LINGUIST List 7.464

Tue Mar 26 1996

Sum: "MUST"

Editor for this issue: Ljuba Veselinova <lveselinemunix.emich.edu>

Directory

1. Edith A Moravcsik, "MUST"

Forwarded message:
>From edith Tue Mar 26 14:58:30 1996
From: edith (Edith A Moravcsik)
Message-Id: <199603262058.OAA12014alpha1.csd.uwm.edu>
Subject: no subject (file transmission)
To: edith
Date: Tue, 26 Mar 1996 14:58:29 -0600 (CST)
X-Mailer: ELM [version 2.4 PL24alpha3]

 On February 6, l996, I posted the following query on LINGUIST:

 *** Of the three sentences below, why is 1. ungrammatical but 2.
 and 3. grammatical?

 l. *The books sell well (in order) to raise money.
 2. The books must sell well (in order) to raise money.
 3. It is necessary that the books sell well (in order) to
 raise money. ***

 Sentence 1. was taken from an exercise in Liliane Haegeman's
 textbook _Introduction to government and binding theory_ (1994
 (second edition), Blackwell's, page 79). The fact that 2. was
 grammatical in spite of its minimal difference from 1. had been
 pointed out to me by William Bellin.

 Many thanks to the following 32 persons who responded:

 David Baxter
 Bill Bennett
 Ginny Brennan
 Wayles Brown
 Annabel Cormack
 Peter Daniels
 Suzette Haden Elgin
 Joseph Foster
 Frank Gladney
 Ted Harding
 Michael Hegarty
 Richard Ingham
 Roumyana Izvorski
 Graham Katz
 Marion Kee
 Luuk Lagerwerf
 Donna Lillian
 Waruno Mahdi
 Mark Mandel
 Kate McCreight
 Michael Niv
 Ellen Prince
 David Powers
 Diego Quesada
 J. Reinhardt
 Larry Rosenwald
 Marilyn Silva
 Wilbert Spooren (through Luuk Lagerwerf)
 Jack Wiedrick
 Debbie Ziegeler
 Magdalena Zoeppritz
 and a linguist who wished to remain anonymous.

 Below is a brief summary of the responses, structured as
 follows:
 l. Well-formedness judgments
 2. Proposed explanations
 3. Additional data with comments
 4. Literature

 l. WELL-FORMEDNESS JUDGMENTS

 The well-formedness judgments that I submitted were as follows:
 (numbers refer to sentence numbers):
 *1. 2. 3.
 The following comments and alternative judgments were reported:
 a/ 1. 2. 3. (by a speaker of Ozark English who is also
 fluent in American Midwestern English; however,
 for another Ozark speaker, 1. was ill-formed)
 b/ ?1. 2. 3. (this speaker said she could get a reading for
 1. where an implicit animate agent was present)
 c/ *1. 2. 3. (if 2. contains no "in order") but
 *1. *2. 3. (if 2. does contain "in order")
 d/ *1. 2. 3. (if 2. has the epistemic "must") but
 *1. *2. 3. (if 2. has the non-epistemic "must")
 e/ *1. ?2. *3. (2. is better if "for us" and "enough (money)"
 are added)
 f/ *1. *2. *3. (by a speaker of Canadian (Southern Ontario)
 English and by a speaker of American English who
 said several others also agreed with him)

 2. PROPOSED EXPLANATIONS

 The general consensus was that 1. was ill-formed not for
 syntactic but for semantic reasons. Two alternative semantic
 conflicts were proposed as the cause:

 a/ One of the conditions under which an infinitival purpose
 clause can occur in a well-formed sentence is if the main
 clause contains a goal-directed action verb and, as a
 consequence, an intentional agent which is then understood as
 the subject of the purposive infinitive. In 1., this condition
 is not met: the main clause is about a non-goal-directed action
 ("sell" in its middle use) and an unintentional agent ("the
 books").

 b/ Another possible function of an infinitival purpose
 clause is to restrict a modal operator, such as "must", saying
 something like 'If the books are to raise money, they must sell
 well.' But in 1. there is no modal operator for the purpose
 clause to restrict.

