sign in sign up
<<
Home , Consonant harmony, Vowel harmony

MFM 21, May 24, 2013, University of Manchester

Consonant harmony and harmony: Local interactions Comparisons in typology and sources for nonlocality • In the temporal dimension, locality serves as a basis for relating segments, such that adjacent segments are more prone to mutual Rachel Walker restrictions. University of Southern California [email protected] • in adjacent syllables can be considered articulatorily adjacent (Gafos 1996, Ní Chiosáin & Padgett 2001). 1. Introduction (Proximity has also been treated in a scalar fashion, with more stringent restrictions levied over more proximate segments. See, e.g., Suzuki 1998, Action-at-a-distance Walker 2000c, Hansson 2001, 2010, Pulleyblank 2002.)

A notable feature of many harmony systems is their capacity for • (2) Concentrate here on sources for nonlocal interactions among action-at-a-distance. segments • Long-standing questions surround what gives rise to nonlocal segmental interactions and the mechanisms that make them i. Similarity – spatial symmetry possible. • Agreement by Correspondence (ABC) approach (Walker 2000a, b, 2001, Hansson 2001, 2010, Rose & Walker 2004). versus ii. Prominence differences – prosodic/positional antisymmetry Prominence-based Licensing approach (Walker 2011). • It would be appealing to unify the analysis of vowel harmony • (VH) and consonant harmony (CH) systems. • Additional topics that will be touched upon as sources of • Nevertheless, VH and CH show typological differences in some nonlocality: areas pertaining to o Inventory structure and contrast. o Interacting segments o Triggers that are weak in quality. o Transparent segments o Blocking effects (3) Preview • The first two areas of difference are of particular interest because • Consonant harmony they relate to harmony systems’ potential for long-distance o Similarity serves as a basis for segmental interactions that can interactions. be nonlocal.

(1) Two-fold aims for this talk o Captured in similarity-driven ABC approach. i. Probe the relational dimensions and structures that enable • Vowel harmony interactions among nonadjacent segments. o Proposals have been made to analyze VH in the ABC approach. ii. Examine areas of typological difference between VH and CH.

1 MFM 21, May 24, 2013, University of Manchester

o Three areas where ABC is problematic for VH patterns are (5) Nasal variants Oral suffix variants discussed. -nik-ini ‘grind’ -suk-idi ‘wash’ i. Inventory-sensitive sources for potentially nonlocal -nat-ini ‘carry’ -bak-idi ‘catch’ interactions. Points to need for forms of sensitivity to contrast. -leem-ene ‘shine’ -sos-ele ‘search for’ ii. Prominence differences as a basis for potentially nonlocal interactions. -futumuk-ini ‘revive, rise’

iii. Biases for triggers with perceptually weak qualities. The consonants that interact with nasals in Kikongo nasal CH are • those that are most similar to nasals. 2. Consonant harmony • This criterion for the interacting segments differs from nasal V-C (4) CH typology harmony, where the segments that are most prone to interact with nasals are vowels, and voiced stops interact with nasals only when • CH refers to patterns in which consonants are required to agree in some property, where the consonants can be separated by at least all sonorant segments do also (Walker 1998).

one . (6) Similarity as a source of nonlocality • A core insight emerging from typological studies of consonant Similarity serves as a type of spatial symmetry between segments harmony is that similar consonants are more prone to interact • (Hansson 2001, 2010, Rose & Walker 2004). in the sense that spatial properties of the segments’ representations are structurally similar (e.g. location of constriction, degree of Example: Kikongo (Bantu) stricture, etc.). (Bentley 1887, Meinhof 1932, Dereau 1955, Ao 1991, Odden 1994, Piggott • A large body of work has suggested that similarity promotes 1996, Rose & Walker 2004) interactions among segments. • Kikongo shows nasal CH between nasals and following voiced • More specifically, it has been proposed that similarity can give oral stops and /l/. rise to the formation of relations between segments that can transcend adjacency. • The perfective active suffix, with oral variants [–idi] and [–ele], shows variants with a when preceded by a nasal in (Walker 2000a, b, 2001, 2009, Hansson 2001, 2006, 2010, Rose & the stem (5). Walker 2004, Gallagher & Coon 2009, Sasa 2009, McCarthy 2007, 2010, Inkelas 2008, Rhodes 2010a, Bennett 2013; related observations are • Nasal harmony operates across intervening vowels and voiceless made in van de Weijer 1994, MacEachern 1997, Suzuki 1998, consonants, which remain oral. Pulleyblank 2002, Zuraw 2002, Frisch et al. 2004, Ajíbóyè & Pulleyblank 2008, Shih 2013, a.o.) In addition to engendering alternations in , the • generalization holds that nasals are not followed by {b, d, g, l} in stems.

