Date: Mon, 23 Jan 2006 00:51:05 -0500 From: Michael R. Marlo Subject: Reduplication: Doubling in Morphology
AUTHORS: Inkelas, Sharon; Zoll, Cheryl TITLE: Reduplication SUBTITLE: Doubling in Morphology SERIES: Cambridge Studies in Linguistics, 106 PUBLISHER: Cambridge University Press YEAR: 2005
Reduplication: Doubling in Morphology, by Sharon Inkelas and Cheryl Zoll, presents a new theory of reduplication, Morphological Doubling Theory (henceforth MDT), which reanalyzes the fundamental identity relation in reduplication as morphosyntactic. Most current theories of reduplication, building on McCarthy and Prince (1995) and earlier work such as Wilbur (1973), assume some version of Base- Reduplicant Correspondence Theory (BRCT), which requires surface phonological identity between the base and the reduplicant. In MDT, reduplicative constructions call for multiple copies of stems, which have independent inputs and are subject to independent phonotactic and morphotactic requirements. Surface phonological identity between the two reduplicative copies, as well as surface phonological non- identity, is therefore an indirect byproduct of identity of morphosyntactic features, as mediated by phonological and morphological requirements on each copy and on the entire reduplicative construction.
In this book, Inkelas and Zoll bring reduplication within the purview of theories of morphologically conditioned phonology, removing it as a special instance of phonological correspondence that has powerful theoretical machinery unto itself. Inkelas and Zoll reanalyze in MDT several examples showing opaque underapplication and overapplication interactions between the base and reduplicant that have been previously argued to provide the basis for BRCT. In many of these reanalyses, Inkelas and Zoll show that on different assumptions about the morphological structure of the examples in question, the application of the phonological processes is transparent and derives from ordering relations inherent in reduplicative constructions. In addition to countering the primary arguments for base-reduplicant correspondence, this book argues that BRCT overgenerates kinds of opacity that are not known to occur in human languages and that MDT correctly predicts as impossible.
1. Introduction Chapter 1 highlights the main claims of MDT, explicitly details differences between MDT and phonological copying theories of reduplication, and provides general information about the theoretical assumptions of MDT. MDT is couched within Cophonology Theory (e.g., Orgun 1996, Inkelas, Orgun & Zoll 1997, Inkelas 1998), in which every morphological construction is associated with a separate cophonology (constraint ranking), due to the theory's ability to account for morphologically conditioned phonology. In the MDT approach to reduplication, every reduplicated stem has three cophonologies: one for each of the copies of the stem (the 'daughters') and one for the entire reduplicated stem (the 'mother'), which dominates the two daughter nodes. In this approach, each daughter has an independent input, and the output of each daughter serves as the input to the mother node, predicting the occurrence of ordering effects. Additionally, divergence in the surface form of each copy of the stem is predicted, as each cophonology has a potentially different constraint ranking.
In MDT, reduplicative constructions call for two daughter nodes that are identical in morphosyntactic or semantic features. In addition to being able to encode morphologically conditioned phonology, the MDT approach can encode idiomatic semantics, since the semantics of the mother node can be specified independently of the semantics of the daughter nodes. This is important in accounting for reduplication whose meaning is non-iconic and, outside of reduplication, for exocentric compounds (''pick-pocket''), where the meaning of the whole differs from the sum of the meanings of the parts.
2. Evidence for morphological doubling Chapter 2 discusses the main evidence for MDT. MDT claims that the each of the reduplicative copies -- the daughters -- has a morphologically and phonologically independent input, which allows them to diverge morphologically and phonologically in both the input and output. MDT therefore predicts the existence of reduplicative constructions in which the two copies are semantically identical but differ phonologically due to internal structural differences. Several types of phenomena have this characteristic: linking markers, melodic overwriting, prosodically beneficial empty morphs, and tier replacement.
Two other phenomena that critically support these basic claims of MDT, but which are difficult to model in phonological copying theories of reduplication, are synonym and antonym constructions and root allomorphy. In synonym and antonym constructions, there is a requirement that the two copies be semantically identical, similar, or different, without any requirement that the copies be phonologically identical. The existence of a phenomenon of this sort supports the essential claim of MDT that doubling is at an abstract semantic or morphosyntactic level. Inkelas and Zoll argue that a grammar that can model synonym and antonym constructions already has the power to handle reduplication, without recourse to Base-Reduplicant Faithfulness or a RED morpheme.
