< Back to previous page
Publication
Contextual cues for particle placement: Multiplicity, motivation, modeling
Book Contribution - Chapter
This paper discusses some contextual factors that influence constituent ordering in the English transitive verb-particle construction. The particle in this construction can generally occur on either side of the object NP--see (1a-b)--but it has to follow an unstressed pronominal object and precede an NP with sufficient complexity--see (2a-b).
(1a) Pete {put down the knife / put the knife down}.
(1b) Kate {googled up the data / googled the data up}.
(2a) They had to {*helicopter out us / helicopter us out}.
(2b) I {made up things about people who don't exist / *made things about people who don't exist up}.
These well-known restrictions on particle placement have been considered as a grammatical reflection of the more basic pragmatic tendency to put discourse-old information before discourse-new information: pronouns typically refer back to contextually identifiable entities and therefore precede the particle, while heavily modified nouns typically refer to entities that are newly introduced into the discourse, and therefore follow the particle (cf. Gries, 2003: 18-20 for a brief overview of the literature on the effect of discourse-newness on particle placement).
While previous mention of the referent of the object NP certainly influences particle placement, other contextual factors have to be taken into account as well.
One factor that can be shown to be relevant is focality, which is to be distinguished from discourse-newness, and which should be viewed as an attribute that can receive one of multiple values (e.g. 'focus on the entire VP', 'focus on the object NP', 'focus on the particle', 'focus on the verb-particle combination but not on the object NP', etc.). Depending on the value, which represents a particular contextual scenario, one or both orderings are felicitous (cf. Dehé, 2002 for the influence of focality on particle placement, but she considers fewer scenarios and matches each of them with one ordering only, something which cannot be supported empirically). To illustrate with only two values (focused constituents appear within square brackets, accented words appear in capitals):
(3a) Focus on the object NP (e.g., as a possible reply to the question, What did Durban turn off, the camera or the microphone?)
Durban {turned off [the CAMERA] / turned [the CAMERA] off}.
(3b) Focus on the particle (e.g., as a possible reply to the question, What did Durban do with the camera, turn it off or turn it on?)
Durban {??turned [OFF] the camera / turned the camera [OFF]}.
Among further contextual factors that need to be accommodated are:
- syntactic priming: an ordering tends to be triggered by its previous use (cf. Gries, s.d.)
- accessibility (perhaps to be treated in terms of Fillmore's Frame Semantics): an NP referring to an entity that is easily evoked in the context of a certain activity may occur more readily in the discontinuous order than an NP referring to an entity that is less obviously associated with that activity (cf. Bolinger, 1971: 55-57):
(4a) Today I {cleaned up my room / cleaned my room up}.
(4b) Today I {cleaned up my website / ?cleaned my website up}.
Some data gathered from web search experiments tentatively corroborate the difference in status between expected and less expected object NPs. (For example, with "look up", the expected NP "a word" has a one-in-four chance of preceding the particle, while the less expected NP "a name" has less than a one-in-ten chance of preceding it. The possible but unexpected NP "a quote" has virtually a zero chance of preceding the particle.)
- (other) 'encyclopedic' knowledge: the discontinuous order is used when the verb is stative (e.g. She {*left up the blinds / left the blinds up}, cf. Fraser, 1976: 11), but only real-world experience with objects, the laws of physics, etc. can tell us whether a stative verb like "keep" is involved in a stative or a dynamic situation in a given context:
(5a) He {??kept on his coat / kept his coat on}. (cp. He {put on his coat / put his coat on})
(5b) This rope will {keep down the hood / keep the hood down} while you drive.
In (5a), the situation referred to is stative, since no particular effort is needed to keep a coat on. By contrast, in (5b), the seemingly stative situation requires a sustained application of a counterforce exerted by the rope: the function of the rope is to prevent the hood from being blown up. This counteraction makes the situation dynamic, despite the resulting steadiness. In the following authentic example, the tendency of non-elastic socks to slip down is also neutralized by a counterforce (thus making the situation dynamic), which again accounts for the possibility of the particle to be put before the object NP:
(6) I suggested that she find a pair of those old fashioned garters that men wore to keep up their socks, in the pre-elastic stone ages. (www)
This paper, in short, adduces evidence that the choice of one word order pattern over another is prompted not just by the assessment of the relative discourse-newness that the referents of the involved words have in a given context. At least in the case of particle placement, the selection of a particular ordering is cued by contextual information that goes far beyond what could be handled by an automatic processor and that can only be gathered via a full linguistic and extra-linguistic understanding of the situation referred to.
References
Bolinger, Dwight. 1971. The Phrasal Verb in English. Cambridge, Mass.: Harvard University Press.
Dehé, Nicole. 2002. Particle Verbs in English: Syntax, Information Structure and Intonation. Amsterdam and Philadelphia: John Benjamins.
Fraser, Bruce. 1976. The Verb-Particle Combination in English. New York: Academic Press.
Gries, Stefan Thomas. s.d. Syntactic priming: a corpus-based approach. Ms. University of Southern Denmark.
Gries, Stefan Thomas. 2003. Multifactorial Analysis in Corpus Linguistics: A Study of Particle Placement. New York & London: Continuum.
