AUTHOR: Blevins, Juliette
TITLE: Evolutionary Phonology
SUBTITLE: The Emergence of Sound Patterns
PUBLISHER: Cambridge University Press
Grover Hudson, Department of Linguistics and Germanic, Slavic, Asian and African
Languages, Michigan State University
The occasion of this review is the 2007 paperback reissue of the 2004 first
edition. Blevins argues that ''the most common sound changes and the most common
types of synchronic alternations are nearly coextensive'' (4), so ''...if we can
demonstrate that principled diachronic explanations exist for particular sound
patterns, considerations of simplicity would seem to dictate that explanations
for the same phenomena should not be imported into, or otherwise duplicated
within, synchronic accounts'' (5). According to Blevins, ''regular sound change is
the locus of naturalness in phonology'' (295), and the only contribution of
innateness to acquisition is ''the feature system, prosodic organization, and
their combinatorics'' (22). Another purpose of the book is to present and argue
for a three-part theory of hearer-driven sound change. The book's title is not
particularly suggestive of these purposes.
The book has three parts: ''Preliminaries'' (3 chapters), ''Sound patterns'' (5
chapters), and ''Implications'' (3 chapters). Following is a brief
chapter-by-chapter review of contents, followed by an evaluation.
Chapter 1 ''What is Evolutionary Phonology'' (EP, below) deplores ''a general
extension of synchronic descriptions and mechanisms to encompass nearly all
patterns and generalizations within a linguistic system, irrespective of their
status or origin.'' This nevertheless ''invariably fails to explain - and often
fails even to describe accurately - many of the sound patterns that recur in the
world's languages'' (3). ''The majority of commonly attested sound changes in the
world's languages are mirrored by synchronic alternations of precisely the same
type,'' and ''if we can demonstrate that principled diachronic explanations exist
for particular sound patterns, considerations of simplicity would seem to
dictate that explanations for the same phenomena should not be imported into, or
otherwise duplicated within, synchronic accounts'' (4). For example, ''grammars do
not need to explain the absence of voiced obstruents in final position'' (5).
Blevins gives a list of would-be phonological universals (9-10), and briefly
explains these as results of sound change; for example, ''in a series of voiced
stops /b d g/, /g/ is most likely to be missing'', and short vowels are not
stressed to the exclusion of long vowels in a language with both.
Blevins critiques some inadequate explanations of synchronic grammarians, such
as for inventories with [b] and [d] but not [g] the (unattributed) constraint
*g, which is unassociated ''with the arguably related fact that in contexts of
devoicing, /g/ is more likely to devoice than /d/ or /b/'' (12), and a constraint
against stress on short and not long vowels, which is ''unlikely to arise'' by
''perceptual principles'' (13). ''[W]here Grammont explains sound change in terms
of the interplay'' of ''the law of least effort'' and ''the need for clarity'', EP
''acknowledges the accidental nature of change'' (16) (well, if not exactly
''accidental'', unpredictable). The latter is a critical point insufficiently
emphasized by Blevins: whereas sound changes explain their resulting sound
patterns, the articulatory and perceptual biases of articulation and perception
can't explain sound change any better than they predict it. ''Absolute universals
and universal tendencies in sound patterns emerge from general pathways of
language change, and have no independent status in the grammar...; [f]ew, if
any, markedness principles proposed in phonology encode anything more than
statistical probability, [and] [t]here are counterexamples to nearly every
universal constraint or principle'' (20).
Chapter 2 ''Evolution in language and elsewhere'' presents language change as
metaphorically like natural selection: small variations are selected in
unpredictable ways. The big difference concerns the nature of selection, which
is (at least ordinarily) adaptive in biology but only very controversially so in
language, where selection is more like cultural change (as noted by Andersen
2006: 170), and the selected variants may even come from another language, for
social not physiological reasons. But Blevins is unconcerned with sound change
owed to contact.
Blevins presents (32-33) a ''general typology of sound change'' in EP, in which
sound changes not owed to contact are one of three types, alliteratively but
rather unhelpfully termed 'change', 'chance', and 'choice'.
(1) 'Change' happens when the phonetic signal is misheard, so that e.g. [anpa]
is analyzed as [ampa].
(2) 'Chance' happens when ''the phonetic signal is accurately perceived'' but,
being ''phonologically ambiguous'', is analyzed differently by speaker and hearer.
The example concerns a speaker's /a?/ (? = glottal stop) pronounced with an
allophonic glottal-stop onset and laryngealized vowel. The hearer analyzes this
as /?a/ vs. the speaker's /a?/.
