Lexical constructionalization Elizabeth Closs Traugott [email protected] in collaboration with Graeme Trousdale University of Santiago de Compostela,

Lexical constructionalization

Elizabeth Closs Traugott [email protected]

in collaboration with Graeme Trousdale

University of Santiago de Compostela, Oct. 18th 2012

2

Outline• My overall objective.• Summary of key points about the model of construction

grammar (CxG) used and types of constructionalization (Cxzn).

• Some prior work on lexicalization (Lxn).• Word formation (W-F) as evidence for expansion in

lexical constructionalization (LCxzn).• Two examples: un-, -ræden.• Snowclones.• Similarities, differences between LCxzn and GCxzn.• The “value-added” of a constructional perspective on

language change.

3

Objective

• Together with Graeme Trousdale I seek to develop a coherent and restrictive account from a usage-based construction grammar perspective of the development of constructions over time that resynthesizes what we know about language/sign change (Traugott & Trousdale Forthc).

• Today I discuss some aspects of thinking about change in contentful Cxns, including lexicalization.

• I emphasize similarities with and differences from GCxzn.

4

Construction grammar: key points

• According to the model of construction grammar

(CxG) I espouse (see Croft 2001, Goldberg 2006):

a) The basic unit of grammar is the construction

(Cxn): a conventional pairing of form and

meaning.

b) Grammar is non-modular—morphosyntax,

phonology, semantics, pragmatics, discourse

function cannot be studied separately.

c) Language, like other cognitive systems, is a

network of nodes and links between them.

d) Language structure is shaped by language use.

5

• There is a gradient from contentful (“lexical”) to procedural (“grammatical”) Cxns at both the specific and the schematic levels.

• The most contentful Cxns (LCxns):

F: N/V/Adj ↔ M: contentful, referential.• The most procedural Cxns (GCxns):

F: TMA markers/case markers (including word order)/connectives/some adverbs (even), etc.

↔ M: indexical/relational/non-referential.

• Many are intermediate (ICxns), e.g. some adverbs (luckily), way-Cxn.

6

Some prior work on lexicalization• Diachronically lexicalization (Lxn) has been

interpreted in many ways, but typically as reduction (LIR):

• Lehmann (2002: 13): “only complex units may be lexicalized”. A complex unit of the type [XY]Z ceases to be complex and is accessed as a unit: Z as a whole is affected, and the dependency relationship between X and Y is obliterated, e.g.

- cup board ‘shelf for cups’ > cupboard ‘cabinet

with shelves’,

- Lat. de ex de ‘from out from’ > Span. desde.

7

• Lehmann contrasts “renunciation of [the unit’s] internal analysis” (Lehmann 2002: 13) in Lxn with retention of internal complexity/asymmetrey in Gzn.

• Although lexicon and grammar are on opposite ends of a pole, L does not allow for gradience in Lxn—any change that is accessed as a unit is lexicalized.

• Brinton & Traugott (2005: 96): “Lxn is the change whereby in certain linguistic contexts speakers use a syntactic cxn (≠ Cxn!) or word formation as a new contentful form with formal and semantic properties that are not completely derivable or predictable from the constituents of the cxn or the word formation pattern. Over time there may be further loss of internal constituency and the item may become more lexical”.

8

• B&T show that Lxn and Gzn share many properties, e.g. in both there is strong evidence of:

- gradualness

- coalescence

- fusion

- decategorialization

but in Lxn little evidence of e.g.:

- bleaching

- subjectification

• Note these are statements about characteristics, not absolutes!

9

Is there evidence of expansion?• These interpretations of Lxn as “increased reduction”

(LIR) need to be revisited and enriched.• Just as there is GE, so there is lexical expansion (LE) in

LCxzn (but they are not identical processes).• The development of word-formation (W-F) patterns = LE

in LCxzn; note this development often involves increase in dependency (contra Lehmann 2002).

• W-F formation results from generalization of specific micro-Cxns as a schema, which then serves as the template for specific new type-formations.

