Understanding Metonymy - OSF

UNDERSTANDING METONYMY

Deirdre Wilson & Ingrid FalkumUCL Linguistics and IFIKK, Oslo

Standard definition of metonymy

• Metonymy is a figure of speech involving substitution of the name of an associated attribute or adjunct for that of the thing meant

e.g. suit for business executive or the crown for the monarchy

• The type of associative relation involved is not resemblance or antonymy but contiguity

Standard metonymic relations

Any contiguity-based associative relation:• Producer for product (Plato, Picasso)• Disease for patient (measles, appendicitis)• Meal for orderer (ham sandwich, pizza)• Place for people (Buckingham Palace)• Instrument for player (violin, saxophone) • Place for event (Watergate, Vietnam)

Standard examples of metonymy

• The appendicitis in bed 3 is threatening to write to the newspapers

• Can you take the peperoni pizza his glass of wine?

• Buckingham Palace is refusing to comment

Features of metonymy

• The word ‘metonymy’ (vs ‘metaphor’) is rarely used in everyday conversation

• Few people can supply a definition• But the ability to exploit metonymic relations

is present in children from a very early age• It is impaired in conditions such as autism,

schizophrenia and Williams Syndrome• A goal of pragmatics is to describe this ability

and explain how metonymy is understood

Standard ‘transfer of meaning’ account (Nunberg, Recanati)

Metonymy involves rules or conventionslicensing the following ‘transfers of meaning’:• Reinterpret appendicitis as denoting the

category of PEOPLE WITH APPENDICITIS

• Reinterpret peperoni pizza as denoting the category of ORDERERS OF PEPERONI PIZZA

• Reinterpret Buckingham Palace as denoting the category of OFFICIALS AT BUCKINGHAM PALACE

Problems with the ‘transfer of meaning’ account of metonymy

• It is a largely code-based account, which might describe the outcome of repeated metonymic uses, but does not explain how they arise or how they are understood when first used

• For any proposed set of metonymic patterns, it is always possible to find new acceptable metonymic uses which do not fit the pattern, so a purely code-based ‘transfer of meaning’ account won’t work

Standard ‘transfer of meaning’ accounts of other figurative utterancesMetaphor (based on resemblance relations)Reinterpret ‘John is a toad’ as meaning (or implicating) that John is like a toadIrony (based on antonymy relations)Reinterpret ‘That was a great party’ as meaning (or implicating) that the party was awful

Problems with ‘transfer of meaning’ accounts of figurative utterances

• They don’t explain why figurative utterances arise naturally and spontaneously across cultures

• They make many figurative utterances sound entirely irrational:

From an internet forum on sarcasm for people with Autism Spectrum Disorders“Sarcasm is easy once you understand the ‘rule’. Basically, think of what you mean, find words that express the opposite, and speak them with a bit of edginess in your voice.”

Alternative programme on figurative utterances

Aims to show that ‘figurative’ utterances arisenaturally and spontaneously, involving no special interpretive mechanisms or principlesnot required for ordinary, literal utterances, e.g.:• metaphor – a type of loose use of language• irony – (a) a type of echoic use of language

– (b) a type of pretence

Explaining metaphor

Metaphor may be seen as a variety of lexical modulation, where the concept communicated by use of a word is broader than, but shares implications with, the encoded concept:• John is a toad• encoded concept: TOAD

• communicated concept: TOAD* • shared implications: repellent, ugly, noxious

Explaining irony/sarcasm

Irony/sarcasm may be seen as a variety of echoing (or pretence) where the speaker ‘echoes’ (or pretends to assert) a thought she attributes to someone else, while expressing her own mocking/dismissive attitude that thought:Peter: That was a great partyMary (enthusiastically): Great!Mary (cautiously): Great?Mary (scornfully): Great!

Question

• These alternative explanatory accounts of metaphor and irony are intended to replacestandard (e.g. classical or Gricean) accounts of ‘figurative’ utterances based on code-like ‘transfer of meaning’ rules – and have been largely successful in doing so

• No such alternative exists for metonymy. How is metonymy to be explained?