The phenomenon of root allomorphy is characterized by the use of different allomorphs of the same root in various morphosyntactic constructions. In reduplicative constructions, which have two copies of the root, the allomorph of the root that is expected in the particular morphosyntactic configuration is used for one of the copies, while the morphological ''default'' allomorph occurs in the other copy. Phonological copying theories of reduplication cannot derive these patterns, since the form of each copy is essentially determined by morphological principles.
3. Morphologically conditioned phonology in reduplication: the daughters
Chapter 3 investigates two of three main phonological predictions of MDT, the ''Generalized Phonology Prediction'', which claims that ''the set of phonological effects found applying within reduplication is equivalent to the set of morphologically conditioned effects found outside of reduplication (69), ''and the ''Independent Daughter Prediction'', which claims that ''the phonological effects associated with the two copies in reduplication are independent'' (69).
Surveying the typical modifications of either daughter node or the mother node in reduplicative constructions, Inkelas and Zoll find cases of assimilation, dissimilation, deletion, insertion, truncation, augmentation, lenition, fortition, neutralization—essentially the same set of phonological input-output modifications that occur outside of reduplication, as shown, for example, in the parallel results of surveys of neutralizations that occur in reduplicants (Alderete et al. 1999, Steriade 1988) and positionally conditioned, non-reduplicative environments (Barnes 2002). Supported by these findings, MDT argues that reduplicative phonology is not qualitatively different from non-reduplicative phonology and that reduplicative phonology can be handled within the theory of morphologically conditioned phonology that is independently necessary. As a result, Inkelas and Zoll assert that there need not be, and therefore should not be, a special theory just to handle reduplicative phonology.
The Independent Daughter Prediction, which claims not only that the daughters in reduplication have morphologically separate inputs but also that the cophonologies of the daughters in reduplication are independent of each other, is borne out by cases in which both copies in a reduplicated stem are modified. In such phenomena, each copy undergoes distinct morphologically conditioned phonology, as in, for example, double melodic overwriting, where both copies of the stem have an independent, fixed melodic overwrite, and parallel modification, where both copies are different, in the same way, from their inputs. An interesting corollary of the Independent Daughter Prediction is that it excludes 'base-dependence' phenomena in which the output form of one copy (the reduplicant) is dependent on the other (the base).
4. Morphologically conditioned phonology in reduplication: the mother node Chapter 4 explores the third main phonological prediction of MDT, the Mother Node Prediction, which claims that reduplicative constructions may be associated, as a whole, with morphologically conditioned phonological rules. The types of phenomena investigated here include those junctural effects that occur only in reduplication, at the boundary between the two reduplicative copies.
Similar to their findings that the phonological alternations that occur in either of the daughter nodes are not unique to reduplication, Inkelas and Zoll claim that ''the range of junctural alternations in reduplication is as broad as the range of junctural alternations generally, including epenthesis, lenition, metathesis, coda sonorization, assimilation, dissimilation, and syncope (99).'' So, even if some language has a phonological process that occurs (or fails to occur) only in reduplication, Inkelas and Zoll argue that it can be handled with the same kind of technology that is used to account for some alternation that occurs (or fails to occur) only in, e.g., pluractional stem formation or in the hodiernal perfective tense. Cases of underapplication occur not only in reduplication but in other kinds of morphologically conditioned phonology, which require reference to constructions, for which BR-Faithfulness is insufficient. Cases of non-identity reduplicative underapplication are reanalyzed in MDT as non- application, where ''the cophonology in question does not enforce the patterns in question'' (108). While such effects are unexpected in BRCT, they are expected in MDT, since the mother node has an independent cophonology. At the end of Chapter 4, Inkelas and Zoll provide a critical discussion and rejection of a variant of BRCT, Existential Faithfulness (Struijke 2000), as a possible means of accounting for such phenomena generally.
5. Morphologically driven opacity in reduplication In Chapter 5, Inkelas and Zoll discuss and reanalyze one of the primary pieces of evidence for BRCT and one of the hallmark properties of reduplication generally: opacity. Several languages appear to show overapplication or underapplication only in reduplication, and as a result several theories (Clement 1985, Mester 1986, McCarthy and Prince 1995) have differentiated reduplicative opacity from other kinds of opacity. McCarthy and Prince (1995), building on Wilbur (1973), attributes the apparent uniqueness of reduplicative opacity effects to identity enhancement between the base and reduplicant.
