forced choice and varieties of default in morphological systems is morphology truly discrete n.
Skip this Video
Download Presentation
Forced choice and varieties of default in morphological systems: is morphology truly discrete?

Loading in 2 Seconds...

play fullscreen
1 / 23

Forced choice and varieties of default in morphological systems: is morphology truly discrete? - PowerPoint PPT Presentation

  • Uploaded on

Forced choice and varieties of default in morphological systems: is morphology truly discrete?. Mark Aronoff Stony Brook University AIMM I University of Massachusetts 22 September, 2012. The takeaway. Morphological systems are usually thought to be discrete

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
Download Presentation

PowerPoint Slideshow about 'Forced choice and varieties of default in morphological systems: is morphology truly discrete?' - renate

Download Now An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
forced choice and varieties of default in morphological systems is morphology truly discrete

Forced choice and varieties of default in morphological systems: is morphology truly discrete?

Mark Aronoff

Stony Brook University


University of Massachusetts

22 September, 2012

the takeaway
The takeaway
  • Morphological systems are usually thought to be discrete
  • The paradigm case of discreteness is the forced choice or complementary distribution of rival realizations, which has been analyzed as a system of default or elsewhere mechanisms, using an idea borrowed from phonology
  • Inflectional competition can be modeled in this way
  • Rivalries between lexeme formation mechanisms do not form discrete systems, first impressions and prejudices to the contrary
  • At least one English inflectional system, the comparative and superlative forms of adjectives, does not exhibit discrete distributional patterns, but instead more resembles the competition found in lexeme formation
competition in imperfect languages
Competition in imperfect languages
  • In a perfect language, there would be only one way to express something
  • In actual languages, there is often more than one way to express the same thing
  • The two or more ways to express the same thing must be sorted systematically if they are all to remain part of the language system
  • Sorting among alternative expressions is done by competition
complementary distribution
Complementary distribution
  • Complementary distribution works well as a sorting mechanism in discrete systems in which one member of a given set maps onto many members of another set, each of which has a distinct distribution
  • In linguistics, complementary distribution was first found in phonology, where it has been recognized as a basic organizing principle for a century
  • Complementary distribution in phonology is inextricably bound up with the notion of the phoneme
  • Each member of the set of abstract phonemes is mapped onto one or more members of the set of more concrete phonetic allophones, each of which occurs in a specific environment
complementary distribution residue elsewhere and default in phonology
Complementary distribution, residue, elsewhere, and default in phonology
  • If the distribution of the allophones does not overlap, then the environment of one allophone does not need to be specified, because its distribution can always be treated as residual
  • The allophone whose distribution is residual and hence not specified will be the elsewhereor defaultallophone
  • The default variant/allophone is sometimes called basic
an example german plural morphology
An example:German plural morphology
  • The distribution of plural markers in German nouns is determined by a number of factors, with a few rules and a few strong tendencies
  • Rules
    • Feminine nouns ending in –eexceptionlessly take a –n suffix: Dame, Damen
    • Certain derivational suffixes are consistent
    • Most feminine derivational suffixes (-heit, -keit, -schaft, -ung) take -n
    • Nouns ending in –ling have a plural in –linge
    • -er, -ler, and –ner are unsuffixed
  • Tendencies
    • Masculine nouns whose last syllable contains a schwa tend to be unsuffixed
    • But there are exceptions, e.g. bauer(n), muskel(n)
    • There are other tendencies of various strengths
  • Many nouns must be marked lexically for which plural marker they take (including umlaut, sometimes combined with other markers)
  • Some nouns borrowed from Latin or Greek borrow their plural markers and, in very learned German, their case markers
an example defaults in german plural morphology
An example:Defaults in German plural morphology
  • The default plural marker is –s (Clahsen et al. 1992)
    • German plural –s is associated with no large class and occurs most commonly with borrowed words, nonce forms, acronyms, and proper names: autos, parks, kiosks
  • The default plural marker –s is a pure default that cleans up residue; it is not more frequent in any sense
  • This analysis is controversial
  • Regardless of whether Clahsen et al. are correct, it doesn’t make much sense to think of the default German plural marker as underlying in any sense of the term
default gender and inflectional class in arapesh nouns
Default gender and inflectional class in Arapesh nouns
  • Arapesh has thirteen genders and twenty-two inflectional classes for nouns
  • Gender and noun class are assigned according to semantic or phonological criteria
  • A noun that does not fit into the regular assignment system or is a lexical exception is placed in gender viii and noun class 8b