 2. in turn is well-formed (even though it differs from 1. only
 by the added "must") for the following alternative reasons:

 A/ The "must" of 2. implies intention: it implies an
 assessment of the situation on behalf of the speaker and it
 presupposes that the speaker (or the hearer) is responsible for
 the selling of the books. Putting it differently; "must"
 is associated with an implicit external argument; by
 introducing the perspective of the speaker it suggests the
 meaning 'I find it very important that these books sell well if
 they are to raise money.' Thus, with "must", the sentence
 fulfils the condition mentioned in a/ above.

 B/ 2. is well-formed because there is a modal which the
 purpose clause can serve to restrict. 2. therefore fulfils
 the second condition, mentioned in b/ above.

 C/ The "must"-containing main clause in 2. expresses a
 precondition for the event referred to in the purpose clause.
 Thus, 2. means something like 'If the books sell well, money
 will be raised.'.

 3. ADDITIONAL DATA WITH COMMENTS

 4. *This car drives smoothly (in order) to increase
 sales/prevent accidents/ prevent mechanical damage...

 Ill-formed for the same reason as 1.

 5. a/ ?Max wounds easily (in order) to come across as a
 sensitive guy.
 b/ *Max wounds easily (in order) to help him make stronger.

 5.a/ is OK if Max "is the orchestrator of a series of public
 events where he is seen to be easily wounded" since in this
 case there is an intentional agent and goal-directed
 activity. 5.b/ is ill-formed because the agent of the
 purpose clause is different from Max.

 6. a/ *John must be a murderer (in order) to explain these
 facts.
 b/ John must be a murderer (in order) to ensure that Brenda
 gets away. /said in a context where a plan is being
 hatched to kill somebody/
 c/ John must register before tomorrow (in order) to take
 classes this semester.
 d/ *Rock conducts heat (in order) to maintain its chemical
 structure.
 e/ John runs fast (in order) to increase his heart rate.

 These sentences illustrate that regularly occurring events
 (as in 1.) as well as deductions (6.a/) and natural laws
 (6.d/) fall outside the domain of intentionality.

 7. a/ Eggs are broken (in order) to make omelettes.
 b/ The ship was sold to collect insurance.
 c/ The books were sold (in order) to raise money.

 These sentences are well-formed even though they are very
 similar to ill-formed 1. The reason is that the passive
 verbs in the main clause imply an intentional agent, which
 then is interpreted as the subject of the infinitival
 purpose clause. A middle verb in the main clause (as in 1.)
 does not similarly imply an intentional agent. The general
 point is that control is determined by thematic relations
 and not by grammatical relations.

 8. The covers must be attractive (in order) for the book to
 sell well.

 This sentence shows that "must" does not (always) supply an
 animate agent for the purpose clause; in this instance it
 does not since the purpose clause has an explicit non-
 animate subject!

 9. a/ The staff work well (in order) to raise money.
 b/ The engine runs hot (in order) to save fuel.

 9.a/ is well-formed even though it is similar to 1. because
 the main clause includes an intentional agent. 9.b/ is also
 well-formed because it is implied that the engine was
 designed by somebody to run hot and this person is
 understood as the subject of the purpose infinitive. It is
 possible to design a car to run hot but less possible to
 design books to sell well.

 10. a/ *The books were sold without reading them.
 b/ The books can be sold without reading them.
 c/ *The books might have been sold without reading them.
 d/ ?The books might sell well in order to raise money.
 e/ *The books may sell well in order to raise money,

 Non-epistemic modals other than "must" can also save a
 sentence like 10.a/ from ill-formedness. 10.a/ is ill-formed
 since, because of the attachment site of the adjunct, the
 external argument of the passive cannot be a controller of
 the PRO of the infinitive. 10.c/, d/, and e/ are ill-formed
 because the modals are epistemic.

 11. a/ *The boat sank in order to raise money.
 b/ The boat must sink in order to raise money.
 c/ It is necessary that the boat sink in order to raise
 money.
 d/ *The rock rolled down the hill in order to raise money.
 e/ The rock must roll down the hill in order to raise money.

 11.a/ and 11.d/ show that not only middle verbs but other
 kinds of semantically non-agentive predicates exhibit the same
 pattern.

 12. The Dutch translations of sentences 1.-3. are associated with
 the same well-formedness judgments as their English
 equivalents:

 a/ *Deze boeken verkopen goed om geld op te brengen.
 b/ Deze boeken moeten goed verkopen om geld op to brengen.
 c/ Het is noodzakelijk dat deze boeken goed verkopen om
 geld op to brengen.

