LINGUIST List 29.3446
Fri Sep 07 2018
Review: Historical Linguistics; Morphology; Phonology; Syntax; Typology: Samuels (2017)
Editor for this issue: Jeremy Coburn <jecoburnlinguistlist.org>
***************** LINGUIST List Support *****************
Fund Drive 2018
28 years of LINGUIST List!
Please support the LL editors and operation with a donation at:
Michael Cahill < mike_cahill
Beyond Markedness in Formal Phonology E-mail this message to a friend Discuss this message
Book announced at https://linguistlist.org/issues/28/28-5171.html
EDITOR: Bridget D. Samuels
TITLE: Beyond Markedness in Formal Phonology
SERIES TITLE: Linguistik Aktuell/Linguistics Today 241
PUBLISHER: John Benjamins
REVIEWER: Michael C. Cahill, SIL International
Many phonologists take the concept of “markedness” as self-evident and an accepted part of phonology. This volume shows that this is far too simple a view, and that there are a significant number of linguists who reject the concept as unnecessary as part of formal phonology. There have been various attempts to define what the characteristics of markedness are (e.g. Steriade 1995, Hume 2003, Rice 2007). These characteristics generally focus on frequency and structural or phonetic complexity. The Optimality Theory enterprise, of course, generally divides constraints into Faithfulness and Markedness (Prince & Smolensky 1993) where Markedness means roughly “disfavored structure” of some sort. Samuels has done phonologists a favor by collecting contributions from various scholars who critically examine the concepts of phonological markedness, and in many cases, conclude by rejecting the notion. The nine chapters are as follows.
David Odden writes on “Markedness in substance-free and substance-dependent phonology.” His opening sentence appropriately sets the tone for the book: “‘Markedness’ has a decent claim to being the most disagreed-on notion in phonological theory.” Odden observes that markedness has in general been applied to rules and representations, but also to frequency of occurrence – and that these are two unrelated domains which resist a unified treatment. He traces the notions of markedness in detail from Trubetzkoy to Jakobson to early generative theory and Chomsky and Halle (1968). Their SPE approach to markedness boils down to computational simplicity: some relationships, whether segment inventories or rules, are formally simpler than others, and so are unmarked. Odden points out that in SPE a simpler rule is expected to be more common; however, this confuses the domains, and the formal theory really makes no claims on probabilities. For example, changing a plus to a minus in feature specifications of a rule is formally allowed by the mechanism of SPE, but this often leads to highly unnatural and unattested rules.
Turning to more contemporary models like Optimality Theory, Odden asserts that the concept of markedness is even less coherent than in previous models. OT constraints are generally of two types: Correspondence and Markedness. Correspondence constraints deal with relations between two levels, and Markedness “simply refers to anything that isn’t Correspondence.” So in OT, markedness has no relationship to previous notions of markedness. Inside and outside of OT circles, markedness concepts are still being discussed, especially in questions of asymmetry, underspecification and privativity, in which the feature that is present is by definition marked; and this may still be worth exploring. Odden concludes that phonological patterns which have been explained on the basis of markedness are also explainable on the basis of functional causes, possibly phonetically or historically based ones. Since these are independently needed anyway, it calls into question the need for the existence of an independent notion of markedness in phonological theory.
Charles Reiss’ chapter is on “Contrast is irrelevant in phonology: A simple account of Russian /v/ as /V/” The title of this paper is a bit misleading, as it depends on a specific definition and domain of contrast. For Reiss, “phonology” is the phonological computation system, but does not include segmental inventories, which he relegates to the domain of the “lexicon.” He writes (p. 26) that “the phonology doesn’t ‘care’ about contrast, because it has no mechanism to do so.” He notes that a rule of the form s → z / __ d could be either an allophonic rule or a neutralizing rule, depending on the underlying consonantal inventory. So the form of the rule does not “care” about whether there is an underlying contrast between s and z. However, he does admit that contrast is used as a method to determine different underlying forms (e.g. maz vs. mas, p. 28). He has argued before that markedness is incoherent (Hale and Reiss 2008).
Reiss spends the bulk of the chapter in a detailed discussion of Russian. For final devoicing, he adopts a strategy of first deleting the [+voiced] feature, and then by separate rule, inserting [-voiced]. He does a similar two-step process for voicing assimilation before another obstruent. The Russian /v/ acts sometimes as an obstruent and sometimes like a sonorant. Reiss proposes that /v/ is [-sonorant], but underlyingly unspecified for [voice]. It does undergo final devoicing by insertion of [-voice], but crucially, does not trigger voicing assimilation, by having no [+voicing] specification at that point of the derivation. It does, however, undergo voicing assimilation. All this follows from underspecification of [voice]. He ends with the statement that his analysis of Russian lack of contrast “perhaps constitutes an argument against markedness as a useful notion.”