2 MFM 21, May 24, 2013, University of Manchester

Agreement by Correspondence approach • Evaluated locally over chain-adjacent pairs (Hansson 2007). In (Walker 2000a, b, 2001, 2009, Hansson 2001, 2007, 2010, Rose & Walker other , in a corresponding sequence […S1…S2…S3…], 2004) IDENT-XX(F) enforces identity for [F] between S1 and S2 and between S and S . Correspondence is established between similar segments in an 2 3 • output. (10) Ranking structure for harmony driven by ABC • Feature identity requirement in corresponding segments brings about agreement. CORR-XX IDENT-XX(F) (Details of analysis here follow Rose & Walker 2004.) IDENT-IO(F) (7) Agreement by Correspondence

(11) Illustration: a. ni i k i di i b. ni i k i ni i | | | | CORR-XX(+cons, –cont, +voi), IDENT-XX(nas) >> IDENT-IO(nas) [+nas] [–nas] [+nas] [+nas] /-nat-idi/ CORR-XX IDENT-XX IDENT-IO (+cons, –cont, +voi) (nas) (nas) [F] identity * (violated) ✓ (obeyed) in correspondents a.  niatinii * b. niatidii *! (8) CORR-XX constraints c. niatidji *! • Assign a penalty to any pair of Xs that are not in correspondence

in the output. (12) Voiceless stops do not participate • Correspondence is similarity driven. Xs can be restricted so that IDENT-IO(F) dominates the constraint that enforces they share certain features. CORR-XX(αF), enforces • correspondence between X’s that are specified [αF]. correspondence between nasals and all oral stops: (See McCarthy 2010 for a different proposal.) ni a t i ni i ≻ ni a ni i ni i • Implicational relationship: If two segments sharing certain features j are required to correspond, then segments that are more similar in via IDENT-IO(nas) >> CORR-XX(+cons, –cont) the same dimension are also required to correspond. Captured by

fixed rankings or stringent constraints. (13) Transparency versus blocking

(9) IDENT-XX(F) constraints • Non-harmonizing segments in consonant harmony are typically transparent (Hansson 2001, 2010, Rose & Walker 2004). • Assign a penalty to any pair of corresponding Xs in the output that do not match in specification for [F].

3 MFM 21, May 24, 2013, University of Manchester

• ABC has the capacity to capture blocking behavior in harmony in 3.1 Transparency in VH that appears to be inventory-sensitive specific circumstances. The highly specific nature of the ranking Example: Lokaa (Niger-Congo) (Akinlabi 2009) configuration is suggested by Hansson (2010) to minimize the attestation of blocking in CH. (16) Vowel inventory • The blocking segment stands in correspondence with a potential [+ATR] [–ATR] trigger or a potential target, but it interferes with a correspondence i u chain linking segments that are expected to harmonize (Hansson e ə o ɛ ɔ 2007, 2010, Walker 2009, Rhodes 2010a). a

(14) Summary: Agreement by Correspondence Lokaa ATR harmony • Makes three basic predictions about harmony: • Lokaa exhibits regressive ATR harmony in the PrWd among paired mid vowels.1 i. Harmony is favored among segments that are similar (Agreement By Correspondence; ABC). (17) [+ATR] [–ATR] lè-kó ‘war’ lɛ̀-dɛ̀ ‘greeting’ ii. Segments that are transparent are dissimilar from triggers for ò-bèlè ‘water pot’ l -t ‘giant ant’ harmony (Transparency by Lack of Correspondence; TLC). ɔ̀ ɛ́ è-tòm ‘life’ ɛ̀-kɔ̀ ‘cloth’ iii. Segments that block harmony show some similarity with triggers or targets (Blocking By Correspondence; BBC) • [ə] and [a] trigger ATR harmony in preceding mid vowels (18a-b).

3. Vowel harmony • [a] blocks [+ATR] harmony from following vowels (18c).

• The status of [ə] is “indeterminate” with respect to whether it is (15) VH and ABC transparent or blocks harmony (Akinlabi 2009:197). Studies have shown that it is possible to analyze some VH systems • (18) a. [ə] triggers harmony for [+ATR] using ABC (Hansson 2006, Sasa 2009, Walker 2009, Rhodes 2010a). è-də́l ‘louse’ • This work has shown that the ABC approach has the capacity to ò-də́m ‘man, male’ capture triggering, blocking, and transparency effects, even when they all occur in the same system. b. [a] triggers harmony for [–ATR] ɛ̀-tál ‘kite’ • Despite these capabilities, there are questionable aspects of the -k ‘needle’ predicted typology when it comes to VH. ɔ̀ à

