<p>Linguistics 251 B. Hayes Phonological Acquisition Spring 2002</p><p>Amahl at Two Years, 60 Days: A Partial Sketch</p><p>1. Strategy</p><p>This is a first plunge, using constraint ranking software, which facilitates speed but not care...</p><p>THE CONSTRAINTS RATIONALIZED</p><p>2. Constraints based on Articulatory Difficulty</p><p>*C = *C This could be just *STRUC. *CC = *CC This is dicey as an articulatory constraint, and I think about revising it. Conjecture: close timing is difficult; in any CC, two things must modulate in carefully synchronized timing. *FRICATIVE = *FRIC Conjecture: maintenance of a precise degree of partial closure is hard. Adults are different, showing lenition of stops to fricatives. FINAL DEVOICING = *[-SON +VC] ## INTERVOCALIC VOICING = *VKV These enforce the vegetative outcome. I’ve ignored the treatment of initials, which are obligatorily voiceless unaspirated—hence no voicing contrast. *[-ANTERIOR] = *[-ANT] By this is meant, only coronal nonanteriors. [+anterior] is closer to the default tongue blade position. Various languages (Tagalog?) lack nonanterior coronals. ONE LINGUAL ARTICULATOR ONLY = 1 LING. ONE ARTICULATOR ONLY = 1 PLACE That is, one per word. The lingual articulators are COR and DORS. As expected, ONE LINGUAL ARTICULATOR ONLY is stricter. Small children seem to employ a “jaw strategy” for articulating consonants (or at least, mine did), which means that ONE ARTICULATOR ONLY would allow the articulators not to move at all in the course of a word. This is my story for why adults don’t have consonant harmony. Sketch of Amahl Apr. 29, 2002 p. 2</p><p>3. Faithfulness Constraints for Perceptually Salient Segments and Features</p><p>Here, structural descriptions refer to the adult form: “if the adult form has this salient characteristic, keep it in the output.”</p><p>MAX(VOICELESSSTOP) = MAX(PTCK) Note that child’s voicing is, however, allophonic. Many languages with no phonemic voicing in obstruents make intervocalic obstruents voiceless, in the face of the articulatory default (Keating, Linker and Huffman 1984, Journal of Phonetics) MAX(PRESONORANT STOP) = MAX(STOP / _ [SON]) See Steriade’s work for extensive discussion of this crucial, highly salient environment. MAX(LIGHTL) = MAX(LIGHTL) Amahl let only the light (prevocalic) l’s survive. See Hayes (2000) “Gradient Well-Formedness in OT” for salience of light [l]. MAX(LONGNASAL) = MAX(LONGNAS) The “short” ones referred to by implication are those that are extremely short, as in sent. MAX(DORS AUTOSEGMENT) = MAX(DORS) MAX(LAB AUTOSEGMENT) = MAX(LAB) Based on the long closure and transitions of dorsals and labials. Note, however, that coronals preferentially survive in other children: Daniel Menn could harmomize dark in either direction, in free variation. See work of Jongho Jun (UCLA diss., Phonology article) on how the greater perceptibility of dorsals and labials governs assimilation patterns. Stated on autosegments, because labiality sometimes hops—cf. squat [gp] ID(PLACE IN PRESONORANT STOPS) = ID(PL:STOP /_[+SON]) See comment under MAX(PRESONORANT STOP) ID(ARTIC IN [-SON]) = ID(ART/[-SON]) Weaker than the above, since more general. ID(PLACE IN POSTTONIC NASALS) = ID(PLACE: N/'V_) This is the weird one, no? So far, I can’t get around this.