Inkelas and Zoll reject these claims that the opacity effects associated with reduplication are unique to it and that they result from an identity requirement holding between the two reduplicative copies. Instead, they argue that the kinds of opacity effects that occur in reduplication derive from the morphological structure of reduplicative constructions, which have independent input-output mappings for each of the daughter nodes that serve as the input to the mother node, which is subject to its own phonotactic and morphotactic requirements. This intrinsic layering or ordering, ''gives rise to the cyclic, or stratal, interactional effects to which Kiparsky (2000) attributes opacity in general (136),'' such that, for example, an alternation triggered at the mother node level renders opaque an alternation occurring in one of the daughter nodes. In particular, truncation of one of the reduplicative copies, common inside and outside of reduplication in morphologically conditioned phonology, often renders opaque the transparently triggering context. Inkelas and Zoll argue, then, that the same kind of approach to opacity generally, often involving morphological layering, with possible reranking between levels, accounts for reduplicative opacity without further reduplication-specific technology.
This chapter also identifies two serious problems with the BRCT approach to reduplicative opacity. First, not all cases of reduplicative opacity involve identity enhancement, so BRCT is therefore insufficient as an explanation for the full range of reduplicative opacity effects and requires additional theoretical machinery. Second, Inkelas and Zoll point out that BRCT, and other identity-based theories, make incorrect predictions about two classes of possible reduplicative opacity effects involving internal junctural and external junctural alternations that are not known to occur in language and that are impossible within MDT.
Finally, Chapter 5 discusses the question of backcopying -- a kind of opacity effect whereby a phonological change in the reduplicant is reflected in the base, which MDT predicts impossible, but which is an important piece of evidence for BRCT. On closer inspection, Inkelas and Zoll find little evidence for backcopying. Their reanalyses take two general tacks: either the effect is non-existent, the result of morphological misanalysis, as in McCarthy and Prince's (1995) analysis of Klamath, or the effect is the result of local phonological assimilation of the kind that is found outside of reduplication, as in Chaha biliteral roots (Kenstowicz and Petros Banksira 1999).
6. Case studies Chapter 6 implements the main ideas of MDT in reanalyses of case studies of Tagalog and Chumash reduplication, two examples that have been cited in the literature as evidence of backcopying overapplication. In each language, prefixal material, assumed by McCarthy and Prince (1995) to be outside the domain of reduplication which targets morphological roots, is expressed in both reduplicative copies. McCarthy and Prince assume morphological structures in which reduplication is prefixing and therefore take the doubling of the prefixal material as evidence for backcopying from the reduplicant (1st copy) into the base (2nd copy).
Inkelas and Zoll reanalyze these phenomena similarly to Downing's (1998) approach to prosodic misalignment. Their reanalyses have four main components: prefixation precedes (or at least is present in the input to) reduplication; reduplication targets not the morphological root, but the prosodic root, which minimally incorporates adjacent phonological material outside the morphological root, such as, e.g., by resyllabification; reduplication is infixing, not prefixing; and the output truncates one of the copies, eliminating in one of the copies the remainder of the prefixal material that provided the basis for the prosodic root.
7. Final issues In the final chapter, Inkelas and Zoll tie up a few loose threads by addressing the issue of phonological copying, which MDT maintains in a restricted way, in addition to morphosyntactic feature doubling. Phonological copying, which is constrained to apply locally (unlike morphosyntactic doubling, which can result in nonadjacent copies in, e.g., opposite-edge reduplication), remains independently necessary due to the ubiquity of processes of phonological assimilation. Inkelas and Zoll provide criteria for distinguishing phonological and morphosyntactic copying but admit that the phenomenon of CV reduplication is often difficult to diagnose as the result of phonological or morphological copying. Since phonological and morphosyntactic copying are formally distinct, some phenomena are predicted to show effects of both processes, as seen in, e.g., Hausa pluractional verbs. Inkelas and Zoll devote the final pages of the book to critically discussing the relationship between reduplication, language games, and rhyme; casting doubt on phonological anti-identity effects as deriving directly from (anti-) correspondence requirements; and identifying avenues for future research within MDT.
