Dennis Perzanowski Naval Research Laboratory Navy Center for Applied Research in Artificial Intelligence Code 5512 Washington, DC 20375-5337 voice: (202) 767-9005 fax: (202) 767-3172 email: dennisp@aic.nrl.navy.mil John Gurney Army Research Laboratory AMSRL-MO-S 2800 Powder Mill Rd. Adelphi, MD 20783 voice: (301) 394-3920 email: gurney@adelphi-msc.army.mil The Message in the Medium: On the functionality of it-clefts in selected discourses Dennis Perzanowski and John Gurney 1. Abstract Information in a discourse can be obtained by analyzing several different linguistic properties of the discourse. Various syntactic and semantic triggers provide clues to the complex informational structure of a discourse. For the purposes of this investigation, we selected several discourses taken from a series of wire service messages dealing with terrorist incidents that occurred in Central America from 1989 to 1991. The corpus is known as the "MUC-3" corpus. Because of their inferential properties, we identified it-cleft constructions in the corpus. We argue that, despite their rarity in the corpus under investigation, it-clefts provide additional information, and do not just contrast or emphasize focused linguistic material, such as noun or prepositional phrases. By substituting so-called "normal" word order, or SVO paraphrases, for the it-clefts in the messages, we determined what information the it-cleft sentences provided in the discourse. Our investigation reveals that, as a subset of the information that can be obtained in a discourse, it-clefts can be used to avoid conflicting or differing interpretations inherent in their SVO paraphrases, thereby minimizing possible confusion regarding the interpretation of what the author is trying to communicate. In some cases the speaker/writer's point of view or attitude about the subject matter being discussed is also revealed. For example, we show how a focused generic noun phrase in an it-cleft sentence provides a clearer statement of the author's intended meaning. 2. Introduction In "The Pragmatics of It-clefts and WH-clefts," (Declerck 1984) observed that a speaker/writer's1 use of a specificational cleft sentence over a predicational sentence can be construed as the speaker/writer's desire to produce "an unambiguously specificational sentence." Such a desire in writing may stem from the fact that certain phonological features, such as stress and intonation, are unavailable2 on the printed page, thereby causing the author to choose a syntactic construction that forces the desired specificational interpretation. This interpretation can be achieved in certain types of cleft constructions, such as it-clefts. Specificational clefts, herein defined as it-cleft sentences such as (1b), like predicational sentences as in (1c), can answer the question embodied in (1a). (1)(a) Who murdered Fred? (b) It was the terrorists who murdered Fred. (c) The terrorists murdered Fred. However, as Declerck notes, "using the former type of structure [i.e. (1b)] enables the speaker[/writer] to give additional information along with the message." In his now historic work, The Medium is the Message (McLuhan 1967), argued that meaning, the idea--the message--was not merely carried by the medium used to convey a message, but the message was the very medium, in McLuhan's case television, carrying the message. Carried over into the syntactic realm, then, what this means for linguists is that the constructions employed to convey meaning can be meaning-bearing in themselves. Thus, something of the meaning is embodied in the very syntactic constructions employed to convey meaning. More specifically, as Declerck argues, certain constructions, such as it-clefts and wh-clefts, which will not concern us here, can convey meaning, additional information, other than what is being "talked about" in a discourse. Declerck (1984) enumerates what some of these meanings might be; such as, contrastiveness, emphasis and exhaustiveness, that would lead a speaker/writer to prefer a specificational cleft over a predicational sentence. In the following paper, we will attempt to show what kind or kinds of additional information are provided by analyzing certain texts containing it-clefts. We will not be presenting some unified theory of it-clefts, nor will we be presenting some singular function of it-clefts in discourse. In fact, if anything, we believe we show that the subject of it-clefts in discourse is a much more complex area than has heretofore been indicated. Instead of narrowing down this construction to one or two "types," we find that it-clefts serve various functions in a discourse, and we have only scratched the surface. Furthermore, we offer no explanatory discussion for why it-clefts function as they do in discourse. We believe that this is a far more complex problem than can be discussed here. We merely wish to open the area for future discussion with the hopes that an adequate explanatory theory of the function of it-clefts in discourse will be forthcoming. For the purposes of this study, we take for granted or assume the world knowledge necessary to fully comprehend and understand texts. For the purposes of this study, we assume that such information as the geography of Central America is generally known to the readers of the wire service messages that constituted the corpus under investigation. Likewise, the identities of current events and figures are also assumed as known by the readers and such information constitutes world knowledge, or a larger context, in which these messages are read. In fact, this type of assumption can produce some interesting results which are immediately subject to further testing. If, for example, some of our conclusions based rather