Review of Hopper & Traugott
Chapter 4:

Pragmatic Inferencing

  1. Given the mechanisms of successive strategies of reanalysis and analogy the question remains what enables these mechanisms to work. Why do these things happen? Can we isolate causes or can we only rely on potential or enabling factors, but not causes.

    H&T deal with first and second later; here focus on speaker/hearer relationships or roles, and notions of 'grammar' and 'rule.'

  2. Various researchers have focussed on two competing motivations in communication:

    Speakers in a given situation may not share these motivations--one may prefer economy, the other may prefer informativeness. They process information differently.

    But ideally (the ideal communicator/hearer? ) people want others to understand them, and they seek ways to facilitate the success, helping interlocutors to process the new information they are imparting, and helping them integrate it wit h old information. This assumes they know what is new and what is old; if interlocutors differ in:

    they may not process the same way, and change may occur. Even in the individual speaker, motivations of simplicity and informativeness may compete (at different times and places: fatigue? time constrains?) so there are conflicts and needs to solve problems.

  3. What constitutes economy? Problems knowing this.

    I.e. what we have here is a competition between motivations.

  4. Some researchers more interested in competition between creativity and routinization.

    If you're interested in You'll focus on:
    abductive reasoning morphosyntactic changes
    expressivity lexical origins of grammaticalization,
    especially pragmatic enrichment of lexical stuff.
    tendency to economize routinization, semantic loss, frequency.

  5. Problem with Teleology or notion that change results from goal-directed motivations. If Grammaticalization is motivated by speaker-hearer interactions and communicative strategies, this is a teleological claim ( telos: direction) which American linguistics mistrusts, because:

    In other words, American linguistics mistrusts functionalism and doesn't see language as "in flux" and "always changing" which is what grammaticalization is focusing on, crucially. H&T admit that

  6. Inferencing and Meaning Change. Most linguists working on grammaticalization don't like the general focus on autonomous syntax found in abstract linguistics, and in fact they actively challenge it. Clearly when a lexical item becomes grammaticalized it is because it has both 'meaning' and other (syntactic, phonological, morphological) properties. Meaning is therefore involved. Also, the kinds of things we see happening in grammaticalization occur across a wide variety of languages and families, affecting similar classes of lexical items. Thus most work on grammaticalization has focused on the kinds of meaning changes involved, and the cognitive motivations behind them. For some these changes are

    Position taken by H&T: meaning changes and cognitive strategies that motivate them are crucially linked to expressivity. Changes are initially pragmatic and associative, arising in flow of speech. Later, as grammaticalization proceeds, meaning loss, "bleaching" occurs, but even then older meanings "hang in there ". and constrain or affect the newer "emptier" (because grammaticalized) meanings.

  7. Difference between 'semantics' and 'pragmatics.' We need to distinguish between the 'semantic' meaning and the 'pragmatic' meaning of
    Can you pass the salt?
    i.e. between the

    Or, as some have put it:

  8. Polysemy, Homonymy, Metaphor, Metonymy.

    In contemplating the role of meaning change and its involvement in the mechanisms and motivations for grammatical change, we have to talk about different elements or aspects of meaning. But as H&T say

    (4.2.2) "there is little agreement ...[on how] to characterize the relationship between the various senses of a form. People don't even always agree on the idea that forms that 'sound the same' but have different meanings should be called homonymy or polysemy .
    [I can't agree with McCawley, hs]. H&T declare themselves in favor of a theory of semantics that
    allows for polysemy wherever some common semantic factor ... is present ... . From this perspective, any two or more conceptually related senses with the same form are regarded as one item.

    1. Polysemy vs. Homonymy

      Polysemy has to do with the fact that a particular lexical item may have multiple meanings, which may have something in common with each other, such as the "will" example on pg. 70. Idea is that 'will' is polysemous, and the 3 meanings have something in common.

      • Prediction
      • Willingness
      • Intention

      Homonymy is the term that defines items that sound alike, but are really different, such as "two, too, and to" all sounding like [tu] but not sharing any meaning. Or different senses of the word down (i.e. 'opposite of "up"; fine feather-like undercoat found on some birds). Or McCawley's "bachelor" example.