1 This description characterizes the basic system of Lokaa’s ATR harmony. For more details, see Akinlabi (2009).

4 MFM 21, May 24, 2013, University of Manchester

c. [a] blocks [+ATR] harmony (21) For an input /e-CiCɔC/ k -plákòn ‘afternoon’ ɛ́ i. Winner vs. candidate where nonhigh vowels do not correspond ɛ́-kàpóóŋ ‘pepper’ ɛi - C ij C ɔi C ≻ ei - C ij C ɔk C • High vowels are transparent to ATR harmony (19). via CORR-XX(–high) >> IDENT-IO(ATR) (19) [+ATR] [–ATR] ii. Winner vs. candidate where nonhigh vowels correspond but do not è-sìsòŋ ‘smoke’ ɛ̀-sísɔ̀ŋ ‘housefly’ agree ó-yínə́ ‘story’ lɔ̀-títàŋ ‘rib’ ɛi - C ij C ɔi C ≻ ei - C ij C ɔi C kè-túkə̀ ‘running’ ɔ̀-túːmá ‘need’ via IDENT-XX(ATR) >> IDENT-IO(ATR) kó-fúkə́ ‘counting’ kɛ̀-fúkà ‘gathering (things)’ iii. Winner vs. candidate where high and nonhigh vowels correspond Why could we expect high vowels in Lokaa to be transparent? and agree • High vowels lack [–ATR] counterparts, i.e. there is no [ATR] ɛi - C ij C ɔi C ≻ ɛi - C ɪi C ɔi C contrast among the high vowels. via IDENT-IO(ATR) >> CORR-XX(+voc)

The history of Lokaa’s vowels (22) Recapitulation of comparisons in (21) • Lokaa’s high vowels are believed to have historically presented a /e-CiCɔC/ CORR-XX IDENT-XX IDENT-IO CORR-XX [±ATR] contrast (Akinlabi 2009), which was found in Benue- (–high) (ATR) (ATR) (+voc) Congo high vowels (Stewart 1971, Williamson 1973). a.  ɛi-CijCɔiC * ** • The historical contrast is evidenced in some high vowel stems of b. i-CijC k *! *** Lokaa which select [–ATR] (20b), analyzed by Akinlabi e ɔ C with a floating [–ATR] feature. c. ei-CijCɔiC *! ** (20) a. è-kpí ‘rat’ b. ɔ̀-kpí ‘viper’ d. ɛi-CɪiCɔkC **! lé-dù ‘trash’ kɛ̀-yú ‘riches’ • According to the intuition behind the ABC approach, high vowels Transparency in Lokaa and ABC do not interact in ATR harmony with mid vowels, because they • Could be captured by enforcing correspondence among nonhigh are not sufficiently similar to them. vowels only, using CORR-XX(–high), and ranking the constraint • Yet this strategy does not make a connection to the lack of ATR that drives correspondence among all vowels (CORR- contrast in Lokaa’s high vowels. XX(+vocalic)) below the relevant faithfulness constraint. (On the feature [vocalic], see Nevins & Chitoran 2008 and Padgett 2008)

5 MFM 21, May 24, 2013, University of Manchester

(23) Integrating a role for contrast? (25) For an input: /e-CɪCɔCuC/ • Rhodes (2010a) has a proposal in development where CORR-XX constraints can reference the feature strength in a segment, which The output could be: ɛi - C ij C ɔi C uj C Arrows highlight can depend on its contrastive status. harmonizing Vs • The question of whether transparent segments are always non- contrastive for the harmonizing feature has a long history. (26) CORR-XX(–high) CORR-XX(+high) IDENT-XX(ATR)

(Work in the last decade alone includes the following, Krämer 2003, Calabrese 2005, Dresher 2009, MacKenzie 2009, 2011, Nevins 2010, and IDENT-IO(ATR) Rhodes 2010b, a.o.) | CORR-XX(+voc) • The issue depends in part on how contrastivity is calculated, which is debated in some of these works. (27) Illustration of constraint hierarchy in (26) • Even if CORR-XX constraints were sensitive to the contrastive /e-CɪCɔCuC/ CORR- CORR- IDENT- IDENT- CORR-XX status of a feature, this leaves unclear whether the similarity- XX XX XX IO (+voc) driven basis for correspondence and action-at-a-distance would (–high) (+high) (ATR) (ATR) remain intact. a.  i-CijC i ujC ** ******** ɛ ɔ C (24) A further issue b. ɛi-CɪjCɔiCukC *! * **********

• The ABC approach predicts a harmony system where high vowels c. ei-CijCɔkCujC *! * ********** that contrast for ATR are transparent to harmony among mid d. ei-CɪjCɔiCujC *!* ******** vowels. At a minimum, this is typologically rare.