exclusively on textual analysis can be falsified by some fact of world knowledge, then we know that our claims and arguments are incorrect. On the other hand, if our conclusions about the data, based largely on linguistic evidence, prove correct in terms of world facts, then we maintain a stronger position in terms of this analysis. However, we do not claim to have discovered some exhaustive procedure on what additional information results from a use of specificational constructions, namely it-clefts, in a discourse. Further, for simplicity's sake, we have limited ourselves to it-clefts. For the purposes of this paper, we have selected a corpus of written text, the so-called "MUC-3" text corpus.3 This corpus was used by the Navy and the Message Understanding Conference (MUC-3) to extract information from texts. While our purpose in using this data is entirely different from theirs, the corpus still lends itself to research of the type we are about. The MUC researchers used the data to produce templates containing factual information extracted from the texts. These templates contained such information as the type of incident that was being reported about or spoken of in the message, such as a bombing of an embassy or a railway station. The templates also included information about perpetrators of these incidents and the human targets of terrorist activities discussed in the messages. The purpose of the MUC task was to glean factual information from the data presented in the message, using natural language processing techniques, so that a database could be compiled based on these analyses. Our purpose in using the MUC-3 data is to extract information as well; however, the information we extract is not contained explicitly in the dialogs or discourses. For, while a sentence like (1b or 1c) may fill a slot in a template, with explicit information from the text, such as the perpetrator or agent of an act, there may also be information about the speaker/writer's intent or point of view, for example, that can come from the use of a specific syntactic construction. In this paper, we discuss some of the implicit information revealed about the discourse through the use of a particular syntactic construction, so-called "it-clefts," in selected discourses. While natural language techniques have been used to extract information into a database for information retrieval purposes, to our knowledge no work has been done on extracting information from specific syntactic constructions. We wanted to test Declerck's idea that certain constructions, such as it-clefts, provide additional information, and to see in a real corpus what, if any, additional information could be gleaned. By concentrating on the it-cleft construction, we hoped to obtain information on presuppositions in the messages. In our preliminary work for this investigation, we had originally been interested in studying it-clefts because they are thought to be syntactic triggers for presupposition (Gazdar 1979). We were interested in tracking presuppositions in a discourse. However, as we studied the MUC-3 corpus, our interests shifted to other matters, since we saw other phenomena resulting from the use of it-clefts in discourses. Therefore, while this study produces little, if anything, of importance in the realm of presupposition, we believe that we show how the use of the it-cleft in these messages does provide specific additional information such as disambiguation of speaker/writer's attitude, specifically in showing speaker's point of view, ameliorating the tone of the discourse, and in one case forces a generic reading on a focused noun phrase. Our chief method in showing how it-clefts function in discourse will be to contrast the it-cleft in context with the simple declarative paraphrase ("normal" or SVO word order4) of the sentence. 3. The Corpus The entire corpus for this study consists of 1300 messages gleaned from wire services during 1990. These messages are groups of sentences obtained "via a keyword query to an electronic database containing articles in message format from open sources worldwide. These articles had been gathered, translated (if necessary), edited, and disseminated by the Foreign Broadcast Information Service (FBIS) of the US Government" (Sundheim 1991). The corpus mostly consists of newspaper articles, TV and radio news, speech and interview transcripts, rebel communiquŽs, etc. The messages concentrate on terrorist activities occurring in Central and South America in 1989-90, several of which deal specifically with the murder of six Jesuit priests. For our study, the first 100 messages, consisting of approximately 6405 sentences, were analyzed for it-clefts. We uncovered 12 occurrences of "it-clefts" in this body of data.5 Clearly, the it-cleft construction (approximately .19% in occurrence) is a rarely used syntactic construction, at least in the corpus we studied. One might then simply argue that such a construction having such a minimal distribution is of no great importance. But an equally persuasive argument can be made that its rarity points to some special usage, so that when it is used--sparingly--it signals something specific. Thus, its scarcity might rather be interpreted as a major vehicle for indicating something noteworthy in a discourse. In searching through the MUC-3 corpus of terrorist messages for all occurrences of it, we of course also uncovered all anaphoric and pleonastic occurrences of the word. We went through the resulting list by hand and removed any sentence containing an anaphoric it. Thus, sentences like (2b) containing anaphoric it were discarded. (2)(a) The terrorist pointed the gun. (b) It was the president's own gun. Next, we searched through the remaining sentences and further removed sentences containing pleonastic instances of the pronoun it, as in (3). (3) It was easy to defuse the bomb. The remaining sentences, we hoped, were the so-called "it-cleft" sentences. 