Green: Pragmatics & Syntax

1

Syntactic Constructions 1

Green: Pragmatics & Syntax

– Eks delivered a rug to Aitchberg.

– A rug was delivered to Aitchberg by Eks [PASSIVE]

– There was a rug delivered (to Aitchberg)(by Eks) [THEREINSERTION]

– A rug, Eks delivered to Aitchberg [TOPICALIZATION]

– It was a rug that Eks delivered to Aitchberg. [CLEFT]

– What Eks delivered to Aitchberg was a rug.[PSEUDOCLEFT]

– …and deliver a rug to Aitchberg, Eks did. [VERBPHRASE PREPOSING]


– Finding typographical errors is never easy.

– It is never simple to find typographical errors.[EXTRAPOSITION]

– Typographical errors are never simple to find. [TOUGH-MOVEMENT]

– Simple to find, typographical errors are not.[ADJECTIVE PHRASE PREPOSING]

– Eks met a woman who said she was the PrincessAnastasia’s governess at Treno’s.

– At Treno’s Eks met a woman who said she was in thePrincess Anastasia’s governess. [ADVERB PREPOSING]

– Eks met at Treno’s, a woman who said she was thePrincess Anastasia’s governess [HEAVY NOUNPHRASE SHIFT]

2


– Eks met her at Treno’s, a woman who said she was thePrincess Anastasia’s governess. [RIGHT DISLOCATION]

– A woman who said she was the Princess Anastasia’sgoverness, Eks met her at Treno’s. [LEFT DISLOCATION]

– The little bunny scampered into its hole.

– Into its hole, the little bunny scampered. [LOCATIVEPREPOSING]

– Into its hole scampered the little bunny. [INVERSION]

• In general, and all other things being equal, the first phrase in asentence tends to be intended to denote familiar (or TOPICAL, orGIVEN, or OLD, or presupposed, or predictable, or THEMATIC)material, while phrases toward the end of the sentence tend to denoteNEW (or asserted, or RHEMATIC) material.


Preposing (Ward 1985)

– The preposed element refers to an entity which isrelated as a BACKWARD-LOOKING CENTER toentities previously evoked in the discourse (the set ofFORWARD-LOOKING CENTERS).

• The Cb must stand in a salient scalar relation to the {Cf}.

• Set/subset, part/whole, type/subtype, entity/attribute, etc.

– The unstressed, presupposed part of the sentence (itsOPEN PROPOSITION) is salient in the discourse.

3


– An Eastern bloc embassy official gave Eks six full-sizeoriental rugs, and directed him to give them to thesenators who had been most cooperatiave. One ofthese rugs Eks delivered to Sen. Aitchberg.

– Eks and Aitchberg played golf together regularly.??An oriental rug Eks delivered to Aitchberg oneday.


Inversion (Green, Birner)

– Subject appears after the main verb.

– Preposed adjectival phrase, participial phrase, orlocative or directional adverbial phrase.

– Most common in writing rather than speech.

– Initial phrase may be explicitly or implicitly anaphoric,so that the inversion describes how the followinginformation relates to previous discourse.:

• Attached to it, as always, is an application blank for next year’slicense.

• At issue is Section 140(a) of the Controlled Substances Act.

4


– A new discourse element may be placed in sentence-final, focused position:

• In a little house lived two rabbits.

• The grounds were lavishly furnished with ceramic, stone, andwrought-metal sculpture. There were an enormous stainlesssteel frog and two tiny elves in the foyer of the guest house,and outside stood a little angel.

– The event or locative relationship may resolve somepuzzle in a narrative as it has been established up to thatpoint:

• Then at the darkest hour dawned deliverance. Through therevolving doors swept Tom Pulsifer.

• One night there was a tap on the window. Mrs. Rabbit peepedthrough the window. Outside stood a little angel.


Reflection of Beliefs and Attitudesabout Context

– The subordinate clause (a) sentences represents apresupposed or otherwise subordinate proposition, but hasits own illocutionary force in the (b) sentences:

• (a) I bet it’ll float if you throw it in the lake.(b) It’ll float if you throw it in the lake, I bet. [SLIFTING]

• (a) That Sandy thought it was Tuesday is obvious.(b) It’s obvious that Sandy thought it was Tuesday.

[EXTRAPOSITION]

• (a) Someone who said the girls were supposed to bring two quarts ofpotato salad called.

(b) Someone called who said the girls were supposed to bring twoquarts of potato salad. [RELATIVE CLAUSE EXTRAPOSITION]

5


– RAISING CONSTRUCTIONS reflect the speaker’sassumption that the experiencer or agent would becapable of interaction with the referent of the objectphrase at the time referred to by the main verb.

• (a) It seemed to MacArthur that Patton/Julius Caesar was thegreater general in history.

(b) Patton/!Julius Caesar seemed to MacArthur to be the greatest general in history.

• (a) Eks asked that Sandy leave.(b) Eks asked Sandy to leave.

– The use of the present tense to refer to future time is possible onlyif the event is mutually understood to be prearranged or scheduled:

• The Celtics play the Bucks tomorrow.

• !The Celtics win tomorrow.


– NEGATIVE TRANSPORTATION constructionsrepresent hedged or weaker claims by implicatingrather than asserting the negative proposition.

• (a) I don’t think Sandy will arrive until Monday.(b) I think Sandy won’t arrive until Monday.

– The use of some in interrogative, conditional, andhypothetical constructions indicates a positive attitudetowards the described situation, while any [NEGATIVEPOLARITY ITEM] reflects a neutral or negativeattitude.

• (a) Does Bill want some spinach?(b) Does Bill want any spinach?

• (a) If you eat some bread, I’ll cook hamburgers all week.(b) If you eat any bread, I’ll cook hamburgers all week.

6


– SUBJECT-AUXILIARY INVERSION in embeddedquestions and truncation of embedded questions[SLUICING] implies that the individual to whom theanswer is implied or assumed to be relevant is in factignorant of the answer.

• (a) She wants to know who did I appoint.(b) !She already knows who did I appoint.(c) She already knows who I appointed.

• (a) John broke something, but he won’t say what.(b) !John broke something, and he said what.(c) John broke something, and he said what he broke.


Discourse Particles– Ah indicates that the following material represents

information that is news to the speaker. • Ah, it says in the paper that Kissinger is a vegetarian.

• Mr. Eks will arrive at, ah, 3:35.

– Oh with high level intonation indicates that the followingmaterial represents a choice made with the knowledge thatat least one other choice would have been equally corrector cooperative.

• A: What’ll we do when our cousins get here?B: Oh—we’ll have a picnic. Or maybe we’ll go to the movies.

• A: When are we going to Chicago?B: Be ready to leave at, oh—2:00, and we’ll be there by 6:00.

7


– Well indicates that something in the immediatelypreceding discourse is hard to reconcile with thespeaker’s model of the world.

• A: What time is it?B: Well, my watch says 2:30.

– Why indicates that the speaker believes that she is nottelling the addressee anything he didn’t already know.

• A: Who is buried in Grant’s tomb?B: Why, General Grant is.

– Why before a statement implies that the content is newsmore to the speaker than to the addressee:

• Why, you’re Clyde’s little girl!


Reflections of Perceived Difficulty– Extraposition and Heavy NP Shift allow the longest or

most conversationally significant constituent to be put last.• Whether Kim will visit museums in France and Dana will go to

concerts in Vienna, or Dana will visit museums in France and Kimwill go to concerts in Vienna is unclear.

• It is unclear whether Kim will visit museums in France and Danawill go to concerts in Vienna, or …

– Longer phrases sound better even if they have no moresemantic content, but of two NPs of equal length, the moresignificant-sounding one sounds better postposed:

• The district attorney considers indictable Montgomery J.Jingleheimer-Smith III/?Rose Budd.

