Hopper & Traugott Final Chapter

GRAMMATICALIZATION

Handout for LING 519/SARS 519

  1. Summary and Loose Ends

    This chapter addresses issues that are either unsettled, unstudied, or areas where grammaticalization still has much work to do. In particular, they deal with the issue of what causes change in language, particularly change in morphological systems, grammars, etc.

    Some of these are:

    1. whether it is through child-language acquisition that we see grammatical change?

    2. whether contact-induced change is also possible?

    3. whether it is a matter of parametric change vs. grammaticalization?

    4. what is the role of pidginization/creolization?

    This is to ignore issues such as spread across styles, registers, dialects, communities; what is the role of literacy, etc.

  2. Grammaticalization vs. parametric change.

    A number of researchers have made claims about language change, what causes it, what factors are irrelevant.

    • Some say can only take place during language acquisition; major changes are made by learners simplifying grammars.

    • Halle claimed that adults can't make major changes in their grammars, except to add rules (which doesn't simplify).

    • Discontinuity between adults and children is enabling; discontinuities in ones one life is not.

    • Lightfoot says that there are parameters, can get set differently by diff. generations.

    • H&T disagree with this; wish to challenge this dichotomy.

    Lightfoot's parameter settings:

    • Changes occur in clusters, simultaneously (e.g. loss of inversion and negative Sentence's; introduction of do-support with. interrogative and negative s's.)

    • changes set off chain reactions , e.g. word order changes, shifts from VO to OV.

    • change more rapidly than others, show S-curve (grad. beginning, rapid rise, slow taper off). Lightfoot says S-curve phenomenon not like grammaticalization and morphological change, e.g. loss of gender, inflections, mood; gradual cumulativeness (paradigmatic leveling?) not parameter setting.

    • Obsolescence of earlier forms/rules: don't drop out for expressive reasons; must be due to "knock-on" effect, by-product of s.t. else, triggered by s.t. avail only to children.

    • significant change in meaning is byproduct of new parameter setting (but H&T say no, look at all the meaning changes in grammaticalization of main verbs into auxiliaries etc.)

    • Parameter setting occurs only with shifts in unembedded clauses (e.g. shift from OV to VO).

    H&T: grammaticalization and its perspectives (GP) challenges all these assumptions.

    • Approach from grammaticalization says grammaticalization of lexical material is enabled by pragmatic factors , the conventionalizing of conversational inferences in certain local contexts, with morphosyntactic reanalysis.

    • meaning change accompanies instead of following syntactic change.

    • GP says (contra Lightfoot) that very few changes are arbitrary, because of pragmatic inferencing. Arbitrariness the exception (not the rule) because of pervasiveness of iconicity in the organization of ling material.

    • GP: Changes are gradual, and along various dimensions:

      • along clines

      • along functional hierarches (animacy, definiteness, thematic relations)

      • changes rarely go to completion (along clines)

      • obsolescence is gradual (leaving residue, detritus in the system)

      Bybee (1985) unidirectionality results from cognitive processes alone.

      H&T: not only cognitive, but discourse production strategies , i.e. over time, certain strategies not felt to serve communicative purposes; no need for "structural knock-on effect."

    As for the idea that changes cluster , occur in S-curves, and have chain reactions: are these evidence for parameter settings? How about a tendency for speakers/hearers to organization info in accessible ways.

  3. Distinguishing parametric changes from minor changes:

    • which ling changes (cross-linguistically) have a "typological ripple" or "cascading effect.": word order?
    • which are more language specific? English modals?
    • having decided this, which changes show rapid spread
    • which don't?
    • Need longitudinal studies (tape, video) to show how child language acquisition influences this.

  4. Otherwise: do children in fact innovate differently from adults? Bybee & Slobin (1982) say no. Evidence from creoles shows no difference. (Cf. Sankov's tables showing children pick up where adults left off.)

  5. Contact-induced Change

    Monogenetic change, evolutive change: assume an idealized homogeneity, ignoring contact situations, especially with speakers of other dialects, especially social, regional, stylistic variants. H&T also confess to having "privileged transmission in ... homogeneous contexts." (H&T also ignore "language admixture across a wide area, such as evidenced by the Balkan and Dravidian languages." What does this mean? Areal linguistics? Why are Dravidian language more 'admixed' than the Munda and Indo-Aryan? In fact, I would say the opposite!)

    • Need to look at pidgins and creoles, which challenge basic assumptions about homogeneity, and the role of adults and children in language change. Quotes Gillian about role of adults, who originated change, and then children who went further than their own parents had gone.

    • Another issue: Bickerton's bioprogram which allows children to create language out of 'bits and pieces' of 'degenerate' material available to them. McWhorter (1992) challenges this also; says many features thought to come (in Surinam) from the 'bioprogrami can be found in substratum of African languages. Bickerton thought that the speech of individuals does not change much after adulthood; this now challenged (e.g. work by Gillian Sankoff).

    • Stabilized pidgins show more morphology, predicate marking, object marking, aspectual particles, some embedding structures. Speakers use certain material in multiple ways, with periphrasis, for expressivity, features that may carry over into creoles, which have more elaborate varieties (basilect vs. acrolect, with many degrees in between).

    • Thus: pidgins are restricted, based on lexicon of lexifier language, with abductive processes going on. Creoles are more complex, but perhaps not so much more complex than the pidgins they are most immediately based on. But rule simplification that is supposed to be the process in child language acquisition may not be working if the input is a pidgin. Ignores also the input to the pidgin, which may be the colloquial, vernacular variety of a language, rather than the 'standard.' Or they may be brought in from the substrate (subordinate) language as well.

    • There is also recreolization observed in some creoles, a movement away from the 'standard' and back toward basilect, or the norms of an older system. Does this lead to structural reanalysis?

    • Expressive material from the subordinate language may be more important than the lexifier language, as many examples they give show. (HS: example of Afrikaans, which has lots of expressive particles, echo-word reduplication etc. which must come from African languages, not Dutch [See Hesseling].)

    • Evidence from Sri Lanka Portuguese, which has Tamil as a substrate, show strong influence of Tamil structures, e.g. OV structure, development of postpositions (from Port. prepositions), and other stuff. (Examples from Smith 1987).

    • Ergo, more importance must be given to multiple origins of grammatical structures, and strictly monogenetic view of grammaticalization is inappropriate.


    Haroldfs@ccat.sas.upenn.edu, last modified 4/5/05