4. Preliminary Discussion For the purposes of this paper, it-cleft sentences can be categorized syntactically as in (4). (4) it (modal) (not) (perf) (prog) be {NP/PP}6 (that/who/when/¯) Sentence.7 The first constituent is the word "it"8 giving the construction its characteristic name. The second constituent is some form of a modal, followed optionally by the negative. The fourth and fifth constituents may be aspectual forms of the verb which occurs in sixth position. The seventh constituent is either a noun phrase or a prepositional phrase. The eighth constituent may be the complementizer "that," one of the relative pronouns "who" or "when," or the null operator "¯" (Heggie 1993). The final or ninth constituent is a "Sentence," grammatically defined as some syntactic subject followed by a verbal complement. Thus, (5a) through (5d) contain examples of sentences that qualify as it-cleft sentences for the purposes of this investigation. (5)(a) It was (not) the terrorist who shot the president. (b) It wasn't the terrorist who shot the president. (c) It was in 1989 (that) the president was shot. (d) It was the president the terrorist shot. (e) It was the one that you saw yesterday. (f) It was the man on the corner that finally shot the president. Sentences like those in (5e) and (5f) can be easily confused with other constructions that look on the surface very much like it-clefts. In (5e) the NP "the one" might be syntactically analyzed as modified by the relative clause "that you saw yesterday." The NP in (5f), "the man on the corner," may be modified by the relative clause "that finally shot the president." To some extent, in an aural discourse, one in which the intonational patterns of the speakers are available, disambiguating it-clefts from other types of seemingly similar constructions can be a simple enough matter. The sentences (5e) and (5f) can easily be transformed into it-cleft sentences simply by stressing the NPs in the verbal complement position. However, in printed material, such as in the wire services we investigated, these sound patterns are not available. Distinguishing it-clefts of the form (4) from sentences containing NPs modified by relative clauses of the form (4) is not an easy matter. However, their differences are noted in the literature, as in (Quirk et al. 1985), and such litmus tests as focal differences, cataphoric9 use of the initial "it" in it-cleft sentences, as well as the rarity of certain wh-words in it-cleft sentences, were used in this investigation to consider only it-clefts in the MUC-3 corpus. However, the difficulty of noting such things as focal differences in the MUC-3 corpus is an item to which we now turn. In aural communications, clarifications can be obtained by means of repetition, tonal emphasis, restatement, and the like. In writing, typographical cues, such as the use of italics or emboldened print, help to clarify, as for example in (6a) and (6b) where italics or emboldened print are used to indicate emphatic and/or contrastive stress in speech. (6)(a) The innocent will pay the price. (b) The innocent will pay the price. In (6a) and (6b) it is not clear whether the italics or emboldened words are to be understood as emphatic or contrastive elements in a discourse. Context is needed to determine which interpretation to give the cued expressions. Thus, for example, when (6a) or (6b) are uttered in a speech, a speaker could be emphasizing the fact that the innocent will pay the price with no overt mention having been made of any guilty parties and whether or not they will suffer. The speaker may merely be emphasizing the fact that innocent people will be involved. Thus, (6a) and (6b) might be given an emphatic reading. On the other hand, (6a) and (6b) could be read with contrastive stress placed on the NP "the innocent." In this case, the speaker may be stating that contrary to what might be expected, namely that the guilty will pay the price, the innocent will suffer. In texts, such typographical cues as italics and emboldened print can be used, but the issue of which interpretation is to be assigned to these cues is still moot. Only context can dictate which is the correct or appropriate interpretation to assign these sentences. In the texts that we examined for this study, the authors did not employ any of the typographical devices, such as those in (6), and tonal emphasis in direct or indirect speech was unavailable to us, since there are no aural tapes of the MUC-3 corpus to our knowledge. Therefore, no phonological information could help us differentiate or disambiguate any of the messages, and typographical cues simply were nonexistent.10 Thus, we had to rely on other elements in the discourse, to which we turn shortly, to determine whether the stress was contrastive or emphatic, and whether the sentence under consideration was truly an it-cleft or one containing an element modified by a restrictive relative clause. Since we assumed that the authors of these wire service messages intended them to be fully comprehensible as written material, we wondered what their function was and what other method(s) could be used to clarify such things as speaker/writer's attitude in written texts. We noticed that it-clefts, despite the fact that they were not marked phonologically or typologically in the messages, aside from their syntactic forms, emerged as prime candidates for indicating where the listener/reader's attention was being drawn for a particular purpose (Jones and Jones, 1985) or focused by the author