• If a language had an inventory like Lokaa’s but additionally (28) No cross-height harmony: retained the ATR contrast among high vowels, high vowels could IDENT-XX(high), IDENT-IO(high) >> CORR-XX(+voc): be rendered transparent to ATR harmony among mid vowels in /e-CɪCɔCuC/ CORR-XX IDENT- IDENT- IDENT- CORR-XX the same way as in (21–22) above. (–high)/ IO XX IO (+voc) (ATR) (high) (high) • In fact, it is possible to generate a height-stratified ATR harmony (+high) and transparency system, where high vowels harmonize but mid- a.  ɛi-CijCɔiCujC ** ******** vowels are transparent, and mid vowels harmonize but high vowels are transparent (see (25–26)). b. ei-CiiCoiCuiC ** *!**

c. ii-CiiCuiCuiC ** *!*

6 MFM 21, May 24, 2013, University of Manchester

(29) Summary: Inventory sensitivity • Intervening vowels can be transparent. • Transparent vowels in unbounded VH tend to be those that lack a • Boxed forms in (32), with disharmonic stems, show that backness contrast for the harmonizing feature. harmony can operate across full vowels with opposite specifications for back. • The mechanics of similarity-driven ABC do not line up well with this typological characteristic of VH, although there may be ways (32) a. Nom. sg. 2 pl. poss. suffix to elaborate the approach to make it a better fit. Front Back • The capacity of the ABC apparatus for conditional transparency, ém-dæ ‘medicine’ uβér-ta ‘news’ where a vowel may be transparent to harmony from one kind of tʃødrǽ-tæ ‘forest’ kutkó-ta ‘ant’ trigger but not another, appears problematic. tʃijǽ-tæ ‘paint’ olək-tá ‘meadow’ Note: Whether this prediction is problematic for CH also needs attention, tyrə-tǽ ‘edge’ rəwəʒ-tá ‘fox’ but it seems to me to be more likely to be attested in CH patterns. b. Dative suffix 3.2 Nonlocal interactions in VH sensitive to prominence asymmetries Front Back meráŋ-læn ‘hare’ keŋéʒ-læn ‘summer’ (30) Interactions among vowels that show prominence asymmetries imɲə-lǽn ‘horse’ olmá-lan ‘apple’ • Vowels in contexts that are asymmetrically weak in prominence kyzə-lǽn ‘knife’ munə-lán ‘egg’ can harmonize with vowels in strong positions (e.g. initial, stressed, root). (33) Why a similarity-driven ABC approach does not seem promising 2 The harmonizing vowels reside in positions of asymmetric • Some patterns show action-at-a-distance (Walker 2011). • prominence rather than in symmetric positions. Example: Eastern Meadow dialect of Mari (Uralic) (Vaysman 2009) o Initial syllables serve as privileged or prominent in numerous languages (see, e.g., Casali 1997, Beckman 1998, Smith 2002, 3 (31) Vowel inventory Nelson 2003, Walker 2011). i y u o Final syllables show mixed effects, and are often associated e ø ə o with weakness (Barnes 2006, Walker 2011). æ a o Further, a separate pattern of round harmony in Eastern Meadow Mari (not shown here) operates between stressed • Vowels in the final syllable assimilate in backness to the vowel in vowels and final syllables, displaying a prosodic asymmetry. the initial syllable (32). • Similarity in vowel quality does not provide a basis for identifying 2 See Majors 1998 on stress-controlled VH patterns suggested to originate in the trigger and target for back harmony. The transparent vowel can , which are not expected to show action-at-a-distance. be of the same quality as a target, e.g. /meráŋ-lan/  [meráŋ-læn] 3 [æ] occurs only in non-initial syllables and is possibly non-phonemic. ‘hare.’

7 MFM 21, May 24, 2013, University of Manchester

(34) Other VH patterns with prominence asymmetries (37) CH and Prominence-based Licensing? • Other systems show VH between positions of asymmetric strength • It has been claimed that prosody does not play a role in CH where harmony is controlled by the vowel in weak position. systems (Hansson 2001, 2010). • Examples that can operate among nonadjacent vowels include the • Apart from stem control, positional asymmetries do not seem to be dialect of Lena (Neira Martínez 1955, 1983, Hualde 1989, 1998) and a major conditioning factor in CH. the dialect of Ascrea (Fanti 1938–1940, Maiden 1991, Walker 2011), Since vowels are the segments that chiefly and most commonly both involving stressed vowels assimilating to unstressed. • express prosodic properties, prosodic sensitivity in harmony is

expected to most often involve vowels. However, this does not (35) Prominence-based Licensing analysis rule out prosody-sensitive CH outright. • It has been proposed that prominence differences can form the basis for segmental interactions that are potentially nonlocal. (38) Summary: Prominence asymmetries • Such interactions have been formalized in terms of Prominence- • VH systems exist that display interactions between vowels with based Licensing. systematic differences in positional prominence. • Prominence-based Licensing constraints assign a penalty to • These patterns are not in accord with the predictions of a distinctive segmental properties that are expressed only in non- similarity-driven ABC approach. prominent contexts. • Instead, proposals have been made where prominence • Such constraints can drive between strong and weak asymmetries can serve as a source for potentially nonlocal positions, operating from weak-to-strong or strong-to-weak. interactions, distinct from similarity-based interactions.