Aim of this talk

• To argue that metonymic uses arise through natural, spontaneous processes of neologism, or word coinage;

• that they show important continuities with other neologisms which are clearly not figurative;

• that they are produced for similar reasons, and understood in the same way, as other neologisms/word coinages

How do neologisms arise?

“One remarkable aspect of our capacity to use language is our ability to create and understandexpressions we have never heard before.” (Clark & Clark 1979, ‘When nouns surface as verbs’, Language 55: 767-811)

Two types of neologism

• Opaque: the form of the new word gives no clue to the intended meaning (e.g. Exxon, Persil), which must be explicitly stipulated or else inferred by the hearer using contextual information and pragmatic principles alone

• Transparent (or ‘motivated’): the linguistic properties of the new word provide a clue to the intended meaning (e.g. blackbird, chillax, mouse) without necessarily determining it completely

Our claim

Metonyms are transparent neologisms, and show important continuities with other transparent neologisms that are not normally regarded as figurative:• Denominal verbs (e.g. to hole the ball)• Noun-noun compounds (e.g. door mat)• Nicknames (e.g. Alex ‘Hurricane’ Higgins)

Denominal verbs

Clark & Clark (1979) analysed 1,300 attested novel and familiar denominal verbs (i.e. verbs created from existing nouns) and concluded:• “People readily create and understand

denominal verbs they have never heard before, as in to porch a newspaper and to Houdini one's way out of a closet”.

Denominal verbs

Some of Clark & Clark’s ‘novel’ verbs (in 1979) are now established lexical items:• wait-list the passenger; hole the ball; • footnote her colleagues, site the power plant;• turf the yard, blot the ink; • minute the amendment; meter the water;• spread-eagle his opponent(cf. to empty chair, to no platform)

Denominal verbsClark & Clark roughly categorise the possible relations between existing noun and new verb (e.g. the noun may provide a LOCATION, INSTRUMENT, OBJECT, DURATION, AGENT, EXPERIENCER … for the action of the verb), but comment:• “These categories don't really do justice to

denominal verbs. Many examples don't fit neatly into these categories, and others have the characteristics of more than one category at a time.” (ibid: 781)

Our hypothesis

Denominal verbs arise as transparent neologisms, whose links to existing words provide clues to the speaker’s meaning. No ‘transfer of meaning’ rule is needed: a hearer will treat the new word as an ostensive stimulus, assume that the clues were intentionally provided, and infer the intended meaning using pragmatic mechanisms such as the relevance-guided comprehension heuristic, based on the clues provided, plus contextual information.

Relevance-guided comprehension heuristic

Follow a path of least effort in looking for implications (and other cognitive effects): • Test interpretive hypotheses in order of salience;

• Stop when you have enough implications to satisfy the expectations of relevance raised by the utterance.

Noun-Noun compounds (Bauer 1979, 2001)

“The compound noun structure is extremely varied in the types of meaning relations it can indicate. It can be used to indicate what someone does (language teacher), what something is for (waste-paper basket, grindstone), what the qualities of something are (whiteboard), how something works (immersion heater), when something happens (night frost), where something is (doormat), what something is made of (woodpile), and so on.” (Carter & McCarthy 2006, Cambridge Grammar of English)

Noun-Noun compounds

As with denominal verbs, no code-like analysis of these relations seems entirely adequate:• Bauer (1979) argues that no finite categorisation

will do: new relations can always be added.• Early experiments (e.g. Gleitman & Gleitman

1970, Downing 1977) show that novel N-N compounds are easily understood

• apple juice chair, black house-bird, house-bird black, appendicitis chap, pizza guy, BP lot …

Our hypothesisNoun-noun compounds arise as transparentneologisms: a hearer will treat the new word as an ostensive stimulus and infer the intended meaning using a relevance-guided comprehension heuristic, based on clues provided by the meanings of the constituent words, plus contextual information and expectations of relevance(Bezuidenhout, 2019. Noun-noun compounds from the perspective of relevance theory, in Scott, Clark & Carston, Relevance, Pragmatics & Interpretation)