of the message. However, drawing focus does not seem to be the only use of it-clefts in a discourse. We assumed that a reader of these messages, with the world knowledge necessary to understand any references that might be made in them and with a grasp of English, can interpret the sentences so that the phonological stress patterns characteristic of an it-cleft emerge. But why and how a reader knows that a sentence of the pattern in (4) is to be read/interpreted as either an it-cleft with some kind of characteristic stress pattern versus a sentence containing a noun phrase or prepositional phrase modified by a restrictive relative clause remained somewhat of a mystery. We believe we have found a partial answer to this. In the following discussion we will also discuss some of the ways in which it-clefts can be identified and interpreted. Certain types of it-clefts,11 namely those that exhibit contrastive phonological stress, frequently occur in negative contexts in the discourse that we investigated. To our knowledge, researchers have not mentioned this fact. We noted that it-clefts receiving contrastive stress are no doubt acting contrastively because of the negative environment that is set up around them. This type of it-cleft, the contrastive focus it-cleft, is being used in the entire negative context as a final marker to signal the contrast. Thus, some negative or contradictory marker precedes the it-cleft in a discourse, and both the negative marker and the cleft emphasize the non-negotiability of the correction as it is stated in the proposition of the cleft (Delin 1992). In the following we will discuss how certain contexts in the discourse facilitate the interpretation of (4) as it-cleft sentences. Ultimately, the purpose for using an it-cleft in any context, however, rests on how well, if at all, the author's message is comprehended. Therefore, we will show that once an it-cleft is identified, its use clearly facilitates an interpretation which might be otherwise missed or more difficult to obtain through some other syntactic device. For the purposes of this study, we contrast the it-cleft with so-called "normal" word order in order to determine the it-cleft's usage in the discourse. 5. Discussion of the Corpus In Message 0400 [Appendix A],12 (7a) exhibits one characteristic of it-clefts that has been noted in the literature (Jones and Jones 1985), namely that it focuses the reader's attention clearly and specifically on the main idea of the passage. (7)(a) Sanguinetti added that it was "with pain and with indignation" that he received the news of the death of the Colombian senator, "murdered by criminals at the service of spurious interests." (b) Sanguinetti added that he received the news of the death of the Colombia senator "murdered by criminals at the service of spurious interests" "with pain and with indignation." (c) Sanguinetti added that "with pain and with indignation" he received the news of the death of the Colombia senator "murdered by criminals at the service of spurious interests." (7a) is written in what is known as indirect address; i.e. it is a paraphrase by the reporter, but key phrases, such as "with pain and with indignation," are directly quoted. The reporter might have written (7b) or (7c) or some such paraphrase, which are written in so-called "normal" or SVO word order. Instead, the reporter chose the it-cleft.13 We must ask ourselves, then, why the reporter deliberately chose this syntactic construction over another. As we see in the paraphrases (7b) and (7c), the SVO word order actually obfuscates the main point; namely that the senator was killed by individuals with ulterior motives. However, the it-cleft still maintains focus on the phrase "with pain and with indignation," while allowing the phrase "murdered by criminals at the service of spurious interests" to occupy the final position in the sentence; the main point remains clear. The functional load of the it-cleft sentence therefore carries the weight of clarity, while the supposedly simpler word order does not bear up and obscures the main idea. Curiously, the SVO word order of (7c) also permits the phrase "murdered by criminals at the service of spurious interests" to occur in sentence final position as in the cleft version (7a). However, the main point, which seems to be that Sanguinetti received the news of certain events in a particular way, is not as clearly stated in (7b or c) as it is in (7a). It seems that the it-cleft sentence (7a) is the clearest expression of the author's intention to portray the feeling or mood with which Sanguinetti reacted to the events depicted. Thus, the it-cleft (7a) offers the clearest statement of the main idea of any of the possibilities considered. Aside from its use as a syntactic device to focus on the main idea of a discourse, it is also our contention that the speaker/writer may use the it-cleft as a kind of discourse marker to contrast some untenable position or contradictory idea in a discourse. We discovered that the it-cleft was used for clarifying misconceptions, specifically where the speaker/writer's point of view might be misunderstood. As we argue in this study, such misconceptions may arise from ambiguities about particular points in the discourse or even from misconceptions about the speaker/writer's point of view, but they are clarified or resolved through the use of the it-cleft. By substituting "normal" or SVO word order paraphrases for it-cleft sentences in the MUC-3 data, we show that the anaphoric function of the it-cleft is totally lost. Therefore, one of its functions in the discourse, which is crucial to understanding, is lacking. If misconceptions, misunderstandings, or ambiguities arise or are permitted to