• The committee has attributed to Margaret Thatcher anextraordinary poem/?the 27-line poem.

8


Some Other Cases

– Passive is used to place the agent at the end of thesentence, to accommodate the fact that the agent is notknown, or to suppress the fact that the agent is known.

• (a) The car was stolen by two young men wearing designer jogging suits.

(b) My bike was taken between 3:00 nad 5:30 on Monday.(c) This idea has been attacked as simplistic and naïve.

– There are truth-conditionally equivalent alternatives topractically every describable construction

– The conditions are so strongly linked to the syntacticconstruction that it sounds irrational to deny them:

• !The Celtics win tomorrow, but it’s not preordained or fixed;there’s an off chance the Pistons might beat them.


Hurewitz 1998: A quantitative lookat discourse coherence

– Centering, as a theory of local discourse coherence,seeks to explain our intuitions that some utterancessound good together while others do not.

– Centering transitions can be used in a corpus-basedquantitative analysis in order to test hypotheses aboutdiscourse structure and speech production.

– The passive construction has different uses in spokenand written discourse.

9


Constructions and Topics

– Passives are said to be marked for topic (Davison),meaning that the subject of a passive is more topical thanthat of the corresponding active.

– Researchers agree that topics tend to fall at the beginning ofutterances; the topic is what the sentence is about; topicsare used to maintain or establish discourse coherence

– But it is unclear how syntactic constructions influence andare influenced by discourse coherence, the identification of‘topic’ can’t give us all the answers.

– This chapter examines whether constructions mark certainconstituents as more likely to be linked to preceding orsucceeding utterances, and how discourse genre influencesthese coherence relations.


Differences in Discourses

– In speech production we are often limited by memoryconstraints and by time constraints, and we have theoption of using prosodic signals that contribute to theaddressee’s understanding of the discourse structure.

– Written discourse is typically highly edited. Certainforms of prescriptive grammar are expected to beobserved. Various genres of writing (e.g. newspapertext, computer manuals, recipes) may carry with themrigid notions of style and coherence

– Here we look at conversational speech and written text.

10


Centering– Centering theory allows us a method of algorithmically

measuring the nature of the transition between oneutterance and the next.

– The Cb is presumed to be ‘the most salient entity in thediscourse’ and ‘the entity that an individual utterancemost centrally concerns’, hence is a topic-like entity.

– If certain pragmatic-syntactic structures are in somesense different from canonical word order in terms oftopicality or discourse coherence, a statistical analysisof centering transitions should expose this.

– We will compare transition statuses of a random sampleof utterances against a sample of syntactically markedutterances


Quantitative Analysis

– It is assumed that within a given corpus, theproportional number of each transition type should beroughly equivalent across all discourses when averagedamong a large number of tokens.

– If a control group of utterances contains 40%Continues, then any large number of tokens taken froma similar style of discourse should also containapproximately 40% Continues.

11


Procedure– The Brown corpus is made up of excerpts from various

written texts, including newspaper and magazinearticles, novels, announcement notices and theaterreviews, etc.

– The Switchboard corpus contains 2,430 taped telephoneconversations. Strangers were dialed by a computerand assigned to talk about a specific topic (e.g. guncontrol, working mothers, golf, etc.) for 5 to 10minutes.

– For each corpus, approximately 200 utterances wererandomly chosen from at least twenty different fileswithin the corpus.


Coding

– These utterances were coded according to sentencetransition (Continue, Retain, Smooth-Shift, Rough-Shift, No Cb).

– The grammatical subject was ranked as the preferredcenter (Cp) and other entities were ranked according toword order.

– In the control sample there were no instances of by-passives, topicalizations, left-dislocations or inversions.

– An utterance was defined as a finite clause exceptrelative clauses, where only unrestrictive relativeclauses modifying a clause-final NP were counted asseparate utterances.

12


– Speaker interjections such as I know were counted asseparate utterances only if they interrupted the flow of theother speaker’s turn.

– Possessive pronouns in subject position were coded asContinues: ‘the man’…’his hand’.

– ‘instance of’ and ‘poset’ relations were coded as Smooth-Shifts: dogs…one dog, Cocker Spaniels. So werefunctional dependencies: house…door. Discourse deicticpronouns (this, that, it referring to chunks of discourse)were collapsed into the Smooth-Shift category.

– When pushes or pops to another discourse segment tookplace, the transition was coded as No Cb, only 2 percent ofutterances were Rough Shifts and they were collapsed withNo Cb.


Control Chart: Transitions for a RandomSample

Transition Brown Corpus(N = 206)

Switchboard(N = 200)

CONTINUE 43% (88) 41% (81)

SMOOTH-SHIFT

27% (56) 27% (53)

RETAIN 9% (18) 6% (11)

NO Cb 21% (44) 27% (55)

13


– Mostly Continues and Smooth Shifts (70% BrownCorpus, 68% Switchboard).

• Most of the time if there is a coherence link between twoutterances, the linking entity is in subject position.

• Fits well with theories of a given-new contract between thespeaker and hearer. The speaker places given information firstto assist the addressee in attaching the new information toprevious discourse.

• Not only does material at the beginning of the sentence tend tobe given, but this material also tends to provide the linkingstructure with the previous utterance.

– The data show a lot of Shifts between utterances.• Other studies have shown Retains are used infrequently.

• There may be a processing cost assocaited with not putting themost salient information at the beginning of the currentutterance, so the grammatical subject is not what the utteranceis about but what links it to previous discourse.


A Test for Markedness

– In order to analyze whether a particular position in asentence is marked, we need to measure how often thatposition is used to link the utterance to the previousutterance.

– Observe how often the position is the Cb in selectedinstances as compared to a control group.

– Centrality Index = % total transitions that are Smooth-Shifts and Continues.

– The CI for by-passives should be higher than the CI forthe control group is by-passive subjects are marked fortopic.

14


Passive Transitions

– Tokens of by-passives were extracted from the corpora,and coded in the same way.

– By-passives were defined as those utterances that usedthe finite form of the passive participle of a verb,contained the sequence by+agent, and could betransformed into an active utterance.

• The game was won by default. NO*Default won the game.

– Relativized and subordinate passives were excluded.


Written passive transitions

Transition Written passive(N=149)

Written control(N = 296)

CONTINUE 40% (60) 43% (88)

SMOOTH-SHIFT

34% (51) 27% (56)

RETAIN 10% (15) 9% (18)

NO Cb 15% (23) 21% (44)

CI 74% 70%

15


Spoken passive transitions

Transition Spoken passives(N = 174)

Spoken control(N = 200)

CONTINUE 56% (97) 41% (81)

SMOOTH-SHIFT

19% (33) 27% (53)

RETAIN 11% (19) 6% (11)

NO Cb 14% (25) 28% (55)

CI 75% 67%


– The CI for written passives is not significantly differentfrom the control utterances. Spoken passives do showsuch a tendency, although it is not quite at the 5%significance level.

– However, there do seem to be a lot more Continues inthe spoken passives, but the effect is dampened by adecrease in Smooth-Shifts.

– If the percentage of spoken Continues in the controlversus passive utterances is compared, the effect ishighly significant.

– This effect is not seen in a comparison of writtencontrol and passive utterances.

16


Analysis– Spoken passives are used to Continue with the same

center, Shifts are rarely found. This would not bepredicted by simple theories of given-new informationstructure, nor from Davison’s markedness theory sincein a Smooth Shift the grammatical subject is oldinformation contained in a previous sentence.

– Perhaps there are two conflicting strategies in sentenceproduction—a tendency towards putting salient entitiesearly in the sentence and a tendency towards canonicalactive constructions. When the logical direct object ofa sentence was the Cb of the previous utterance, thetendency to begin with the most salient overrides theactive default, and a passive sentence is produced.


– However, non-Cb old information doesn’t have enoughprominence in the discourse model to compete againstthe default active tendency, therefore few spokenpassives are the result of centering Shifts.