    2. Metaphor. Metaphor is a figure of speech where something is referred to as if it were something else, e.g. He's a stuffed shirt. One thing is referred to in terms of another, usually the direction being from concrete to more abstract meaning. "You take the high road and I'll take the low road."

    3. Metonymy: there is confusion in the literature on the meaning and use of this term. For some, it has to do with contiguity (adjacency?) i.e. the kinds of meaning shift when some element acquires part of the meaning of an element that is near to it, spatially or figuratively. E.g., the borrowing of the French joue 'cheek' into English as "jaw".

      Another business is that of the phenomenon of defining something because of "part for whole" i.e. someone/something is identified by some (prominent?) feature of its characteristics, e.g. 'Mr. Stuffed Shirt' (for a pompous, self-absorbed person), 'Mr. High-pockets' for a tall man; "Four eyes" for someone who wears glasses; `Big-mouth' for someone who talks a lot, etc.

      Here's a cartoon playing on the expression Hanky-Panky. Or see this set of metonyms for Disney characters. Metonymy can get confused with synecdoche by some...

    When it comes to the fine-grained continuum of meaning distinctions involved in the stages ( cline ) of grammaticalization, we may need a theory that allows for meanings to be closer, or more distant, i.e. multi-gradient (see various proposals p. 70). For grammaticalization, we must assume polysemy if there is a plausible semantic relationship, because if meaning is changing along the continuum, there has to be overlap of meaning at some (or multiple) points: it would not make sense for there to be discontinuity in the semantic "chain".

    But what constitutes a plausible semantic relationship? H&T list some work on this:

  9. Conversational and Conventional Inferencing

    Grice and his conversational implicatures (1975) outlined a number of maxims, but there are questions about whether they are the right ones; H&T say maybe the maxim of Relevance alone, defined to include informativeness, may be adequate to account for pragmatic meaning. Processes of grammaticalization draw on Relevance. (See also the review of Levinson's new book Presumptive Meanings: The Theory of Generalized Implicature.

    H&T: most conversational implicatures are abductive since a speaker hears an utterance, concludes it is motivated by a particular Maxim, then guesses speaker's intent. But maxims can be flouted:

    Pragmatic effect of implicatures across utterances illustrated by examples where once clause may be assumed to be related in cause and effect when in fact it may be unrelated, as in:

    1. Ned Devine got a telegram saying he'd won the Irish Sweepstakes.
    2. Ned Devine died of a heart-attack.

    3. Denise Rich contributed $1.5 million to Clinton's Presidential Library fund.
    4. Clinton pardoned Denise Rich's former husband.

    Speakers assume that (2) results from (1) (and 4 from 3) but if the sentences are reversed:

    1. Ned Devine died of a heart-attack.
    2. Ned Devine got a telegram saying he'd won the Irish Sweepstakes.

    3. Clinton pardoned Denise Rich's former husband.
    4. Denise Rich contributed $1.5 million to Clinton's Presidential Library fund.

    it could be that Ned died before he learned the news about the lottery, and that March Rich was pardoned before Denise forked over the dough. If connectives such as and (then), subsequently, and so, whereupon we can assume causality, or at least "constrain the relevance of the proposition it introduces".

    Conversational implicatures are contrasted with conventional implicatures, which are unpredictable and arbitrary, and must be learned (and are not cancelable ). Example from Tamil of use of self-benefactive ko(L) which implies "success" as contrasted with lack of ko(L) which implies nothing; the latter can be cancelled; but the former can't be cancelled. (H&T give example of English manage to Vb and its implicature, which cannot be cancelled.)

    Without ko(L) With ko(L)
    raman veele-ye teeDinaan raman veele-y-e teeDikiTTaan
    "Raman work-ACC. sought R. work-ACC sought-Self-Benefactive
    ...aanaa keDekkalle * aanaa keDekalle
    but available-not but available-not
    but didn't find any." *but didn't find any
    "R. looked for work but didn't find it." *"R. looked-for-and-found work
    but didn't find work."

    We can therefore include conventional implicatures/meanings in the polysemies of a form. Examples of temporal and causal meanings of since which can be usually figured out (non-past event/state, causal; events in past, temporal); but sometimes ambiguous, as in

    Since Rocker was suspended for xenophobic and homophobic remarks, a foreign-born judge was appointed to hear the arbitration case.

    which could mean either that there is a sequence of events (appointment of the foreign-born judge followed the suspension), but no causality or causality is implied; or the appointment of the foreign-born judge was made precisely because Rocker had demonstrated xenophobia and homophobia. Thus since is semantically ambiguous and polysemous. Next question: Are there pragmatic polysemies as well as semantic ones?

