Mangarrayi
0 Followers
Recent papers in Mangarrayi
Reduplication occurs in almost every human language, yet for languages with CV(C*) syllables, only a minority elect to reduplicate by infixing a VC* string. Reduplicants copy some or all of a base, where the base itself is some... more
Reduplication occurs in almost every human language, yet for languages with CV(C*) syllables, only a minority elect to reduplicate by infixing a VC* string.
Reduplicants copy some or all of a base, where the base itself is some contiguous string which sits to the left or right of the reduplicant. Reduplicants are often only partial copies, and often contain unmarked segments (e.g. short vowels) in place of base segments which are more marked (e.g. long vowels). Data like (1–3) raise the questions: (i) why does the reduplicant take on a VC* shape, and (ii) why is the reduplicant an infix within the word as a whole?
McCarthy & Prince (1993) analyse such reduplication as driven by the placement of the reduplicant within the word: it is attracted to the left edge, but a higher-ranking constraint denies it the absolute leftmost position. Consequently, the first segment of the reduplicant is the second segment of the word, and in order that syllables all retain a CV(C*) structure, the reduplicant will begin with a vowel. Pensalfini (1998) presents an alternative analysis, driven by shape: the reduplicant is attracted to the left edge but must begin with a vowel, and consequently, in order that all syllables have an onset, it becomes an infix. Also required for this analysis, is that reduplicants copy as many contiguous consonants as possible.
Intriguingly, Pensalfini’s account is driven by constraints which, if ranked high enough, would push a language to undergo initial consonant loss — a process which is historically attested in many Australian languages. Thus it would be enlightening to ascertain which analysis is ultimately correct: is VC* reduplication a placement-driven phenomenon, or is it a shape-driven process which contains the seeds of initial-dropping?
Kuuk Thaayorre (Paman, Gaby 2006) is situated on the south-west of Cape York peninsula, not far from many languages which have undergone initial dropping. Thaayorre itself has CV(C*) syllables. It also possesses infixing VC* reduplication for most verbal stem shapes (4–5), but not for stems containing a long first vowel, whose reduplication is CV (6–7). Significant here is that in Thaayorre, underlying vowel length is always preserved in the base. However, only the initial vowel of a Thaayorre word can be long, which means that a long vowel in the base cannot afford to be shunted to the right by an infix. Consequently, the infixing reduplicant starts at the third segment in the word. This fact allows us to contrast, and thus test, the placement-driven and shape-driven analyses. The former analysis predicts that the infix will stay as far to the left as possible, as in (7a); the latter predicts it will drift rightwards if by doing so it increases its number of copied consonants, as in (7b) [note: kt̪ in (7b) would be a perfectly legal cluster]. In fact, (7a) is the attested form.
Kuuk Thaayorre:
(4) REDUP; /ŋeɻnkan / → ŋ<eɻnk RED>[eɻnkan BASE]
(5) REDUP; /kal/ → k<al RED>[al BASE]
(6) REDUP; /koːpe/ →
a. [koː BASE]<ko RED>pe
b.* k<oːp RED>[ope BASE]
(7) REDUP; /ti̪ ːk/ →
a. [ti̪ ː BASE]<ti̪ RED>k
b.*[ti̪ ːk BASE][ti̪ k RED]
This advances our understanding of the nature of reduplication: in a CV(C*) language, infixing VC* reduplication in the general case is driven not by shape, but by placement.
Reduplicants copy some or all of a base, where the base itself is some contiguous string which sits to the left or right of the reduplicant. Reduplicants are often only partial copies, and often contain unmarked segments (e.g. short vowels) in place of base segments which are more marked (e.g. long vowels). Data like (1–3) raise the questions: (i) why does the reduplicant take on a VC* shape, and (ii) why is the reduplicant an infix within the word as a whole?
McCarthy & Prince (1993) analyse such reduplication as driven by the placement of the reduplicant within the word: it is attracted to the left edge, but a higher-ranking constraint denies it the absolute leftmost position. Consequently, the first segment of the reduplicant is the second segment of the word, and in order that syllables all retain a CV(C*) structure, the reduplicant will begin with a vowel. Pensalfini (1998) presents an alternative analysis, driven by shape: the reduplicant is attracted to the left edge but must begin with a vowel, and consequently, in order that all syllables have an onset, it becomes an infix. Also required for this analysis, is that reduplicants copy as many contiguous consonants as possible.
Intriguingly, Pensalfini’s account is driven by constraints which, if ranked high enough, would push a language to undergo initial consonant loss — a process which is historically attested in many Australian languages. Thus it would be enlightening to ascertain which analysis is ultimately correct: is VC* reduplication a placement-driven phenomenon, or is it a shape-driven process which contains the seeds of initial-dropping?
Kuuk Thaayorre (Paman, Gaby 2006) is situated on the south-west of Cape York peninsula, not far from many languages which have undergone initial dropping. Thaayorre itself has CV(C*) syllables. It also possesses infixing VC* reduplication for most verbal stem shapes (4–5), but not for stems containing a long first vowel, whose reduplication is CV (6–7). Significant here is that in Thaayorre, underlying vowel length is always preserved in the base. However, only the initial vowel of a Thaayorre word can be long, which means that a long vowel in the base cannot afford to be shunted to the right by an infix. Consequently, the infixing reduplicant starts at the third segment in the word. This fact allows us to contrast, and thus test, the placement-driven and shape-driven analyses. The former analysis predicts that the infix will stay as far to the left as possible, as in (7a); the latter predicts it will drift rightwards if by doing so it increases its number of copied consonants, as in (7b) [note: kt̪ in (7b) would be a perfectly legal cluster]. In fact, (7a) is the attested form.
Kuuk Thaayorre:
(4) REDUP; /ŋeɻnkan / → ŋ<eɻnk RED>[eɻnkan BASE]
(5) REDUP; /kal/ → k<al RED>[al BASE]
(6) REDUP; /koːpe/ →
a. [koː BASE]<ko RED>pe
b.* k<oːp RED>[ope BASE]
(7) REDUP; /ti̪ ːk/ →
a. [ti̪ ː BASE]<ti̪ RED>k
b.*[ti̪ ːk BASE][ti̪ k RED]
This advances our understanding of the nature of reduplication: in a CV(C*) language, infixing VC* reduplication in the general case is driven not by shape, but by placement.