Defaults and Morphological Structure in Linguistics, Summaries of Linguistics

Download Defaults and Morphological Structure in Linguistics and more Summaries Linguistics in PDF only on Docsity!

“Defaults”

and Morphological Structure

Stephen R. Anderson Dept. of Linguistics, Yale University

Default: The default aspects of a state of affairs are those that follow from its intrinsic nature. Observed properties that do not have this character must be stipulated as over-riding or supplementing the default conditions, and thereby constitute complexity.

Complexity is the other side of the coin from defaults. So an understanding of morphological complexity illuminates our understanding of defaults.

What about Morphology?

Surely the default ought to be that the elements combined in the syntax map directly onto those that form the input to the phonology. But morphology involves a system for arranging meaningful material within larger units (‘morphotactics’) To the extent this is distinct from the way the syntax organizes meaningful elements into larger units, morphotactics ought not to be necessary.

What about Morphology?

The ‘same’ morphological element can have a variety of overt realizations (‘allomorphy’) To the extent this is distinct from the modifications required by the phonology, allomorphy lacks independent motivation. Specific principles of morphotactics or allomorphy – and thus any morphology at all – thus constitute overrides of the most basic defaults of the system of natural language.

Morphotactics ≠ Syntax

kwakw’ala-exsd-@n speak.Kwakw’ala-want-1sg ‘I want to speak Kwakw’ala’

Morphological composition:

ax. -exsd-@n q-@n kwakw’ala ;-want-1sg that-1sg speak.Kwakw’ala ‘I want to speak Kwakw’ala’

Syntactic composition:

Allomorphy ≠ Phonology

In Kwakw’ala (and other Wakashan languages) suffixes are of three types, not predictable from their phonological shape, in terms of their effect on the final consonant of the preceding stem:

Hardening (glottalizing), e.g. /qap+alud/ → [qap’alud] ‘to upset on rock’ Softening (roughly, voicing), e.g. /qap+is/ → [qabis] ‘to upset on the beach’ Neutral (no change), e.g. /qap+a/ → [qəpa] ‘(hollow thing is) upside down’

Similar to, e.g., Celtic mutations, these changes no doubt have their origin in segmental accommodations, but in synchronic terms, they are arbitrary morphology.

Some Systems are More

Complex than Others

Kwakw’ala: hux. w-sanola-gil-eì vomit-some-continuous-in.house ‘some of them vomit in the house’

Mohawk:

Wa’koniatahron’kha’tshero’ktáhkwen. wa’-koni-at-ahronkha-’tsher-o’kt-ahkw-en FACTUAL- 1 SG/2SG-MIDDLE-speech-NMZR-run.out.of-CAUS-STATIVE ‘I stumped you.’ (left you speechless) Central Alaskan Yupik: Piyugngayaaqellrianga-wa. pi-yugnga-yaaqe-lria-nga=wa do-able-probably-INTR.PARTICIPIAL- 1 SG=suppose ‘I suppose I could probably do that.’

[Thanks to Marianne Mithun for Mohawk and CAY examples]

Dimensions of Complexity

System Complexity: Number of distinct affixes (non-root meaningful elements) in the system Number of meaningful elements in a single word Predictability of ordering relations among the elements that make up a single word Complexity of exponence: Deviation from the classical morpheme Number of word forms corresponding to a single lexeme, and vice versa Complexity of allomorphy

Number of Affixes in a

Word

“[Central Siberian Yupik] postbases are most often productive and semantically transparent, and can be added one after another in sequences of usually two or three, the maximum encountered being seven. These sequences are relatively short in comparison to other Eskimo languages, such as CAY, where one can find more than six postbases in a work, and where it is possible to have more than a dozen.” (deReuse, 1994) Kwakw’ala is similar to CSY in the degree of observed complexity.

Element Order

Compositional (scope-based) order in Kwakw’ala:

ne’nakw’-exsda-mas-ux. w^ John gax-@n go.home-want-cause-3sg John to-1sg ‘John made me want to go home’

a. cause to want

q’aq’o¨a-madz-exsd-ux. w^ John gax-@n q-@n gukwile learn-cause-want-3sg John to-1sg that-1sg build.house ‘John wants to teach me to build a house’

b. want to cause

Here the order follows from the content properties of the elements involved, a situation we can think of as the default.

What factors are ‘default’ predictors of element order? Semantic scope Grammatical function (e.g. derivation is ‘inside of ’ inflection) More detailed “Bybee effects” (mood inside of tense inside of agreement, etc.) Is some version of this a theorem rather than a tendency? Phonological shape (element size and prosodic status; high vowel before low, V-initial before C-initial as in Sanskrit 2P clitics)

Default Element Order

Complexity of Exponence

The ‘default’ morphological element, corresponding to the classical structuralist morpheme, is a discrete, indivisible unit of form linked to exactly one discrete unit of content. Real morphology is not like that. Circumfixes (e.g. Slavey ya--tį ‘preach, bark, say’; cf. yahtį ‘s/he preaches, barks, says’, xayadatį ‘s/he prayed’, náya’ewítį ‘we will discuss’; Rice 2012) Infixes (e.g., Mẽbengokre [Je] fãgnãn ‘to spend almost all (pl.)’, sg. fãnãn ; Salanova 2012) Multiple exponence (e.g. Choctaw akíiyokiittook ‘I didn’t go’; cf. iyalittook ‘I went’; Broadwell 2006)

Complexity of paradigms (mapping from lexemes to word forms) Relations between morphosyntactic words (pairings of a lexeme and a morphosyntactic representation) and overt word forms that are not one-to-one Syncretisms (e.g. [hIt] as both present and past of {HIT}) Variation (e.g. either [dajvd] or [dowv] as past of {DIVE})

Complexity of Exponence

Complexity of Allomorphy

A range of degrees to which the behavior of an element can follow by default from its other properties:

Phonological variation under phonological conditions Lexically specified variation (“allomorphy”) under phonologically specified conditions (e.g. Warlpiri ergative - rlu/-ngku; Surmiran stems) Allomorphy conditioned by specific morphological categories or semantically/grammatically coherent sets of categories Allomorphy conditioned by semantically/grammatically arbitrary sets of categories (“morphomes”)