Reduplication: Doubling in Morphology is an extremely important and timely contribution to the theoretical discussion of reduplication, a long- standing and central issue in theories of phonology, morphology, and their interaction. This book, which is well written, well edited, and readable, advocates an important methodological moral—that morphological analysis should precede and guide phonological analysis, leveling a nontrivial criticism against some phonological theorizing that fails to adequately investigate phonologies in their larger morphosyntactic settings. (In fact, the advice has even been successfully applied to Inkelas and Zoll's own work, as in Marlo and Pharris (2004), which strengthens Zoll's (2002) analysis of Klamath reduplication by more fully considering the implications of assumptions regarding the directionality of affixation inside and outside of stems in reduplication.) The book provides solid conceptual and empirical arguments for rejecting reduplication-specific Base-Reduplicant Correspondence Theory and for approaching reduplication within the general domain of morphologically conditioned phonology. MDT is couched within the theories of Sign-Based Morphology and Cophonology Theory, but the main results of the book can relatively easily be translated into other theories of morphology and approaches to morphologically conditioned phonology, such as lexical phonology and indexed constraint theory.
While in general the arguments put forth are well argued and supported, there are a few places where questions remain. One case involves the synonym and antonym constructions discussed in Chapter 2. While these examples are quite interesting and appear to constitute strong evidence for the claim that grammar must have the power to require abstract semantic (or morphosyntactic featural) identity and non-identity, there are several lingering uncertainties about the grammatical properties of these constructions and their productivity. It is not known, based on the discussion in Chapter 2, whether these constructions are synchronically productive, and if, for example, new synonym constructions can be coined. This is an important point in probing the formal properties of the constructions and what semantic or morphosyntactic features are involved. In other words, what formally, makes two items synonyms, and is synonymy the only requirement for participation in reduplication? And several other questions arise. Would, for example, 'glance-stare', 'notice- regard', 'watch-see' be grammatical? What are the relevant semantic features? Do the elements have to match for morphosyntactic category?
Another question involves the Thesis of Morphological Targets, introduced in Chapter 2, which claims that ''a reduplicative construction calls for morphological constituents (affix, root, stem, or word), not phonological constituents (mora, syllable, or foot) (25)''. The issue is that many of the critical MDT analyses, reanalyzing opacity effects previously argued to provide evidence for BRCT, crucially involve the doubling of prosodic constituents, such as the PRoot or the PStem, which are based on but crucially not identical to the morphological constituents. These later analyses therefore seem to violate the Thesis of Morphological Targets, since the analyses are untenable if the reduplicative targets are the morphological roots and stems (unless certain other assumptions such as Consistency of Exponence are rejected and there is actually no distinction between MStems and PStems except the level of derivation at which they are inspected).
The gravest concern regarding MDT, in this reviewer's eyes, involves the claim that the input to the two copies of the stem in all reduplicative constructions is independent. As Inkelas and Zoll point out, this makes 'base-dependence' effects, in which the form of the reduplicant (in traditional terms) is determined in relation to the form of the base, arbitrary properties of the cophonologies of each copy, since the two copies are not in a correspondence relation. Inkelas and Zoll discuss the base-dependence problem in Section 3.4.1 and conclude that the evidence for base-dependence is slim. They look at the evidence from cases of CV ~ VC reduplicant allomorphy where the reduplicant has a CV shape with C-initial bases and a VC shape with V-initial bases and reanalyze the cases as involving infixation, not base-dependence, following McCarthy and Prince's (1999) reanalysis. Inkelas and Zoll also reject a few other cases of alleged base- dependence based on the shape of the reduplicant, suggesting that such templatic effects can be accomplished at the mother node cophonology, without need for correspondence between the two reduplicative copies.
There are, however, more serious challenges to the claim that phonological or morphological requirements on one copy are completely unrelated to the requirements on the other. At least two Bantu languages, Kikerewe (Odden 1996) and Lusaamia (Marlo 2004), have patterns of asymmetric morphological reduplication of the verb stem in which the first copy of the stem has all of or a subset of the suffixes of the second copy of the stem. When there are multiple suffixes in the second copy of the stem, variants of the first copy of the stem are possible that have all of, none of, or some subset of the suffixes in the second copy. For example, ni-ba-lim-il-an-e-lim-il-án-é, ni-ba-lim-il-an-a-lim-il-án-é, ni-ba-lim-il-e-lim-il-án-é, ni-ba-lim-il-a-lim-il- án-é, ni-ba-lim-e-lim-il-án-é, and ni-ba-lim-a-lim-il-án-é are all possible forms of 'they should not cultivate for each other' (-a is a default suffix that is essentially the zero equivalent of -e). Odden (1996) shows that there is a contiguity requirement in such examples in Kikerewe such that examples with discontinuous copying within the derivational stem are ungrammatical (e.g., *ni-ba-lim-an-e-lim-il-án-é), although the form of the first copy of the stem would be grammatical (if not idiomatic for this particular root), outside of reduplication (ku-lim-an-a 'to cultivate each other') or inside of reduplication with a different structure of the second copy of the stem (ni-ba-lim-an-e-lim-án-é 'they should cultivate each other'). Therefore, although these patterns may not necessarily be the result of an Identity requirement in the usual sense of BRCT, they show that the morphosyntactic and phonological features of the first copy (the reduplicant) are dependent on those of the second copy (the base), and that the Independent Daughter Prediction does not hold in its strongest form.