• Specific micro-Cxns generated by W-F may undergo LIR-type reduction as specified in Brinton & Traugott (2005).

10

A recent example: un-N• A recent ex. is development of un-N W-F (derived

from an already existing W-F pattern).• According to Horn (2005) un-N originates with

Humpty-Dumpty’s unbirthday present (1872 Lewis Carroll, Alice in Wonderland) where birthday is a modifier, i.e. adjectival, and 7-Up’s Uncola ad (1962).

• It developed gradually (occasional exs. attested), then became highly productive after Uncola appeared (Cxzn).

• F: N not derived from Adj (untruth),

M: ‘not your typical/’bad’ ex. of’ (unbank 1999, unelection 2000 re Gore-Bush election).

11

• Horn (2005) draws on:

- Aristotelian notions of markedness and privation

(marked exception to category) found in un- ‘not’,

- scalarity of un- (Algeo 1971: 90-91 noted that anyone

of any faith (or none) can be non-Christian, but only a

Christian can be un-Christian).• Horn gives exs. of expansion to phrases (a common feature of

W-F in the 20thC), with:

- modifier Ns (Un-Petroleum Lip Jelly, un-Mister

Rogers tone of voice) (p. 357).

- idioms, routines (un-fooled-around-with, un-what-I-

expected) (p. 362).

• No evidence of reduction yet.

• But clearly an ex. of contentful schema expansion (LE).

12

An old example: -ræden• The next ex. involves development of schema,

increase in dependency, and LIR reductions.• In OE there were several Ns meaning ‘status’

(Dalton-Puffer 1996, Dietz 2007, Haselow 2011):

- dom ‘judgment, doom, dignity, power, status’,

- had ‘state, rank, condition’. In ME merged with

-hede (= -hood, e.g. knighthood, neighborhood),

- ræden ‘condition’; mainly in compounds in OE.

• All could appear as free nouns in early period.

13

• All were used in compounding patterns in later OE.

• All developed affixoid properties (bound, somewhat bleached semantics: more abstract semantically than in compound, but less abstract than affix, Booij 2010: 57),

e.g. c900 Adj + dom compounds and c950 N + dom became productive affixoid W-Fs (Dietz 2007, Haselow 2011).

• A fourth N scipe ‘condition, state, office’ was used as an affixoid throughout OE.

• This resulted in a network of productive compound and affixoidal nominal W-F Cxns.

14

• Development of compound W-F pattern is Cxzn (development of morphological dependency, accompanied by meaning change), e.g. in N + dom compounds, N was decategorialized (no morphology), dom was agglutinated, and meaning was restricted (cyningdom ‘king’s jurisdiction/sovereignty’). This is development of affixoid (Cxzn).

• Further development of affixoid W-F pattern is third Cxzn (greater morphological dependency, bleaching) (kingdom ‘king’s domain/realm’ < cyningdom). This is development of affix.

• This means LCxzn may recur.

15

• Ræden used as N (pre-Cxzn, rare):

(1) Sume mid aþum gefæstnod, þæt hi on hyre

some with oaths secured, that they in their

rædenne beon woldon.

service be would

‘had made some swear that they would be in service

to them’. (918 Chron C [DOEC])

• Compound schema developed c1000 referring to

judicial and social relations (Cxzn):

(2) lufige annysse & broðorrædene.

love oneness and brotherhood (c1000 ÆCHom I.xi. [DOEC])

16

• LE as new schema and relatively productive; but also LIR as increased dependency, and some reduction as N1 decategorialized:

(3) [[[X]Nk [ræden]Ni]Nk ↔

[[SEMi with relation R to SEMk] ENTITY]j]

• In ME there appears to have been affixoiding and further generalization of meaning (Dietz 2007), cf hatræden ‘hatred’, a second Cxzn.

• This was LE as schema was further expanded; but also LIR as morphological boundaries reduced:

(4) [[[X]Nk [-ræden]i]Nj ↔

[[condition with relation R to SEMk]ENTITY]j]

• X-ræden competed with X-dom, X-had, X-scipe, as well as X–ness, borrowed X–ity, etc.

