a weight problem - vowel length in classical latin (jon brennan).pdf

Upload: ron-gonzales

Post on 14-Apr-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    1/12

    A Weight Problem: Vowel Length in Classical Latin

    Jon BrennanNew York University

    [email protected]

    8th SUNY/CUNY/NYU MiniconferenceNovember 18th, 2006

    Introduction

    Classical Latin exhibits a regular vowel length alternation apparent in nominal, adjectival,and verbal paradigms:1

    (1) Noun (masc. singular): honora. nom honorb. gen hono:risc. dat hono:ri:d. acc hono:rem

    e. abl hono:re:(2) Adjective (masc. singular): stronger

    a. nom fortiorb. gen fortio:resc. dat fortio:rid. acc fortio:reme. abl fortio:re

    (3) Verb (subjunctive): make, doa. 1s feramb. 2s fera:s

    c. 3s feratd. 1p fera:muse. 2p fera:tisf. 3p ferant

    I would like to thank Tuuli Adams, Maria Gouskova, Kyle Major, Jason Shaw, and Cara Shousterman, as wellas three anonymous reviewers, for their thoughtful input regarding the material presented here.

    1Examples are adapted from Lindsay (1896/1963) and Lewis and Kingery (1918), who also offer detailed discussionof the sources of information for vowel quality and length in Classical Latin (see also Allen, 1973).

    1

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    2/12

    Vowel Length in Latin Jon Brennan

    Goals: Characterize the phonological context that conditions this phenomenon, and identifythe relevant markedness and faithfulness constraints responsible for the alternation withinOptimality Theory (Prince and Smolensky, 1993/2002).

    Proposal: The effect of a cross-linguistic markedness constraint against tri-moraic (super-

    heavy) syllables surfaces in non-initial contexts. This is explained in terms of an interactionbetween a constraint against tri-moraic syllables (), and position-specific faithfulnessconstraints (Beckman, 1998).

    Several apparent exceptions suggest that not all coda segments count for weight, that is, somesegments resist being given a mora. The distribution of these exceptions suggests that therelevant dimension characterizing those segments is perceptibility (Steriade, 2001).

    While vowel length in Latin has been extensively discussed in the generative literature withrespect to metrical structure, the vowel length observation noted here is shown to reflect theinteraction of cross-linguistic constraints on syllable structure while also offering insight intonon-sonority-based weight restrictions.

    Outline:

    1. Background

    2. A generalization and the analysis

    3. Exceptions, and a refined analysis

    4. Conclusions

    1 Background

    1.1 Morphology

    The alternation is phonologically conditioned and not due to idiosyncratic morphology.

    For verbs, the vowel immediately preceding the affixed endings belongs to the stem and notto the inflectional endings which are consistent across verb classes and moods:2

    (4) 3p, si. present active indicative 3p, si. present subjunctive indicativeopta-t he/she hopes opte-t he/she should hopeimple-t he/she fills implea-t he/she should fill

    du:ci-t he/she leads du:ca-t he/she should lead

    The weight-alternation could still result from a floating mora associated with the endings,Ill address this in a bit.

    2The relevant vowel is not necessarily part of the root; (4) shows that indicatives and subjunctives differ preciselyin form of the relevant vowel. At present it only matters that this vowel is the same throughout the inflectionalparadigm.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    2

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    3/12

    Vowel Length in Latin Jon Brennan

    For nouns and adjectives, identifying the stem in nominal forms is harder as inflection fornumber/gender/case varies significantly across noun classes.

    While its possible to argue that the vowel quantities here are morphologically fixed, thecontexts in which a final vowel must be short are exactly the same as for verbs. Because the

    alternation is highly predictable, it is desirable to capture the pattern in the grammar.

    1.2 Shortening, not Lengthening

    The alternation is note due to lengthening associated with the inflectional morphology (afloating mora), or to phonological lengthening.

    The following examples are phonologically equivalent to (1) and (3) respectively, but showshort vowels throughout:

    (5) Noun (masc. singular): grove, wood3

    a. nom nemusb. gen nemorisc. dat nemori:d. acc nemorume. abl nemoro:

    (6) Verb (indicative): begin4

    a. 1s incipio:b. 2s incipisc. 3s incipitd. 1p incipimuse. 2p incipitisf. 3p incipiunt

    In other words, short vowels are predictable in the paradigms but long vowels are not.