</p><p>4. A Rhythmic Constraint</p><p>*UPBEAT SYL *UPBEAT Specifically: an initial, unstressed, pretonic syllable. This is famous; cf. work of Gerkin, Carter, and others; suggestion that it is responsible for dropping function words. It seems to hold true in languages like Dutch and English, with big differences in syllable duration, but perhaps not in syllable-timed languages like Spanish and Italian. Sketch of Amahl Apr. 29, 2002 p. 3</p><p>5. Constraint Conjectured to be Part of MD Phonology-in-Progress</p><p>FINAL POST-ATONIC NASAL = N -'IN Generalizing MD’s casual-speech phenomenon to obligatoriness. Since not part of Phonology of Amahl</p><p>6. Weak Faithfulness Constraints, Included Only for Explicitness </p><p>(The analysis would still work without them.)</p><p>MAX(C) MAX(C) ID(CONTIN) ID(CONT) ID(VOICE) ID(VCE) ID([ANTERIOR]) ID([ANT]) ID(ARTICULATOR) ID(ARTIC)</p><p>7. Algorithmically-Generated Ranking</p><p>Stratum Constraint Name Abbr. Stratum #1 MAX(VOICELESSSTOP) MAX(PTCK) MAX(LIGHTL) MAX(LIGHTL) *FRICATIVE *FRIC *UPBEAT SYL *UPBEAT MAX(PRESONORANT STOP) MAX(STOP / _ [SON]) *CC *CC *[-ANTERIOR] *[-ANT] FINAL DEVOICING *[-SON +VC] ## ID(PLACE IN POSTTONIC NASALS) ID(PLACE: N/'V_) FINAL POST-ATONIC NASAL = N -'IN INTERVOCALIC VOICING *VKV ONE LINGUAL ARTICULATOR ONLY 1 LING. MAX(DORS AUTOSEGMENT) MAX(DORS) Stratum #2 MAX(LONGNASAL) MAX(LONGNAS) ID(CONTIN) ID(CONT) ID(VOICE) ID(VCE) MAX(LAB AUTOSEGMENT) MAX(LAB) ID(PLACE IN PRESONORANT STOPS) ID(PL:STOP /_[+SON]) Stratum #3 *C *C ID(ARTIC IN [-SON]) ID(ART/[-SON]) Stratum #4 MAX(C) MAX(C) ID([ANTERIOR]) ID([ANT]) ONE ARTICULATOR ONLY 1 PLACE Stratum #5 ID(ARTICULATOR) ID(ARTIC)</p><p>THE PHENOMENA</p><p>8. Voicing is Allophonic</p><p> following “vegetative” settings (Westbury and Keating, J. Ling. 1986): voiceless unaspirated initially, voiced medially, unvoiced finally. Sketch of Amahl Apr. 29, 2002 p. 4</p><p>Account: because the phonotactics that distributed it (following vegatative settings) outrank faithfulness for [voice]. apple /bu/ bu pu *VKV *! ID(VCE) *</p><p>9. /h/ is Everywhere Deleted</p><p> because there is no Faithfulness constraint strong enough to protect it from *C. As in many dialects, /h/ suffers from its poor audibility. Note also the effect of final devoicing (again, voicing allophonic)</p><p> head: hd t /hd/ t ht d *[-SON +VC] ## *! ID(VCE) * * *C * **! * MAX(C) * *</p><p>10. Nasal-Stop Clusters</p><p> are resolved in favor of the stop if it it is voiced, of the nasal if the stop is voiceless. This accords, I think, with the relative perceptibility of the various elements, especially in English where the nasal is very short before a voiceless consonant.</p><p> bump /bmp/ bp bm MAX(PTCK) *!</p><p> mend /mnd/ mn md MAX(LONGNAS) *! *C ** ** 1 PLACE * *</p><p>11. Dark /l/ is Deleted; Light /l/’s Survive</p><p>Strategy: Faithfulness for light l >> *C >> Faithfulness in general Sketch of Amahl Apr. 29, 2002 p. 5</p><p> hello /hlo/ lo o MAX(LIGHTL) *! *C *</p><p>Compare: ball /bl/ b b l *C * **!