Heine & Claudi and Huennemeyer, Chapter 2

Handout for LING 519/SARS 519

GRAMMATICALIZATION

Go to print version here.

  1. Cognitive Processes

  2. 2.1.2 Creativity and Problem Solving.

    HCH use the notion of problem solving to mean reaching a goal that is not readily available.

    Three aspects:

    1. original state
    2. goal state
    3. the rules (or the processes)

    Example: Suppose that we need a (new) way to express future tense.

    • Original state: there is no way to express future (or the old way has become inadequate in some way).

    • Goal state: we arrive at a way to express futurity, using some existing means or expression (often a verb of volition, or of motion) to express future.

    • The rules (or the processes): draw on relevant (!) pool of existing terms, making a conceptual link between two domains

      • (concrete lexical concepts, in the source domain)
      • more abstract grammatical concepts (target domain)

    • the activity drawn on is referred to as creativity

  3. What is creativity?

    • Ability to bring something new into existence?
    • involve diverse associations formed into new combinations?
    • ability to see things in an original way, but also fitting a worthwhile purpose.
    • involves finding a solution that is unusual and useful.
    • For their purposes, creativity is the ability to conceptualize abstract domains of cognition in terms of concrete domains, e.g. domain of time in terms of spatial domains, etc.

    Result: lexical sources are used to express grammatical meanings. Unidirectional! Unusual, and Useful; the inventive type of creativity, using old stuff for new.

  4. Should we distinguish three types of creativity?

    • Universal: common to all humankind. Example: all groups can conceptualize spatial organization through use of body parts (e.g. back); use of verbs to express aspect, tense, mood.

    • Communal: common to certain social groups, units: example: some use body parts for spatial organization (e.g. the "head" of an organization); some use environmental (sky, ground) ones.

    • Individual: ordinary vs. "great" Example: Use of a term from handball e.g. "off the wall" to express something random, out of control, spontaneous (this term developed in the early 1970's). This is more "idiomatic" and doesn't translate easily.

    HCH say grammaticalization begins with individual creativity, is then taken up communally, then influenced by universal strategies of conceptual manipulation. All these things interact.

  5. Source Concepts: basic cognitive structures in the projected world, used for grammaticalization.

    • concrete objects
    • processes
    • locations

    "Source" is relative; it may be concrete objects (body parts, e.g. "back") used for space ("back of the house") and then derived from that for time ("back in 1984"). The concrete use (body part) is the source for the spatial, but the spatial is then the source for the temporal. Basic source concepts can't be derived from anything else (body parts). We may say therefore we have:
    basic --> derived2 --> derived3
    Basic stuff are the most useful names for things, or most usual states or verbal activities ('sit, stand, lie';). Some source concepts are of higher order ('person, thing, do, go'; 'have, be, take, give, make, come')

    Egocentricity, or relationships-to-ego, are important (remember the speaker-centered strategy?)

    Linguistically, we are talking about lexemes i.e.

    • the basic vocabulary, the Swadesh list;
    • the culturally independent stuff:
    • Body parts, the basics of the world (sky, earth),
    • kinship terms;
    • dynamic verbs;
    • mental process verbs like say ,
    • demonstratives.
    • Refer to elementary human experiences;
    • set of universal semantic primitives (however defined.)
    • They provide concrete reference points for human orientation: everybody has a back; everybody stands upright and looks up at the sky; everybody looks forward towards that which is coming; the past is behind, etc.
    • the head is the center of mental activity, thus of human organization
    • the hand is used to hold and do things; it may be used for CAUSE/PURPOSE.

    Wierzbicka (1989) uses two main parameters, defining power and universality

    • Thus the lexical item say is more universal than ask, demand, curse, scold, criticize... Go is a more universal verb of motion than totter, leap, stagger, ooze etc.

    • Others: do/make, take/ hold, finish; go, come, leave, arrive; be/exist, be at, sit, stand, lie, stay/live; want/like/ shall/ought

    • Some others that are lexicostatistically common are not usu. used, e.g. 'hit' (but cf. Tamil aDi 'beat' that has come to be used as a "pejorative" verb, e.g. the contrast kaappi paNNu 'copy something. (legitimately) vs. kaappi aDi 'copy something. (cheating, looking over someone's shoulder, illegally) and taNNi kuDi 'drink' (normal substances, e.g. water) vs. taNNi aDi 'drink excessively; abuse alcohol.')

