24.962 Page 1 2/14/05 Reduplication and its basic laws (1) Summary •main issues in analysis of reduplication; connection to correspondence theory•templates and theories of templates•the prosodic theory of templates and its consequences for Base RED corresp.•markedness and RED•over and under rule application in RED and correspondence-based analyses.(2) Reduplication types (examples from Madurese: Austronesian) •simple reduplication: copy everything change nothingsakola/an-sakola/an ‘schools’; •partial reduplication: copy a fragment, change nothingdus-garadus ‘fast-and-sloppy’; •partial and prespecified reduplication: copy a fragment, change a segment or moredas-garadus ‘fast-and-sloppy’; •total, prespecified: (not Madurese) table schmable (3) Partial RED parameters and corresponding constraints: (a) size: RED = (specify a prosodic unit: s ¤ or s* , F, PrWd) (b) location wrt base or edge of PrWd: Align Edge, RED, Edge, PrWd (c) source in base (preliminary): Anchor Edge:RED/Base (d) segmental contents: phonotactics interacting with Contiguity, MAX + the above (4) Base of RED vs. Input of RED and Base /lexical entry/ [RED Base] the reduplicated word/stem RED-I corr B-I corr RED-B corr(5) Why study Reduplication? • Output-to-Output correspondence: surface RED corresponds to surface Base [debated]•syntagmatic (B-RED) + paradigmatic (RED-I, B-I) correspondence in one construction•unique evidence of markedness effects•interactions with both prosodic and segmental phonology•a self contained universe with rich typology(6) Size constraints on partial RED: why is just dus copied in dus-garadus RED is subject to a template constraint (7) What is a possible template? (McCarthy and Prince 1986, now on ROA) a. A sequence of segmentally empty other skeletal slots? (CVCCVCCCV?)b. A sequence of segmentally empty syllable nodes? (ssssss?) c. A sequence of syllables with specified internal structure (CVC.CV:CC.CV?)d. A foot, a (heavy/light) syllable: “the units of authentic prosody”24.962 Page 2 2/14/05 (8) Reminder: Semitic stems (McCarthy 1979 MIT diss) [ <y >= IPA [j]) Form Past active Non-Past act Past Pass Non-Past Pass Imperative 1 write katab-a ya-ktub-u kutib-a yu-ktab-u ktub-2 caus kat:ab-a ya-kat:ib-u kut:ib-a yu-kat:ib-u kat:ib-3 recip ka:tab-a yu-ka:tib-u ku:tib-a yu-ka:tab-u ka:tib-4 dictate /a-ktab-a yu-ktib-u /u-ktib-a yu-ktab-u /a-ktib-6 write letters to e.o. ta-ka:tab-a ya-ka:tab-u tu-ku:tib-a yu-ta-ka:tab-u ta-ka:tab-7 sub-scribe n-katab-a ya-n-katib-u n-kutib-a yu-n-katab-u n-katib-8 copy ktatab-a ya-ktatib-u ktutib-a yu-ktatab-u ktatib-9 ask s.o to sta-ktab-a ya-sta-ktib-u stu-ktib-a yu-sta-ktab-u sta-ktib-(9) Compare: Form Past active 1 study daras-a 2 caus dar:as-a (10) Morpheme shapes (not exhaustive) (a) Derivational morphemes (b) Non-derivational morphemes form 1: CVCVC (syncope in yV-kVtVb-u) Active: (u)-a-i1 form 2: CVCCVC Passive: u-a form 3: CV:CVC Non-Past yV-2 form 3: /a- CCVC form 6: ta-CV:CVC form 7: n-CVCVC (c) Invariant root shape: here ktb form 8: CtVCVC form 9: sta-CCVC --- CV slots: dominated by syllable nodes; dominate segments (11) Autosegmental assembly of morphemes: •L-R association of segments to template slots yields actual syllables.(in some cases the strict L-R order is replaced by other conventions) •Too many slots, not enough segments? E.g. sm ‘poison’: samam-, sam:am Spread the rightmost segment. •Too many segments, not enough slots? E.g. magnati:S ‘magnetize’ magnat. Delete the segments, add a coda position. •Critical assumption: features of each morpheme occupy distinct tiers from features of other morphemes. Affix u a i CV-CVVCV C k t b yields yu-ka:tib (form 3, non-past active) Affix: y Root:========================================= 1 Vocalism is subject to further rules. Thus in the non-past active: u-a-i -> a-i in 7-9; u-a-i -> a in 5, 6. 2 Past and non-past vowels are subject to dissimilatory conditions. Thus ya-ktub-u but katab-a24.962 Page 3 2/14/05 Affix u a CV-CVVCVC k t b yields yu-ka:tab (form 3, non-past passive) Affix: y Root:========================================== •Biliterals work similarly but here the C’s must spread too: cf. sm ‘poison’ Affix u a i CV-CVVCVC sm yields yu-sa:mim (form 3, non-past active)3 Affix: y Root:========================================== • LR association (+ OCP) explains why sa:mim and not *sa:sim •Quadriliterals have form 2 without gemination.(12) Templatic morphology without consonantal roots: Miwok (Smith 1984), English Basic Derived 2 Derived 5 Derived 6 Derived 7 Derived 8 polaat polat pollat polta poolat polaat kelti kelit kellit kelti keelit keliit halh halÈh hallÈh halhÈ haalÈh halÈÈh tappu tapu/ tappu/ tap/u taapu/ tapuu/ • [È] and [/] are epenthetic segments. • V and C on same tier, hence polat vs. polta difference requires reordering. • But same “LR” effect in mapping: running out of segments at R edge. (13) What is a possible template? (McCarthy and Prince 1986, now on ROA) a. A sequence of segmentally empty other skeletal slots? (CVCCVCCCV?)b. A sequence of segmentally empty syllable nodes? (ssssss?) c. A sequence of syllables with specified internal structure (CVC.CV:CC.CV?)d. A foot, a (heavy/light) syllable: “the units of authentic prosody”(14) Why (13. a) could not be right: English nicknames Base Truncated left Truncated and augmented Truncated right William Will, Willie Elizabeth Liz, Lizzie Beth Barbara Barb, Barbie Gabriel Gabe (Gabi) Herbert Herb Herbie Bert Diana Di Stuart Stu Arthur Art Artie Christine Chris Chrissie, Christie Agnes Aggie 3 An invented form.24.962 Page 4 2/14/05 •template can’t be CVC: Herb, Stu •can’t be (C(C))V(C(C)): Gabe, *Gabrie; Aggie *Agnie •must be: 1 syllable (+ [i]), plus rules/principles on how to map the full word onto template. •Contiguity (nickname to base): segments in nickname are adjacent only if their full form correspondents are (Di, *Dine, *Dinie; Stu, *Stur, *Sturt; Will, *Wilm) •Ident stress (nickname to base): syllables in nickname are stressed iff their full form correspondents are (Liz, Beth, *El; Stu; *Art) •MAX (base to nickname): the largest number of segments from the base must be present in the nickname, (subject to the