</p><p>12. Fricatives are Stopped</p><p> by the standard means, *Fricative >> Faithfulness</p><p>Smith /sm/ mt m *FRIC *! ID(CONT) *</p><p>13. Upbeat Syllables are Lost</p><p> banana: bnan banNote the survival of the stop here, fairly common. Other children preserve adjacency, with [nan] winning. You can’t keep both the /b/ and the /n/ (*[bnan, so the more salient one wins. *[bam candidate bears on harmony, discussed below.</p><p> banana /bna:n/ bam bnan nan bnan ban *UPBEAT *! MAX(STOP / _ [SON]) *! *CC *! ID(PLACE: N/'V_) *! MAX(LONGNAS) * * 1 PLACE * * *</p><p>ID(PLACE: N/'V_) >> 1 PLACE *CC >> MAX(LONGNAS) MAX(STOP / _ [SON]) >> MAX(LONGNAS) *UPBEAT >> MAX(LONGNAS) Sketch of Amahl Apr. 29, 2002 p. 6</p><p>14. Final Post-Atonic Nasals must be [n]</p><p>This guessed to be literal imitation of prestigious adult behavior.</p><p> bottom /btm/ bodin 'bobim bodim -'in *! *! MAX(LAB) * ID(PL:STOP /_[+SON]) *</p><p>15. The Harmony System for Consonant Place</p><p>Alveolars assimilate long-distance to non-alveolars. This depend on three factors: Is the trigger velar or labial? Is the target a nasal or an obstruent? Relative position of trigger and target</p><p>Target Target first Target second velar trigger labial trigger velar trigger labial trigger Obstruent duck [gk] stop [dap] good [guk] Smith [mit] motorcar [mugaga] table [debu] curtain better [bd] [ggn] Nasal snake [ek] clean [gin] workin’ [wgin] nanga [g] nipple [mibu] corner burn [bn] [gn]</p><p> nipple vs. burn is quite interesting, given what we think we know about cues…</p><p>16. Strategy</p><p> Obstruents show a trigger asymmetry; here obtain by penalizing words with two different lingual articulations more harshly than words with two different articulations in general. Thus velars are more stringent triggers. But there was apparently an earlier stage where labials also triggered harmony on alveolars: table [bebu] ~ [debu] at very beginning of the study; this presumably a slightly different ranking. Nasals are the hard nut: my wild guess is that they are happier in post-tonic position. In final post-atonic position, nasals have to be /n/ in any event, imitating adult casual speech. Sketch of Amahl Apr. 29, 2002 p. 7</p><p>17. Target is Obstruent, Target First, Trigger is Velar: Harmony</p><p> dark /dak/ gak dat dak 1 LING. *! MAX(DORS) *! ID(PL:STOP /_[+SON]) * ID(ART/[-SON]) * *</p><p>18. Target is Obstruent, Target First, Trigger is Labial: No Harmony</p><p> table /teb/ debu bebu dedu MAX(LAB) *! ID(ART/[-SON]) *! * *C ** ** ** ID(PL:STOP /_[+SON]) * * 1 PLACE * ID(ARTIC) * *</p><p>1 PLACE is not strong enough to override the Faithfulness constraints on LABIALs and obstruents.</p><p>19. Target is Obstruent, Target Second, Trigger is Velar: Harmony</p><p> good /gud/ gug gud dud MAX(DORS) *! 1 LING. *! ID(ART/[-SON]) * * *C ** ** ** ID(PL:STOP /_[+SON]) * 1 PLACE * ID(ARTIC) * *</p><p>No Faithfulness constraint protects the final /d/ enough to save it from 1 LING, requiring a single lingual articulation.. Sketch of Amahl Apr. 29, 2002 p. 8</p><p>20. Target is Obstruent, Target Second, Trigger is Labial: No Harmony brush: /br/ bt 1. bp 2. bt- 3. rt MAX(STOP / _ [SON]) *! *[-ANT] *! ID(ART/[-SON]) *! 