    1.3 Metrical Structure

    1.3.1 Stress Placement

    Vowel length is not determined by stress.

    The location of main stress can be succinctly stated as follows: A long penult was ac-cented. . . but if the penult was short, the antepenult received the accent. . . [d]isyllables werenecessarily accented on the penult. Kent (1932), cited in Allen (1973).

    (7) shows the paradigm in (3) with stress and syllabification:

    3 Nemus belongs to a group of nouns in which which intervocalic [s] shifted to [r] in all inflected forms. Forpresent purposes, the critical comparison is in the oblique forms, where short [o] surfaces throughout.

    4The first-person singular marker surfaces alternatively -o and -m, correlating with mood.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    3

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    4/12

    Vowel Length in Latin Jon Brennan

    (7) Verb (subjunctive): make, doa. 1 Si. /fera: -m/ fe.ramb. 2 Si. /fera: -s/ fe.ra:sc. 3 Si. /fera: -t/ fe.ratd. 1 Pl. /fera: -mus/ fe.ra:.mus

    e. 2 Pl. /fera: -tis/ fe.ra:.tisf. 3 Pl. /fera: -nt/ fe.rant

    Note in particular (a), (c), and (f), where vowel length of the final syllable varies though eachis unstressed.

    1.3.2 Extrametricality

    Final-syllable extrametricality (Hayes, 1995) does not condition vowel length.

    Latin foot structure is a moraic trochee, and degenerate feet are banned (Prince and Smolen-sky, 1993/2002). So, bisyllabic words will be prosodified into a single foot: ( ) not ().

    Therefore, minimal paris such as fera:s / ferat are both fully footed and dont differ withregards to final syllable extrametricality.

    1.3.3 Iambic Shortening

    Vowel shortening isnt example of iambic shortening under the influence of a preceding shortsyllable (e.g Lindsay, 1896/1963; Allen and Greenough, 1888/1903; Allen, 1973; Mester, 1994).

    (8) cani:s canis (dog), ego: ego (I), male: male (bad)

    The alternation occurs even when the shortened vowel comes after a long vowel 5:

    (9) Verb: leada. 1s du:camb. 2s du:ca:sc. 3s du:cat. . .

    2 Analysis

    2.1 Descriptive Generalization

    The relevant context is apparent when syllabification and prosodic structure are considered:

    5In addition, brevis brevians is most frequently associated only with pre-classical dramatic meters, while thepresent data is characteristic of Classical Latin.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    4

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    5/12

    Vowel Length in Latin Jon Brennan

    (10) Noun (masc. singular): honora. nom /hono:r/ (ho.nor)

    b. gen /hono:r -is/ ho.(no:).risc. dat /hono:r -i:/ ho.(no:).ri:d. acc /hono:r -em/ ho.(no:).rem

    e. abl /hono:r -e:/ ho.(no:).re:(11) Adjective (masc. singular): stronger

    a. nom /fortio:r/ (for.ti).orb. gen /fortio:r -is/ (for.ti).(o:).resc. dat /fortio:r -i:/ (for.ti).(o:).rid. acc /fortio:r -em/ (for.ti.(o:).reme. abl /fortio:r -e/ (for.ti).(o:).re

    (12) Verb (subjunctive): make, doa. 1 Si. /fera: -m/ (fe.ram)b. 2 Si. /fera: -s/ (fe.ra:s)

    c. 3 Si. /fera: -t/ (fe.rat)d. 1 Pl. /fera: -mus/ (fe.ra:).muse. 2 Pl. /fera: -tis/ (fe.ra:).tisf. 3 Pl. /fera: -nt/ (fe.rant)

    While long vowels surface freely in open syllables, vowels are always short if the syllable isclosed by a consonant such as [r], [m], [t], or the cluster [nt]. Long vowels do surface insyllables closed by [s].