They also can drive neutralization, deletion, and . • 3.3 Biases for VH triggers that show weakness in vowel quality (Ringen & Vago 1998, Walker 2005, 2011, Revithiadou et al. 2006, Jiménez & Lloret 2007, Lloret 2007, Kaplan 2008; for foundation, see Itô (39) Example: Round harmony typology (Kaun 1995, 2004) 1986, Goldsmith 1990, Lombardi 1991, Steriade 1995, Zoll 1998.) • Front vowels are more prone to trigger round harmony (also (36) Nonlocal licensing-driven interaction nonhigh vowels and short vowels). • Specification for [–back] in strong (s) initial position licenses • Typologically preferred triggers for unbounded round harmony are [-back] in weak (w) final position those for which rounding has comparatively weak acoustic cues.

m es r á ŋ –l æw n • Nonhigh round vowels tend to be avoided, as the product of | | harmony and otherwise (but note Hungarian; Jurgec 2011, Törkenczy [–back] [–back] 2011).

8 MFM 21, May 24, 2013, University of Manchester

Example: Kazakh (Turkic) (Korn 1969, Kaun 1995) (42) Weak trigger analysis • Kazakh shows front/back harmony. • The prevailing insight about asymmetries in this pattern derives from a bias for weak triggers, i.e. front round vowels in Kazakh. • Round harmony among front vowels is unrestricted (40a). o Front vowels trigger harmony even if they produce rounding • Among back vowels, round harmony operates only to high targets in a following nonhigh vowel. (40b). Boxed words show where harmony does not occur. • Accounts sensitive to triggers that are weak in quality have been (40) a. køl-dy ‘lake’ ACC pursued in various treatments of VH. køl-dø ‘lake’ LOC yj-dy ‘house’ ACC • These include claims that the perceptual advantage achieved by yj-dø ‘house’ LOC extending or repeating cues for a segment whose quality is perceptually weak can provide a basis for nonlocal interactions. b. koj-du ‘sheep’ ACC son-dan ‘rubble’ ABL *son-don • Various constraints have been proposed that promote harmony kul-du ‘servant’ ACC between a trigger that is weak in quality and potentially nonlocal kul-da ‘servant’ LOC *kul-do targets. Exx. ALIGN(F-domain) (Cole & Kisseberth 1995a, b), EXTEND(F) (Kaun (41) Difficulties for an ABC approach 1995), ALIGN(F) (Kaun 2004), SPREAD(F) (Walker 2005, Kimper 2011), LICENSE(F, V) (Jiménez & Lloret 2007, Walker 2011). • A lack of correspondence between back round vowels and /a/ does not seem feasible on the basis of similarity. (43) Nonlocal interaction triggered by a weak vowel quality o Correspondence relations would exist for parallel front vowels • Schematic illustration: [+round] in a short vowel is related to in the inventory. adjacent and nonadjacent vowels (Baiyina Oroqen; Walker to appear). o Harmony operates from [o] to high vowels, necessitating correspondence between back vowels that differ in height and rounding. V • Vː • V | | | • The failure of back nonhigh vowels alone to be targets does not [+rd] [+rd] [+rd] seem to be markedness based.

o Nonhigh round vowels are generally accepted to be more (44) Weak triggers in harmony involving consonants marked than high round vowels (*ROLO, Kirchner 1993, Kaun 1995), but [ø] is more marked than [o] (*ROFRO, Kaun 1995). • Weak trigger effects have been suggested for nasal V-C harmony (Cole & Kisseberth 1995b, Walker to appear). • Long-distance laryngeal restrictions in consonants have been suggested to have a basis in strengthening cues for contrasts