    After Rocker was suspended for xenophobic remarks, a foreign-born judge was appointed to hear the arbitration case.

    H&T indicate that this sentence (with "after") can be interpreted generally to mean:

    Because Rocker was suspended for xenophobic and homophobic remarks, a foreign-born judge was appointed to hear the arbitration case.

    because the relationship between the two clauses is "enriched" with the causal interpretation, and relevance is increased i.e. the relationship is both temporal and causal. But there are no conventional or regular implicatures and no syntactic correlation, so this suggests that there can be pragmatic ambiguities/polysemies as well as semantic ones.

    This problem of ambiguity illustrates the logical fallacy known in Latin as Post hoc, ergo propter hoc `After the fact, therefore because of the fact.'

    Thus, if you want to avoid the pragmatic ambiguities, you have to say something like:

    Precisely because Denise Rich contributed $1.5 million to Clinton's library fund, Clinton pardoned her former husband.
  10. Role of pragmatic inferencing in grammaticalization

    Grice stated tentatively that conversational implicatures may become conventionalized. Others hypothesized that the "secondary meanings" of tense and aspect (e.g. "present relevance" of the perfect) are derived by conventionalization of implicatures.

    H&T state that in early stages of grammaticalization, conversational implicatures often become 'semanticized,' i.e. become part of the semantic polysemies of a form. Inferences, to play a role, must be frequent and must be standard in order to play a role, especially cross-linguistically. Standard would be such things as "since" having both temporal and causal meanings, i.e. the inference of causality from temporal sequencing. If something is local, idiosyncratic, highly contextualized (perhaps specific to a culture) we won't expect inferences to become conventionalized.

    Another question: When can we recognize conventionalization to have occurred? H&T give examples of development of causality with English "since", some of which are ambiguous (could be causal, could be temporal; could be causal, but probably not conventionalized in OE, but only in late ME, i.e. attested in stative and other non-completive environments where the temporal reading is blocked.)

  11. Metaphorical and Metonymic Processes

    Metaphor heavily involved in meaning change. Metaphoric processes involve:

    • inference across conceptual boundaries.
    • Metaphoric processes are seen as mappings or associative leaps (itself a metaphor) from one domain to another;
    • Such leaps are motivated by analogy and iconic relationships;
    • These relationships can be observed cross-linguistically.

    Traditionally, metaphoric processes seen as semantic, but lately, seen more as pragmatic (not being truth conditional, based in communicative use.) H&T:

    We accept the view that metaphor is pragmatic, and ... that in so far as metaphor is primarily analogical in character, is different from the kinds of conversational processes based on maxims mentioned above... .
    As Green says, metaphors are intended to be interpreted as literally false, (a person is not literally a stuffed shirt) whereas conversational implicatures are not. (But note that people often say things like He keeps his nose to the grindstone, and I mean literally.)

    Earlier, metaphor was seen in the lexicon, but now some people see it as an early motivation in grammaticalization: (Bybee and Pagliuca 1985)

    Rather than subscribe to the idea that grammatical evolution is driven by communicative necessity, we suggest that human language users have a natural propensity for making metaphorical extensions that lead to the increased use of certain items.
    For H&T, best examples are development of spatiotemporal forms, e.g. the development, in many languages, of temporal forms out of spatial ones, e.g. 'back' (in time) out of 'back (behind us)' based on the metaphors SPACE IS AN OBJECT, TIME IS SPACE, so this happens in Dravidian, too: meele 'on top of' can be used temporally: adukku meele 'on top of that' or 'upon that; whereupon' (i.e. right after):
    adukku meele, enna senjirukkiinga?
    that upon, what done-have-you?
    After that, what have you done?

    Note also development of expressions like inimee from inru-meele 'today-top of' (Fabricius gives ini meaning 'hereafter') which now means 'from now on.'

