in Avery et al. (Reference Avery, Dresher and Rice2008) have re-evaluated the place of contrast in phonology, including its role in motivating underspecification theories. Lodge's focus is similarly on representational issues. He largely steers clear of arguments about processes, presenting declarative accounts of selected data from a range of languages (including German, Icelandic, English, Irish and Scots Gaelic) without elaborating a case against derivational alternatives.
According to Lodge, phonological difference can be defined on two levels: ‘phonemic’ and ‘systemic’. Phonemic difference is established in the familiar way, by testing whether a given phonetic difference gives rise to a lexical distinction. Lodge takes inspiration from Trubetzkoy (Reference Trubetzkoy1939), Firth (Reference Firth1957), Fudge (Reference Fudge1969) and Silverman (Reference Silverman2006) on the notion of ‘systemic difference’, which is established by comparing the functions of two sounds in their respective phonological environments – for example, by establishing how many other sounds each contrasts with, or to what extent each is involved in alternations. Unfortunately, Lodge does not elaborate more concrete criteria, but acknowledging the importance of both levels in establishing degrees of phonological difference is useful. Phonemic analyses, and much subsequent generative work, have tended to neglect the systemic dimension. Explicitly polysystemic work, on the other hand, such as that of the London School (Firth Reference Firth1957, Palmer Reference Palmer1970), has downplayed the relevance of the phonemic dimension. As Lodge points out, the risk of the latter approach is that it can lead to ‘unbridled polysystemicity’ (p. 41): an outlook in which the recognition of phonological sameness across structural environments is all but impossible, so that, for example, ‘[s] in see is considered to be different from the [s] in same, sieve …, where the following vowel is ‘the same as’ the one in see’ (p. 41). Lodge's version of polysystemicity is kept in check by his acknowledgement of the relevance of phonemic distinctions, while maintaining that systemic considerations play a crucial role in the establishment of phonological representations.
Of course, it is desirable for a framework of representation to incorporate both a consistent way of encoding phonemic sameness and difference and a consistent way of encoding systemic sameness and difference. Lodge's framework accommodates both, firstly because it incorporates underspecification and secondly because it allows for the non-terminal attachment of features.
Lodge's main motivation for adopting underspecification is that a phonology that disallows both structure-changing and destructive rules can use it to handle basic alternations. For example, a structure-changing rule such as d→t can be made structure-building by removing the voicing specification from the input representation: T→t, where T is an alveolar plosive unspecified for voicing. An important corollary of this approach is that sameness and difference are encoded at multiple levels of representation: phonemic identity is encoded in fully specified representations, while systemic considerations determine the nature of the underspecification and formulation of structure-building statements. This is clearly visible in Lodge's treatments of German vowels. Lodge points out that in German, a vowel quality such as [yː] can occur in lexical items which are part of an umlaut alternation, such as Füße [fyːsə] ‘feet’ (singular Fuß [fuːs]); and in non-alternating lexical items, such as für [fyːʌ] ‘for’ (p. 47). In his analysis, the difference between the two is marked at the level of lexical representation: the umlaut vowel is specified as [round], and receives the feature [front] by rule; the non-alternating vowel is specified as [round, front] to start with (p. 113). The fully specified representations are identical. Similarly, Lodge observes that [f] in Scots Gaelic has three ‘sources’: it occurs in mutation alternations as the lenited correspondent of [pʰ] and [m], and with the lenited correspondent ∅ (p. 29). All three instances of [f] are associated with the same features, [voiceless], [oral], [friction] and [labial], but they differ both in their lexical representations and in the nature of the structure-building processes that give them their full specifications (p. 117).
Lodge's representational framework is similar to that of Coleman (Reference Coleman1998), among others, in that features are not necessarily organised into segment-size bundles, but ‘may be attached to any syllable node, not just terminal nodes, or even at levels higher than the syllable’ (p. 102). For example, Lodge offers the partial representations in (1) in an analysis of Muskerry Irish data (pp. 109–110). In (1a), [front] is attached to the syllable node to signal that its phonological domain is the syllable as a whole. In (1b), [front] and [back] are attached at Onset and Rhyme level. Whatever the rationale for these particular analyses, it should be clear that this approach exploits the expressive potential of non-linear representation to the full (and allows for accounts of assimilation and harmony patterns without destructive processes; p. 106), and can result in analyses in which the phonetic segment has no obvious unit status.