    • Nouns ending in b, k, or s have no phonological gender assignment and so fall into gender viii and the inflectional class 8b: mib ‘thigh’, mibehas ‘thighs’
    • Sex-neutral terms for persons show default gender because the two genders for persons are male and female: arapeñ ‘friend’, arapeʃ ‘friends’; batouiñ ‘child’, batouiʃ‘children’
    • A few nouns do not follow their expected inflectional class assignment. They show noun class 8b plurals and gender viii agreement; lim ‘roller’, limehas ‘rollers’
    • A few nouns show regular inflection for their class but agree as if they do not belong to the gender that the class corresponds to. Instead they show gender viii agreement: diliat ‘side post’, diliatogu‘side posts’; gender xi, class 8
  • Gender viii and noun class 8b are both default classes that emerge when the normal methods of assigning gender or noun class fail, for whatever reason
last resort default realization
Last-resort default realization
  • There are two reliably reported cases of languages with ‘literal alliterative concord’ (Dobrin 1995, 1998), Bainouk (Atlantic) and Arapesh
  • Both languages have large numbers of genders and inflectional noun classes
  • In both languages, loanwords that lie outside the noun class system condition agreement in which segments of the controller noun are copied onto the target
    • In Bainouk, the initial #CV is copied as a prefix
    • In Arapesh, the final (V)C# is copied as a suffix
  • In both cases, there is a last-resort default mechanism that captures those cases that are not caught by the normal default system
complementary distribution as default inheritance in inflectional morphology
Complementary distribution as default inheritance in inflectional morphology
  • Brown and Hippisley (2012) provide a computationally implementable account of complementary distribution in inflectional morphology
  • The most important mechanism in this account is default inheritance within a network or hierarchy
  • Default inheritance encodes the elsewhere principle very elegantly
  • The elsewhere variant is the default
  • More specific variants or lexical specifications override the default
  • Inflectional systems are nicely described as competing realizations organized in terms of the domain of their application, with the realization with the largest domain usually being the default
  • The default realization emerges both in the elsewhere environment and when lexical specification cancels the realization that would normally emerge
c omplementary distribution in lexeme formation
Complementary distribution in lexeme formation
  • Competition in lexeme formation appeared at first to be organized in the same way as in inflection
  • Aronoff (1976) showed that the allomorphs of English noun-forming –ion are in complementary distribution, with –ation as the default form
  • Aronoff claimed that the productivity of one suffix in a restricted environment (e.g. –cyafter adjectives ending in –ent, -ant, and –ate) blocks the occurrence of the competing default suffix –ness
  • Aronoff noted that phonological restrictions on the suffixation of –ation to verbs ending in palatal stridents (ʃ, ʒ, tʃ, dʒ) feeds–ment in precisely this environment (e.g. abridgement, astonishment)
defaults and overrides in lexeme formation within network morphology
Defaults and overrides in lexeme formation within Network Morphology
  • Brown and Hippisley (2012) devote a chapter of their book on network morphology to competition in lexeme formation, mostly in Russian
  • B&H use complementary distribution to distinguish among a number of competing suffixes, based on syntacticand morphological conditions: “When there is more than one rival affix available to realize a particular derivational function, the base’s properties can be accessed to resolve the competition.” (p. 277)
  • B&H note that “the default/override machinery in Network Morphology is ideal for teasing out the transparently productive only from the non-productive altogether.” ( p. 278)
  • B&H use “semantic-level override” for lexically listed senses of individual lexemes that result from non-productive lexeme formation rules
  • B&H do not attempt to account for non-lexical facts about relative productivity
the distribution of rivals in lexeme formation is not complementary
The distribution of rivals in lexeme formation is not complementary
  • In the last forty years, it has been found that rival LFRs are in fact rarely in absolutely complementary distribution
  • Differences in productivity between rival LFRs are seldom all-or-none
    • More common are pairs of rival LFRs where one member of the pair is relativelyless productive than the other, in a given environment.
      • English –ity is less productive than –ness when preceded by –ive.
      • English –ity is more productive than –ness when preceded by –able.
the use for less productive lfrs
The use for less productive LFRs
  • An LFR that is less productive overall is likely to be especially productive in its specific niche
    • -ity is much less productive overall than –ness

-ity is much more productive than –ness ever is under any conditions when preceded by –able, -al, -ic, -id

    • -ical is much less productive overall than –ic

-ical is much more productive than –icwhen preceded by –ology: biological, psychological

But in the narrow context of medical discourse, -ic is more productive with –ology: “gynecologic cancer”heard on a TV ad on 9/17/12