      But some things can't be explained: frequency of use, conceptual simplicity, semantic unmarkedness, pragmatic silence, may not be sufficient to explain what are used as source concepts... Some languages use unusual sources for pronouns (e.g. southeast, East Asian languages) esp. when there are marked social hierarchies that must be observed in those societies. The Spanish pronoun usted may be derived from Arabic ustad 'master, expert' (instead of the commonly thought source, a shortening of vuestra merced 'your grace.')

  6. Beyond source concepts, we also needs source propositions i.e. basic states or processes common to human experience:

      ExampleType of
    Proposition
    (1) "X is at Y"Locational
    (2) "X moves to/from Y"Motion
    (3) "X does Y" Action
    (4) "X is part of Y" Part-whole
    (5) "X is (like) a Y" Equational
    (6) "X is with Y" Comitative

    These state things like where Ego is, where Ego is moving to or from; what is moving toward or from Ego, what kind of spatial relations exist between Ego and others or the world; what kinds of temporal relations exist; which are dynamic, which are static.

    The locational proposition may be used for aspectual/mood distinctions, such as progressive etc. e.g. Tamil pooy-irukeen 'I have gone' Kannada hoogiddiini 'ibid.' or French je suis en train d'aller 'I am in (the process of) going' or je suis sur le champ d'aller 'I'm about to go ('on the field of') German Ich bin im Begriff zu gehen 'I am in (the concept of) going';

  7. From Source to target The transition from source to target is via the logical distinction between intension and extension : the intensional is reduced, and the extension is increased. Another way to say this: bleaching and abstraction : the source meanings are bleached out or weakened, and the target meanings become more abstract. [We have covered this elsewhere thoroughly, in H&T (hs)]

    Note Sapir's types of concepts:

      Type of Concept Degree of Abstractness
    I Basic (concrete) Concrete
    II Derivational Less concrete
    III Concrete relationalMore abstract
    IV Pure Relational Purely abstract

    Another treatment of abstraction: Diehl (1975) correlating a continuum of egodeictic remoteness with the concrete/abstract continuum. Proposes 4 types of space:

    • Social space (SOC)
    • Spatial space (SPA)
    • Temporal space (TEM)
    • Logical space (LOG)

    which are increasingly progressively remote unlike, or 'distant from' EGO.

    Other kinds/uses of abstraction.

    • Generalizing abstraction as in Folk taxonomies: central characteristics/nucleus vs. more abstract: generalizing abstraction is inclusive, with most general category (e.g. 'life form') including the less general ('animal' vs. 'plant') and on and on ('fish' vs. mammal, etc.)

    • Isolating abstraction: separates one property, not necessarily the 'core'

    • Both types may be present in grammaticalization as the generalizing becomes more abstract, but also perhaps isolating one particular meaning

    • Metaphorical abstraction relates more abstract concepts to more concrete concepts across conceptual domains , the concrete domains being conceptual vehicles for the abstract ones.

    This latter kind is what HCH feel underlies grammaticalization; concerns the way we conceptualize the world around us.

    Table 2.2 Characteristics of Metaphorical Abstraction
    Domain Vehicle Replaced by Topic
    Ideational Clearly delineated, compact
    ==>
    Fuzzy, diffuse
      Physical (visible, tangible)
    ==>
    Nonphysical, mental
      Thing-like
    ==>
    Qualities
      Sociophysical interactions
    ==>
    Mental processes
      Process
    ==>
    State
      Space
    ==>
    Time, cause, manner
     Individual
    ==>
    Mass, class, noncountable
      Autonomous
    ==>
    Relational
    Textual "Real World"
    ==>
    "World of Discourse"
     Less discourse-based
    ==>
    More discourse-based, more speaker-based
     Referential
    ==>
    Nonreferential
      New Information
    ==>
    Old Information
    Interpersonal Expressive
    ==>
    Nonexpressive

    Of the kinds of metaphorical abstraction that exist, for linguistics, 2 are important:

    • Structure-preserving abstraction--doesn't affect the categorial status of the linguistic entities involved; Examples: preserving: a verb remains a verb, a noun remains a noun.
    • Structure-changing abstraction: changes the categorial status of the linguistic entities involved; Examples: a verb becomes a postposition; a noun becomes an adverb (e.g. Latin mens, mente 'mind' becomes an adverbial marker in French, Spanish e.g. clairement 'clearly.')

  8. METAPHOR

    As we have seen, metaphor is defined/used differently, sometimes broadly, sometimes including metonymy, synecdoche, hyperbole etc. HCH prefer a narrow definition, but see it as one of the main processes underlying grammaticalization

    They quote Bybee and Pagliuca, acc. to which metaphorical extension

    "A concrete lexical item is recruited to express a more abstract concept ... this emptying of lexical content is a prerequisite to grammaticalization because grammatical functions in themselves are necessarily abstract."