 13. a/ The girl works hard (in order) to raise money.
 b/ *The girl works well (in order) to raise money.
 c/ *The girl functions well (in order) to raise money.
 d/ *The solution works well (in order) to raise money.
 e/ The girl is diligent (in order) to raise money.

 These sentences show that the subject of the main clause has
 to be something that may carry motivation and that engages
 in a goal-directed activity.

 14. The animals sell well (in order) to raise money.

 This sentence is ill-formed in isolation since animals
 cannot intend to raise money. But in an appropriate context,
 where animals can be construed as having intentions, the
 sentence becomes well-formed. Suppose a pet store owner is
 complaining about the indifference of the people in town
 and their unwillingness to buy pets. The animals in the
 store overhear this and make a point of acting cute and
 frisky in order to attract passsers-by. In this context, one
 can say "The animals were selling well to raise money for
 the owner."

 15. a/ *The book is a best-seller to raise money.
 b/ These bagels cut easily to allow for quick buttering.
 c/ The sleeves are long to show the cuffs under a jacket.
 d/ The boat was sunk to collect the insurance.
 e/ *The boat got sunk to collect the insurance.

 All of these sentences show if an agent is inferrable in the
 main clause, the purpose clause is appropriately added and
 the whole sentence is well-formed.

 16. Our subway doors close slowly (in order) to minimize the
 chances of injury. /said by company representative in sales
 pitch to city officials/

 This is syntactically like 1. but it is well-formed since it
 is possible for the listener to find a sentient agent
 involved whose presence justifies the purpose clause.

 17. a/ The books are being sold to raise money.
 b/ The books must be sold to raise (the/some/enough...)
 money.
 c/ ?It is necessary that the books be sold to raise money.

 These are true-passive variants of 1.-3. Once again, true
 passives involve an implied agent which makes them
 well-formed.

 4. LITERATURE /given as cited in the responses; only a few
 additional data have been supplied by me)

 Carlson, 1989.
 Fox, Denny. /a paper on passives and on the _by_- phrase/
 Gross, Derek. /unpublished CLS paper/
 Hale, Kenneth and Samuel J. Kayser. 1993. "On argument
 structure and the lexical expression of syntactic relations".
 In: K. Hale and S. J. Kayser, ed., _The view from building 20_,
 53-109. Cambridge: MIT Press.
 Hegarty, Michael. l989 or later. WECOL 2.
 Jones, Charles. l985. /dissertation on syntax/ University of
 Massachusetts.
 Kehler. l995. _Interpreting cohesive forms in the context of
 discourse inference._ Harvard University Dissertation.
 Kratzer, Angelica. 1986.
 Kratzer, Angelica. 1991. "Modality". In A. von Stechow, D.
 Wunderlich (ed.), _Semantics: an international handbook of
 contemporary research_. Berlin: Mouton de Gruyter.
 Lagerwerf, Luuk. 1995. "The implication of Dutch 'hoevel'
 (although)" _Proceedings of the Third ISSA Conference on
 Argumentation_, volume 3.
 Lascarides and Asher. 1991. "Discourse relations and
 defeasable knowledge". _Proceedings of the Association of
 Computational Linguistics_.
 Minkoff, Seth. /MIT dissertation from a couple of years ago/
 MIT WPL.
 Mishigauchi. l994. /article/ _Language_.
 Sanders, Jose. 1994. _Perspectives in narrative discourse_.
 Tilburg University Dissertation.
 Schlesinger, I.M. 1988. "The origin of relational
 categories." In Y. Levy, I.M. Schlesinger, and M.D.S. Braine
 (ed.), _Categories and processes in language acquisition_.
 121-178. Hillsdale: Erlbaum.
 Williams, E. 1985. "PRO and Subject of NP" _Natural Language
 and Linguistic Theory_, 3.
 Williams, E. 1987. "Implicit arguments, the Binding Theory,
 and Control". _Natural Language and Linguistic Theory_, 5.



 ************************************************************************
				 Edith A. Moravcsik
				 Department of Linguistics
			 University of Wisconsin-Milwaukee
				 Milwaukee, WI 53201-0413
 USA

				 E-mail: edithcsd.uwm.edu
				 Telephone: (414) 229-6794 /office/
					 (414) 332-0141 /home/
			 Fax: (414) 229-6258