– There were also more Retains amongst the passives, but13/19 had first- or second-person pronouns in subjectposition. It is likely that is due to an empathy effect.

• Speakers relate information from a camera angle (Kuno)

• ??John walked down the street. He was followed by me.

– Contemporary studies of speech production indicatethat as concepts become available, they areincrementally assigned a surface structure and uttered.

– Written discourse doesn’t follow these constraints.Conceptual accessibility isn’t a factor in choosingbetween active and passive constructions.

17


Partial Shifts

– In coding transitions, there are many ambiguous casesin which is it is difficult to determine if an entity iscoreferent to an element in the previous utterance.

– E.g. discourse deictic pronouns, which can refer tochunks of discourse ranging from events or clauses towhole paragraphs:

• [1]And Rochester was in a state of emergency for three weeks,[2]and everyone lost their power. [3]And we had to pay overfive hundred dollars to have a tree removed. [4]And it wasn’tcovered by insurance. (Switchboard file 2697)

• These cases were coded as “deictic shifts” and then collapsedwith Smooth-Shift or Rough-Shift.


– A similar issue arises in the case of functionaldependencies.

• [1]On November 4, 1952, an earthquake occurred under the sea offthe Kamchatka Peninsula. [2]At 17:07 that afternoon (Greenwichtime) the shock was recorded by the seismography alarm inHonolulu. (Brown corpus, file cf21)

• According to Joshi & Weinstein (1981), the Cb(U2) is the shock,but the earthquake is still present in U2 as a backgrounded memberof the Cf list, therefore in U3 another aspect of the earthquake suchas the rumble could be discussed without creating a No Cbtransition.

– Same goes for instance-of or poset relations:• [1]The whole official City apparently has an intense hatred towards

birds.

[2]Starlings and blackbirds are scared off by cannon from CityHall. (Brown corpus, file cb19)

18


– The problem is with what it means to be realized in theprevious utterance.

– Psycholinguistic evidence such that part-of relations are inhigh salience at least some of the time.

– When two utterances have the relationship such thatCp(Un) is a subset of several elements or pieces ofelements in Cf(Un-1), the transition between the utterancescan be termed a Partial-Shift.

– In Heim’s (1983) file card card semantics, we might saythat Partial-Shift file cards are created out of eithercombining or segmenting file cards that are already extantin the discourse model.

– It may turn out that Partial-Shifts carry a heavierprocessing load than smooth shifts.


Passives and Partial Shifts– Written passive utterances have a significantly higher rate

of Partial-Shifts than written control utterances (19%passives, 9% control).

– Spoken passives show no such tendency (16% control,13% passive).

– Therefore it seems that written passives have a propensityto shift or reorganize the focus of attention.

– Unlike spoken passives, written passives are not used tolink with the most salient entities in the previous utterance.

– Spoken passives as compared to controls have an increasednumber of pronouns in subject position, but writtenpassives have significantly fewer pronouns than thecontrol.

19


Percentage of Pronouns in subject position

Spokendiscourse

Writtendiscourse

Control 68% (118) 53% (109)

Passive 63% 126) 28% (42)


Passives and Beyond

– Perhaps the main coherence function of a written passiveis comprised by its transition to the subsequent utterance.

– The topic once established is said to be in focus insubsequent utterances (Gundel et al. 1993). If passiveswere marked for topichood, we would expect thegrammatical subjects of passives to more likely than thecontrol sample to be the Cb of the successive utterance.

– Surprisingly, a statistically significant difference can befound but in the opposite direction: the utterance followinga written passives was less likely to Continue, more likelyto Smooth-Shift; this effect was not found for spokenpassives.

20


Transitions for utterances following passives

Transition Utteranceafter spoken

passives(N = 174)

Spokencontrol

(N = 200)

Utteranceafter written

passive(N = 149

Writtencontrol

(N = 206)

CONTINUE 38% (66) 41% (81) 25% (37) 42% (88)

SMOOTH-SHIFT

14% (25) 10% (20) 30% (45) 18% (38)

PARTIAL-SHIFT

18% (31) 17% (33) 13% (20) 9% (18)

RETAIN 10% (18) 6% (11) 14% (21) 9% (18)

NO Cb 20% (34) 28% (55) 17% (26) 21% (44)


– In 42 out 45 instances of Smooth-Shifts after writtenpassives, the antecedent of the Cb was the agent of thepassive utterance. That is, 15% of the time, the Cb ofpassive (Un+1) is coreferential with the grammatical subjectwhile 28% of the time it is coreferential with the agent.

– Typically agents are animate and themes are not, andanimates are high in conceptual accessibility and onKuno’s empathy scale. Also many languages have a strongtendency to put agents in subject position. Perhaps agencyshould have an impact on the ranking of the Cf list incentering.

– One reason why written passives were not found to behighly coherent, is because the more common agentlesspassive was not considered. Writers may tend to usepassives to avoid expressing the agent.

21


– When a writer does use a passive with an agent, itmight be to shift attention onto that agent, so the role ofpassive in written discourse is as a transitional elementto shift focus from one entity to another:

• [1]In order to save the mystery story, they [writers] haveconverted the private detective into an organization man. [2]Thefirst of two possible variations on this theme is symbolized byMickey Spillane’s Mike Hammer. [3]At first glance, this heroseems to be more rather than less of an individualist than anyof his predecessors. [4]For Hammer, nothing is forbidden.

(Brown corpus, file cg19)

– The writer uses the by-passive in order to smoothlyshift attention from a general discussion of mysterystories to one detective in particular. The passiveconstruction allows this shift to take place in a highlycoherent way, without necessitating No Cb transitions.


Conclusion– Written and spoken by-passives do not have the same

coherence function.

– Given the luxury of unlimited processing time andeditorial freedom, writers are able to manufactureutterances that utilize passives as transitional structures.

– Speakers, who have neither time nor opportunity forrevision, plan their utterances clause by clause accordingto the constraints imposed by levels of accessibility.

– Centering has more to contribute to linguistic andcomputational theory than its typical application as a toolfor anaphora resolution.

– Through quantitative use of Centering, we can highlightsubtle discourse tendencies unavailable to our intuitions.

22


Birner: Information Status and WordOrder: An Analysis of English Inversion

• Labor savings are achieved because the crew is put to better usethan cleaning belts manually: <also eliminated> is <the expenseof buying costly chemicals>. [WOODEXTRA]

• You can drive as fast as you like in the outside lane on a WestGerman highway and may feel like t he king of the road—untilyou look in the rear mirror. <Zooming in on you like a guidedmissile> comes <a rival contender, bullying you to get out of theway> [Chicago Tribune]

– It will be argued that inversion is an information-packaging mechanism, allowing the presentation ofrelatively familiar information before a comparativelyunfamiliar logical subject.


Definitions and Terms

– An INVERSION is a sentence in which the logicalsubject appears in postverbal position while some other,canonically postverbal, constituent appears in clause-initial position.

• <Sitting in the garden> was <an old man>.

• <In the garden> sat <an old man> unhappily.

– Excludes some other constructions:• Enter John Smith, a 40-year-old lawyer from Brooklyn.

• Gladly would he now have consented to the terms.

• Into this derived structure, then, lexical items are inserted.

• There’s only one hallway in the West Side school.

23


Previous Analyses– Inversion is ‘exclamatory’, ‘emphatic’, signals a

‘counter-expectation.• Loud-speakers had been erected along the garden: <through

them> emerged <the bland informal voice of CommanderStephen King-Hall describing the scene as he saw it from St.Paul’s Cathedral.> [Holtby 1936]

– Marks an Open Proposition as salient shared knowledge• The giant leader roared and shouted and cheered on the guests.