    Willet compares three ways to account for grammaticalization :

    • containment hypothesis
    • implicature hypothesis
    • metaphorical extension hypothesis

    and concludes that metaphorical extension hypothesis offers the best explanation.

    Hopper & Traugott of course prefer metonymy to metaphor; but others talk about "semantic assimilation" (Schlesinger) etc. but they really mean metaphor. HCH then give the example of going to used first spatially/locationally ('he is going to town') vs. futuratively ('he is going to work; the rain is going to come').

    (We also know that the 'locational' use of going to cannot be truncated phonologically to gonna while the future use can:

    • He's gonna work hard from now on.
    • He's going to/(*gonna) town tonight.

    Reasons for the explanation of metaphor are as follows:

    • There is a literal meaning and a transferred meaning.

    • Metaphor involves transfer, here from domain of space to domain of deictic time.

    • More 'concrete' meaning of go is more easily grasped than more abstract meaning of tense.

    • Metaphors involve anomalies, violations of truth conditions; there must have been a time when go required a human subject, but in (14: 'the rain is going to come') there is an inanimate subject.

    • Metaphors, when all is said and done, are usually ambiguous, and can be understood in both literal and abstract meaning, with result of semantic ambiguity or homonymy (of 'going to') in sentence

      (15) I am going to work.

    Other explanations: (HCH quote Jackendoff; Traugott; Peirce quoted by Brinton). But HCH feel that these explanations are somehow evasive, or attempting to avoid 'metaphor' when what it really is, is metaphor, or more specifically, metaphorical transfer. The argument of metonymy, offered by Traugott (and Hopper and Traugott) is, HCH say, too strong; actually what is happening (they say) is a combination of metaphor and metonymy. The strong involvement of a continuum with transfer along the chain of the continuum, is characteristic of metonymy, but metonymy is not the whole picture.

  9. Categorial Metaphors

    There is a scale or chain of metaphorical abstraction, thus:

    PERSON > OBJECT > ACTIVITY > SPACE > TIME > QUALITY
    These categories are

    • prototypical;
    • each includes a variety of linguistically defined concepts,
    • and each can be seen as representing a domain of conceptualization important for structuring experience.
    • The relation(ship) among them is metaphorical in that any one can be used to conceptualize any one on its right .

    • This chain thus represents some categorial metaphors e.g. OBJECT-TO-SPACE or SPACE-TO-TIME, the first being the metaphorical vehicle and the 2nd the metaphorical topic.

    E.g. in many languages go is a vehicle to express a temporal concept 'future.' In some languages PROCESS (not shown in this scale) verbs are reanalyzed as LOCATIVE particles, e.g. Ewe.

    The kind of metaphor we are dealing with here is experiential rather than expressive or taboo (e.g.). Experiential metaphors describe abstract/complex phenomena in terms of concrete/simpler phenomena . Further, only experiential metaphors seem to be involved in such use of grammatical categories. Developments in the lexicon don't follow this 'rule.'

    Examples of categorial metaphor of SPACE-to-QUALITY is Lakoff & Johnson's up vs. down metaphorical vehicle, which encompasses many expressions, e.g.

    HAPPY IS UP; SAD IS DOWN

    CONSCIOUS IS UP; UNCONSCIOUS IS DOWN

    MORE IS UP; LESS IS DOWN

    GOOD IS UP; BAD IS DOWN

    VIRTUE IS UP; DEPRAVITY IS DOWN

    RATIONAL IS UP; EMOTIONAL IS DOWN

    They refer to categorial metaphors as root metaphors whereas conceptual metaphors are typically conveyance metaphors The latter are more limited or special in scope; the former are more universal.

    The point of this distinction is to establish a difference between what one might call grammatical metaphors (the categorial) that are probably more universal, and perhaps 'built-in' to language ["used to comprehend an entire area of human experience or of the physical world"], and the conceptual metaphors that are more specific ["isolated, limited"] examples of the categorial (but in many ways also universal.)

    Example of the use of 'with' (comitative) in many languages for difference purposes:

    • comitative: "Seymour sliced the salami with Sheila."
    • instrumental: "Seymour sliced the salami with a knife."
    • manner: "Seymour sliced the salami with great care."

    (Seymour examples from G. Lakoff).

    These are entirely different categories and experiential domains, but metaphor is used to include all three. (In English at least; in Tamil, comitative would be with -ooDu and the others would be with -aale .)