Juliette Blevins’ contribution is titled “What are grammars made of?” She starts by debunking three ideas of phonological universals that have been proposed for phonological grammars: Distinctive Feature Theory, the sonority hierarchy, and the prosodic hierarchy. Distinctive features are not seen as innate and universal, as in SPE (Chomsky & Halle 1968). She gives examples where Distinctive Feature Theory fails. Features do exist, but can be viewed as emergent and possibly language-specific, as in Mielke (2008). For the sonority hierarchy, Blevins demonstrates that different languages do not treat a segmental sequence the same in terms of sonority, and again suggests that sonority scales are language-specific and thus learned properties. The prosodic hierarchy (Phonological Phrase > Prosodic Word > Foot > Syllable), with strict layering, has been claimed to be universal and innate, but Blevins cites cases where the Foot level, for example, can be skipped. Again, her proposal is that prosodic hierarchy can be language-specific. These proposed universals are viewed rather as broad and common tendencies.
Turning specifically to markedness, Blevins notes that the phonological properties of markedness listed, for example, in Haspelmath 2006 do not follow automatically from the type of markedness constraints common to Optimality Theory. She examines two specific claims of unmarkedness – that of coronals and of CV syllables, and cites Northwest Mekeo, which has no coronals, as counter-evidence to the former claim. She discusses several phonetic qualities of coronals that would make them extremely common, but points out difficulties with attempts to incorporate these phonetic explanations into a formal phonological grammar. The tendency to have elements that are stable and easy to produce and perceive can also be extended to sign languages. For open syllables, Blevins cites two languages that require closed syllables word-finally. The tendency for CV syllables can again be explained in terms of stability and ease of production and perception, rather than markedness. These are grammar-external factors, not built into the formal grammar.
Bert Vaux and Bridget Samuels write on “Consonant epenthesis and markedness.” This concrete contribution focuses in on one particular phenomenon that directly relates to markedness: if a consonant is epenthesized in a language, which one is it, and why? The answers often have been related to the assumed unmarked status of coronals, or, if available, pharyngeal, especially glottal stop (e.g. Lombardi 2002). This has been undermined by the study of Hall (2013) on the rarity or even non-existence of epenthetic [r] outside of English and Uygur (but see Cahill 2007 for an OT account of both epenthetic [ʔ] and flap [r] in a single language).Vaux and Samuels discuss the analysis of r-insertion in English dialects and in Uygur in some detail, concluding that it is indeed insertion, but not glide insertion (and that English is a case of hypercorrection of the r-deletion rule).
They discuss various formal proposals on markedness by Lombardi, de Lacy, and Steriade, finding both theoretical and empirical defects in them. Crucial for any views of markedness is the extremely wide variety of consonants that have actually been found to be epenthetic, including those with velar and labial places. The proposals examined are all constraint-based systems which thus include markedness as an inherent part of their theoretical apparatus. Vaux and Samuels, in light of the seemingly endless variety of epenthetic consonants, instead propose a rule-based system which is not limited by putative universal markedness constraints. They conclude with a proposal for English that includes a rule of r-deletion ordered before a rule of r-insertion, reflecting the historical hypercorrection that is the source of the phenomenon. This accords with their view that much of synchronic phonology is the result of historical processes that are frequent because of the frequency of their phonetically-motivated sound changes, a la Blevins (2004).
Edoardo Cavirani and Marc van Oostendorp contribute a chapter called “On silent markedness.” These authors, unlike the previous ones, maintain a type of markedness as a useful concept, but propose a different domain for it. Parallel to empty nodes in morphosyntactic theory, they propose empty positions in phonology, and a markedness hierarchy for these. They begin by reviewing the case for empty categories in syntax, such as an empty PRO subject, concentrating on the reasons scholars have proposed empty structures. They then review the Government Phonology (GP) approach to empty positions. For example, the Arabic lexeme ktb has empty nuclei that can sometimes be filled, e.g. ktəb ‘he/she writes.’ (Note: I do object to their labeling the vowel-less ktb in Arabic a “word”, since it is never pronounced without vowels.) The Final Empty Nucleus (FEN) assumption is that there is an (empty) nucleus after a final consonant cluster. In words like “drink,” the final [k] is regarded as the onset of a syllable followed by an empty nucleus, and this empty nucleus explains why clusters like [ŋk] are followed by vowels.