remain in a discourse, then the speaker/writer's main idea is misunderstood or lost. We turn now to some additional examples of it-clefts from the MUC-3 corpus of messages. Our next example of an it-cleft sentence (8) is taken from Message 0022 [Appendix B], which is the written transcript of an interview between the Central American head of state, Ruben Zamora and correspondent Alessandro Oppes (8). (8) But in the summit's final document, it is the FMLN and the Nicaraguan Contras who pay the price for the agreement among the presidents: these are diplomatic games that do not facilitate the restoration of peace in Central America. For the purposes of this discussion, we will ignore to a large extent the remainder of the sentence from the colon to the end, since it does not affect our discussion of the it-cleft. As we stated above, the purpose of it-clefts in discourse can be understood in the context in which they are embedded; therefore, a brief review of that context is necessary here. In Message 0022, an international incident is referred to: the slaying of six Jesuit priests. In conjunction with this slaying, five low-ranking officers were arrested, and Salvadorian President Cristiani is cited as the person responsible for selecting these officials in order to end what might have become an international incident. The correspondent Oppes then asks the Central American leader, Zamora, his opinion of this action. Zamora states that while Cristiani's action was an important first step, it is insufficient unless those higher up in the government, those in positions of power and are therefore responsible, are singled out. He even cites a previous incident, the murder of Monsignor Romero, which went without any government sanction, other than the military's comment that it was an "isolated action 'divorced from any institutional policy.'" Zamora notes that such a conclusion cannot be supported historically. At this point in the interview, Oppes interrupts Zamora by stating that a recent summit of Central American leaders ended with an exoneration of President Cristiani who "secured an explicit condemnation" of guerrilla activity, and by doing so, President Cristiani appears to have been working on securing a condemnation of such activity, but no human rights violation was leveled at Christina's government. Zamora states that Christiani's condemnation of the government's activity is in a sense positive, but in the it-cleft of (8), he says that the FMLN (the Farabundo Marti National Liberation Front) and the Nicaraguan Contras "pay the price" for the summit's actions. As a way of clarifying what "paying the price" means, Zamora finishes this statement (8) by stating that the summit's actions are really "diplomatic games." We must now ask ourselves what purpose the it-cleft in (8) serves. It-clefts, which result in contrastive stress, frequently occur in either/or contexts, or in contexts in which some position held previously in the discourse is being contradicted. Consider (8) in the context of certain questions (9) and (10), which might be raised to create these different contexts just mentioned. (9) Speaker 1: Who's going to pay for this action? The government or the FMLN and Contras? Speaker 2: It's the FMLN and the Contras who will pay for this action. (10) Speaker 1: The government is the one who should pay for this action. or Speaker 1: Should the government be the one to pay for this activity? Speaker 2: It's the FMLN and the Contras who will pay for this action. Judging from the context of Message 0022, it seems that Oppes is not offering Zamora a choice, as in the either/or context of (9). Instead, as Oppes seems to be implying in (11), taken from the paragraph immediately preceding (8) in Message 0022 [Appendix B], the government is not being blamed for some act of omission. (11) Yet last month the summit of Central American heads of state in San Jose ended with a success for President Cristiani, who secured an explicit condemnation of the offensive launched by the guerrillas in November, whereas no specific charges were leveled at the government for human rights violations. So, in some sense, Oppes in taking a position very similar to Speaker 1 in (10). In this dialog, the position is implicitly stated in (10), but what Speaker 1 explicitly states in (10), Oppes implies in (11). He states that "no specific chargers were leveled" against the government. Therefore, the it-cleft of (8) is to be interpreted in some sense as a retort of the previous proposition implied in the dialog. However, we must ask ourselves if the use of the it-cleft in (8) conveys any more information in its usage than if the statement had been made differently, namely as a simple declarative sentence, for example, as in (12). (12) But in the summit's final document, the FMLN and the Nicaraguan Contras pay the price for the agreement among the presidents: these are diplomatic games that do not facilitate the restoration of peace in Central America. In (12), so-called "normal" word order is followed; namely SVO word order for English where, very simply, the subject of the sentence is "the FMLN and the Nicaraguan Contras"; the verb is "pay"; and the object is "the price." One might ask what difference, if any, exists between (8) and (12). Stated thusly, as in (12), Zamora's statement sounds so much more a matter-of-fact statement in the context of the entire passage. Whereas, in (8), there seems to be a more emotional undertow to the statement. Of course, this may result from the intonational contour that is characteristic of an it-cleft, whether they are cued on the printed page or not. However, note further that in (8) there also seems to be an underlying idea that Zamora is somewhat or