  • A less productive LFR can be used to coin technical terms: productivity, biologic, excellency, excellentness
  • These phenomena only make sense when both alternatives are possible
paradigmaticity and compositionality
Paradigmaticity and compositionality
  • Why are inflection and lexeme formation different with respect to complementary distribution?
  • Inflection is paradigmatic and compositional: all the cells in the paradigm of a lexeme need to be fillable for reasons of effability and all are semantically compositional
  • The need to fill all cells leads to automaticity: the means for filling all cells in a paradigm must be automatically available
  • Lexeme formation is driven instead by onomasiological considerations. Words like paradigmaticity, onomasiological, and compositionality are needed only when a concept arises that they can name
  • Naming is conditioned to a great extent by nonlinguistic needs
    • Canadianization
    • Systemness
  • The differences in the importance of complementary distribution may be rooted in the two different uses to which morphology is put
a case of noncomplementary distribution in inflection the english comparative
A case of noncomplementarydistribution in inflection: The English comparative
  • The comparison of adjectives (degree) in English is famously expressible by two means, the suffixes –er, -est and the adverbs more, most
  • Degree is usually considered to be syntactic rather than lexemic and hence inflectional (Zwicky 1989)
  • the adverbial expressions of degree might accordingly be thought of as periphrastic
  • Periphrastic forms are usually treated members of the lexemic paradigm
    • Latin perfect passive
    • Romance perfect
the distribution of the rival realizations of degree in english
The distribution of the rival realizations of degree in English
  • The two means of expressing the comparative/superlative degrees in English appear at first glance to be in complementary distribution, like other inflectional realizations
  • Words of one syllable generally take –er/-est
  • Words of three or more syllables take only the periphrastic form
  • Two-syllable words ending in-y take -er/-est: sillier, livelier
  • Predicate-only adjectives take only the periphrastic form: *awarer, *afraider, *contenter
  • Elsewhere, only periphrastic forms occur
not so simple
Not so simple
  • There are many exceptions and uncertainties
  • Some one-syllable words avoid –er: ?apter
      • Clearly borrowed words avoid –er/-est: *loucher
  • Most exceptions and uncertainties occur among two-syllable words
    • Many two syllable words not ending in unstressed syllable other than –y prefer -er/-est: stupid, narrow, noble, simple, clever
    • But some words of this type prefer periphrasis: vapid, callow, ample
    • The one word likely accounts for most cases of periphrasis among –y words (Kyto and Romaine 1997)
  • Linguists differ sharply on individual words
    • Zwicky (1989) says that words with tense (unstressed) vowels in their final syllable take -er/-est: profound, polite, sincere, obscure
    • As far as I can tell, words of this type accept both forms, with some lexical preference for one form or the other but a great deal of uncertainty
not so simple1
Not so simple
  • Zwicky quotes Evans and Evans (1957): “But this is a description of what usually happens, not of what must happen. Mark Twain wrote: the confoundedest, brazenest, ingeniousest piece of fraud.”
  • Jespersen (1949, p. 347) writes that “a good deal is left to the taste of the individual speaker or writer” and that the “rules given in ordinary grammars are often too dogmatic.”
  • “Disyllabic words have always been subject to more variation.” (Kytö and Romaine 2000, p. 180)
  • Frequency plays an important role among two-syllable words (Graziano-King 1999)
  • A number of authors claim that there are stylistic differences between the two, with the periphrastic form more common in written registers.
a bit of history
A bit of history
  • Germanic had only the suffixed forms
  • Modern German continues to have only the suffixed forms
  • The periphrastic construction entered the language in Middle English, modeled on French and Medieval Latin, which had lost the Latin (I-E) suffixed forms
  • Kytö and Romaine (1997) show that the modern distribution developed gradually over a period of centuries
  • Kytö and Romaine suggest that the periphrastic form originated in written English
  • Since records of the earlier spoken language are rare to nonexistent, it is difficult to know how much the periphrastic form was originally restricted to written language
what should we learn from english degree morphology
What should we learn from English degree morphology?
  • The Modern English comparative is a real instance of inflectional competition that is not describable as complementary distribution
  • Complementary distribution may not be fundamental, even in inflectional systems
  • Even in inflectional systems, the distribution of rival forms may be an emergent process governed by principles of competition
  • Complementary distribution in inflectional systems is a final state, with complementarity driven by morphosyntactic paradigms, not an initial state
  • We may have to rethink the centrality of complementary distribution in languages
  • The tendency to model languages as entirely discrete systems may result at least in part from privileging stable synchronic systems
  • Linguistics has not paid attention to systems that are not discrete and has even denied their validity as linguistic systems
  • It is clear now that LFR rivalries are not discrete or profitably modeled in terms of complementary distribution
  • Linguistics has not paid much attention to how morphological systems emerge
  • Looking more closely at both nondiscrete systems and the emergence of systems of all types may shed light on some fundamental questions about language and languages