Following Turbidity Theory (originally proposed to account for opacity in Optimality Theory), the authors assume the independent and potentially asymmetric relations: (a) an element projects to a segment in the lexicon, but (b) a segment projects to an element in the phonetics. Thus when an element is inserted, there is a pronunciation relation (b) but no projection relation (a). This schema enables theoretical differentiation between a purely empty vowel, a “phonologically contentful silent vowel,” and a pronounced schwa, for example. They exemplify their approach by examining word-final devoicing in Dutch and vowel reduction and deletion in Italian. They conclude that some empty positions are “emptier” than others, that is, there is a markedness hierarchy present, and suggest future research investigate the relation between empty content in phonology and in syntax.
Kuniya Nasukawa contributes “The phonetic salience of phonological head-dependent structure in a modulated-carrier model of speech.” Like Cavirani and van Oostendorp above, Nasukawa finds a connection between morphosyntax and phonology in the area of markedness. After reviewing some common diagnostics for unmarked segments, he notes that in the foot structure of the word “water,” the first syllable is considered the head, as evidenced by stress. That is, the perceptually more salient element is the head. However, in more modern syntactic theory, in syntactic structures like the DP “the backyard,” the head is the determiner “the” and it is the non-head “backyard” that is more important than the head in conveying information. And it is the non-head “backyard” rather than the head “the” that is stressed and thus more prominent. He contrasts a sonority-driven approach with his modulated carrier model. In sonority, the loudness or amplitude is paramount, but in the modulated carrier model, it is the magnitude of modulation that counts. So in sonority terms, a plosive consonant is not very sonorous, but in modulated carrier terms, the modulation of the signal is bigger for plosives than vowels or even other consonants.
This chapter maintains that phonetic prominence of head/dependent structures functions analogously in both phonology and syntax. He proposes that the head/dependent role in “water” and such be reversed to match syntactic properties. Thus the first syllable is the dependent, though information-rich, and the second syllable, though information-poor, is the head. The size of modulation is less in the structural head. The mechanism proposed avoids the issue of markedness altogether, and presumably, though not explicitly stated, provides another angle on the superfluity of markedness.
Shanti Ulfsbjorninn writes on “Markedness and formalising phonological representations.” This paper is the only one which goes beyond segments, dealing with syllables. Ulfsbjorninn acknowledges the criticisms of markedness made by those in this volume, as well as others such as Haspelmath 2006, but maintains that markedness, at least “markedness by complexity,” still has explanatory value, at least for syllable structure. Though markedness is emergent, it still has its source in the phonological component. Ulfsbjorninn’s markedness is extragrammatical in that it is not used to directly compute surface forms, unlike markedness constraints in OT.
After a summary of other scholars’ lists of putative functions and properties of markedness, the author dives into specifics of syllable structure, specifically, Strict CV theory, in which the fundamental unit of syllables is CV. Word-final consonants thus end in empty nuclei, and vowel-initial words begin in empty onsets. (Like Cavirani & van Oostendorp’s paper in this volume, Strict CV theory has its origin in Government Phonology.) In this approach, syllable typology can be stated in terms of onset and coda parameters: the more positive parameter settings, the more marked the syllable. Markedness is thus described in terms of complexity of description: the more empty categories, the more marked. He presents an extended example of how this would apply in the case of consonant clusters, which are traditionally viewed as marked structures, as evidenced by language acquisition and language pathology. He details consonant clusters in English in this schema, and eventually presents a detailed table of eight types of languages ranked in ascending complexity of CC behavior, from Yoruba (no parameters positive) to Polish (all 8 parameters positive). He goes through all eight types of languages in his schema, and concludes that markedness as complexity does have a formal phonological definition.
Mathias Sharinger contributes “Are there brain bases for phonological markedness?” This paper goes beyond the usual phonology to explore various brain imaging methods and the light they may shed on markedness. He forthrightly says in his first paragraph that “Such results… cannot be taken as neuroscientific proof for the existence of markedness.” However, his results show that reducing markedness to pure frequency measures is not warranted by most of the data, and that markedness is not so easily dispensed with.
After a review of what various researchers mean by “markedness,” Sharinger reviews brain imaging methodologies. Functional magnetic resonance imaging (fMRI) can measure blood deoxygenation in very specific spatial areas of the brain, and this is associated with neural activity. There have been a multitude of fMRI studies that Sharinger cites that deal with processing abstract phonological units such as features, phonemes, syllables, and words. Electroencephalography (EEG) is more temporally-based than fMRI, and measures electrical activity of the brain. Studies show increased electrical brain activity when the person is presented with deviant stimuli, as opposed to more standard ones.