9 MFM 21, May 24, 2013, University of Manchester

(Gallagher 2010), pointing to possible weak trigger effects in some o Whether this long-standing insight can be effectively CH phenomena. integrated in a similarity-driven ABC approach remains an open question. (45) Summary: Triggers that are weak in quality ii. Positional prominence differences • The typology of VH for rounding shows biases for weak triggers. o Patterns of bounded VH show interactions between vowels in positions of asymmetric prominence (e.g. Eastern Meadow • Round harmony in Kazakh does not show the characteristics of a pattern based on similarity. Mari). o Harmony driven by prominence differences can be local, but • Enlarging the opportunities for perception of a weak segmental transparency can occur for intervening elements that are not quality (even apart from non-prominent positions) has been relevant to the prominence asymmetry, suggested as another source for potentially nonlocal interactions. iii. Weak trigger biases 4. Discussion and closing o Unbounded VH shows biases for triggers in which the harmonizing feature is perceptually weaker, as in the typology (46) The ABC approach to harmony of round harmony (e.g. Kazakh). • Originally developed for CH, but has been suggested as a basis for o The perceptual advantage earned by expressing the at least some patterns of VH. harmonizing feature in contexts beyond the trigger can promote harmony in adjacent and nonadjacent elements. • Predicts harmony will preferentially operate among similar segments, even if they occur at a distance. (48) Two general results • Predicts transparent segments will be comparatively dissimilar a. Differences exist between CH and VH from triggers. Typological differences between CH and VH of the kind (47) ABC and VH highlighted here are suggestive that CH and VH are not easily unified in full under a single mechanism, such as ABC. • The typology of VH departs from what is predicted about which segments harmonize and which can be transparent if it were b. There are distinct sources of nonlocality consistently attributed to ABC. Factors that play into nonlocal interactions include similarity, positional prominence asymmetries, improving perceptibility, and i. Inventory sensitivity contrast. o In unbounded VH, vowels that are prone to be transparent are those that lack a contrast for the harmonizing feature (e.g. Lokaa).

10 MFM 21, May 24, 2013, University of Manchester

(49) Locality asymmetries in CH and VH Beckman, J. 1998. Positional Faithfulness. PhD dissertation, University of Massachusetts, Amherst. (Published by Garland, New York, 1999.) • VH is more common in the languages of the world than CH. Bennett, W. 2013. , Consonant Harmony, and Surface Correspondence. PhD dissertation, Rutgers University. • This could have a basis, at least in part, from intrinsic locality Bentley, W. H. 1887. Dictionary and Grammar of the Kongo Language. London: differences in vowels versus consonants. Baptist Missionary Society. Calabrese, A. 2005. Markedness and Economy in a Derivational Model of • Vowels in VCV sequences are usually articulatorily adjacent, Phonology. Berlin: Mouton de Gruyter. despite the overlaid consonantal articulation (unless the C involves Casali, R. 1997. Vowel in contexts: Which vowel goes? Lg 73, 493– an interfering articulation) (Gafos 1996, Ní Chiosáin & Padgett 2001, 533. Mahanta & Grijzenhout 2004). Cole, J. & C. Kisseberth. 1995a. Paradoxical strength conditions in harmony systems. NELS 25, 17–31. • However, consonants in CVC sequences are not articulatorily Cole, J, & C. Kisseberth. 1995b. Nasal harmony in optimal domains theory. adjacent. This difference, plus the cross-linguistic markedness of WECOL 7, 44–58. clusters of three or more consonants, severely diminishes the Dereau, L. 1955. Cours de kikongo. Namur: A. Wesmael-Charlier. opportunities for local CH in contrast to local VH.4 Dresher, B. E. 2009. The Contrastive Hierarchy in Phonology. Cambridge University Press. Fanti, R. 1938; 1939; 1940. Note fonetiche e morfologiche sul dialetto di Ascrea (50) Outlook (Rieti). L’Italia dialettale 14, 201–218; 15, 101–135; 16, 77–140. • In future research it would be valuable to probe further into how Frisch, S., J. Pierrehumbert & M. Broe. 2004. Similarity avoidance and the OCP. typological properties of CH and VH are connected to intrinisic NLLT 22, 179–228. differences with respect to consonants and vowels Gafos, A. 1996. The Articulatory Basis of Locality in Phonology. PhD i) in their adjacency dissertation, Johns Hopkins University. (Published New York: Garland 1999.) ii) in their relevance to different sources of nonlocality. Gallagher, G. 2010. The Perceptual Basis of Long-Distance Laryngeal Restrictions. PhD dissertation, MIT. Gallagher, G. & J. Coon. 2009. Distinguishing total and partial identity. NLLT 27, References 545–582. Goldsmith, J. A. 1990. Autosegmental and Metrical Phonology. Oxford: Ajíbóyè, Ọ. & D. Pulleyblank. 2008. Mọ̀bà nasal harmony. Ms., University of Lagos and University of British Columbia. Blackwell. Akinlabi, A. 2009. Neutral vowels in Lokaa harmony. CJL 54, 197–228. Hansson, G. Ó. 2001. Theoretical and Typological Issues in Consonant Harmony. Ao, B. 1991. Kikongo nasal harmony and context-sensitive underspecification. LI PhD dissertation, University of California, Berkeley. 22, 193–196. Hansson, G. Ó. 2006. Locality and similarity in phonological agreement. Paper Barnes, J. 2006. Strength and Weakness at the Interface: Positional Neutralization presented at the workshop on Current Perspectives on Phonology. Indiana in Phonetics and Phonology. Berlin: Mouton de Gruyter. University, June 23, 2006. Hansson, G. Ó. 2007. Blocking effects in agreement by correspondence. LI 38, 395–409.