    Other typical metaphorical things are the use of English particles up, down on, over for aspectual meanings; the development of gonna , the development of modals, where obligation --> possibility, probability (epistemic meanings). Sweetser takes a different approach: "socio-physical concepts of forces and barriers" i.e. tendency to experience the physical, social, and epistemic worlds in similar ways, allowing the mapping of sociophysical potentiality onto the world of reasoning. E.g. obligative must and epistemic must

    • You must marry the girl and do the right thing.
    • It must have rained last night (= it must be the case that...)

    H&T criticize those who think only "lexical item > grammatical item" i.e. independent of context, instead of (what they prefer):

    lexical item-in-discourse --> grammatical item

    i.e. in utterance contexts. They say it is not go that is grammaticalized in gonna but rather the phrase be going to in the local context be going to in order to V. But see the development of Tamil poo (Kannada. hoogu ) meaning "go" that becomes a change of state aspect marker with certain verbs, such as die, rot, cool, spoil, break etc. e.g.

    koDe kaaladalli haNNu beega keTT-hoogatte antare
    hot season-in fruit quickly spoiled-goes say-they
    Fruit spoils quickly in the hot season, they say.

    As for context here, it seems to be the verbs that have a meaning involving transitional state (in fact they're all involved with a kind of semantic 'cline') which can be conjoined with hoogu to give the notion 'go to complete the change of state.' So hoogu meaning simply 'motion away from speaker' with these verbs is grammaticalized as an aspect marker meaning 'completive' (and usually with a nuance 'pejorative' i.e. the effect is undesirable.)

  12. Metonymic Processes

    H&T take the stand here that metonymy is more important than metaphor, since metaphor is merely analogical. Metonymy, in contrast, depends on contiguity and reanalysis. They try to indicate (p. 81) that much of what has been called metaphoric in grammaticalization is perhaps not correct: the metaphoric use of 'go' becoming grammaticalized in gonna is, (they say) improper; it's not 'go' that was grammaticalized, but the context be going in order to V and it was the contiguity (adjacency?) with 'to' in the purposive sense that made the difference.

    Meaning changes arising out of contiguity in Linguistic/pragmatic contexts are associative or conceptual metonymic changes. For short, H&T call them "(conceptual) metonymy".

    Problems with Metonymy:

    • originally, used for contiguity in the non-linguistic world, e.g. French joue "cheek" --> Engl. jaw (and other Latin to French. ex's). These are parts of the body that are contiguous. (Perhaps there are some politeness and euphemistic factors here, e.g. the French soutient-gorge (lit. 'support-throat') for "brassiere") Another famous ex: ME (ge)bed 'prayer' transferred to the actual 'bead' of the rosary.

    • Other examples above, of part-for-whole, e.g. things/characters being named after a part (albeit distinctive) of their perceived physical or behavioral characteristics.

    • tendency to confuse metonymy with synecdoche already mentioned.

    • tendency by many linguists to include metonymy under metaphor, which H&T don't like.

    • Or, if distinguished, consigned to an insignificant place.

    • See also Jakobson & Halle's 'classic distinction' of metaphor as paradigmatic vs. metonymy as syntagmatic, thus metaphor leads to coherence, metonymy leads to juxtaposition and potential incoherence.

    H&T reject these uses as useless (?), preferring the restrictive sense of conceptual association. In this process, meaning (say element A) permutes to another element that is adjacent to it. They refer to Brinton (1988) and his analysis of E. considering, supposing and concessive 'while' (e.g. 'While I do admit that Mr. Gore has a point there, I otherwise reject his proposal.') Note also Brinton's analysis of development of English aspect markers, esp. have which sees this as metonymically motivated, rather than metaphorically motivated.

    More useful views (acc. to H&T): Antilla, saying metaphor is analogical and iconic, while metonymy is (semantic transfer through) contiguity and thus is imminent future. Thus semanticization + grammaticalization of be going to is shown best when following subject and/or verb is incompatible with purposiveness , yielding to only possibility: immediate future.