(1)
It also has major implications for the encoding of sameness and difference. One can imagine two segments associated with the same set of features in full specification, but different in the location of some of the features in the prosodic hierarchy, in addition to possible differences in how many features are specified lexically, and how the segments receive their full specifications. One could also imagine talking about sameness and difference at the Rhyme level, the Syllable level or the Foot level, establishing ‘phonemic’ and ‘systemic’ differences between units of this size. The representation in (1a) suggests as much: it allows, in principle, for the existence of minimal pairs such as [mʲaːgʲ]~[mˠːɑgˠ], in which syllable-level frontness vs. backness is the only active distinctive feature.
Throughout the book, Lodge emphasises the complexity of the relationship between sameness and difference at the phonetic and phonological levels. Many of his examples involve a single phonological contrast associated with differences along multiple phonetic parameters. In these cases, marking the contrast with a single distinctive feature seems like an unwarranted simplification from a phonetic point of view, although appropriate phonologically. Lodge's conclusion is that phonological representations need to be accompanied by statements of the phonetic implementation of the features that constitute it. These statements are language-specific and context-sensitive: for example, an analysis of Icelandic aspiration requires at least three statements of the phonetic implementation of the feature [voiceless]: one for attachment at Rhyme level, one for Coda and one for Onset (p. 108). Crucially, they can be complex, referring to multiple phonetic correlates of a single feature. In the case of [front] in the representation in (1a) above, for example, the statement should refer to the quality of the Nucleus vowel as well as the palatalisation observed on the Onset and Coda consonants.
Unfortunately, Lodge's failure to specify exactly what constitutes systemic sameness and difference is symptomatic of a general lack of rigorous exploration of the ideas outlined so far. With reference to sameness and difference, Lodge's approach offers a finite number of combinations of phonetic and phonological sameness and difference: phonetic difference, phonemic sameness, systemic sameness; phonetic difference, phonemic sameness, systemic difference; and so on. Some of these combinations warrant discussion: for example, if two sounds are phonetically the same, how can we establish that they are phonemically – as opposed to systemically – different? But Lodge does not develop the model, and returns at various points in his discussion to what would seem a binary view on phonological sameness. The phonemic level of sameness appears to be the sticking point. This sits rather uncomfortably in a ‘polysystemic’ and ‘non-segmental’ phonology, and perhaps for this reason Lodge glosses over sameness at the phonemic level in several places, while emphasising systemic differences. For example, when Lodge asks ‘what reason is there for identifying as the same the stressed vowels of Canada and sanity, or those of Oberon and verbose?’ (p. 35), the answer seems obvious: phonetically, they are very similar, and what difference there may be between them is not lexically distinctive. Similarly, Lodge asserts that ‘I specifically do not identify the vowel system in the final syllable of Malay disyllabic words with that in the first syllable; the former is subject to constraints on tongue height depending on whether the syllable has a coda or not, the latter is not so constrained’ (p. 36). In both cases the sounds involved would appear to be phonemically equivalent, although different systemically. In Lodge's treatment, the systemic difference appears to outweigh the phonemic sameness.
In fact, in his own discussion of the example of [s] in see, same, sieve and so on mentioned above, Lodge suggests that ‘the guard against unbridled polysystemicity resides in the acknowledgement of phonetic similarity as a classifier when no other functional considerations are relevant’ (p. 41). Taken literally, this means that phonological sameness follows from phonetic similarity when systemic considerations do not suggest a different analysis. This skips the crucial question of whether observed phonetic differences are potentially lexically distinctive or not. In English, such differences as exist between alveolar fricatives in different vowel contexts are not lexically distinctive, so the instances of [s] are the same on the phonemic level. To suggest that the fricatives are phonologically the same simply because of phonetic similarity and the fact that there is not much more to say about them phonologically seems counterproductive.
With reference to phonological representation, Lodge's analyses employing underspecification are clear and parsimonious, but his use of non-terminal feature attachment is less convincing. In demonstrating his representational framework, Lodge refers to non-terminal attachment primarily as a way of accounting for phonological contrasts whose phonetic correlates span more than one segment. For example, he suggests that in German, alternating [yː] is lexically specified [round], and [round] is attached at the Syllable level, to account for the presence of lip-rounding throughout syllables with rounded vowels. As indicated above, [front] is added by predictive statement, and it is attached at the Nucleus level (p. 112). This is presumably because ‘frontness’ is not necessarily observed across the syllable as a whole – unlike in the Muskerry Irish case illustrated in (1a).