4 Hansson, G. Ó. 2010. Consonant Harmony: Long-Distance Interaction in Note that vowel-consonant harmony could operate among temporally adjacent Phonology. Berkeley: University of California Press. segments, as in V-C nasal harmony or emphasis harmony, and plausibly also Hualde, J. I. 1989. Autosegmental and metrical spreading in the vowel-harmony coronal harmony (for an articulatory study, see Walker et al. 2008). systems of northwestern Spain. Linguistics 27, 773–805.

11 MFM 21, May 24, 2013, University of Manchester

Hualde, J. I. 1998. Asturian and Cantabrian . Rivista di Linguistica 10, Maiden, M. 1991. Interactive Morphonology: Metaphony in Italy. New York: 99-108. Routledge. Inkelas, S. 2008. The dual theory of reduplication. Linguistics 46, 351–401. Majors, T. 1998. Stress Dependent Harmony: Phonetic Origins and Phonological Itô, J. 1986. Syllable Theory in Prosodic Phonology. PhD dissertation, University Analysis. PhD dissertation, University of Texas, Austin. of Massachusetts, Amherst. (Published, New York: Garland, 1988.) McCarthy, J. J. 2007. Consonant harmony via correspondence: evidence from Jiménez, J. & M.-R. Lloret. 2007. Andalusian vowel harmony: Weak triggers and Chumash. Papers in Optimality Theory 3, ed. by, L. Bateman, M. O’Keefe, perceptibility. Paper presented at the 4th OCP, Rhodes, January 18–21, 2007. E. Reilly & A. Werle, 223–237. Amherst, MA: GLSA. Jurgec, P. 2011. Feature Spreading 2.0: A Unified Theory of Assimilation. PhD McCarthy, J. J. 2010. Agreement by Correspondence without CORR Constraints. dissertation, University of Tromsø. Ms., University of Massachusetts, Amherst. (Available as ROA-1089.) Kaplan, A. 2008. Noniterativity is an Emergent Property of Grammar. PhD Meinhof, C. 1932. Introduction to the Phonology of the (trans. dissertation, University of California, Santa Cruz. by N. J. van Warmelo). Berlin: Dietrich Reimer/Ernst Vohsen. Kaun, A. 1995. The Typology of Rounding Harmony: An Optimality Theoretic Neira Martínez, J. 1955. El habla de Lena. Oviedo: Instituto de Estudios Approach. PhD dissertation, UCLA. Asturianos. Kaun, A. 2004. The typology of rounding harmony. In Phonetically Based Neira Martínez, J. 1983. La metafonia por /-i/ en los Bables centrales. In Phonology, ed. by B. Hayes, R. Kirchner & D. Steriade, 87–116. Cambridge Philologica Hispaniensia in Honorem M. Alvar, Vol. I. Editorial Gredos: University Press. Madrid. Kimper, W. 2011. Competing Triggers: Transparency and Opacity in Vowel Nelson, N. 2003. Asymmetric Anchoring. PhD dissertation, Rutgers University. Harmony. PhD dissertation, University of Massachusetts, Amherst. Nevins, A. 2010. Locality in Vowel Harmony. Cambridge, MA: MIT Press. Kirchner, R. 1993. Turkish vowel harmony and disharmony: an optimality Nevins, A. & I. Chitoran. 2008. Phonological representations and the variable theoretic account. Paper presented at the Rutgers Optimality Workshop 1, patterning of glides. Lingua 118, 1979–1997. Rutgers University, New Brunswick, NJ, October 22, 1993. Ní Chiosáin, M. & J. Padgett. 2001. Markedness, segment realization and locality Korn, D. 1969. Types of labial harmony in the . Anthropological in spreading. In Segmental Phonology in Optimality Theory, ed. by L. Linguistics 11, 78–92. Lombardi, 118–156. Cambridge University Press. Krämer, M. 2003. Vowel Harmony and Correspondence Theory. Berlin, Odden, D. 1994. Adjacency parameters in phonology. Lg 70, 289–330. Germany: Mouton de Gruyter. Padgett, J. 2008. Glides, vowels, and features. Lingua 118, 1937–1955. Lloret, M.-R. 2007. On the nature of vowel harmony: Spreading with a purpose. Piggott, G. L. 1996. Implications of consonant for a theory of Proceedings of the XXXIII Incontro di Grammatica Generativa, ed. by A. harmony. CJL 41, 141–174. Bisetto & F. E. Barbieri, 15–35. Dipartimento di Lingue e Letterature Pulleyblank, D. 2002. Harmony drivers: No disagreement allowed. BLS 28, 249– Straniere Moderne, Università di Bologna. 267. Lombardi, L. 1991. Laryngeal Features and Laryngeal Neutralization. PhD Revithiadou, A., M. van Oostendorp, K. Nikolou & M.-A. Tiliopoulou. 2006. dissertation, University of Massachusetts, Amherst. (Published, New York: Vowel harmony in contact-induced systems: The case of Cappadocian and Garland, 1993.) Silly. In Proceedings of the 2nd International Conference on Modern Greek MacEachern, M. 1997. Laryngeal cooccurrence restrictions. PhD dissertation, Dialects and Linguistic Theory, ed. by M. Janse, B. Joseph & A. Ralli, 350– UCLA. (Published 1999, New York: Garland.) 365. University of Patras. Mackenzie, S. 2009. Contrast and Similarity in Consonant Harmony Processes. Rhodes, R. 2010a. Vowel harmony as Agreement by Correspondence. Ms., PhD dissertation, University of Toronto. University of California, Berkeley. Mackenzie, S. 2011. Contrast and the evaluation of similarity: Evidence from Rhodes, R. 2010b. A survey of vowel harmony with special focus on neutral consonant harmony. Lingua 121, 1401–1423. segments, glides, and diphthongs. Ms., University of California, Berkeley. Mahanta. S. & J. Grijzenhout. 2004. Nasals blocking harmony in Assamese. In Ringen, C. O. & R. M. Vago. 1998. Hungarian vowel harmony in Optimality UiL-OTS Yearbook, ed. by A. Kerkhoff, J. de Lange, & O. Sadeh Leicht, 25– Theory. Phonology 13, 393–416. 35. Rose, S. & R. Walker. 2004. A typology of consonant agreement as