<Beneath the chin lap of the helmet> sprouted <blackwhiskers>. [Upfield, 1988]

– Inversion marks ‘presentational’ or ‘contrastive’ focus,• As the skipping rope hit the pavement, so did the ball. As the

rope curved over the head of the jumping child, the child withthe ball caught the ball. <Down> came <the ropes>. <Down>- <the balls>. [L’Engle 1962]


– Subjects are postposed when they are formally heavierthan the verb.

• In the grass little jeweled lizards darted. [L’Engle 1978]

– The preposed element serves a connecting function.• <Down the dusty Chisholm Trail into Abilene> rode <taciturn

Spit Weaver,> his lean brown face an enigma, his six-gunswinging idly from the pommel of Moisshe, the wonder horse.

• <In a little white house> lived <two rabbits>.

– The postposed subject represents new information(information not in consciousness)

• Yes, this is no ordinary general election. ‘Evans is a Democrat; Daley is a Democrat. DifferentDemocrats have different p;oints of view about the city ofChicago and its politics,’ Jackson noted. ‘The war between forceswithin the party continues, and within our coalition. <Standing in the middle of it all> is <Jesse Jackson>. [ChicagoTribune]

24


Information Status– Chafe 1976

• Given information is ‘that knowledge which speaker assumesto be in the consciousness of the addressee at the time ofutterance.’

– Prince 1981• Assumed familiarity: what the speaker assumes to be familiar

to the addressee.

• Scale of familiarity (evoked….unused….brand new, etc.)

– Prince 1992• Hearer Old/Hearer New; Discourse Old/Discourse New

• Provides a more concrete way of getting at many of the sameintuitions that motivate the theme/rheme distinction.


Analysis of the Corpus

– Seldom did the entire preposed constituent representfamiliar information, typically only a subset of theinformation represents familiar information, here thepreposed constituent was still coded as familiar.

– Often the preposed constituent describes an evokedconcept in new terms, here it was still coded asfamiliar.

– In many cases, the preposed familiar element was notan entity, but rather a description or activity, here it wasstill coded as familiar.

– Postposed element was almost always an NP.

25


Results and Discussion

FinalElement

InitialElement

H-old, D-old

H-old, D-new

H-new, D-new

Total

H-old, D-old

29 0 0 29

H-old, D-

new

100 4 6 110

H-new, D-new

433 28 103 564

Total 562 32 109 703


FinalElement

InitialElement

D-old D-new Total

D-old 29 0 29

D-new 533 141 674

Total 562 141 703

26


– In postposed position, discourse-new constituentsoutnumbered discourse-old constituents by more than20 to 1.

– In preposed position, discourse-old constituentsoutnumbered discourse-new constituents by 4 to 1.

– 76% (533) tokens contained both a preposed discourse-old element and a postposed discourse-new element.

• We have complimentary soft drinks, coffee, Sanka, tea, andmilk. <Also complimentary> is <red and white wine> Wehave cocktails available $2.00. [stewardess on Midway airlines]

– The opposite ordering is infelicitous:• A: Hey, mom, have you seen my gym shirt? I’m in a big

hurry to get to the bus stop.B: #In the hall closet is your gym shirt.


– Such infelicity doesn’t result from combining apreposed hearer-new element with a postposed hearer-old element:

• You won’t believe what I saw yesterday when I was walkingpast the park. Sitting and talking with an elderly man was yourlittle brother. I think they were feeding the squirrels.

– Pragmatic constraint on Inversion: the preposedelement must not be newer in the discourse than thepostposed element.

– This allows the new element to be processed in terms ofits relationship to the (preceding) familiar element.

27


– In 29 cases both elements were discourse-old; in 21 ofthese the initial element had been mentioned morerecently than the final element.

• Each of the characters is the centerpiece of a book, doll andclothing collection. The story of each character is told in aseries of six slim books, each $12.95 hardcover and $5.95 inpaperback, and in bookstores and libraries across the country.More than 1 million copies have been sold; and in late 1989 aseries of activity kits was introduced for retail sale. <Complementing the relatively affordable books> are <thedolls, one for each fictional heroine and each with acomparatively pricey historically accurate wardrobe andaccessories>. [Chicago Tribune]

– As Prince (1992) notes, saliency is an additional featurethat may be relevant for distinguishing betweendiscourse-old elements.


Inferrables

• She got married recently and <at the wedding> was <themother, the stepmother and Debbie>. [Conversation]

– Prince (1992) notes that inferrables, although they aretechnically both hearer-new and discourse-new, dependupon a discourse-old ‘trigger’ element. She suggeststhat inferrables may be collapsible with discourse-oldnonpronominals.

– Prince (1981):• Evoked > Unused > Inferrable > Containing Inferrable >

Brand-New Anchored > Brand-New

28


– Inferrable elements to precede unused elements ininversions:

• Enclosed Are the GEICO Home Insurance Rates You Requested.[on front of envelope]

– Both orderings of evoked and inferrable elements arefound.

• She’s a nice woman, isn’t she? <Also a nice woman> is <ournext guest> [David Letterman]

• Nusseibeh’s unusual predicament causes concern all around. Hisfriends fear that Arab hard-liners will turn on Nusseibeh, thinkinghe is an Israeli ally. The Israelis, who certainly want to squelch the 17-month-olduprising in the West Bank and Gaza Strip, are under intensepressure from the United States not to jail moderates who mayfigure in the their election proposal for the territories occupiedsince the 1967 war. <Most immediately affected> is <Nusseibeh himself> [ChicagoTribune]


— Inferrable information may be collapsed with evoked information.

FinalElement

InitialElement

Evoked Inferrable Total

Evoked 29 26 55

Inferrable 42 41 83

Total 71 67 138

29


— Discourse-familiarity including inferrables (excluding 20containing inferrables and 468 insufficient context cases)

FinalElement

InitialElement

D-old D-new Total

D-old 138 3 141

D-new 1008 141 1149

Total 1146 144 1290


Containing Inferrables

• <On the ground occupied by this building> once stood <the housewhere lived in 1791 Alexander Hamilton Secretary of theTreasury of the United States> [inscription on wall of building]

– It seems that the discourse status of the containinginferrable is dependent on the discourse status of theelement inferenced from.

30


Definiteness– Of the 1485 relevant tokens, 1332 (90%) had definite

preposed constituents, and only 153 (10%) hadindefinite preposed constituents.

– There were 763 (51%) postposed definites and 722(49%) indefinites.

– While inversion is sensitive to discourse-familiarity,definiteness is sensitive to hearer-familiarity.

– There does seem to be a constraint against postposedanaphoric pronouns, but not deictic:

• Among the guests of honor was sitting HER [pointing]


Verbs

– Of 1778 tokens in the corpus, 654 were instances ofinversion around be, while the remaining 1124contained some other main verb.

– The verb must represent evoked or inferrableinformation in context, i.e. must be ‘informationallylight.’

• A vase of wildflowers sat in the middle of the table. <Fromthe kitchen> wafted <aromatic smells of fresh-cooked meat,spices, garlic, and onions>. [Kellerman, 1990]

31


Conclusions

– Prince (1992) found that canonical subjects tended tobe discourse-old.

– But it is the preposed constituent rather than the subjectwhich tends to be discourse-old in inversions.

– Therefore, the generalization is that discourse-oldinformation tends to precede discourse-new informationin the sentence.


Ward & Birner: Definiteness and theEnglish Existential

• Have you seen the dog or the cat around?Not lately. #There’s the dog running around loose in theneighborhood.

– The class of formally definite or possessive determiners(e.g. the, this, those, our, her) or certain quantifiers (all,every, most) are constrained from appearing in post-verbal position in a there-sentence but are not forbiddenacross the board: Definiteness Effect.

– What is the real restriction? Many writers assume acognitive constraint, information must be new in somesense.

32


Previous Accounts

– Rando & Napoli (1978)• Postverbal NP must be nonanaphoric in Kuno’s 1972 sense,

i.e. unfamiliar.