somehow siding with Oppes and those who think the summit may have fallen short of their purpose in condemning a government action. In (12), this tone is not so clearly derivable, and as such is ambiguous between a matter-of-fact reading and the one in which Zamora and Oppes share similar ideas, namely they are in some sense condemning the actions of the meeting of the presidents. Thus, we see that the use of the it-cleft in a discourse may indicate the speaker/writer's attitude underpinning a proposition. This finding seems promising, but the way in which certain attitudes and it-clefts, among other linguistic elements, can be mapped into a coherent theory of discourse, is yet to be worked out. Furthermore, the specific speaker/writer's attitude discussed here is context dependent. Additional information was needed to determine the specific attitude of the speaker/writer. In another context, an it-cleft might be used to exonerate rather than condemn. Therefore, very simply, the point being made here is not that the it-cleft is being used to condemn or exonerate any proposition stated or implied in the discourse: the it-cleft is being used as a device to reveal a speaker/writer's attitude. Additional information from the context will determine what kind of attitude is being revealed. We turn next to a consideration of contrast and the interpretation of it-cleft constructions in a discourse. In the remaining part of this discussion, we will be deriving these interpretations of the text based largely on the substitution of "normal" word order, or SVO word order, paraphrases of it-cleft constructions in the various discourses. In the standard interpretation of an it-cleft, some constituent phrase in a discourse is being contrasted, emphasized, or is being given more prominent focus in the discourse.14 For example, as we see in Message 0213 [Appendix C], the it-cleft (13) emphasizes and singles out the perpetrator of certain acts through the repetition of the noun phrase "Shining Path," and thereby focuses on this noun phrase in the focus position of the it-cleft. (13) If the Shining Path is operating in Bolivia, it should announce that it is in fact Shining Path that is committing these crimes. In Message 104 [Appendix D], sentences (14a) and (14b) below, contrast is used to focus on two constituents, "the majority [of armed forces members]" and "the minority [...]," the latter of which is placed in focus position in the it-cleft (14b). (14)(a) We think this indicates [words indistinct] behind the backs of the majority of armed forces members. (b) It is a minority who have decided on these provocations against the popular movement. Despite the garbled message in (14a), the noun phrase "the minority" in (14b) is being contrasted with the phrase "the majority" in the immediately preceding sentence. Our examples thus far have been it-cleft constructions in which the focused element is a noun phrase. Let us turn now to an example or two that contain prepositional phrases. Another example of an it-cleft that is used to call attention to the contrast between certain phrases in the sentence is (15) from Message 0430 [Appendix E]. (15) During a radio program on Monday, 4 September, journalist Jose Vicente Rangel, former movement toward socialism presidential candidate, maintained that it is not at this time that drug traffickers are attempting to enter Venezuela because they entered a long time ago. In (15), the phrases "at this time" and "a long time ago" are being contrasted. This effect is clearly achieved through the use of the it-cleft, since it creates focus on the phrase in the syntactic position of the complement of the it-cleft construction and contrastive stress is achieved. Message 0064 [Appendix F] ends with an it-cleft (16). (16) The communique ends by stating: "It is within that framework that the Argentine government is confronting extremist actions, and we therefore wish to express our recognition." Message 0064 is a wire service release in text form and ends with a direct quote from a communiquŽ that is being summarized. The it-cleft in this sentence is used to conclude the message, as it did the communiquŽ that this message directly quotes. (16) reiterates the main idea of the communiquŽ, namely that certain political parties are acting legally and in a democratic fashion. The use of the it-cleft in (16) repeats a key phrase from a previous sentence (17), which is also a direct quote from a previous communiquŽ. (17) "They reassert their profound conviction that democratic societies may effectively defend themselves from the scourge of terrorism within the framework of the law and the exercise of political pluralism." By employing an it-cleft, the phrase "within the/that framework" receives prominence in the message. Both the author of Message 0064 and the author of the original communiquŽ are stressing the importance of the point that the political parties are acting in a particular manner which is to be sanctioned. This framework is mapped out in (17) and stressed in the concluding sentence of the communiquŽ (16). Thus, in this instance of an it-cleft sentence, a phrase embodying the main point of a discourse gets phonological stress because of its syntactic position in an it-cleft construction, namely the focused constituent. Granted, (16) could have been written in SVO word order. However, without typographical cues, the reader might not be able to determine that it is specifically the phrase "within that framework" that is being emphasized. The positioning of the phrase "within that framework" in the position of prominence in an it-cleft clearly emphasizes the phrase, and once again indicates something about the author's attitude about the subject matter