    Latin closed syllables are heavy as indicated by weight-sensitive stress placement: CVC andCVV syllables attract stress (e.g Allen, 1973; Mester, 1994).

    Adopting a moraic theory of syllable weight, coda consonants may be assigned a mora whenin coda position (weight-by-position; Hayes, 1989).

    Consequently, (C)VVC syllables are superheavy (Prince and Smolensky, 1993/2002; Zec, 1995;Moren, 1999).

    Now, closed syllables with long vowels appear elsewhere:

    (13) Noun (masc. si.): enjoymenta. nom fru:c.tus

    b. dat fru:c.tu:sc. gen fru:c.tui:d. acc fru:c.tume. abl fru:c.tu:

    (14) Generalization(C)VVCi (trimoraic) syllables are disallowed except word-initially (where Ci doesntinclude [s])

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    5

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    6/12

    Vowel Length in Latin Jon Brennan

    Can we confirm that the positional contrast is initial vs. non-initial?

    1. There are some derivationally complex forms that show no long vowels in medial closedsyllables, though the base form does have a long vowel:

    (15) a. Hope (present active indicative)optam, opta:s, optat, opta:mus, opta:tis, optant

    b. future passive participle: op.tan.dus

    2. Cross-linguistically, initial syllables show more contrasts when compared with medialand final syllables (e.g. Beckman, 1998).

    2.2 OT Analysis

    The generalization in (14) reflects the combined effects of three different kinds of constraints:those enforcing weight-by-position, those mitigating against superheavy syllables, and thoseensuring that the vowel alternation only surfaces non-initially.

    2.2.1 Weight-by-Position

    Weight-by-position effects can be captured using the following constraints (Moren, 1999):

    (16) a. WbyP: Coda consonants be moraic.6

    b. *Mora(C): Do not associate a mora with a consonant.7,8

    c./CVC/ WbyP *Mora(C)

    a.

    CVC *b. CVC *!

    2.2.2 Superheavy Syllables

    Zec (1995) argues that superheavy syllables are cross-linguistically dis-preferred. Within amoraic theory of weight this can be translated into the following constraint:

    (17) : A syllable cant have three moras.

    The fact that the effects of this constraint emerge only in only certain positions suggests aninteraction with position-specific constraints.

    6Equivalently, *Append (Rosenthall and van der Hulst, 1999) or MoraicCoda (Broselow et al., 1997).7*Mora(C) generalizes a set of markedness constraints ranked in a fixed sonority-based hierarchy.8Equivalent to Rosenthall and van der Hulsts (1999) /cons.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    6

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    7/12

    Vowel Length in Latin Jon Brennan

    2.2.3 Positional Constraints

    Within the theory of Positional Faithfulness, (e.g. Beckman, 1997, 1998) faithfulness to oneposition, e.g. Faith(1 ), is ranked above a general markedness constraint which then isranked above general faithfulness.9

    (18) Faith-1 >> Mark >> Faith

    For the present analysis, the relevant faithfulness constraints ensure that vowel length isidentical between the underlying and surface form.

    (19) Ident-Length(1): The length specifications in the first syllable of the inputmatch the length specifications in the output.10

    (20) Ident-Length: The length specifications in the input match the length specifica-tions in the output

    Putting these constraints together, the following tableau illustrates the necessary rankings:

    (21) a. WbyP, Ident-Length(1) >> >> Ident-Length

    b.

    /hono:r/ WbyP Ident-Length(1) Ident-Length

    i. ho.nor *

    ii. ho.no:r *!

    iii. ho.no:r *!

    /fru:ctu:m/

    iv. fru:c.tum * *v. fru:c.tu:m **!

    vi. fruc.tum *! **

    vii. fru:c.tum *! * *

    3 Exceptions and a Refined Analysis

    The ranking in (21b) ensures that no long vowels will surface in closed non-initial syllables,but we have already seen one exception to this generalization:

    vowel shortening is insensitive to syllables closed by [s]

    9Positionally specified markedness (Zoll, 1998; de Lacy, 2001; Gordon, 2004) can also capture the generalization,though only by allowing a markedness constraint to target non-initial syllables. The Latin data alone do not providecompelling evidence for adopting one account of positional constraints over another. It is worth noting, however,that allowing markedness constraints to target non-initial segments conflicts with the conclusion reached by Gordons(2004) positional markedness account of weight effects which addressed data for which initial-syllables more likelycontained moraic codas than non-initial syllables. This was captured by a constraint ensuring moraic codas syllableinitially was ranked above faithfulness constraints. If a positional markedness account were pursued, something of areconciliation between these two accounts would need to be made.