(3) 'Choice' happens when ''multiple phonetic...variants of a single phonological
form are [again] accurately perceived'' but differently analyzed by speaker and
hearer, exemplified by /kakata/ with penultimate stress and three
pronunciations: with (a) syncopated first vowel, (b) reduced first vowel, and
(c) full first vowel. The speaker's phonemic form has the full vowel, but the
hearer chooses (a), /kkata/. As an argument against 'optimality' in sound
change, Blevins thinks ''ease of articulation favors /kkata/...[but] maintenance
of perceptual contrast favors /kakata/'' (45).
Blevins emphasizes the source of all three changes in the hearer, referring to
much work by John Ohala. (All three, by the way, are 'abductive changes'
according to Andersen 1973). 'Choice' is over cases of intraspeaker variability
on the scale of hyper-to-hypo-articulated (casual-to-careful) speech (referring
to 'H&H theory' of Lindblom 1990). Even though both 'chance' and 'choice' have
no necessary effect on form, consisting only of differences of analysis by
speaker and hearer, Blevins thinks they do prejudice change of form consistent
with the change of analysis. All three types of change emphasize the role of the
hearer vs. that of the speaker, although Blevins acknowledges that, in 'choice',
the speaker's changing token frequency is the 'catalyst' (37) of change, a point
to which we shall return.
Chapter 3 ''Explanation in phonology: a brief history of ideas'' reviews some
history of linguistics, especially the rise of markedness theory, from the
neogrammarians to Grammont, Baudoin, Saussure, Trubetzkoy, Chomsky and Halle,
and Kenstowicz. Blevins refers to Paul’s (1886) claim that ''features of modern
languages were subject to seemingly arbitrary variation, and could only be truly
understood in terms of their history” (6), and quotes Kenstowicz (1994: 1) for
the generativist claim that learners appear to know more in acquisition than
they can get from the data. Blevins (219) refers to the argument of Pullum and
Scholz (2002) against 'poverty of the stimulus' arguments for innateness.
Grammont is credited with the claim that language change satisfies competing
tendencies of 'least effort' (speaker's grammar) and 'need for clarity'
(hearer's grammar), and Ohala (1990: 266) is quoted (74) that ''neither speaker
nor hearer chooses - consciously or not - to change pronunciation... Rather,
variation occurs due to ''innocent'' misapprehensions about the interpretation of
the speech signal'' (which sounds like unconscious choice to me).
Here arises what seems to be the main argument for historical explanation of
synchronic sound patterns and against markedness and other claims of innateness:
that lots of phonology is not natural, for example British English r-insertion
and the spread of uvular /r/ in Europe; both, however, are well-established
sound changes. ''By modeling sound change as a mapping between one synchronic
grammar and the next, synchronic phonology remains a pure non-teleological
system of abstract features and categories, with phonetic explanation limited to
the diachronic dimension'' (85).
Part II ''Sound patterns'', chapters 4-8, discusses five sorts of phenomena all
argued to show synchronic effects owed to sound change, and not to synchronic
markedness expressed as OT-type constraints or synchronic phonological rules:
laryngeal features (ch. 4), place features (ch. 5), ''other common sound
patterns'' (ch. 6), geminates (ch. 7), and ''some uncommon sound patterns'' (ch.
8). Here, briefly, are some important points in these chapters.
Blevins (107) lists and explains four laryngeal-feature sound changes according
to the triad of 'change', 'chance', and 'choice'; but here these are described
as the rise of variants, not the hearer's response to these. Final devoicing is
described as owed to phrase-final lengthening, ''since voiceless consonants are
typically longer than voiced ones''. Regarding place features and although
'syncretization' (indeterminateness) is a recognized characteristic of unmarked
categories (Battistella 1990: 40; cf. sg. he/she/it vs. pl. they), Blevins
argues against the unmarkedness of coronals, noting (126) that their frequency
is partly owed to their several places of articulation from dental to retroflex.
Among ''other common sound patterns,'' Blevins notes that synchronic
vowel-insertion rules, sometimes thought to create unmarked CV sequences, are
often the residue of syncopy rules (e.g. the plural vowel of English 'horses').
Blevins asserts, probably controversially, that ''[i]n the majority of the
world's languages where closed syllables occur, there does not seem to be a
strong preference for sonorants over obstruents'' (162). Surprisingly, Blevins
says ''there is nothing intrinsically difficult about the production or
perception of...CCCCCC in Georgian'' (214). Concerning geminates, Blevins argues
(183-191) against the synchronic determination of geminate integrity and
inalterability (by constraints on linking) and antigemination (by the Obligatory
Part III proposes some implications of EP for synchronic phonology (ch. 9),
diachronic phonology (ch. 10), and, very sketchily, morphology and syntax (ch.
11). Blevins' review of evidence for innateness in child language acquisition is
critical, concluding (231) ''there is very little evidence from child language
acquisition for innate phonological constructs apart from distinctive features
and the prosodic units which function as the domains for stress and intonation
contours. The actual content of phonological representations appears to be
acquired through data-driven learning.''