under discussion. Such analyses as in (13) through (17) are typical of the past work dealing with it-clefts. Much of the previous research on it-clefts has merely focused on this one particular aspect of it-clefts, contrast, and simply left the matter at that. Of course, other researchers, such as Prince (1978), Declerck (1984), and Delin (1992), have gone beyond this, but even their work stops short at considering some related issues in discourse analysis. We do not fault them in this regard. Their collective works contribute to certain ways in which it-clefts contribute to the informational structure of a discourse. We turn now to additional ways in which it-clefts contribute to the information structure of a discourse. Contrasting elements in it-cleft constructions can assist in determining the author's point of view, as in Message 0195 [Appendix G]. In (18a) the author of the text uses the it-cleft to emphatically state his/her position concerning the establishment of guilt in a particular situation. (18)(a) "That debt was paid a long time ago, and we must ask whether it is the developed capitalist countries that owe us because of their looting of our natural resources." (b) "That debt was paid a long time ago, and we must ask whether the developed capitalist countries owe us because of their looting of our natural resources." The author's condemnation of the United States is emphasized by the use of the it-cleft in questioning whether or not the more advanced countries of the world are not taking advantage of third world countries. The use of the it-cleft clearly places phonological emphasis on "developed capitalist countries" to be contrasted with "us" in the same sentence.15 The implication here is that the speaker believes that the developed capitalist countries of the world are indeed guilty of taking advantage of the lesser developed countries of the world by robbing these lesser developed countries of their own natural resources. The it-cleft seems to embody a slightly more accusatory tone than its SVO counterpart (16b). The question becomes much more rhetorical in light of the use of the it-cleft. This may be the result of the fact that with it-clefts readers know where to assign stress, given the syntactic pattern of it-clefts. This kind of stress might easily lend itself to being construed as accusatory in tone. With the SVO word order pattern, assigning an accusatory tone to an appropriate word or phrase is not so clear cut. Our point here is not that it cannot be done. Surely it can. Our point is that it is more difficult to assign the necessary stress pattern to achieve the accusatory tone in (18b) versus (18a). Similarly, in Message 0027 [Appendix H], we see another instance of an it-cleft (19). (19) Merino said that the FMLN [Farabundo Marti National Liberation Front] should not claim credit for the resumption of a dialogue because it was President Alfredo Cristiani who took the initiative in resuming the talks. With only two previous sentences in the discourse, a fairly small context is set up. In the first two sentences of the report, the narrator, Nancy Perla, establishes that the possible resumption of talks between the Salvadoran government and a group of rebels has been called a positive act by Vice President Francisco Merino. But she further reports in (19) that the FMLN should not claim credit because President Cristiani took the initiative in resuming the talks. We claim that the reporter's use of the it-cleft in (19) forces an exclusivity or uniqueness interpretation (cf. (Gurney and Perzanowski, forthcoming)) which might be lost if normal word order had been used. Thus, (20), the SVO paraphrase, contains an inherent ambiguity. (20) Merino said that the FMLN should not claim credit for the resumption of a dialogue because President Alfredo Cristiani took the initiative in resuming the talks. Given SVO paraphrase in (20), let us ask ourselves the questions in (21). (21) True or False? Did both the FMLN and Cristiani initiate the resumption of talks? It seems that the answer to these questions is debatable. Certainly, given (20), Cristiani did initiate the action, but the FMLN is also claiming this distinction. However, given the it-cleft in (19), a different set of consequences results. Again, asking ourselves the questions of (21), we determine that Christiani did initiate the action, but we might infer that the FMLN is falsely claiming this distinction as well, since the it-cleft evokes an exclusive reading of who actually initiated the action. Message 0034 [Appendix I] poses a problem in interpretation because of the incomplete transmission of the message. In Message 0034, we have two sentences containing it-clefts (22) and (23). (22) [Words indistinct], as Cristiani said, that it was Col Benavides, but it was not Col Benavides who issued the order. The context set up in Message 0034 is that Cristiani has announced the perpetrators involved in the killing of the Jesuit priests and their maids. Col. Benavides has been involved because the crime was committed in the sector under his command, and the accusation is that everyone is directly involved in the crime, except, given the incomplete information of the message, Col. Benavides. Because part of the message is indistinct, we ignore the first it-cleft of (22). It is not altogether clear what is being attributed to Col. Benavides by Cristiani in the first part of the sentence. However, because the proposition is complete in the second part of (22), we can interpret the it-cleft of (22). Therefore, we turn now to determining what purpose the it-cleft serves in this discourse. The it-cleft in (22) can be interpretable as (23). (23) Col. Benavides did not issue the order. Because of the