17

• During ME -ræden lost out to the other affixoidal patterns, i.e. the network lost a Cxn.

• -ræden was reduced phonologically and eventually restricted to: hatred, kindred, i.e. atomic monomorphemes—LIR (Brinton & Traugott’s reduction to L3 in Lxn).

• The remaining affixoidal W-Fs gradually underwent division of semantic labor (Dalton-Puffer 1996: 125):

(5) -ship: ‘social status of a N’ (kingship),

-dom: ‘jurisdiction/territory of a N’ (kingdom),

-hede: ‘abstract or inner qualities making up a N’

(motherhood).

18

An example of gradience in post-LCxzn

• Post-LCxzn CCs can be gradient with respect to compositionality: see X-fire.

• bonfire < bone fire (compound):

(6) the people waked at home, & made iij maner of fyres. One was clene bones and noo woode, and that is called a bone fyre.

‘the people awoke at home and made three kinds of fire. One used clean bones and no wood, and that is called a bone fire’. (1493 Festyvall (1515) [OED, bonfire])

• By EModE bonfire means ‘funeral or celebratory fire’.

19

• Bonfire is a member of a compound Cxn X-fire

including logfire, campfire:

(7) [[[Xi fire]Nk ↔

[[fire with relation R to SEMi]ENTITY]k]

• Each member of the X-fire Cxn is slightly different, so this is not a strongly entrenched schema:

- in logfire, X is what is burned,

- in campfire, X is the location of the fire,

- in bonfire, X is phonologically reduced and less

compositional than in other cases.

20

Snowclones• Snowclone: a term for a type of cliché that is:

“a multi-use, customizable, instantly recognizable, time-worn, quoted or misquoted phrase or sentence that can be used in an entirely open array of different jokey variants by lazy journalists or writers.” (Pullum 2003, 2004)

(8) X is the new Y: Green is the new black, Fake is the new real, Old is the new new (The Times Oct 4 2012)

• Zwicky (2006) argues that snowclones arise in several stages:

- A pre-formula stage with variations on an expression,

all understood literally, and requiring no special

knowledge (What one person likes, another person

detests),

21

- A catchy fixed formula is used (with similar meaning) often drawing on a proverb, title, or quotation (One man's meat is another man's poison),

- The fixed expression may be quickly extended with the development of open slots or playful allusion to it, e.g. via puns (One man's Mede is another man's Persian),

- Snowcloning, a second fixing as variants become (relatively) routinized as formulas with open slots in them (One man's X is another man's Y).

• On this analysis, snowclones can be said to arise from LCxzn of a schema after a number of CCs.

22

• Kay (Forthc) argues that:

“unlike the construction that licenses red ball”, patterns like [A as NP, e.g. dumb as an ox, flat as a pancake] “are neither necessary nor sufficient to produce or interpret any set of expressions of the language: each expression that exemplifies one of these patterns has to be learned and remembered on its own”, i.e. snowclones ≠ Cxns.

23

• But without experimentation we cannot know whether each is produced independently. In most cases there clearly is a pattern which qualifies as a Cxn.

• Snowclone patterns are productive, have all the characteristics of Cxns; see also Gonzálvez-Garcia (2011) on Spanish snowclones: “language users store both the parts and the wholes, and retrieve them when they need them (Bybee and Eddington 2006, Bybee 2010)”.

24

Instantaneity vs. gradualness

• Development of schemas and subschemas is gradual, therefore Cxzn.

• However, the individual expressions (the specific micro-Cxns) are created instantaneously, and are therefore not Cxzns.

• This is also true of the various kinds of “extra-grammatical W-F”, e.g. (Mattiello Forthc):

- clippings (‘tude < attitude, (to) diss < disrespect)

(clippings were claimed to be deGzn in Ramat

1992, 2001),

25

- (compound) clippings (sitcom < situation comedy,

‘hood < neighborhood),

- initialisms (OTT ‘excessive’ < over the top),

- acronyms (AIDS < acquired immunodeficiency

syndrome),

- blends (motel < motor + hotel, chortle < chuckle

+ snort),

- back-formations (edit < editor, emote < emotion),

- phonaesthemes (e.g. forms in –ump denoting

heaviness or clumsiness, e.g. clump, dump).