• In list-sentences, it is the list itself that is nonanaphoric:

• Q: What’s worth visiting here?A: There’s the park, a very nice restaurant, and the library.

– However, nonlist formal definites are also possible• A: Don’t forget that Kim will be bringing a salad.

B: Oh right—there is that. (Abbott 1992)


– Abbot (1992, 1993)argues that all existential theresentences are contextualized existentials.

• A: I guess we’ve called everybody.B: No, there’s still Mary and John.

• Speaker B is not asserting the existence of a list, but is drawingaddressee’s attention to the existence of Mary and John asfilling the predicational slot ‘people for us to call’.

• It will be generally be anomalous to assert the existence of afamiliar entity, but not when the existence of the entity ispointed out as a response to a request for entities of a certaintype or as fulfilling a certain role:

• A: What could I give my sister for her birthday?B: There’s John book on birdwatching.

– For Abbott, use of the definite, implies hearerfamiliarity, hence use of the contextualized existential‘becomes a polite way to suggest that [familiar] entityas suitable for the purposes at hand.’

33


– But there are some cases of post-verbal familiar NPsthat don’t seem to filling such a role:

• The worst one that existed was 10 thousandths on the single O-ring on the Titan, and there are 20 of the five-segment. Thatwas the earliest version. There were four of the seven-segment, which never went into production, but was just adevelopment: and then two five-and-a -half segments, whichwas a way of getting a little additional performance. And Ibelieve every one of the flying now is the five-and-a-half-segment device. And there is not any evidence, but there wasthis 10 thousandths. [Challenger Commission Transcripts, 2/10/86]


– Holmback (1984): The post-verbal noun phrase can beuniquely identifiable (Hawkin’s 1978 ‘inclusive’) if it isalso presentational (i.e. unfamiliar):

• There were both major political parties represented at theconference.

• There is the village idiot at the front door.

– But there-sentences don’t have to be presentational:• There was that deaf comedienne I was telling you about on TV

today.

• I’d love to get away from my job, the kids, the bills…I’vethought of chucking it all and going to Hawaii. But there arethe kids to consider.

34


– Hannay (1985): Post-verbal NP may be definite if itdoes not represent the sentence topic.

– But:• I think there was one flight where we had one problem. It

wasn’t ours, but there was that one flight. Other than that., Ibelieve the answer to the remaining flights is yes. [ChallengerCommission transcripts, 4/2/86]

• Corpus for the present paper:– Over 1.3 million words of transcribed oral data drawn

from the transcripts of the Presidential Commission onthe Space Shuttle Challenger Accident.

– There were approximately 100 there-sentences with beas the matrix verb.


The information status of definitepostverbal NPs in there-sentences– The speaker treats the post-verbal NP in there-sentences

as representing a hearer-new entity (Prince 1992), wherea hearer-entity is one that the speaker does not assume toexist within the hearer’s knowledge store.

– Inferrables are treated here as hearer-old.

– It is possible for an entity to be treated by the speaker asbeing hearer-old in one respect yet hearer-new inanother.

– An entity may be both hearer new and uniquelyidentifiable, hence definite.

35


– Prince 1992, hearer-new, formally definite NPs:• There were the same people at both conferences.

• There was the usual crowd at the beach.

• There was the stupidest article on the reading list.

– Ward & Birner: five classes of hearer new, uniquelyidentifiable, formally definite NPs in there-sentences.

I. Hearer-old entities treated as hearer-new.

II. Hearer-new tokens of hearer-old types.

III. Hearer entities newly instantiating a variable.

IV. Hearer-new entities with uniquely identifying descriptions

V. False definites.


Hearer-Old Entities treated asHearer-New

– Reminder there sentences:j• Mr. Rummel: Well, didn’t the designer of the orbiter, the manufacturer,

develop maintenance requirements and documentation as part of thedesign obligation?Mr. Collins: Yes, sir. And that is what we showed in the very first part,before the Pan Am study. There were those other orbiter maintenanceand requirement specifications, which not only did processing of thevehicle, but in flow testing, pad testing, and what have you, but alsoaccomplished or was in lieu of an inspection plan. [Challenger transcritps,3/31/86]

– Referent here has been at least momentarily forgotten, so it can be treated ashearer-new; demonstrative here conveys the entity is part of private sharedknowledge, based on prior co-presence, as well as uniquely identifiable.

– Demonstrative signals H can recognize the entity, use of there-sentencesignals that H is not expected to recall it.

36


Hearer-New Tokens of Hearer-OldTypes

– Known type: same, usual, regular, traditional,obligatory, expected.

– Inferrable type: ideal, correct, perfect, necessary,required.

• The Woody Allen-Mia Farrow breakup, and Woody’sdeclaration of love for of Mia’s adopted daughters, seems tohave everyone’s attention. There are the usual sleazy reasonsfor that, of course—the visceral thrill of seeing the extremelyprivate couple’s dirt in the street. [San Francisco Examiner]

• The real trouble is, of course, that there isn’t the necessaryintelligence. [Huxley, 1968]


– The post-verbal NP has dual reference, both to a typeand a token. The definite is licensed by the hearer-oldtype, while the there-construction is licensed by thehearer-new status of the current instantiation of thattype.

• There was the usual crowd at the beach today. They werethere yesterday too. Today for the first time they sat around afire and roasted marshmallows.

37


Hearer-Old Entities NewlyInstantiating a Variable

– List reading:• And there’s two components in [Division H], which his the

operations division: the people that do the flight activityplanning procedures work, provide for the crew activityplanning and the time line support and integrated proceduresdevelopment and overall flight data file management; and thenthere is the payload support folks, who provide for customeroperations integration and support of their onboard interfaces.[Challenger Commission transcripts, 4/8/86]

– X is a component in Division H.


– Here the individuals listed are uniquely identifiable, but theirmembership in the set being enumerated is new to the hearer;the individuals constitute hearer-new [focus] instantiations insome salient open proposition.

• What a great time I had last night! #There was/were John Mary,Fred, Susan, Hilda, Xavier, and Ethel at this party I went to. Wedanced for hours.

• A: Who was at the party last night?B: There was John, Mary, Fred, Susan, Hilda, Xavier, and Ethyl.

38


Hearer-New Entities with UniquelyIdentifying Descriptions

• Postponing the investigation will increase the chance that we’lluncover something additional that is significant.

• In addition, as the review continues, there is always the chancethat we’ll uncover something additional that is significant.[Challenger Commission transcripts, 3/18/86]

– Such there-sentences are infelicitous when negated,because it is odd to deny the existence of some entitythat is simultaneously being introduced into thediscourse:

• #In addition, as the review continues, there isn’t the chancethat we’ll uncover something additional that is significant.


– Superlatives, deictics, and cataphoric references may alsorepresent hearer-new, uniquely identifiable entities:

• There was the tallest boy in my history class at the party last night.

• You can see the runway and the HUD that overlays the Edwardsrunway, and then there is this line which comes out to the outerglide slope aim point. It is hard to see the PAPIs there because ofthe lights that are here. [Challenger Commission transcripts, 4/9/86]

[Spoken with a gesture]

• There are the following reasons for this bizarre effect.

– Containing inferrables:• There was the wedding picture of a young black couple among his

papers.

• A young black couple is the trigger; the wedding picture is rendereduniquely identifiableby the relationship to the trigger.

• #There was the picture of a young black couple among his papers.

• Birner 1994: the discourse status of a containing inferrable dependson the discourse status of its trigger, here hearer-new.

39


False Definites– Brand-new uses of this:

• One day last year on a cold, clear, crisp afternoon, there wasthis huge sheet of ice in in the street.

• Whether one wishes to call this definite or indefinite dependson whether definiteness is being defined as a formal orcognitive category.

– Other cases:• There are all sorts of other false definites.