This emphasis on the use of non-terminal attachment of features to reflect the temporal extent of the phonetic correlates of those features leads to the question of whether any phonological features can be justified as ‘terminal’, given the extent of segment-to-segment coarticulation, and is somewhat at odds with other work in Declarative Phonology. As indicated by Coleman (Reference Coleman1998), Simpson (Reference Simpson1998) and Ogden (Reference Ogden, Hulst and Ritter1999b), in a framework that requires explicit statements of phonetic implementation, the temporal alignment of correlates of particular features is specified in those statements, and therefore does not need to be encoded at the phonological level as well. Moreover, in a framework that accepts a notion of headedness, default phonetic implementation models can be set up, in which, for example, features associated with a Nucleus are expected to have an extent that includes the Onset and Coda (see Coleman Reference Coleman1992). In this way, a clear distinction is maintained between the (phonetic) extent and the (phonological) domain of features, and ‘distributional criteria remain the best basis on which to determine the domain of each distinctive feature’ (Coleman Reference Coleman1998: 209). These distributional criteria include feature co-occurrence constraints, instances of positional neutralisation and so on: the kinds of things that cause systemic differences between segments, and the kinds of things Lodge accounts for in terms of underspecification. Given the theme of this book and Lodge's phonological framework, then, leaving the division of labour between underspecification and non-terminal attachment in encoding sameness and difference largely unexplored seems something of a lost opportunity.
On the related issue of abstractness, while arguing in favour of a clear division of labour between phonetic description and phonological representation, Lodge chooses not to explore the ramifications of banning all intrinsic phonetic interpretation from phonology, opting instead for an approach with ‘partly phonetic’ unary features such as [lips: neutral] and [manner: stop] (p. 103). For example, in accounting for German forms such as Ernte [ɜʌnˠtˠə] ‘harvest’, in which ‘postvocalic’ /r/ is associated with the vowel quality [ʌ] and a degree of pharyngealisation, Lodge uses [pharynx: narrow] attached at the Rhyme level (p. 93). The ‘clear separation of phonetics and phonology’ advocated by Lodge (p. 87) is difficult to see in this analysis, which differs markedly from the abstract declarative analysis of German /r/ developed by Simpson (Reference Simpson1998).
The issue of abstractness is pertinent to the theme of the book. One related question is whether phonological contrast is best encoded using binary or unary features: in short, is the perceived absence of a particular feature a sufficient basis for phonological equivalence or not? Lodge argues that, from a phonetic point of view, binary features are suspect, since generally the articulatory configurations associated with ‘+’ and ‘–’ are both best described in positive terms (p. 76). Unsurprisingly, Lodge's own ‘partly phonetic’ features are unary. But when outlining an alternative analysis of the German /r/ data referred to above, with fully abstract features, Lodge replaces [pharynx: narrow] by [±r] (p. 93). Of course, his phonetically motivated objection to binarity is irrelevant in the latter approach, so Lodge cannot be accused of inconsistency. Still, as in the case of non-terminal attachment of features, not exploring the implications of this choice for the representation of sameness and difference – here potentially doubling the number of equivalence classes defined by one's feature set – seems somewhat short-sighted.
A final, more general criticism is that the ‘retrospective’ nature of this book is rather prominent. Lodge does not stray far from analyses he has written about before, and some sections read rather like summaries. Moreover, Lodge's criticisms of other analyses mostly cover sources published over a decade ago, focusing heavily on contributions to Goldsmith (Reference Goldsmith1995). This is not necessarily a flaw, but discussion of, for example, Morén's (Reference Morén2006) Parallel Structures Model of feature geometry or Steriade's (Reference Steriade and de Lacy2007) discussion of contrast in phonology would certainly have been welcome. Finally, exemplar-based approaches are referred to in passing (e.g. p. 142), but not explored in any detail, despite the suggestion by Bye (Reference Bye, Toivonen and Nelson2007) that they are compatible with an abstract version of Declarative Phonology not unlike that advocated by Lodge.
In sum, Lodge's book is a timely reminder of a range of issues and theoretical positions worth exploring in relation to phonological contrast and representation. It arguably lacks rigour and engagement with current literature, and it remains to be seen how many readers will be persuaded that Lodge's phonological framework will provide answers to the questions of how to delimit phonological sameness and difference, and how to encode these notions in phonological representations. But Lodge is right to stress that these fundamental questions are worth asking.