    The reject the metaphoric explanation, the metaphor being "projection of a trajectory through space mapped onto a trajectory in time" doesn't explain why E. progressive (-ing) and "to" are involved. Compare cross-linguistically:

    ange saappiDa poo r-een
    there to-eat go pres/PNG
    I'm going to eat there (immed. future)

    ange saappiDr-ad-ukku poo r-een
    there eat-VN-DATive go pres/PNG
    I'm going there (in order) to eat

    Tamil's grammaticalization of poo 'go' as an immediate future doesn't have the same morphology as English, i.e. no progressive marker (-ing), only an infinitive, which can have the inference 'purposive' but tends to not have it, since the second example, with a dative-marked VN, is more common as purposive in modern Tamil. (In LT, infinitive has a number of possible readings, such as purposive, but also 'simultaneous' and maybe some others.) Note that in Kannada, however, the distinction between use of dative-marked verbal noun (for purposive) and the 'other' "old" or "second" infinitive (Kittel 1903:122) for other uses has broken down--the old infinitive is now used only with modals and negative, but can't be used (e.g. with hoogu 'go') to mean "going to Vb" in the way that Tamil has grammaticalized poo . So Kannada -okke (dative-marked VN) is used for both purposive and directional as in this example (Schiffman 1983:53.)

    naan sinimaa nooDokkeuurigehoogttiini
    I moviesee-dativeVNtown-dat.gopresPNG
    I'm going to town (in order) to see a movie.

    This sentence has two datives in it, one dative-of-Verbal-Noun for purposive, another directional dative (to town) and a verb 'go' which only means 'move in the direction of', not 'immediate future" etc.

    Semanticization of 'concessive' while illustrates how something that starts out as a marker of temporal simultaneity was reduced phonologically to wile and with the lost of the demonstrative that signalled the simultaneity, less precise conversational inferences could arise, one of them being causality (or the reason for s.t.), i.e. 'all the time that' or 'all the while that' 'all during s.t. Then finally, a new inference of 'even though' e.g.

    (Even) while others slept, you kept a watch out.

    And finally, a concessive inference, i.e. there is a contrast of values, a "subjective construal of the world" in linguistic terms, i.e.

    (Even) while others wasted their time sleeping, unconcerned with the real dangers around us, you valiantly, bravely, and unselfishly kept a watch out for everybody's interest.

    This is, of course, what H&T refer to often as the speaker-centered nature of these inferences and subsequent grammaticalizations, the subjectification of them. (E.g. development of epistemic modals.)

  13. Metaphor and Metonymy as Problem Solving (Heine, Claudi and Hünnemeyer 1991). Semantic change in general, not just grammaticalization, is a response to a problem, i.e. a solution to a perceived problem. What are the problems?

    The problem? The solution Main Direction Which Axis?
    Representing things in one semantic domain in terms of another Metaphor toward informativeness Paradigmatic: specify one (complex) thing as another not present in context
    How to find ways to regulate communication and negotiate speaker-hearer interaction Metonymy Toward informativeness Specifying one meaning in terms of others present, though covertly, in the context.

    Another way of looking at it:

    • Metaphor: correlated with solving problem of representation linked with analogy; (analogy is overt .

    • Metonymy: correlated with shifts to meanings situated in subjective belief state or attitude toward situation, incl. linguistic situation. Solve problem of expressing speaker attitudes (reanalysis which is covert.)

    • Metaphor and Metonymy are complementary

    • Metaphor is common enough, but

    • Metonymy predominates in early stages, at least.

    • Competing motivations of routinization and expressivity are involved at various stages, perhaps even cyclically (hfs).

  14. Problem of bleaching

    Many researchers use the metaphor of Verbleichung, affaiblissement, bleaching or a process whereby linguistic units lose in semantic complexity, pragmatic significance, syntactic freedom, and phonetic substance. (Heine and Reh 1984). But H&T have talked about the opposite: pragmatic enrichment, strengthening, etc.

    Summary of principles:

    • weakening etc. occurs as grammaticalization progresses
    • strengthening etc. occurs at the outset; shift, not loss of meaning; or demotion of some meanings, promotion of others.
    • over time, items lose semantic/pragmatic meaning, becoming completely grammaticalized and thus empty morphosyntactic elements.
    • meaning changes in grammaticalization are not arbitrary but can almost always be reconstructed or derivable by inferencing.
    • there are no sudden losses or shifts of meaning (with some questions about causative do in ME)
    • traces of lexical meanings and their constraints usually are carried along, so that e.g. Tamil postpositions that are derived from transitive verbs still take the accusative case, while other postpositions may not (cf. also example from Gã given by H&T.)
    • Polysemies may be the result, then, as in English will

  15. Generalization and Unidirectionality, cont'd

  16. Development of Nominal Categories and of Verbal Categories: two typical paths. These are clines:

    • A cline of grammaticality, making reference to hierarchical categories relevant to constituent structure.