There is the most curious discussion of them in our paper.There is every reason to study them.

• There are other false definites.#There are the other false definites.

– In all cases of false definites, they are hearer-new at thetime of utterance..


Conclusion

– The constraint on post-verbal definite NPs in there-sentences has been termed ‘the definiteness effect’.

– But the constraint is not on formal definiteness, rather itis a cognitive constraint: the post-verbal NP in a there-sentence must represent an entity which is hearer-new.

– Definiteness, on the other hand, requires the referent ofthe NP to be uniquely identifiable to the hearer.

– Post-verbal position in there-sentences may be filled byprecisely those definite NPs that are construable ashearer-new in context.

– The term ‘definiteness effect’ is a misnomer.

40


Prince: On the Limits of Syntax: Left-Dislocation and Topicalization

– Syntax neither follows from discourse function, norencodes discourse function.

– Rather syntax is autonomous, and the relation betweensyntactic form and discourse function is arbitrary.

– We will examine tokens of left-dislocation andtopicalization, which have been widely assumed to mark‘topic’.

– We will see that these forms can be said to mark ‘topic’only if ‘topic’ remains a vague and undefined notion;instead they have a far more interesting and richerdiscourse function.


Left-Dislocation: Form andPreviously Claimed Functions

– [CP [NPi Xi] [IP …[NPi-pro Xi] … ] ]

• The man my father works with in Bostoni, hei’s going to tellthe police that the traffic expert has set that traffic light on thecorner of Murk Street far too low.

• My fatheri, hei’s Armenian, and my motherj, shej’s Greek.

– Most previous discussions of Left-Dislocation havecrucially involved the notion of “topic”, introducing anew topic or marking a topic (Gundel 1974, 1985, etc.).

– Halliday 1967: topic (“theme”) is in the initial position,but this is entirely circular.

41


– Reinhart 1981: The topic is what the utterance is about,the proposition is retrieved/stored under that entity inthe mental model, but how exactly does this work?And this doesn’t work at all in some cases.

– Chafe 1976, Geluykens 1992: Left-Dislocation marks acontrast.

– Here: no single function can account for all LD data inEnglish, rather there are three:

• Simplifying discourse processing

• Triggering a (po)set inference

• Amnestying an island violation


Function 1: Simplify DiscourseProcessing: “Simplifying” LDs

• My sisteri got stabbed. Shei died. Two of my sisters wereliving together on 18th Street. They had gone to bed, and thisman, their girlfriend’s husband, came in. He started fussingwith my sisteri and shei started to scream. The landladyj, shej

went up, and he laid herj out. So sisteri went to get a washcloth to put on herj, he stabbed heri in the back. But shei sawher death. Shei went and told my mother when my brother wasburied. “Mother,” shei said, “your trouble is not over yet.You’re going to have another death in the family. And it’sgoing to be mei.” And sure enough, it was. [Welcomat, 12/2/81, p.81].

– Information is probably not stored under “landlady”, the discourseis not “about” the landlady.

42


– Gundel’s (1974, 1985) and Reinhart’s (1981) tests fail:• A: He started fussing with my sister and she started to scream.

B: #What about the landlady?A: The landlady, she went up and he laid her out. So sisterwent to get a washcloth to put on her, he stabbed her in theback.

• #He started fussing with my sister and she started to scream.As for/Speaking of the landlady, she went up and he laidher out. So sister went to get a washcloth to put on her, hestabbed her in the back.

• #She said he started fussing with her sister and she started toscream. She said about the landlady that she went up and helaid her out. So she said her sister went to get a washcloth toput on her, he stabbed her in the back.


– Other counterexamples to topic tests:• …there won’t be any dead up there. There’ll just be

tombstones setting there. Because the coal is under the graves.An old preacher down there, they augured under the gravewhere his wife was buried. And he’s nearly blind and heprayed and everything. [Terkel, 1974]

• I know what this piece of equipment’s raised to do. Anycompany, if they’re worth 150 million dollars you don’t needto think for a minute they’re not gonna know what you’redoin’. They didn’t get there that way. [Terkel, 1974]

• Everybody talk about it all the time. Especially AuntKatherine up here, that’s all me and her talk about—whatthey done to us. My father and mother sold their land out.”[Terkel, 19974]

– Definiteness is also not relevant, pace Gundel 1985 andReinhart 1981, nor is hearer-status.

43


– What does seem relevant is discourse-status: all LD’scrucially involve discourse-new entities.

– But discourse newness is not a sufficient condition:• My sister got stabbed. She died. Two of my sisters were living

together on 18th Street. They had gone to bed, and this man, theirgirlfriend’s husband, came in. He started fussing with my sisterand she started to scream. The landlady went up, and he laid herout. #So a wash clothi, sister went to get iti/onei to put on…

– In all cases involving the Left Dislocation of a Discourse-New entity, the NP would canonically be in a position thatis strongly disfavored for NPs evoking Discourse-newentities: possessive, subject.


– Discourse Processing Function of Left-Dislocation:“Simplifying LDs:A “Simplifying Left-Dislocation serves to simplify thediscourse processing of Discourse-new entities byremoving the NPs evoking them from a syntactic positiondisfavored for NPs evoking Discourse-new entities andcreating a separate processing unit for them. Once thatunit is processed and they have become Discourse-old,they (or, rather the pronouns which represent them) maycomfortably occur in their canonical position within theclause.

– Characteristic of unplanned, spontaneous, oral-stylediscourse.

44


Function 2: Trigger a PosetInference: ‘Poset” LDs

• (c) ‘Individuals are going to be asked to make a greatercontribution, but the overwhelming burden on the Americanpeople is going to fall on higher-income Americans, not onmiddle-income Americans,’ Foley said. ‘I think most middle-class Americansi, when they look at the costs plus thebenefits, theyi’re going to be much, much better off,’ Clintonsaid.

• (f) She had an idea for a project. She’s going to use threegroups of micei,j,k. Onei, she’ll feed themi mouse chow, justthe regular stuff they make for mice. Anotherj, she’ll feedthemj veggies. And the thirdk she’ll feed ek junk food.


– Not characteristic of spoken, unplanned discourse.

– Can occur in postverbal position, easy positions to processdiscourse new information.

– Probably intonationally distinct from simplifying LDs inthat they have a fall-rise contour.

– Don’t straightforwardly pass Gundel/Reinhart topic tests.

– Set-Inference Triggering Function of Left-Disloction:POSET LDs.A “Poset” Left-Dislocation serves to trigger an inferenceon the part of the hearer that the entity represented by theinitial NP stands in a salient partially ordered set relationto some entity or entities already evoked in the discourse-model.

45


– Posets are defined by a partial ordering R on some set ofentities, e, such that, for all e-1, e-2, and e-3 that are elementsof e, R is either reflexive, transitive, and antisymmetric oralternatively, irreflexive, transitive, and assymetric.

• a. REFLEXIVE: e-1 R e-1 TRANSITIVE: (e-1 R e-2) and e-2 R e-3) —> (e-1 R e-3) ANTISYMMETRIC: (e-1 R e-2 and e-2 R e-3) --> (e-1 = e-2)

• b. IRREFLEXIVE:¬ (e-1 R e-1) TRANSITIVE: (e-1 R e-2) and e-2 R e-3) —> (e-1 R e-3) ASYMMETRIC: (e-1 R e-2) --> ¬ (e-2 R e-1)

• Poset relations include set relations, identity relations, is-a-part-of(song and CD), is-a-subtype-of (Siberian huskies and dogs), but nothas-a relations (book and its cover).

– #I went by his house and the door, I knocked on it.


Function 3, Amnesty IslandViolations: “Resumptive Pronoun

Topicalization LDs”

– Cases where left dislocation is topicalization indisguise, where the extraction site is difficult orimpossible to extract from.