12 MFM 21, May 24, 2013, University of Manchester

correspondence. Lg 80, 475–531. Weijer, J. van de. 1994. Stem consonant cooccurrence restrictions in Ngbaka. Sasa, T. 2009. Treatments of vowel harmony in Optimality Theory. PhD Linguistics in the Netherlands, ed. by R. Bok-Bennema & C. Cremers, 259- dissertation, University of Iowa. 266. Amsterdam: John Benjamins. Shih, S. 2013. Consonant- interaction as Agreement-by-Correspondence. Ms., Williamson, K. 1973. Some reduced vowel harmony systems. Research Notes 6, Stanford University and University of California, Berkeley. 145–169. Department of Linguistics and Nigerian Languages, University of Smith, J. 2002. Phonological Augmentation in Prominent Positions. PhD Ibadan. dissertation, University of Massachusetts, Amherst. (Published, New York: Zoll, C. 1998. Parsing Below the Segment in a Constraint-Based Framework. Routledge, 2005.) Stanford: CSLI. Steriade, D. 1995. Underspecification and markedness. In The Handbook of Zuraw, K. 2002. Aggressive reduplication. Phonology 19, 395–439. Phonological Theory, ed. by J. A. Goldsmith, 114-174. Oxford: Blackwell. Stewart, J. M. 1971. Niger-Congo, Kwa. In Current Trends in Linguistics, Vol. 7: Linguistics in Sub-Saharan Africa, ed. by T. A. Sebeok, 179–212. Berlin: Mouton. Suzuki, K. 1998. A Typological Investigation of Dissimilation. PhD dissertation, University of Arizona. Törkenczy, M. 2011. Hungarian vowel harmony. The Blackwell Companion to Phonology, ed. by M. van Oostendorp, C. J. Ewen, E. Hume, & K. Rice, 2963-2989. Oxford: Wiley-Blackwell. Vaysman, O. 2009. Segmental Alternations and Metrical Theory. PhD dissertation, MIT. Walker, R. 1998. Nasalization, Neutral Segments, and Opacity Effects. PhD dissertation, University of California, Santa Cruz. (Published, New York: Garland, 2000.) Walker, R. 2000a. Long-distance consonantal identity effects. WCCFL 19, 532– 545. Walker, R. 2000b. Yaka nasal harmony: Spreading or segmental correspondence? BLS 26, 321–332. Walker, R. 2000c. Yaka nasal harmony: Spreading or segmental correspondence? Handout of talk presented at BLS 26, Feb. 9, 2000. Walker, R. 2001. Consonantal correspondence. Proceedings of the Workshop on the Lexicon in Phonetics and Phonology. Papers in Experimental and Theoretical Linguistics, Vol. 6, 73–84. Edmonton: Department of Linguistics, University of Alberta. Walker, R. 2005. Weak triggers in vowel harmony. NLLT 23, 917–989. Walker, R. 2009. Similarity-sensitive blocking and transparency in Menominee. Paper presented at the annual meeting of the Linguistic Society of America, San Francisco, CA, January 9, 2009. Walker, R. 2011. Vowel Patterns in Language. Cambridge University Press. Walker, R. To appear. Nonlocal trigger-target relations. LI. Walker, R., D. Byrd & F. Mpiranya. 2008. An articulatory view of Kinyarwanda’s coronal harmony. Phonology 25, 499–535.

13