(1a) Pete {put down the knife / put the knife down}.
(1b) Kate {googled up the data / googled the data up}.
(2a) They had to {*helicopter out us / helicopter us out}.
(2b) I {made up things about people who don't exist / *made things about people who don't exist up}.
These well-known restrictions on particle placement have been considered as a grammatical reflection of the more basic pragmatic tendency to put discourse-old information before discourse-new information: pronouns typically refer back to contextually identifiable entities and therefore precede the particle, while heavily modified nouns typically refer to entities that are newly introduced into the discourse, and therefore follow the particle (cf. Gries, 2003: 18-20 for a brief overview of the literature on the effect of discourse-newness on particle placement).
While previous mention of the referent of the object NP certainly influences particle placement, other contextual factors have to be taken into account as well.
One factor that can be shown to be relevant is focality, which is to be distinguished from discourse-newness, and which should be viewed as an attribute that can receive one of multiple values (e.g. 'focus on the entire VP', 'focus on the object NP', 'focus on the particle', 'focus on the verb-particle combination but not on the object NP', etc.). Depending on the value, which represents a particular contextual scenario, one or both orderings are felicitous (cf. Dehé, 2002 for the influence of focality on particle placement, but she considers fewer scenarios and matches each of them with one ordering only, something which cannot be supported empirically). To illustrate with only two values (focused constituents appear within square brackets, accented words appear in capitals):
(3a) Focus on the object NP (e.g., as a possible reply to the question, What did Durban turn off, the camera or the microphone?)
Durban {turned off [the CAMERA] / turned [the CAMERA] off}.
(3b) Focus on the particle (e.g., as a possible reply to the question, What did Durban do with the camera, turn it off or turn it on?)
Durban {??turned [OFF] the camera / turned the camera [OFF]}.
Among further contextual factors that need to be accommodated are:
- syntactic priming: an ordering tends to be triggered by its previous use (cf. Gries, s.d.)
- accessibility (perhaps to be treated in terms of Fillmore's Frame Semantics): an NP referring to an entity that is easily evoked in the context of a certain activity may occur more readily in the discontinuous order than an NP referring to an entity that is less obviously associated with that activity (cf. Bolinger, 1971: 55-57):
(4a) Today I {cleaned up my room / cleaned my room up}.
(4b) Today I {cleaned up my website / ?cleaned my website up}.
Some data gathered from web search experiments tentatively corroborate the difference in status between expected and less expected object NPs. (For example, with "look up", the expected NP "a word" has a one-in-four chance of preceding the particle, while the less expected NP "a name" has less than a one-in-ten chance of preceding it. The possible but unexpected NP "a quote" has virtually a zero chance of preceding the particle.)
- (other) 'encyclopedic' knowledge: the discontinuous order is used when the verb is stative (e.g. She {*left up the blinds / left the blinds up}, cf. Fraser, 1976: 11), but only real-world experience with objects, the laws of physics, etc. can tell us whether a stative verb like "keep" is involved in a stative or a dynamic situation in a given context:
(5a) He {??kept on his coat / kept his coat on}. (cp. He {put on his coat / put his coat on})
(5b) This rope will {keep down the hood / keep the hood down} while you drive.
In (5a), the situation referred to is stative, since no particular effort is needed to keep a coat on. By contrast, in (5b), the seemingly stative situation requires a sustained application of a counterforce exerted by the rope: the function of the rope is to prevent the hood from being blown up. This counteraction makes the situation dynamic, despite the resulting steadiness. In the following authentic example, the tendency of non-elastic socks to slip down is also neutralized by a counterforce (thus making the situation dynamic), which again accounts for the possibility of the particle to be put before the object NP:
(6) I suggested that she find a pair of those old fashioned garters that men wore to keep up their socks, in the pre-elastic stone ages. (www)
This paper, in short, adduces evidence that the choice of one word order pattern over another is prompted not just by the assessment of the relative discourse-newness that the referents of the involved words have in a given context. At least in the case of particle placement, the selection of a particular ordering is cued by contextual information that goes far beyond what could be handled by an automatic processor and that can only be gathered via a full linguistic and extra-linguistic understanding of the situation referred to.
References
Bolinger, Dwight. 1971. The Phrasal Verb in English. Cambridge, Mass.: Harvard University Press.
Dehé, Nicole. 2002. Particle Verbs in English: Syntax, Information Structure and Intonation. Amsterdam and Philadelphia: John Benjamins.
Fraser, Bruce. 1976. The Verb-Particle Combination in English. New York: Academic Press.
Gries, Stefan Thomas. s.d. Syntactic priming: a corpus-based approach. Ms. University of Southern Denmark.
Gries, Stefan Thomas. 2003. Multifactorial Analysis in Corpus Linguistics: A Study of Particle Placement. New York & London: Continuum.
Book: Contexts and Constructions
Series: Constructional Approaches to Language
Pages: 145-192
Number of pages: 247
ISBN:978-90-272-0431-8
Publication year:2009
Keywords:phrasal verbs (verb-particle combinations), particle placement, weight, discourse-familiarity, focus structure, syntactic priming, allostruction, register variation, region variation