– The Double Discourse Function of (English)Topicalization.

• [CP [NP1 Xi] [IP … [ei] …] ]

• She had an idea for a project. She’s going to use three groupsof micei,j,k. Onei, she’ll feed themi mouse chow, just the regularstuff they make for mice. Anotherj, she’ll feed themj veggies.And the thirdk she’ll feed ek junk food.

46


– All topicalizations trigger an inference that the entityevoked by the leftmost NP is in a salient poset relationwith something in the prior context:

• And at Andover, which is really a jungle—any kind of highschool, I suppose, is a jungle—the things that counted weremoney, athletic prowess, clothes, and power in the outsidesociety, connections to the Greenwich, New York, sort ofdebutantes, things like that. All of these things I didn’t have.[Haj Ross, interview.]

– Topicalization also structures the proposition into afocus and a presupposition, marking the openproposition as being salient “shared knowledge” at thatpoint in the discourse and marking the instantiation asnew in the discourse.


– Double Discourse Function of Topicalization• 1. Topicalization triggers an inference on the part of the hearer

that the entity represented by the initial NP stands in a salientpartially ordered set relation to some entity or entities alreadyevoled in the discourse model.

• 2. Topicalization triggers an inference on the part of the hearerthat the proposition is to be structured into a focus and a focus-frame as follows. First, if the entity evoked by the leftmost NPrepresents an element of some salient set, make that set-relationexplicit. Then in all cases, the open proposition resulting from thereplacement of the tonically stressed constituent (in the clause)with a variable is taken to represent information saliently andappropriately on the hearer’s mind at that point in the discourse,the tonically stressed constituent representing the instantiation ofthe variable and the new information in the discourse.

• OP: She’ll feed the third (∈ {the three groups of mice}) XInstantiation: X = junk food.

47


– Return to Resumptive Pronoun Topicalization LDs• The book I had I had got from a guy who got it from a very good call

girl. We kept a copy of that book in a safe deposit box. The standardprocedure was that somebody new gave half of what they got the firsttime for each number. You’d tell them: ‘Call so-and-so, that’s a fifty-dollar trick.’ They would give you twenty-five dollars. Then thenumber was theirs. My first book, I paid half of each trick to theperson who gave it to me. After that, it was my book. [Terkel 1974]

• Lucky for Cap, Ike was easygoing and soon went away, while theshah—he kept coming back. [Wall Street Journal]

– OP:• I paid X to the individual who gave me the first member of the set of

books I had. (X = half of the amount I earned from the men listed inthe books)

• The Shah, member of the set Ike and the Shah, Xed. (X = keptcoming back).

– But extraction is impossible out of relative clause or subject.


Cross-Linguistic Comparison: Yiddish

– Yiddish has the Poset discourse function for leftdislocations:

• Di daytshishe yidn zenen groyse amoratsim. Es iz farangeveyntlekh lomdim oykh, ober doeds rov zenen amoratsim unapikorsim. Mer nit, vi groys a daytshhisher apikoryres zol nitzayn, hot er zikh zayn tog, ven er geyt davenen. Dos rovdavenen zey yomkiper, nor es iz faran, vos davenen nortishebov, tsi nor sukes…

• German Jews are big ignoramuses. There are generally somescholars also, but the majority are ignoramuses and heretics.Only, no matter how great a German heretic may be, he has hisday when he goes to pray. The majority, they pray on theDay of Atonement, but there are some that pray only on the9th of Av or on the Feast of Tabernacles…

48


– No Simplifying Left-Dislocations are found in Yiddish,where no poset inference is warranted and where theentity evoked by the Left-Dislocated NP is Discourse-new and would canonically occur in a position favoredfor NPs evoking Discourse-old entities.

– Yiddish has a syntactic form, Subject Postposing, thatis very frequent in all registers, and that serves toremove Discourse-new subjects from preverbalposition..

• [Discourse-initial]Es hobn zikh a mol getrofn tsvey yidn in an akhsanye.It have REFL a time found two Jews in an inn‘Two Jews once met in an inn’

• #Tsvey yidn zey hobn zikh a mol getrofn in an akhsanye.Two Jews, they have REFL a time found in an inn.‘Two Jews, they once met in an inn.’


• Git der keyser a kvetsh oyf a glkel, un es loyft arayn Vite.Gives the czar a squeeze on a bell and it runs in Vite‘So the czar rings a bell and Vite runs in.’

• #Git der keyser a kvetsch oyf a glekl, un Vite, er loyft arayn,‘So the czar rings a bell and Vite, he runs in.’

– This suggests that discourse function does not followiconically from linguistic form, but is rather arbitrarilyassociated with it.

– Secondly it supports treating Poset LD and SimplifyingLD as constructional homonymy

– Thirdly, it refutes Keenan’s 1977 notion thatSimplifying LD is somehow not part of the grammar,but is some sort of generalized performance errorprevalent in the speech of children and linguisticallyunsophisticated adults.

49


Hedberg: The Referential Status of Clefts

– It was CLINTON who won cleft + copula + clefted + cleftpronoun constituent clause

– a. It seems to me that you are wrong.b. It is snowing.c It was John that I saw.d. It’s not true.

– a. *This/*that seems to me that you are wrong.b. *This/*that is snowing.c This/that was John that I saw.d. This/that’s not true.


– German: (a) es/?das (b) es/?das, (c)?es/das (d) ?es/das

– French: (a) il/?ce/??cela (b) il/*ce/?ça(c) *il/ce (d) *il/ce

– Russian (a) (*èto) (b) (*èto)(c) (èto) (d) (èto)

– DP DP

D NP Dthis/ this/that/ NP PP that/ the dog next it

door

50


– Givenness Hierarchy

– FOC > ACT > FAM > UID > REF > TIDit that that N the N indef. a N

this this Nthis N

• The man wins this time, and the fish that he selects is a biggoldfish which is, at the point when he selects it, hidden in arocky formation in the tank, and it’s impossible for the manconducting the game to get at the fish with the net.

that fish this fish it


• The fish is swimming around in a large glass bowl on the tableright next to the bird cage. And the scene jumps back andforth between the bird, the fish and the cat, who’s outsideroaming around the streets. The bird and fish seem to beplaying, turning themselves upside down and doing almost akind of dance, the bird spinning around on his perch and thefish swimming upside down and jumping out of the water.

#that water #this water #it

• Karen: You know I’ve tried several times to take, take pictures of the library in Linden Hills. I, sometimes I think it must be haunted or something, because it won’t uh … develop on some film.

Neil: Weird…I took in the=Karen: = Isn’t that supposed to be symbolic of something?.

51


Cleft pronoun + cleft clause as adiscontinuous referring expression

– DP

D CPthis/that/it who won

– Truncated cleft:• A couple of generations ago, wealth was measured in material

like indoor plumbing or a matching set of silverware. When Iwas a kid, it was a television set or a single-family home.

(Ellen Goodman)


– In Focus• My heart beat fast, for I had thought that as the discoverer of the

body I would be the first to be called; but to my surprise,it was Marcel. He stepped forward, neat, dark, debonair…??This/that was Marcel (Mary Fitt, Death & the Pleasant Voices)

??This/that was Marcel who was calledIt was Marcel who was called

– Activated but not In Focus• I wasn’t surprised by the massacre in China. [pause]

This is not Iowa we’re talking about—this is a different society.#It is not Iowa [Eric Sevareid, 12/31/89]

This is not Iowa.?it’s not Iowa we’re talking about.

52


• NF: And then, one morning, about three or four or five mornings before I was due to get out, I was lying in bedand someone, one of, one of my fellow soldiers came byand shook my bed and said, ‘Come on Fredzo, get up’…and the Sergeant himself said, ‘Leave him alone, he’stoo short’.

KF: Hmm.NF: I mean, the, that was the platoon sergeant that said that.