Sharinger examines 1) feature-based representational markedness, relating to underspecification and markedness (e.g. Steriade 1995), 2) markedness measured by co-occurrence patterns and violations, and 3) markedness caused by morphological variations. For the first, EEG measures show an asymmetry when the order of unspecified/specified stimuli was varied. There was no evidence for the claim that the less marked segment was more frequent, but an underspecification account fared better than a pure markedness account. The other two cases are not examined in as much detail, but illicit co-occurrences such as word-initial [lb] were marked by decreased brain processing, as does irregular morphology such as English past tense “keep/kept” which produce marked phonological structures such as coda clusters. He cautions that more definite conclusions of future studies must be based on a firm understanding of phonological markedness, of brain function, and the connection between these, and we have not reached that point yet.
Pedro Martins writes the last chapter on “There is no place for markedness in biologically-informed phonology.” This briefest paper in the volume, like some others, begins by citing Haspelmath’s 2006 paper against markedness, and notes that though it has not been refuted, most phonologists continue to refer to markedness. Martins asserts that language is a biological system and can be understood in biological terms. On this basis, he maintains that claims that markedness is unique to language are not only implausible, but difficult if not impossible to investigate; there is no link between genes and any linguistic properties. His view that “markedness” in biological terms may not be unique to language or indeed, to humans at all, also has drawbacks. In these expositions, he mostly is speculating rather than adducing concrete evidence. In response to a reviewer, Martins discusses the syntactic notion of Merge, which is unexplainable in biological terms, but he takes Merge to be more established than markedness.
I interpret his “biological” as to be understood as purely “materialistic: rather than any non-physical component, though he doesn’t use that specific term. And he writes (p.225) “In fact, linguistic properties at large, even those that we can define in a much more satisfactory manner than markedness, stand in the way of creating linking hypotheses between linguistic behavior and biology and neuroscience.” The lack of a direct connection between biology and linguistics is a frustration to him, since he proceeds primarily from a biological starting point.
The contributions to the volume are generally well-written and clear, though some terms like “substance-free phonology,” “I-phonology,” and “extra-descriptive grammars” may not be transparent to those who have not been engaged with some aspects of current phonological discussions.
The volume deals only with phonological markedness, where the bulk of linguistic discussion has centered. There also have been contributions to the concept of morphological markedness, from Zwicky 1978 to Bale et al 2011 and Pertsova 2015, and undoubtedly others. It would be interesting to see how the contributors to this volume deal with markedness in this other linguistic domain.
Also, all the contributions deal with segmental markedness (though Ulfsbjorninn includes syllables). Having a contribution that deals specifically with tone would round out the contributions nicely. For example, in many Bantu languages, tone is privative, with only High tone being phonologically active (e.g. Hyman 2001). How does this contribute to the markedness debate?
Even in the segmental realm, there is a limited selection of markedness phenomena examined. Specific topics examined in detail include coronal place (Blevins, Vaux & Samuels), CV syllables (Blevins), consonant epenthesis (Vaux & Samuels), devoicing (Cavirani & van Oostendorp), vowel reduction/deletion (Cavirani & van Oostendorp), and syllables (Ulfsbjorninn). Besides tone, other phenomena could have been examined in detail, such as vowel qualities (and inventories), nasality, etc. Each of these authors rejects a traditional notion of markedness, based on the particular phenomenon they present. Because there are empirical shortcomings to markedness in some areas, and there are other explanations for at least some putative markedness, then the conclusion is that markedness does not exist. Of course, since there are phenomena they do not examine, it is a jump to reject all markedness for all phenomena. The purpose of the book is not explicitly to disprove markedness, as some papers do not do that. If such were the case, then a more exhaustive set of topics would need to be examined.
A question that deserves more discussion is markedness as universal vs. language-specific markedness. Several authors in the volume discuss other literature that asserts that markedness can be language-specific, e.g. Hume & Tserdanelis (2002), Hume (2003), where a labial rather than a coronal is unmarked. How many of the arguments against markedness here and elsewhere are in actuality arguments against universal markedness? Tone offers an example: usually High is marked (by criteria of active spreading, not being default, etc.) but in Bora and Hausa it is reversed – the Low is the tone with these behaviors.
Reiss’ paper only mentions markedness as an afterthought, not directly, and thus seems a bit out of sync with the main topic of this volume. Martins’ paper is the most general (almost philosophical) and the least evidence-based paper.