Prior reviews of this book are De Boer 2006 and Brown 2006. A thorough review
and critique of EP is _Theoretical Linguistics_ 32.2, with a synopsis of the
theory by Blevins and seven critiques to which Blevins replies.
EP's basic claim, that diachrony (sound change) explains sound patterns
(including alternations), which don't then need explanation in synchronic
grammars, was argued by Sampson 1975, an article which Blevins seems unaware of,
and McMahon 2000 argued against the synchronic overuse by Optimality Theorists
of diachronically-valid explanation. See Haspelmath 2006 for arguments that ''the
term 'markedness' is superfluous, because some of the concepts that it denotes
are not helpful, and others are better expressed by more straightforward, less
ambiguous terms'' (Haspelmath, p. 25). Blevins' presentation, however, is a
broad, timely, and effective argument for a thorough reevaluation of markedness
in generative phonology.
Perhaps the principal objection to EP is its absolute claim against innate
markedness constraints. De Lacy 2006 argues for the coexistence of parallel
diachronic and synchronic constraints; see similarly Greenberg's position as
described by Blevins (6), and the argument of Kiparsky 2006 for the absolute
innate constraint that only marked features may be suppressed in weak positions,
so against Blevins' claim of rare final obstruent voicing. Blevins (2006: 251)
says Kiparsky's ''designation of [final] voicing as ‘phonetic implementation’ is
purely arbitrary'', and Iverson and Salmons 2006 discuss ''a direct parallel to
final voicing in structural terms (involving addition of a marked feature)'', a
''robustly and securely attested'' sound change of final aspiration. Andersen 2006
also discusses a ''robustly and securely attested'' sound change of ''final
aspiration, a direct parallel to final voicing in structural terms (involving
addition of a marked feature).'' Andersen 2006 also doubts Blevins' analysis of
the 'final voicing' facts, and faults the EP theory of sound change for failing
to distinguish innovation and change: innovations may happen for natural
reasons, but change only happens when the innovations are adopted (and spread),
as cultural change.
Even if much generative phonological analysis commits the worse error of
exaggerating naturalness in phonology, EP seems to err by restricting innate
principles of articulatory and perceptual naturalness to diachrony. The main
interest of generative phonology has always been to explain productivity, not
sound patterns (grammars are generative). Blevins (22) seems to deny the
importance for innateness of productivity, which ''appears to involve levels of
gradation which are also highly suggestive of learned knowledge''. But
productivity is not graded for allophonic rules, and graded for non-allophonic
rules only over time or over a population. For a single speaker at a moment of
creative language use, productivity is all or none.
Blevins(12), for example, objects to an OT constraint *g (better: [stop, velar,
*voiced]) intended to explain the absence of /g/ in the presence of /b/ and /d/;
Blevins admits a ''simple aerodynamic explanation'' for this. But isn't the same
explanation possibly if not presumptively at work, and reasonably always
lurking, and reinforced in the acquisition of languages which lack [g]? A sound
change can't explain itself.
Furthermore, EP seems to admit synchronic markedness constraints, provided these
arise via ''data-driven learning'' (231), as ''emergent properties'' (237). Kawahara
2006 presents experimental evidence from Japanese for an impossibly emergent
markedness constraint, for which there is no data in the language. Second
language learning presents more such evidence, such as word-final obstruent
devoicing in the English of first-language speakers of Japanese, a language
having no word-final obstruents so no evidence for such a rule (Eckman 1977).
See Wilson 2006 and Berent et al 2007, both evidence against a strong version of
Fortunately, Blevins has left some wiggle-room: ''principled diachronic
explanations for sound patterns replace...synchronic explanations, unless
independent evidence demonstrates, beyond reasonable doubt, that a separate
synchronic account is warranted'' (5). Possible even are ''underlying structures
which do not surface in the corpus of utterances available to the learner''
(absolute neutralization), if these are ''explicitly justified by some type of
explicit evidence or argumentation'' (313).