    10See Moren (1999) for a more sophisticated approach to moraic faithfulness.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    7

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    8/12

    Vowel Length in Latin Jon Brennan

    There also exists one other set of exceptions:

    Present active participles and some numerical adverbs have a final syllable with a long vowelfollowed by coda [ns].11

    (22) a. opta:ns hoping

    b. imple:ns fillingc. duce:ns leadingd. incipie:ns beginninge. sentie:ns feeling

    (23) a. sexie:ns six timesb. optie:ns eight timesc. novie:ns nine times

    There are two possible ways to these exceptional segments might be represented in a moraictheory of weight:

    a. The coda consonant is adjoined to the syllable, that is it does not participate in weight-by-position (24a)

    b. The coda consonant shares a mora with the preceding vowel (Broselow et al., 1997) (24b)

    (24) a. b.

    C V V C C V V C

    In OT terms:

    (24a) represents the case where a constraint against non-moraic codas (e.g. WbyP) is violated

    (24b) represents the case where a constraint against mora sharing is violated (e.g. *SharedMora

    Two questions

    1. Can we decide which representation is right?

    2. What makes [s] and [ns] exceptional?

    3.1 Minimal Word Effects In Classical Latin monosyllabic words must be CVV(C) or CVC (word minimality; McCarthy

    and Prince, 1986)

    Prince and Smolensky (1993/2002) provide an analysis in which all content words must beassigned prosodic structure (LxPr), and a constraint forcing feet to be binary at some levelof analysis (FtBinarity). Accordingly, all content words in Latin must be at least bi-moraic.

    11The forms in (23) occurred in free variation with surface forms with final [s] (e.g. sexies, opties)

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    8

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    9/12

    Vowel Length in Latin Jon Brennan

    Prediction If coda [s] and [ns] are adjoined, than mono-syllabic words with an underlyingshort vowel will surface with a long vowel to satisfy FtBinarity.

    This prediction appears to be correct.

    In the examples below the mono-syllabic nominative singular form is given first followed by the

    disyllabic genitive form which is indicative of the vowel quantity in rest of the paradigm. Thepresence of a short vowel in disyllabic forms is taken to indicate that the vowel is underlyinglyshort.

    (25) a. ge:ns, gentis. . . people

    b. mo:ns, montis. . . mountain

    c. pe:s, pedis. . . foot

    d. a:s, assis. . . coin

    The vowel is lengthened just in the case when the when the appears in a mono-syllabic word.

    Mono-syllabic content words closed with other segments dont show the alternation

    (26) a. dux, ducis... leader

    b. sors, sortis. . . sister

    c. urbs, urbis... city

    Something is forcing [s] and [ns] not to be dominated by a mora and isnt violated to satisfyminimality effects.

    A mora-sharing analysis (24b) does not predict that a coda segment must share a mora in an

    otherwise unmarked circumstance, just that it may do so if necessary. Clearly, FtBinaritydoes not force mora sharing.

    On the other hand, mora-bearing consonants are themselves marked (e.g. *Mora(C)) andthe only way to satisfy a constraint of this type is by direct adjunction to the syllable.

    (27)/pes/ *Mora([s]) FtBin WbyP Ident-Length(1

    a. pe:s * *b. pes *!c. pes *!