Nicknames

• Nicknames are typically transparent word coinages

• Here again, although there may be regularities governing the creation of certain classes of nickname, there is no limit to the possible relations that may be exploited in producing and understanding novel nicknames, which rules out the possibility of a ‘transfer of meaning’ account

Nicknames“Harry was the eighth person I'd met with the name Jones. Everyone seemed to be called Jones, Williams, Davies, or Evans. To prevent confusion, the Welsh use nicknames, and nowhere is their teasing wit and love of language more apparent. Most nicknames derive from a person's occupation, like the builder I heard of called Will Five Bricks or the baker Dai Bread in Dylan Thomas's Under Milk Wood. My favorites were the two Evansesfrom a village in Carmarthenshire. One was an undertaker, the other a travel agent. The travel agent was known as Evans There and Back, the undertaker as Evans One Way.”

Nicknames

“Over the years I have come across many nicknames and have had a few myself. With so many Jones`s, a baker would be known as Jones the Bread and the butcher would be known as Jones the Meat. I recall a baker being known as Dai the Crust whilst somebody posh living nearby was Dai Upper Crust.” http://www.forumwales.com/fwforum/viewtopic.php?f=37&t=5736

Our claim about metonymy• Typical cases of metonymic use (e.g. appendicitis

for APPENDICITIS PATIENT) originate as transparent neologisms – ‘denominal nouns’, as it were –which provide clues to their intended meaning and involve the same interpretive principles or mechanisms used in interpreting other transparent neologisms.

• That is, their meanings are not (initially) decoded, but pragmatically inferred (though the new word may become part of the language in due course).

Evidence from acquisition

Pre-linguistic infants spontaneously produce and understand ostensive acts based on ‘metonymic’ associative relations of contiguity:• smacking lips (for FOOD)• clapping hands (for BASEBALL GAME)• vroom vroom (for CAR)(Acredolo & Goodwyn 1988, ‘Symbolic gesturing in normal infants’. Child Development 59)

Early metonymic reference/naming (Falkum, unpublished diary data)

V (1;4) is at the dinner table with his mother and grandfather. At this stage, V has only a handful of words in his expressive vocabulary, and they don’t include the form for ‘grandfather’. Suddenly, V’s grandfather sneezes loudly. V is very amused by this, and imitates the sneeze, “Pff”, laughing, several times during the rest of the meal. A couple of days later, on seeing a photo of his grandfather, V points to the photo while uttering “Pff”. Soon the form “Pff” is consistently used in making reference to his grandfather, as well as in vocative uses. (Today, at 2;4, the form “Poff” has replaced the early “Pff” in V’s vocabulary. “Poff” has become the name of V’s grandfather, and V consistently uses it even though he now also has the conventional term for ‘grandfather’).

Early ‘metonymic’ referential uses

• nose for handkerchief

• Nonno for grandfather’s apartment building

• cat for cat’s usual location on top of TV

• nap for cot/crib blanket

• Lisa for swing Lisa used earlier

(Rescorla, L. 1980. Overextension in early language development. Jn’l of Child Language)

Early use of transparent neologisms

Children commonly produce and understand novel denominal verbs from around age two: • ‘to gun’, ‘to noise’, ‘fire the candle’ (Clark 1993)• ‘Make the boy seatbelt himself’ (Bushnell &

Maratsos 1984 Child Development 55)They also commonly use novel N-N compounds based on ‘metonymic’ associations:• clown boy for BOY WHO IS A CLOWN, • daddy seed for SEED THAT IS A DADDY

Experimental evidence on early metonymy production

(Falkum, Recasens & Clark 2016)

Children of 3-5 were shown two games with

easily distinguishable components (e.g. stickers).

The games were then put away and the children

were asked which game they wanted to play.