Finally, the book suffers from lax editing: much poor phrasing as in ''when
careful speech is elicited, one finds that non-released stops are released''
(98), absence of page numbers in references, and 40+ simple errors such as
Venneman for Vennemann. ''Although this is a matter of editing rather than
content'', as de Boer (2006: 705) said of errors just in the references, ''it is
nevertheless regrettable'', and the more so in this reissue three years after the
One error should be corrected here. Blevins (14) says that 'Zipf's Law', ''holds
that the less complicated the phonetic realization of a phoneme, the greater its
frequency''. In fact what is ordinarily termed 'Zipf's Law' in linguistics is his
'Law of Abbreviation' (Zipf 1935: 38): ''the length of a word tends to bear an
inverse relationship to its relative frequency'' (and: ''it seems a plausible
deduction that, as the relative frequency of a word increases, it tends to
diminish in magnitude''). In fact, Zipf (1935: 80-81 - not 96-97, as Blevins
says) qualified his generalization about phoneme complexity and frequency as
''wherever the magnitudes of complexity of phonemes are determinable'', and he
confesses uncertainty about the relative complexity of voiced and voiceless
stops, even though the voiceless are more frequent. Consistent with Zipf's Law
(of word length and frequency) and with Jespersen's (1922: 330) theory of
reductive sound change, it seems impossible to ignore the critical importance in
sound change of the casual-speech reductions of speakers (vs. misanalyses of
hearers), who somewhat teleologically follow Grammont's 'law of least effort' in
circumstances where 'need for clarity' is minimal. Casual-speech forms
reasonably spread (cultural change) because child learners, unaware of their
social significance, extend these beyond the casual register which they first
Andersen, Henning. 1973. Abductive and deductive change. _Language_ 49: 765-793.
Andersen, Henning. 2006. Comments on Juliette Blevins, ''A theoretical synopsis
of Evolutionary Phonology''. _Theoretical Linguistics_ 32: 167-174.
Battistella, Edwin L. 1990. _Markedness: the evaluative superstructure of
language_. Albany: State University of New York Press.
Berent, Iris, Tracy Lennertz, Jongho Jun, Miguel A. Moreno, and Paul Smolensky.
2007. Language universals in human brains. _Publications of the National Academy
of Sciences_ 105.14: 5321-5325.
Blevins, Juliette. 2006. Reply to commentaries. _Theoretical Linguistics_ 32:
Brown, Jason. 2006. Review of Blevins, _Evolutionary phonology_ (2004). LINGUIST
List 17.743 (Mar. 10).
De Boer, Bart. 2006. Review of Blevins, _Evolutionary phonology_ (2004).
_Journal of Linguistics_ 42: 703-708.
De Lacy, Paul. 2006. Transmissibility and the role of the phonological
component. _Theoretical Linguistics_ 32: 185-196.
Eckman, Fred R. 1977. Markedness and the contrastive analysis hypothesis.
_Language Learning_ 27: 315-330.
Haspelmath, Martin. 2006. Against markedness (and what to replace it with).
_Journal of Linguistics_. 42: 25-70.
Iverson, Gregory K. and Joseph C. Salmons. 2006. On the typology of final
laryngeal neutralization: Evolutionary Phonology and laryngeal realism.
_Theoretical Linguistics_ 32: 205-216.
Jespersen, Otto. 1922. _Language: its nature, development, and origin_. New
York: Henry Holt.
Kawahara, Shigeto. 2006. Mimetic gemination in Japanese: a challenge for
Evolutionary Phonology. _Theoretical Linguistics_ 32: 411-424.
Kenstowicz, Michael 1994. _Phonology in generative grammar_. London: Blackwell.
Kiparsky, Paul. 2006. The amphichronic program vs. Evolutionary Phonology.
_Theoretical Linguistics_ 32: 197-204.
Lindblom, Björn. 1990. Explaining phonetic variation: a sketch of the H&H
theory. In _Speech production and speech modeling_, William Hardcastle and Alain
Marchal, eds., 403-439. Dordrecht: Kluwer.
McMahon, April. 2000. _Change, chance, and optimality_. Oxford: Oxford
Ohala, John. 1990. The phonetics and phonology of aspects of assimilation. In
_Papers in laboratory phonology, vol. I, Between the grammar and physics of
speech_, John Kingston and Mary Beckman, eds., 258-275. Cambridge: Cambridge
Paul, Hermann. 1886. _Prinzipien der Sprachgeschichte_. Halle (trans.:
_Principles of Language History_. 1889. New York; Macmillan).
Pullum, Geoffrey K. and Barbara C. Scholz 2002. Empirical assessments of
stimulus poverty arguments. _The Linguistic Review_ 19: 9-50.
Sampson, Geoffrey. 1975. One fact needs one explanation. _Lingua_ 36: 231-239.
Wilson, Colin. 2006. Learning phonology with substantive bias: An experimental and
computational study of velar palatalization. _Cognitive Science_ 30: 945 - 982.
Zipf, George Kingsley. 1935. _The Psycho-biology of language: an introduction to
dynamic philology_. Boston: Houghton Mifflin.
ABOUT THE REVIEWER
Grover Hudson taught phonology, historical linguistics, and Ethiopian
linguistics incuding Amharic language at Michigan State University. He is author
of a comparative dictionary of Highland East Cushitic languages, an introductory
linguistics textbook, with Anbessa Teferra a recent book on Amharic, and
articles on phonology and Ethiopian descriptive and historical linguistics.