    Word-minimality requirements favor a syllable-adjunction analysis over a mora-sharing anal-ysis of the exceptional segments.12

    12In response to a question posed by Ellen Broselow during the Miniconference, the possibility of syllable adjunctionis limited to word-final contexts (Hung, 1994), ensuring that medial syllables ending in coda [n] (as in (15b), above)and [s] do count as heavy (i.e. penultimate syllables ending in [s] and [n] do count as heavy for stress placement). Inother words, coda [s] and [n] are only exceptional word-finally.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    9

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    10/12

    Vowel Length in Latin Jon Brennan

    3.2 What is so special abut [s] and [ns]?(a tentative exploration)

    To summarize, the following classes of segments act differently w.r.t. weight effects:

    (28) Exceptional Segments/Sequences:a. Count for Weight: [r, m, t, b, nt]

    b. Dont Count for Weight: [s, ns]

    How can these two classes of segments be characterized? Two approaches: Sonority (Zec,1995) and Perceptibility (Steriade, 2001).

    3.2.1 Sonority

    Zec (1995) and Moren (1999) argue that weight-by-position is conditioned by sonority such

    that less sonorous segments are more marked when counted for weight13

    :

    (29) a. Relative Sonority: Vowels Sonorants Obstruents

    b. Constraint Hierarchy: *Mora(Obs) >> *Mora(Son) >> *Mora(V)

    But, this hierarchy would incorrectly predict that [t] and [s] group together in contrast to [m]and [n].

    What about a finer-grained analysis of sonority (Gouskova, 2004 citing Jespersen, 1904)?

    (30) glides > rhotics > laterals > nasals > vcd frics > vcd stops > vcless frics > vcless stops[r] [n] [m] [b] [s] [t]

    Still cant distinguish the segments [s] and [n] from [t] and [m]!

    Sonority doesnt seem to be able to make the necessary distinctions.

    3.2.2 Perceptibility

    Steriade (2001) argues that the relative perceptibility of segments is reflected in constrainthierarchies.

    Strident segments in particular are highly perceptible, and thus less subject to markednessconstraints.14

    For example, Steriade notes that inter-obstruent stops were generally elided in ColloquialLatin, but interobstruent [s] and nasals was preserved at the expense of a preceding stop. Thehigh perceptibility of [s] making it less susceptible to phonological change.

    13Gordon (2002) argues that coda weight is a function of segment intensity. With regards to the present data,however, this account makes the same predictions as sonority (see Parker, 2002 for discussion)

    14Engstrand and Ericsdotter (1999) offer a phonetic analysis in support of this argument.

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    10

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    11/12

    Vowel Length in Latin Jon Brennan

    Perhaps perceptibility is also responsible for the exceptional behavior of [s] (and [n]) withregards to moraicity. In other words, the hierarchy of constraints prohibiting consonants frombearing moras (i.e. *Mora(C)) is built off of perceptibility. Such a hierarchy explains why*Mora([s]) can be ranked above WbyP in (27) above, while other coda consonants areassigned a mora.

    Intuitively, what is happening in Latin is not that dissimilar from the exceptional capacityfor [s] to appear in complex consonant clusters in English (e.g. sixths).

    What about [n] vs. [m]? In a phoneme identification task Cutler et al. (2004) found [m]was more often misidentified than [n] in coda position. In other words, [n] was more easilyperceived than [m].

    The approach is promising, but would benefit from further evidence that [s] and [n] form anatural class in Classical Latin and elsewhere.

    4 Summary & Conclusions An alternation in non-initial vowel length evidences the emergence of an otherwise inactive

    phonological markedness constraint against trimoraic syllables.

    Vowel length was shown not to be determined by the prosodic conditions of the non-initialsyllable.

    The alternation was captured by ranking a cross-linguistically attested markedness constraint* in relationship with positional faithfulness constraints:

    WbyP, Ident-Length(1

    )>>

    >>

    Ident-Length Examining several apparent exceptions revealed a small set of segments that are forced to be

    non-moraic. The distribution of these segments seems best characterized by a hierarchy ofconstraints built off of perceptibility.