1 PLACE * * ID([ANT]) * *</p><p>ID(ART/[-SON]) >> 1 PLACE *[-ANT] >> ID([ANT]) MAX(STOP / _ [SON]) >> 1 PLACE</p><p>Note also the ban on non-anteriors.</p><p>21. Target is Nasal, Target First, Trigger is Velar: Harmony</p><p> snake /snek/ eek nek net seek xek *FRIC *! *CC *! MAX(DORS) *! 1 LING. *! * *C ** ** ** *** ** 1 PLACE * * MAX(C) * * * * ID(ARTIC) * * * *</p><p> Since the nasal is not posttonic, no Faithfulness constraint is strong enough to block the effects of 1 LING (1 lingual articulation). The final /k/ is inviolate, being velar. The last two candidates represent forlorn attempts to preserve the initial /s/.</p><p>22. Target is Nasal, Target First, Trigger is Labial: Harmony</p><p> knife /naif/ maip naip 1 PLACE *! ID(ARTIC) *</p><p>23. Target is Nasal, Target Second, Trigger is Velar: No Harmony</p><p> corner Sketch of Amahl Apr. 29, 2002 p. 9</p><p>/kn/ dn g gn ID(PLACE: N/'V_) *! MAX(DORS) *! 1 LING. * 1 PLACE *</p><p>Both DORSALs and posttonic nasals are protected by robust Faithfulness constraints from the ban on having more than on lingual articulation.</p><p>24. Target is Nasal, Target Second, Trigger is Labial: Harmony</p><p> see above under (13)</p><p>25. Mixed Labials and Velars: No Harmony</p><p> escape /eskeep/ kep kek MAX(LAB) *! 1 PLACE *</p><p>26. /w/ Migration</p><p>“In a word of the structure /CwVC.../ the second consonant becomes a labial.” Most examples come from later, e.g. from stage 9: squat [gp] squeeze [gib] twice [daif] queen [gim]</p><p>But even at Stage 1, probably was fully in force:</p><p> bird [bibip] Sketch of Amahl Apr. 29, 2002 p. 10</p><p> tweet /twit/ dip bit dit dwit wit MAX(PTCK) *! MAX(STOP / _ [SON]) * *CC *! MAX(LAB) *! ID(VCE) * * * * ID(PL:STOP /_[+SON]) *! ID(ART/[-SON]) * * ID(ARTIC) * *</p><p>ID(PLACE IN PRESONORANT STOPS) is needed MAX(LAB) >> ID(ART/[-SON]) MAX(LAB) >> ID(ARTIC) *CC >> ID(ART/[-SON]) *CC >> ID(ARTIC) { MAX(PTCK), MAX(STOP / _ [SON]) } > ID(VCE) { MAX(PTCK), MAX(STOP / _ [SON]) } > ID(ART/[-SON])</p><p>THINGS NOT YET ANALYZED</p><p>27. Sibilant Deletions</p><p>“In a word of the structure /sVC/ the /s/ is optionally deleted if the C is labial or alveolar.” soup up sun n, dn sing gi</p><p>“In a word of the structure /VC/ the // is optionally deleted if the C is labial or velar.” sharp ap sugar ug</p><p>28. Very Local Harmony in Final Post-Atonic Nasals</p><p>This is triggered only by //. Output is adult English, but remember that Amahl generally goes for [n] in this position.</p><p> singing [gii] banging [bæi]</p><p>29. Summary So Far</p><p> Amahl looks like a pretty sensible child, in setting up his personal Phonological Translation scheme. Sketch of Amahl Apr. 29, 2002 p. 11</p><p> ...provided we assume articulatory difficulty for him is not quite what it is for adults: deploying multiple (lingual) articulations per word, avoiding fricatives. These involve fine control. But there are mysteries, notably in the “posttonic nasal” position for favored Faithfulness, and in Sibilant Deletion.</p>
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages11 Page
-
File Size-