    • clines of decategorialization i.e. starts with a full category (N, V) with then loss of morphological trappings of the category.

    But do we have unidirectionality along the cline? Are the clines continua, or rather clusters at points with grammatical properties resembling ("family resemblance") certain other kinds of things:

    • Auxiliaries
    • prepositions
    • articles
    • something seen in some other languages.

    Clusters are not rigid, fixed points, but gathering places, like iron filings around magnets.

  17. Period of Overlap between older and newer functions, so not a situation with everything lined up like cars of a train. Some ppl. use term 'chaining' to show non-linearity; H&T prefer 'layering' allowing for multiple origins of a grammaticalized form, i.e.


    Remember that once embarked upon, grammaticalization does not have an inevitable result; may be arrested etc. We can't work back from any one form to see a clear path; and we can't illustrate a particular cline with only one form. (Not enough historical record, they say; I would bet we might find examples in Tamil with its long time depth, hs). What is important is the issue of directionality between adjacent forms on the cline not demo. of complete sequence of events for a given form.

  18. What are some typical clines? Lehmann 1985 has given following:

    • Relational Noun -->

    • Secondary Adposition --> (kind of prep/postposition)

    • Primary Adposition --> (kind of prep/postposition)

    • Agglutinative Case affix -->

    • Fusional case affix.


      • Relational Noun: nouns such as 'top, side, way' or Tamil pakkam 'side'

      • Secondary Adposition: short phrases such as 'in/on the way', 'beside', 'ahead of'

      • Primary Adposition: reduced, maybe monosyllabic, indicating purely grammatical relations (instead of locational) such as 'by, of, with, to.' Some of these may be themselves locational and relational, e.g. 'by'. Can be easily cliticized.

      • Agglutinative Case affix: Hungarian examples; Tamil (see below)

      • Fusional case affix. Latin example -ibus , Russian -yx, -ov, -amy etc.

      Again, not strictly discrete, but arbitrary clustering points on a cline. Most forms we find will not fit clearly into one or another of named categories, but will land somewhere on the cline.

    • Example of the relational Noun: development of Tamil case and postpositions.

      • Relational Noun: pakkam 'side' as in maDraas pakkam 'Madras-side' (the Madras area, the Madras region)

      • Secondary Adposition: pakkattule which is pakkam plus loc. ule e.g. maDraas pakkattule 'near Madras'

      • Primary Adposition: eduttaaple 'opposite' from the verb edir 'opposite' + taan 'itself' + poola 'like.' with phonological reduction of edir-taan-poola incl. deletion of long vowels in poola which otherwise doesn't happen.

      • Agglutinative Case affix --> 'Ablative' marker -lerundu, -lendu which is the LT loc. -il plus irundu past participle of iru 'having been' e.g. maDraas-lerundu 'from Madras.

      • Fusional case affix: Dravidian doesn't get fusion in the way that IE lgs. do, but maybe the accusative -e which can't be analyzed as having any source, i.e. whatever its source, it's now obliterated and shortened (from LT ai ) to -e : avan-e paatteen 'him saw-I'.

    • Verb to Affix Cline:

      • Full Verb -->

      • (Vector Verb -->)

      • Auxiliary -->

      • Clitic -->

      • Affix

      That is:

      • Full Verb: verbs with full lexical meaning, all grammatical characteristics, full morphology etc. like other verbs.

      • Vector Verb: may be a special category? See Hindi, Kannada, Tamil (below)

      • Auxiliary Verbs are finite, have tense/mood etc., but syntatically specialized (e.g. English Aux inverted in Q's); maybe can't be marked for tense, or (in Dravidian) tend to not be marked for PNG. Many examples in many lgs. Can have examples of both full and aux status, e.g. Engl. have as full vb, aux (have gone etc.) and new development of hafta .

      • Cliticization: have becomes 've e.g. we've gone and done it again or the French cliticization of forms of avoir added to future, e.g. j'aurai, ils sauront 'I'll have, they'll know .,
last modified 26/1/2001