The children used three referential strategies to

pick out the chosen game

• metonyms: the stickers• noun-noun compounds: the stickers game• full descriptions: the game with the stickers

Results

• All the children used significantly more metonyms than compound nouns or full descriptions, suggesting that metonymy offers young children a useful strategy for filling vocabulary gaps

(Falkum, Recasens & Clark, 2016, ‘The moustache sits down first’: On the acquisition of metonymy. Journal of Child Language 43)

Experimental evidence on early metonymy comprehension

Falkum, Recasens & Clark (2016) tested 3-5 year olds using picture stories like the following: • This story is about these two girls. They are

standing outside talking before going home to work. After they have been talking for a while, the helmet gets on her bike and goes home.

The children were then shown three pictures and asked to choose the one matching the story

Experimental evidence on early metonymy comprehension

• Picture corresponding to literal use:a bike with a helmet on the saddle

• Picture corresponding to metonymic use:a girl wearing a helmet riding a bike

• Picture of the other girl mentioned in the story

Results: Most 3-year-olds correctly distinguished literal from metonymic uses of helmet

Metonymy and stylistic effects

• Metonymy is often seen as having a ‘witty’, ‘deprecating’ or ‘dehumanising’ effect (e.g. referring to someone as the appendicitis in bed 3 is somehow dehumanising)

• How might this arise?

Illustrations: PG Wodehouse

“The conversation in the bar parlour of the Angler’s Rest, which always tends to get deepishtowards closing time, had turned to the subject of the Modern Girl; and a Gin-and-Ginger-Alesitting in the corner by the window remarked that it was strange how types die out.” (P.G. Wodehouse: Mr Mulliner Speaking)

Illustrations: PG Wodehouse

“It was the hour of the morning snifter, and a little group of Eggs and Beans and Crumpets had assembled in the smoking room of the Drones club to do a bit of inhaling. There had been a party of sorts overnight, and the general disposition of the company was towards a restful and somewhat glassy-eyed silence. This was broken at length by one of the Crumpets.

‘Old Freddie’s back,’ he observed.Some moments elapsed before any of those present felt equal to commenting on this statement. Then a Beanspoke.

‘Freddie Who?’” (Young Men in Spats)

Illustration: David Foster Wallace“It so happens that just as work on Everything and Moregot underway, a certain book came out, a pop bio of Cantor by a certain author whom I won’t name except to say that his initials are the same as those of a well-known commercial airline. For a certain publisher whose own name sounds like an autistic person’s description of a room. This unnamed book had two theses … [12 lines of increasingly negative description]…The origin, motives, and contexts of Cantor’s actual achievements got little serious treatment in this unnamed book, basically I think because airline-initials and/or autistic room descriptionfelt the math would be too dull for a mainstream audience.” (The Last Interview and other Conversations)

Illustration: David Foster Wallace

• Aczel, Amir D. (2000), The Mystery of the Aleph: Mathematics, the Kabbala, and the Search for Infinity, New York: Four Walls Eight Windows Publishing.

• Described as “a popular treatment of infinity, in which Cantor is frequently mentioned”.

General feature of metonymy

Metonymy is typically used in sorting, where the domain is given (or easily inferable) and the aim is to pick out particular individuals from that domain• doctors sorting patients by their diseases• waiters sorting diners by their orders• reporters sorting stories by their sources• bartenders sorting customers by their drinks…

When is a metonymic use optimally relevant?

• When there is only a single property of the referents that matters (e.g. for waiters, bartenders, doctors, sorting people by their food, drink, disease etc.)

• When the speaker wants to implicate that there is only a single property of the referent that matters, since e.g. sorting story characters by their drinks or favourite breakfast food suggests a certain perspective on things

Concluding remarks

We have argued – as does everyone else in the field – that associative relations (in cognitive terms, spreading activation patterns) play a central role in metonymy comprehension. The challenge was to show how these could be dealt with in a properly inferential approach, without relying on code-like ‘transfer of meaning’ rules, which have invariably proved inadequate

Concluding remarks

Our solution was to treat metonymic uses not as old words with new meanings, as on standard ‘transfer of meaning’ accounts, but astransparent word coinages, which can be produced and understood ‘on the fly’, and can legitimately exploit associative relations.