    The categories, of course, are related to word-type and constituent type in an almost (as my son would say) DUH! manner:

    CategoryWord Type Constituent Type
    PERSON Human Noun NP
    OBJECT Concrete Noun NP
    ACTIVITY Dynamic Verb VP
    SPACE Adverb, adposition Adv. Phrase
    TIME Adverb, adposition Adv. Phrase
    QUALITY Adjective, Stative Verb, Adverb Modifier

    Taking another view, Croft (1984) argues for correlation between syntactic categories and their pragmatic and semantic behavior, combining Jakobsonian markedness with prototype research, to get these "natural" correlations:

    Syntactic Category(Pragmatic)
    Discourse Function
    Semantic Class
    Noun Reference (Physical) Object
    Verb Predication (Physical) Action
    Adjective Modification (Physical) Property

    Or, correlate the prototypical categories with case functions, a la Givon:

    Case FunctionPrototypical Category
    Agent PERSON
    Benefactive PERSON
    Dative PERSON
    Accusative OBJECT
    Locative SPACE
    Instrument and others QUALITY

    HCH give a visual representation of the 'expansive' nature of the chain of categorial metaphors, (Fig. 2.2,) beginning with a core ("egocentric"?) with the person, and moving out ("egodeixis?"), a progression with human category at the center, moving out to more and more abstract categories.

  10. Categories and Pronouns

    Interrogative pronouns tend to show this very nicely, with a different pronoun for each (with some exceptions) in many languages. Here their chart with Tamil examples added (and also a box for QUANTITY, which they omit):

    CategoryEnglish Gloss Tamil example
    PERSON 'who?' yaaru
    OBJECT 'what?' enna
    ACTIVITY 'what?' enna
    SPACE 'where?' enge
    TIME 'when?' eppoodu
    QUALITY/MANNER 'how?' eppaDi
    PURPOSE 'what for?' edukku
    CAUSE 'why?' een
    QUANTITY?'how much/many?' evvaLavu/ettane

    As they note, most languages don't distinguish object from activity, and use the same pronoun for it, indicating some cognitive connection? Moreover, the pronouns for PERSON, OBJECT/ACTIVITY, SPACE are 'monomorphemic' (except for 'submorphemic' stuff like wh- (and e- in Tamil)., while the categories TIME, MANNER are more complex, esp. in Tamil, where we can detect grammaticalization of some N's ( paDi, poodu, aLavu ) incorporated into these PN's. Is this "complexity in thought reflected in complexity of expression.'? Are PURPOSE and CAUSE more complex and abstract? Cf. Jackendoff's ontological categories.

    Pidgin/Creole languages tend to be multi-morphemic (examples from KPSwahili and Ewe)

  11. Creative vs. Emerging Metaphor

    Summation: we are dealing here with

    • experiential metaphors that are an

    • inclusive type of root metaphor, which they call categorial metaphor.

    • Another distinction: creative vs. emerging.

      • Creative are brand-new, can't be accounted for under the system;

      • emerging metaphors are built on existing predications, but used in new contexts or situations, acquiring extended meaning.

      • Emerging metaphors are pragmatically motivated

      • and involved with

        • conversational implicatures and

        • context-specific reinterpretation;

      • and are much more metonymic.

  12. Metaphor and Metonymy

    Metonymy involves contiguity of some sort, either transfer of meaning from contiguous unit in the discourse (e.g. evolution of negation in French from the ne particle to other particles such as pas, personne, point etc. and then deletion of ne ), or "part for whole" usage (referring to someone as 'bigmouth' or 'Goofy' etc.). HCH want to use a broader definition, involving also synecdoche .

    HCH claim that "grammaticalization is the result of an interaction of both metaphor and metonymy" which they provide more evidence for later on. Based on Jakobson and Halle (1956) who note that these two expressions reflect a "bipolar structure of language" that is fundamental for all verbal behavior. Both processes are always operative (say J&H) but different cultures, personalities, styles result in different emphases. (Note different loss with different aphasic disorders etc.) Goossens shows that metaphor and metonymy are different, but are not always kept separate in figures of speech etc. such as human vs. non-human sounds:

    Metonymic (human?)Metaphoric (non-human?)
    "Why yes," she giggled. "Not at all," he barked.
    "Absolutely never," he sniffed. "Come right in," she purred.
    "And why not?" he snorted. "Be right with you," she squealed.


    haroldfs@ccat.sas.upenn.edu, last modified 3/17/05.