24.962 Advanced phonology 22 Feb, 2005 Reduplicant size and placement (1) “Canonical” reduplication patterns Ilokano (Hayes and Abad 1989) Manam (Lichtenberk 1983, Buckley 1998) (Malayo-Polynesian; Philippines; 8M spkrs) (Malayo-Polynesian; New Guinea; 7000 spkrs in 1998) Sg. Pl. Gloss Unredup. Redup. usa pusp´ aga ‘be long’ salagal´p´usa ‘cat’ sal´aga ‘long-sg.’ aldiN kalk´oita ‘knife’ moita´ıta ‘cone shell’ k´aldiN ‘goat’ m´j j ˇy´oN ‘flying fox’ malabomb´ˇyanitor ˇyanj anitor ‘janitor’ malab´oN ‘flying fox’ oPot ro:r´r´oPot ‘leaves’ Pulan- ‘desire’ Pulanla´N ‘desirable’ (2) Marantz (1982): • RED = an affix (prefix or suffix) • Shape of RED defined by a CV template • Material associated from the same edge (“Marantz’s generalization”): “In the unmarked case, reduplicating prefixes associate with their melodies from left to right, reduplicating suffixes from right to left.” (p. 447, Condition D.i.) (3) The OT equivalent: (sort of ) • ALIGN-L/R(Red,PrWd): RED is a prefix/suffix • RED = X (templatic constraint) • LOCALITY(by seg/σ): RED is not separated from its corresponding string in the base by any segments/syllables • MAX: copy as much as possible (from base? from input?) • BASE-CONTIG: no skipping when copying from the base (Unique predictions from treating these as rankable constraints) (4) Goals today: examine a variety of reduplication patterns, seeing what additional elaborations are needed to capture three basic aspects of reduplication: 1. Where do you copy from 2. How much do you take 3. Where do you put it (We’ll take these questions out of order: (1) and (3), then (2)) Where do you copy from, and where do you put it? (5) Marantz (1982): copying from L/R edges • p. 447: begin at L and associate L→R, or begin at R and associate R→L (6) McCarthy & Prince (1995): ANCHOR-L/R(Base, RED) • The X edge of the base corresponds to the X edge of the reduplicant • How is this not quite the same from L→R or R→L association? What are some unique predic-tions of an ANCHOR-based approach? – Hint: both constraints freely rerankable w/Contiguity24.962—22 Feb, 2005 p. 2 (7) Edge-in association as alignment + locality /RED-badupi/ ALIGN-L(RED) ALIGN-R(RED) LOCALITY(σ) RED = ˘σ ☞ a. ba-badupi *** b. du-badupi *** * c. pi-badupi *** * d. ba-ba-dupi * ** e. ba-du-dupi * ** f. badu-du-pi ** * ☞ g. badupi-pi *** h. badupi-du *** * i. badupi-ba *** * Factorial typology: edge-in candidates harmonically bound nonlocal and infixing candidates • What rankings are needed to get Ilokano and Manam? (8) Infixing reduplication in Timugon Murut (Prentice 1981, McCarthy & Prince 1993, McCarthy 2000) (Malayo-Polynesian; Malaysia; around 8,000 spkrs) Unredup. Redup. a. bulud ‘hill’ bubulud ‘ridge’ b. limo ‘five’ lilimo ‘about five’ c. ulampoy ulalampoy (no gloss) d. abalan ‘bathes’ ababalan ‘often bathes’ e. ompodon ‘flatter’ ompopodon ‘always flatter’ /RED-ulampoy/ *V.V, Onset/DEP/etc. RED = ˘σ ALIGN-L(RED) a. u-ulampoy *! ☞ b. u-la-lampoy * c. ul-ul-ampoy *! * d. ula-la-mpoy **! e. ulam-po-poy **! • How is this similar to other cases of infixation that you are familiar with (e.g., Tagalog -um-infixation) How is it different? • Prosodic circumscription (McCarthy & Prince 1990, 1993) – Skipping initial onsetless syllable is awkward in an operational framework • ANCHOR-L is no help here (it is violated by [ulalampoy], since [l] doesn’t correspond to [u]) • ALIGN-L is doing the work of keeping the reduplicant to the left (just as with infixation of segmentally specified affixes) (9) Another infixation case: Samoan (Broselow & McCarthy 1983) (Malayo-Polynesian; Samoa; 400,000 spkrs 1999) Unredup. Redup. Gloss aa tat´ a. t´aa ‘strike’ uu tut´ b. t´uu ‘stand’ ofo non´ c. n´ofo ‘sit’ oe mom´ d. m´oe ‘sleep’ ofa alol´ e. al´ofa ‘love’ ali savav´ f. sav´ali ‘walk’ g. mal´ıu malil´ıu ‘die’ ue galul´ h. gal´ue ‘work’ au fanan´ i fan´au ‘be born’24.962—22 Feb, 2005 p. 3 (10) Marantz (1982), fn. 15: Points out existence of infixing reduplication, suggests that a similar copy-ing mechanism could handle them a l o f a a l o f a l o f a | | | | | → | | | | | | |V + CV + CVCV V + CV + CVCV • Does not provide an explicit mechanism to position reduplicant as an infix • Direction of copying also unexplained: why is [lofa] copied and not [a]? (Broselow & Mc-Carthy 1983) – Are infixes suffixed to what comes before them, or prefixed to what comes after? • Prosodic circumscription: target final (stressed) foot Possible constraints: • – ALIGN(RED,R,"σ,L) (“ALIGN-to-´σ”) ∗ Alignment to the foot containing main stress is not limited to reduplication; Ulwa possessive marking is aligned immediately after the main stress foot (McCarthy & Prince 1993) – ANCHOR-L("σ, RED) (11) Alignment + locality is not enough to capture Samoan: /RED-sav´ali/ ALIGN-to-´σ RED=˘σ ALIGN-L a. sa-sav´ali *! ☞ b. sa-sa-v´ali * ☞ c. sa-va-v´ali * • σ captures locus of reduplication, but can’t get the direction (same problem as the ALIGN-to-´ Marantz footnote) • Nelson (2003 Rutgers diss.): a positional faithfulness effect (MAX("σ)) /RED-sav´ali/ MAX("σ) ALIGN-to-´σ RED=˘σ ALIGN-L a. sa-sav´ali *! b. sa-sa-v´ali *! * ☞ c. sa-va-v´ali * (12) Other examples of copying from strong positions a. Yareba (Weimer & Weimer 1970, Riggle 2003) (Trans-New Guinea; Papua New Guinea; 750 spkrs 1981) Sg. Pl. Gloss a. boroy-a borob-a ‘reveal it!’ b. fomuy-a fomuf-a ‘break it!’ c. doroy-a dorod-a ‘go through the hole!’ • Stem-final [y] in sg. is epenthetic (to avoid hiatus; a strictly CV lg.) • Plural marked by copy of stem-initial consonant • ALIGN-R(RED), but MAX(Initial C) • We’ll talk about the placement of the reduplicant shortly (point here is to discuss source of RED material) b. Levantine Arabic intensive/pejorative reduplication (Broselow & McCarthy 1983) (Semitic; Lebanon, Syria, Palestine, Jordan) Simple vb. Derived vb. Gloss (of derived) a. faraè farfaè ‘rejoiced’ b. baèaS baèbaS ‘sought’ c. barad