Concluding remarks

On this account, there is no need for special rules or conventions for producing and interpreting metonymy: what matters are spreading activation patterns, which affect the salience of potential coinages and interpretations, and a relevance-guidedcomprehension heuristic (or other suitable pragmatic mechanism), which filters the results.

Possible objection• Does metonymic use really create new words?“For an innovative use to count as a new coinage proper, it must involve the creation of a new linguistic form as well as a new meaning. Noun-noun compounds meet this criterion because they are overtly formally novel. Even denominalverbs and deverbal nouns, although identical to the ‘parent’ noun/verb in bare form, nevertheless appear with verbal/nominal inflections, such that the surface form of the new word shows evidence of the underlying category change. Yet in referential metonymy, the metonymically-used expression does not appear to undergo any morpho-syntactic operations, whether overt (e.g. compounding, affixing) or covert (e.g. category change).” (Bowerman, 2018, ‘What’s really going on with the ham sandwich?’ Int’l Rev. Pragmatics

Evidence A: denominal common nouns derived from proper names

• Inventor for object: macadam, diesel, ohm, watt, ampere, Biles, …[It vs he/she]

• Wearer for garment: mackintosh, raglan, wellington … [It vs he/she]

The ‘denominal’ nouns are syntactically and semantically distinct from their parent nouns [inanimate vs human], and there is no overlap in their encyclopaedic entries or ‘files’.

Evidence B: ‘denominal’ human nouns derived from common nouns

• “A perfumer is an expert on creating perfume compositions, sometimes referred to affectionately as a Nose (French, le nez) due to their fine sense of smell and skill in producing olfactory compositions.”

• “For nearly three hundred years, Lloyd's of London insurance policies were backed by wealthy British investors, who came to be known as "Names" because, in the early days, their signatures were written on the face of each Lloyd's policy.”

The denominal noun is human, the parent noun inanimate [he/she/who vs. it/which], and there is no overlap in encyclopaedic entries or ‘files’

Further evidence

“Meaning transfer exhibits a number of distinctive linguistic properties. The thread running through these is that different types of linguistic behaviour—many of which are often taken to be syntactic and/or semantic—track transferred meaning rather than non-transferred meaning. As will become clear, we can use these properties to distinguish meaning transfer from related phenomena as well as identify its locus.” (Liebesman & Magidor, 2019. Meaning transfer revisited. Philosophical Perspectives)

Further evidence of syntactic/semantic change accompanying metonymic use• The tie and shorts is leaving (singular vs plural)• The ham sandwich who was wearing green

left without paying (human vs inanimate]• [Hebrew] The pastaF leftM without paying

(masculine. vs. feminine for male customer)• Spanish and Italian ‘fruit for tree’ metonymy

(mela/melo, ‘apple/tree’) also shows a formal change accompanying metonymic use

Conclusion• These examples provide even stronger evidence

for our view that novel metonymic uses create new words, with their own linguistic properties, than it does for the standard ‘transfer of meaning’ account.

• We would then have genuinely distinct explanatory accounts of metaphor, irony and metonymy, which are all seen as falling together on standard ‘transfer of meaning’ accounts.

THANK YOU

ReferencesBezuidenhout, A. 2019. N-N compounds from the perspective of relevance theory. In Scott, Clark & Carston (eds) Relevance, Pragmatics and Interpretation. CUP.Bowerman, J. 2018. What’s really going on with the ham sandwich? International Review of Pragmatics: 1-33Clark, H. & Clark, E. 1979. When nouns surface as verbs.Language 55: 767-811Falkum, I., Recasens, M. & Clark, E. 2016, ‘The moustache sits down first’: On the acquisition of metonymy. Journal of Child Language 43Liebesman, D. & Magidor, O. 2019. Meaning transfer revisited. Philosophical Perspectives (online)