26

Similarities between LCxzn and GCxzn

• Routinization, gradualness (of pre-Cxzn CCs and of post-Cxzn CCs),

• Coalescence,

• Fusion,

• Decategorialization,

• Recategorialization.

This is basically what Brinton & Traugott (2005) found

for Lxn and Gzn.

27

Differences between LCxzn and GCxzn

• Output of LCxzn is contentful, that of GCxzn is procedural.

• In LCxzn (but not GCxzn) some micro-Cxns are created instantaneously without pre-CCs.

• The syntactic and constructional contexts for LCxzn are internal to a complex Cxn (e.g. a compound). Host-class expansion is limited to the choice of base (N, Adj, V). Post LCxzn the syntagmatic contexts do not expand (contrast e.g. BE going to).

• In LCxzn there may be little “bleaching”.

28

• What about schematicity, productivity, compositionality? • Trousdale (2008, 2010) suggested:

- in GCxzn increase in schematicity, productivity,

decrease in compositionality,

- in LCxzn decrease in all three (unlike GCxzn).

• Latter is true of specific contentful micro-Cxns, e.g.

gar-leac ‘spear leek’ (a member of an X-leac compound Cxn) > garlic.

• BUT the LCxzn involving W-F (i.e. development of schemas, e.g. X-leac) involves increase in schematicity, productivity, decrease in compositionality (like GCxzn).

29

A constructional view of change

• A constructional perspective is only one way of thinking about language change. Trousdale and I have attempted to develop a coherent view of change, incorporating much of what is already agreed upon

about language/sign change into a coherent approach to Cxzn.

• The main theoretical assumptions and methodological criteria are:

a) Language/sign change is change in use,

b) The basic structural unit is the Cxn,

30

c) Therefore both form and meaning need to be considered equally in studying language change.

d) Changes need to be considered from both specific

(micro-) and schematic (macro-) perspectives.

e) Since lexical and grammatical changes are on a

gradient from contentful to procedural meaning,

they need to be viewed as complementary, not

orthogonal.

f) Changes need to be understood in terms of networks.

31

g) Innovations occur in individual minds as a result of

language use, and can be recognized as “changes” only when conventionalized and taken up by others.

h) Change typically occurs in small discrete steps

(gradualness over time), resulting in variation (synchronic gradience).

i) Being gradient, conventional patterns and norms of

use allow for innovations (and eventually changes) to emerge over time.

32

What is the “value-added” of a constructional perspective?

(i) An approach based in form-meaning pairings (signs) obviates the need for elaborate “interfaces” between modules.

(ii) The ability to see how networks, schemas as well as

micro-Cxns are created, grow or decline, and the ability to track the development of patterns at both levels allows the researcher to see how each micro-Cxn has its own history within the constraints of larger patterns, most immediately schemas, but also larger related network nodes.

33

(iii) With respect to changes that have been studied in

terms of Gzn and Lxn, evidence of the continuum

between contentful and procedural poles of the

constructional gradient shows that these are not

orthogonal developments.

(iv) GCxzn and LCxzn are not equivalent to either Gzn or

Lxn. Rather, certain aspects of these types of changes

can be incorporated within a more comprehensive view

of language change as sign change in the larger

context of shifts in schematicity, productivity, and

compositionality.

34

(v) Expansion and reduction are intertwined.

(vi) Therefore, directionality of change is more nuanced

than has often been thought.

35

Thank you for your attention!

36

Data sources

DOEC = Dictionary of Old English Corpus. Original release 1981 compiled by Angus Cameron, Ashley Crandell Amos, Sharon Butler & Antonette diPaolo Healey. Release 2009 compiled by Antonette diPaolo Healey, Joan Holland, Ian McDougall & David McDougall, with Xin Xiang. University of Toronto.http://www.helsinki.fi/varieng/CoRD/corpora/DOEC/index.html.