    Appendix

    (31) Summary of Rankings:

    Ranking ResultWbyP >> *Mora(C) Coda syllables are generally moraic* >> Ident-Length Superheavy syllables generally bannedIdent-Length(1) >> * Trimoraic (superheavy) syllables OK initiallyFtBin >> Ident-Length(1 ) minimal words resolved by lengtheningWbyP >> Ident-Length(1) minimal words are not resolved by non-moraic codas*Mora([s],[n]) >> WbyP [n] and [s] are not assigned moras

    8th SUNY/CUNY/NYU MiniConference

    November 18th, 2006

    11

  • 7/29/2019 A Weight Problem - Vowel Length in Classical Latin (Jon Brennan).pdf

    12/12

    Vowel Length in Latin Jon Brennan

    ReferencesAllen, J. H., and J. B. Greenough. 1888/1903. New Latin grammar. Boston: Ginn & Company.Allen, W. Sidney. 1973. Accent and rhythm. Cambridge: Cambridge University Press.Beckman, Jill. 1997. Positional faithfulness, positional neutralization, and Shona vowel harmony. Phonology 14:146.Beckman, Jill. 1998. Positional faithfulness. Doctoral Dissertation, University of Massachusetts, Amherst. URL

    http://roa.rutgers.edu.Broselow, Ellen, Su-I Chen, and Marie Huffman. 1997. Syllable weight: Convergence of phonology and phonetics.

    Phonology 14:4782.Cutler, Ane, Andrea Weber, Roel Smits, and Nicole Cooper. 2004. Patterns of English phoneme confusions by native

    and non-native listeners. Journal of the Acoustic Society of America 116:36683678.Engstrand, Olle, and Christine Ericsdotter. 1999. Explaining a violation of the sonority hierarchy: Stop place

    perception in adjacent [s]. In Proceedings from the XIIth Swedish Phonetics Conference.Gordon, Matt. 2002. A phonetically-driven account of syllable weight. Language 78:5180.Gordon, Matt. 2004. Positional weight constraints in optimality theory. Linguistic Inquiry 35:692703.Gouskova, Maria. 2004. Relational hierarchies in optimality theory: The case of syllable contact. Phonology 21.Hayes, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic Inquiry 20:253306.Hayes, Bruce. 1995. Metrical stress theory. Chicago: University of Chicago Press.Hung, Henrietta J. 1994. The rhythmic and prosodic organization of edge constituents. Doctoral Dissertation,

    Brandeis University.Jespersen, Otto. 1904. Lehrbuch der phonetik. Leipzip & Berlin: Teubner.Kent, R. G. 1932. The sounds of Latin, volume 12 of Langauge Monograph. Baltimore.de Lacy, Paul. 2001. Markedness in prominent positions. In HUMUT 2000, ed. Ora Matushansky, Albert Costa,

    Javier Martin-Gonzalez, Lance Nathan, and Adam Szczegielniak, volume 40 ofMIT Working Papers in Linguistics,5366. Cambridge, MA: MITWPL. URL http://roa.rutgers.edu.

    Lewis, Charlton Thomas, and Hugh Macmaster Kingery. 1918. An elementary latin dictionary. American Book Co.Lindsay, W. M. 1896/1963. The Latin language: An historical account of Latin sounds, stems, and flexions. Oxford:

    Clarendon Press.McCarthy, John, and Alan Prince. 1986. Prosodic morphology. Samizdat, Amherst, and Waltham, MA.Mester, Armin. 1994. The quantitative trochee in Latin. Natural Language & Linguistic Theory 12:161.Moren, Bruce. 1999. Distinctiveness, coercion and sonority: A unified theory of weight. Doctoral Dissertation,

    University of Maryland, College Park.Parker, Stephen G. 2002. Quantifying the sonority hierarchy. Doctoral Dissertation, University of Massachusetts,

    Amherst.Prince, Alan, and Paul Smolensky. 1993/2002. Optimality theory: Constraint interaction in generative grammar.

    Rutgers Optimality Archive ROA-537.Rosenthall, Sam, and Harry van der Hulst. 1999. Weight-by-position by position. Natural Language & Linguistic

    Theory 17:499540.Steriade, Donca. 2001. The phonology of perceptibility effects: The P-map and its consequences for constraint

    organization. Ms. UCLA.Zec, Draga. 1995. Sonority constraints on syllable structure. Phonology 12:85129.Zoll, Cheryl. 1998. Positional asymmetries and licensing. Ms. URL http://roa.rutgers.edu.

    8th SUNY/CUNY/NYU MiniConference

    N b 18 h 2006

    12