The volume is well-edited for the most part, with an occasional typo like “delection” for “deletion” on page 76, and “ffactors” rather than “factors” on page 229. There is an IPA mistake on page 167, with [sɪkəs] rather than [sɪkθs]. More serious is that the references in Reiss’ paper to Reiss (2017), Rice (1993), Samuels (2011), Shapiro (1993), and Wiese (2000) are not included. It appears that the last page of references was simply omitted. Also, Advanced Tongue Root for some reason in Reiss’ paper is abbreviated Atr rather than ATR (p. 29).
The index is nicely done; when I needed to look up several items, they were there.
The majority of contributors to this volume (Odden, Reiss, Blevins, Vaux & Samuels, Martins, perhaps Nasukawa) are convinced that markedness has no place in formal phonology, and their arguments have merit; the phenomena they deal with can be explained by historical developments or perceptual and production factors. But it is often clumsy to refer to these varied factors in write-ups. So I wonder if phonologists, though aware of this, might keep on referring to markedness as shorthand for these features. As Samuels 2011:18 (cited in Martins’ paper) writes, “…when we speak of markedness, we are really using shorthand for a number of deeper factors that are in large part extralinguistic.” This shorthand referral system would be analogous to referring to High tone as H rather than by its tonal features [+upper, +raised], for example. Thus as much as it may frustrate many of the authors of this volume, I suspect that the term “markedness” is not going away in the near future.
Bale, Alan, Gagnon, Michael & Khanjian, Hrayr. 2011. On the relationship between morphological and semantic markedness; The case of plural morphology. Morphology 21: 197–221.
Blevins, Juliet. 2004. Evolutionary Phonology. Cambridge: Cambridge University Press.
Cahill, Michael. 2007. Aspects of the Morphology and Phonology of Konni. Dallas: SIL International.
Chomsky, Noam, and Morris Halle. 1968. The Sound Pattern of English. New York, NY: Harper & Row.
Hale, Mark and Charles Reiss. 2008. The Phonological Enterprise. New York, NY: Oxford University Press.
Hall, Tracy Alan. 2013. How common is r-epenthesis? Folia Linguistica 47:55–87.
Haspelmath, Martin. 2006. Against markedness (and what to replace it with). Journal of Linguistics 42.1: 25–70.
Hume, Elizabeth. 2003. Language specific markedness: The case of place of articulation. Studies in Phonetics, Phonology, and Morphology 9.2: 295–310.
Hume, Elizabeth, and Georgios Tserdanelis. 2002. Labial Unmarkedness in Sri Lankan Portuguese Creole. Phonology 19.3: 441–58.
Hyman, Larry M. 2001. Privative Tone in Bantu. In Kaji, Shigeki (ed.), Crosslinguistic Studies of Tonal Phenomena: Tonogenesis, Japanese Accentology, and Other Topics. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies, pp. 237–257.
Lombardi, Linda. 2002. Coronal epenthesis and markedness. Phonology 19: 219–252.
Mielke, Jeff. 2008. The emergence of distinctive features. Oxford: Oxford University Press.
Pertsova, Katya. 2015. Interaction of morphological and phonological markedness in Russian genitive plural allomorphy. Morphology 25: 229–266.
Prince, Alan, & Paul Smolensky. 1993. Optimality Theory: Constraint interaction in generative grammar. Ms, Rutgers University & University of Colorado, Boulder. (Published 2004, Malden MA: Blackwell.)
Rice, Keren. 2007. Markedness in Phonology. In Paul de Lacy (ed.) The Cambridge Handbook of Phonology. 79-97, Cambridge: Cambridge University Press.
Samuels, Bridget. 2011. Phonological Architecture: A Biolinguistic Perspective, Vol. 2. Oxford: Oxford University Press.
Steriade, Donca. 1995. Underspecification and markedness. In The Handbook of Phonological Theory, John Goldsmith (ed.) 114-174. Oxford: Blackwell.
Zwicky, Arnold. 1978. On Markedness in Morphology. Die Sprache 24.2: 22–27. https://web.stanford.edu/~zwicky/on-markedness.pdf
ABOUT THE REVIEWER
Mike Cahill did field research in Ghana for several years, and is currently SIL's Orthography Services Coordinator. He is a phonologist by training and inclination (Ph.D., Ohio State University), especially focused on tone and labial-velars. He has recently been concentrating on linguistic and non-linguistic factors involved in developing orthographies for previously unwritten languages, and advising on these.
Page Updated: 07-Sep-2018