OED = Oxford English Dictionary. http://www.oed.com/.

References

Algeo, John. 1971. The voguish uses of non-. American Speech 46: 87-105.

Booij, Geert. 2010. Construction Morphology. Oxford: OUP.

Bybee, Joan L. 2010. Language, Usage and Cognition. Cambridge: CUP.

Bybee, Joan & David Eddington. 2006. A usage-based approach to Spanish verbs of 'becoming'.  Language 82: 323-355.

37

Brinton, Laurel J. & Elizabeth Closs Traugott. 2005. Lexicalization and Language Change. Cambridge: CUP.

Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: OUP.

Dalton-Puffer, Christiane. 1996. The French Influence on Middle English Morphology: A Corpus-based Study of Derivation. Berlin: Mouton de Gruyter.

Dietz, Klaus. 2007. Denominale Abstraktbildungen des Altengischen: Die Wortbildung der Abstrakta auf –dom, -had,

-lac, -ræden, -sceaft, -stæf und –wist und ihrer Entsprechungen im Althochdeutschen und im Altnordischen. In Hans Fix, ed., Beiträge zur Morphologie. Germanisch, Baltisch, Ostseefinnisch, 69-172. Odense: U Press of Southern Denmark.

Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in Language. Oxford: OUP.

38

Gonzálvez-Garcia, Francisco. 2011. What snowclones reveal about actual language use in Spanish: A constructionalist view. Paper presented at SLE, Logrono, Sept. 8th-11th.

Haselow, Alexander. 2011. Typological Changes in the Lexicon: Analytic Tendencies in English Noun Formation. Berlin: De Gruyter Mouton.

Horn, Laurence R. 2005. An un- paper for the unsyntactician. In Salikoko S. Mufwene, Elaine J. Francis & Rebecca S. Wheeler, eds., Polymorphous Linguistics: Jim McCawley’s Legacy, 329-365. MIT Press. 

Kay, Paul. Forthc. The limits of construction grammar. In Thomas Hoffmann & Graeme Trousdale, eds., The Oxford Handbook of Construction Grammar. New York: OUP. (Originally circulated as Patterns of coining, 2002). 

Lehmann, Christian. 2002. New reflections on grammaticalization and lexicalization. In Ilse Wischer & Gabriele Diewald, eds., New Reflections in Grammaticalization, 1-18. Amsterdam: Benjamins.

39

Mattiello, Elisa. Forthc. Extra-grammatical Morphology in English -Abbreviations, Blends, Reduplicatives and Related Phenomena. Berlin: De Gruyter Mouton.

Pullum, Geoffrey. 2003. Phrases for lazy writers in kit form. Language Log Oct. 27th. 

----. 2004. Snowclones: lexicographical dating to the second. Language Log January 16th.

Ramat, Paolo. 1992. Thoughts on degrammaticalization. Linguistics 30: 549-560.

----. 2001. Degrammaticalization or transcategorization? In Chris Schaner-Wolles, John Rennison & Friedrich Neubarth, eds., Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler Presented on the Occasion of his 60th Birthday, 393-401. Torino: Rosenbach and Sellier.

Traugott, Elizabeth Closs & Graeme Trousdale. Forthc. Constructionalization and Constructional Changes. Oxford: OUP.

40

Trousdale, Graeme. 2008. A constructional approach to lexicalization processes in the history of English: Evidence from possessive constructions, Word Structure 1: 156-177.

----. 2010. Issues in constructional approaches to grammaticalization in English. In Katerina Stathi, Elke Gehweiler & Ekkehard König, eds., Grammaticalization: Current Views and Issues, 51-72. Amsterdam: Benjamins.

Zwicky, Arnold. 2006. Snowclone mountain? Language Log March 13th. http://itre.cis.upenn.edu/~myl/languagelog/archives/002924.html