Despite these possible problems, this book remains a compelling read and a significant contribution to the field.
Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John McCarthy, and Suzanne Urbanczyk. 1999. Reduplication with fixed segmentism. Linguistic Inquiry 30:327-64.
Barnes, Johnathan. 2002. Positional neutralization: a phonologization approach to typological patterns. Ph.D. thesis, University of California, Berkeley.
Downing, Laura. 1998. Prosodic misalignment and reduplication. Yearbook of Morphology 1997:83-120.
Inkelas, Sharon. 1998. The theoretical status of morphologically conditioned phonology: A case study from dominance. Yearbook of Morphology 1997:121-55.
Inkelas, Sharon, Cemil Orhan Orgun, and Cheryl Zoll. 1997. Implications of lexical exceptions for the nature of grammar. In Iggy Roca (ed.), Constraints and derivations in phonology. Oxford: Clarendon Press, 393-418.
Kenstowicz, Michael and Degif Petros Banksira. 1999. Reduplicative identity in Chaha. Linguistic Inquiry 30:573-86.
Marlo, Michael R. 2004. Prefixing reduplication in Lusaamia: evidence from morphology. In Akinbiyi Akinlabi (ed.), Proceedings of the 4th World Congress of African Linguistics. Köln: Rüdiger Köppe.
Marlo, Michael R. and Nicholas J. Pharris. 2004. Which wič is which? Prefixes and suffixes in Klamath full-root reduplication. Linguistic Inquiry 25:639-656.
Marantz, Alec. 1982. Re reduplication. Linguistic Inquiry 13:483-545.
McCarthy, John and Alan Prince. 1995. Faithfulness and reduplicative identity. In Jill Beckman, Laura Dickey, and Suzanne Urbanczyk (eds.), University of Massachusetts Occasional Papers in Linguistics 18: Papers in Optimality Theory. Amherst, MA: GLSA, 249-384.
McCarthy, John and Alan Prince. 1999. Faithfulness and identity in Prosodic Morphology. In Rene Kager, Harry van der Hulst, and Wim Zonneveld (eds.), The prosody-morphology interface. Cambridge: Cambridge University Press.
Odden, David. 1996. Patterns of Reduplication in Kikerwe. Ohio State University Working Papers in Linguistics 48:111-148.
Orgun, Cemil Orhan. 1996. Sign-based morphology and phonology: with special attention to Optimality Theory. Ph.D. thesis, University of California, Berkeley.
Orgun, Cemil Orhan. 1999. Sign-based morphology: a declarative theory of phonology-morphology interleaving. In Ben Hermans and Marc van Oostendorp (eds.), The derivational residue in phonological Optimality Theory. Amsterdam: John Benjamins, 247-67.
Orgun, Cemil Orhan and Sharon Inkelas. 2002. Reconsidering bracket erasure. Yearbook of Morphology 2001: 115-46.
Steriade, Donca 1988. Reduplication and syllable transfer in Sanskrit. Phonology 5:73-155.
Struijke, Caro. 2000. Existential Faithfulness: A study of reduplicative TETU, feature movement, and dissimilation. Ph.D. thesis, University of Maryland.
Wilbur, Ronnie. 1973. The phonology of reduplication. Bloomington: Indiana University Linguistics Club.
Zoll, Cheryl. 2002. Vowel reduction and reduplication in Klamath. Linguistic Inquiry 33:520-27.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Michael Marlo is a Ph.D. Candidate in the Department of Linguistics at the University of Michigan, who has published on reduplication in Klamath (Penutian, Oregon) and Lusaamia (Bantu, Kenya) and is interested generally in phonology, morphology, interactions of the components of grammar, and Bantu linguistics. His dissertation is a study of verbal tonology in three Bantu languages of the Luyia group in western Kenya.