??It was the platoon sergeant this/that was the platoon sergeant this/it was the platoon sergeant that said that

I call that a pretty good guy.


• N: I fly to Michigan on Thursday and come back late on Friday. But I have to leave again on Saturday already for Portland.

Mom: When is that on Thursday that you go?#When is it on Thursday?When is this/that on Thursday?When is this/?it on Thursday that you go?

N: 2:25. (telephone conversation, 2/14/89)

• That’s the French flag you see flying over there, Pierre Dufour,#It’s the French flag??This’s the French flagThat’s the French flag??This is the French flag you see flying over there?It’s the French flag you see flying over there a former legionnaire, pointed out. [New York Times, 5/9/90]

53


– Familiar but not Activated• N: That’s the reason I don’t want to go to Miami!

B: Yeah. Wasn’t that somewhere in Southern Florida where #this it

they thought those people got AIDS from bug bites—gettingbit a hundred times a night or something, because the placewas so roach infested? (breakfast conversation, 2/89)

• K: [answering phone] LinguisticsN: Hi!K: Hi!N: Was that a ‘sign’ or a ‘plaque’ that was on that bridge?

#this itK: Wow! Uh … a ‘sign’ or a ‘plaque’… I would say it was

…a sign. That a good question. I’d say it’s a plaque cause a sign is too ambiguous. (6/8/89)


– Uniquely Identifiable but not Familiar• [Beginning of a newspaper article] It was just about 50 years

#This #That

ago that Henry Ford gave us the weekend. On September 25,1926, in a somewhat shocking move for that time, he decided toestablish a 40-hour work week, giving his employees two days offinstead of one. (Philadelphia Bulletin, Prince 1978)

• The federal government is dealing with AIDS as if the virus was aproblem that didn’t travel along interstate highways and was noneof its business. It’s this lethal national inertia in the

#This #That

face of the most devastating epidemic of the late 20th century thatfinally prompted one congressman to strike out on his own. Gerry Studds, D-Mass, has begun to treat his constituents… (Ellen Goodman, 5/25/87)

54


Existential and ExhaustivenessConditions

– The King of France is bald.There is one and only one King of France, and he is bald.∃x (K-of-F(x) &∀y (K-of-F(y) ⊃ y = x)&bald(x) )

– It was Clinton who won.Someone won, and only one person won, and it wasClinton. ∃x (won(x) &∀y (won(y) ⊃ y = x) & x = Clinton)


• The present queen of France lives in Ithaca.The present queen of France does not live in Ithaca.Does the present queen of France live in Ithaca?If the present queen of France lives in Ithaca, she has probably

met Nelly.There is a unique present queen of France.

• It was Lee who got a perfect score on the semantics quiz.It wasn’t Lee who got a perfect score on the semantics quiz.Was it Lee who got a perfect score on the semantics quiz?If it was Lee who got a perfect score on the semantics quiz,

why does she look so depressed?Someone got a perfect score on the semantics quiz.

55


– É. Kiss (1998): a functional head F (“Focus”) assignsidentificational focus to the clefted constituent in itsspecifier:

• The function of identificational focus: An identificational focusrepresents a subset of the set of contextually or situationallygiven elements for which the predicate phrase can potentiallyhold; it is identified as the exhaustive subset of this set forwhich the predicate phrase actually holds.

• It was a hat and a coat that Mary picked for herself.It as a hat that Mary picked for herself.

• Mary picked a hat and a coat for herself.Mary picked a hat for herself.

– Hedberg: Identificational focus (exhaustivenesscondition) results from disguised definite description inclefts.


Two Basic Syntactic Analyses ofClefts

• Extraposition analysis– Jespersen 1927 (cleft clause is “subject clause” that has

been “transposed” to the end of the sentence):• S* V P 2* (S2

c V)it* was CLINTON *who won.

– Traditional grammarians: Fowler & Fowler 1919,Curme 1931.

– Generative grammarians: Akmajian 1970 & Emonds1976 (Cleft extraposition), Gundel 1977 (rightdislocation), Wirth 1978 extraposition from NP).

56


– Akmajian 1970:

S

NP V NP S’*

it * was Clinton whoi S

ti won


• Expletive analysis– Jespersen 1937 (Cleft pronoun, copula, and relative

pronoun are ‘outside the sentence proper’, he proposes totreat the rest of the sentence ‘as if there had been nointercalation’)

• [sv] S [sc] V[It was] CLINTON [who] won.

– Traditional grammarians: Poutsma 1916, Kruisinga1932.

– Generative grammarians: Chomsky 1977, Williams1980, Delin 1989, Delahunty 1982, Rochemont 1986,Heggie 1988, É. Kiss 1998.

57


– Chomsky 1977

S

NP VP

it V S’’

was TOPi S’i

NP whoi S

CLINTON ti won


– É. Kiss 1998IP

NP I’

it I FP

wask CLINTONi F’

F CP

tk whoi C’

C IP

ti won

58


A New Analysis of Clefts

IP

DP I’

D I VP

it wask V DP

tk DP CP

CLINTON whoj C’

tj won


– It could be Clinton who won.

– The function of Identificational Focus: An identificationalfocus represents a subset of the set of contextually orsituationally given elements for which the predicate phrasecan potentially hold; it is identified as the exhaustive subsetof this set for which the predicate phrase actually hold. (É.Kiss)

– The function of Information Focus: New or nonpresupposedinformation marked by one or more pitch accents. (É. Kiss)

– Comment-Topic Cleft (Hedberg 1990):• M: I want to ask this question: Why is this agreement so bad? I

ask you.G: Because our whole intention was to bring some form of

democracy there; our intention was to make the Sandinistas cryuncle. It is the CONTRAS who have cried uncle.

(McLaughlin Group, 3/25/88)

59


– Topic-Comment Cleft:• The Member for Hertfordshire North East despite his fascist

tendencies, is a notable liberal when it comes to women’s rights.But perhaps women should beware; proximity to this elegantbaronet can be lethal. His first wife was killed in a car accident; hewas driving. Theresa Nolan, who nursed his mother and slept inhis house, killed herself after an abortion. It was he who knewwhere to find the BODY. (P.D. James, A Taste for Death)

– All-Comment Cleft:• It was the Greeks who first made wine around 1500 B.C. They

then took this unique art to all the corners of the ancient world,including Italy, Spain, Russia, and, in about 600 BC, France.(Liquor menu from “It’s Greek to Me’ restaurant, Minneapolis)


– Heggie (1988): cleft clauses are similar to nonrestrictiverelative clauses, not relative clauses:

• It was John that Mary hit.John, who Mary hit…*John who Mary hit…

– Delahunty (1982): cleft clause and clefted constituent actas a constituent:

• I said that it should have been Bill who negotiated the newcontract, and it should have been.

• It could have been—and it should have been—Bill whonegotiated the new contract.

• It must have been, in my opinion, the cyanide that did it.

• It must have been Fred that kissed Mary but Bill that left withher.

• ?I said that it was Bill that argued the case and Bill that arguedthe case it was.

60


– Predicational cleft (as opposed to identificational cleft):• It was an odd televised ceremony that I watched from my living room,

and a touching one … (E. Goodman)

• The televised ceremony that I watched from my living room was an oddone.

– Identificational th-clefts:• It’s the contras who have cried uncle.

• This is Ford and Kissinger we’re dealing with, not two boy scouts.(Ball 1978)

• To the worried motel owner, I said, ‘I know the girl. From the sound ofit. That was her mother and her aunt who came after her. (NancyPickard, Bum Steer)

– Predicational th-clefts:• They’re just fanatics who are holding him. (McLaughlin Group, 3/27/87)

• These are not just notes that you’re playing—they’re phrases (TheWitches of Eastwick)

• Those are real eyeglasses that Mickey is wearing. (Ball 1977)

Green: Pragmatics & Syntax

Documents