incomplete transmission in (22) we can only surmise what possible use the it-cleft might have in this sentence. Clearly, some parallel structure is being maintained in (22). If normal word order, as in (23), is substituted for the second conjunct in (22), we still cannot interpret the first conjunct of (22). We must then ask ourselves why Col. Benavides' involvement is stated as an it-cleft, as in (22), rather than as a simple declarative sentence in (23). Might this be some sort of reportorial style? In other words, had the commentator of this message stated it as (23), the commentator might subsequently be quoted as saying that Col. Benavides was innocent of a particular act, an idea which is basically borne out by the remainder of the message. However, in the context of the entire message, this interpretation is presented as if it is almost hearsay. Furthermore, considering the very next sentence of the same message (24a), the normal word order response (24b), seems strange compared to the cleft response (24c) to the question posed in (24a). (24)(a) So, the question is: Who gave the order to kill the Jesuits and their two maids? (b) ? Col. Benavides did not / (didn't) issue the order. (c) It was not / (wasn't) Col. Benavides who issued the order. (24a) presupposes that someone gave the order, as does the response (24c). (24b) does not. It is, in fact, compatible with a rejection of that presupposition. Therefore, the it-cleft is the appropriate construction for the response in this context, and not some SVO variant that superficially looks the same. However, this is only supposition about an author's attitude in reporting certain information, and a definitive interpretation can never be achieved with this message, because of the indistinct and incomplete information of sentence (22). In the second cleft sentence (25a) of Message 0034, an interesting aspect of the reporter's attitude toward the subject he is discussing is revealed through the use of the it-cleft versus normal declarative word order. Compare (25a) and its SVO paraphrase, (25b). (25)(a) However, in any case and as has always occurred, it will be the lieutenants, the noncommissioned officers, and the soldiers who carried out an order who will pay the consequences. (b) However, in any case and as has always occurred, the lieutenants, the noncommissioned officers, and the soldiers who carried out an order will pay the consequences. (25b) might be interpreted as being much more factual, detached and impartial, compared to (25a) in the context of the rest of the message. Such qualities as factuality, detachment and impartiality, of course, are what one normally anticipates or expects in a news report. In fact, these qualities have almost become stereotypical qualities of news casting. However, the use of the it-cleft seems to be more appropriate in a commentary, as this particular message is (cf. header in Appendix I). Is it, therefore, the commentator's intent to indicate more than to state the facts that someone is guilty for the massacre of the Jesuits? Or is the reporter also trying to express his/her attitude about the massacre, thereby piercing the detached, impartial guise of strictly reporting the event? If we are to be consistent with our interpretation in our work, we maintain that this is precisely the case and captures information not explicitly stated in the message. The construction used to convey the idea is carrying some of the meaning; thus, the medium is also conveying the message which the author wishes to express.16 In (26a) of Message 0308 [Appendix J], we see how an it-cleft subtly elicits a nuance in meaning which might be absent or missed if normal word order had been used. (26)(a) After all, it is the people who pay our salaries and provide the revenue that allows the government to operate. (b) After all, the people pay our salaries and provide the revenue that allows the government to operate. There are two possible interpretations of the sentences in (26). In one interpretation, the speaker believes that the people pay the salaries of the civil servants, and therefore, the people should be repaid as an obligation--a matter of work returned for a salary paid, perhaps nothing more. In the second interpretation, the speaker believes that the people pay the salaries of the civil servants, but beyond just repayment of the money provided, it is incumbent on the civil servants to do their jobs as a patriotic and civic duty. Granted, either interpretation is possible in the context of the entire message in which either (26a or 26b) is embedded. However, the subtlety is almost overlooked in (26b), while (26a) seems to favor the patriotic meaning more readily in the overall context. At this juncture it is necessary to mention that many of these passages from the MUC-3 corpus were probably translated from Spanish. We do not wish to go off onto a tangent about translations, problems associated with them, and conclusions deducible from translations. However, suffice it to say here that we casually polled a native speaker of Spanish about (26a and 26b) in the context of the overall message. The native speaker's reaction about these sentences in Spanish was that in the context of the entire passage, (26b) sounded peculiar, not altogether political, even missing some point; whereas, (26a) with the clefted-construction in Spanish seemed to be the more politically acceptable way to make the statement.17 In (27a) from Message 0378 [Appendix K], we have an example of an it-cleft that we argue is used to soften or ameliorate a particular reading. (27)(a) In this case, unfortunately they did not participate; it was our party platform that was chosen by the people. (b) In this case, unfortunately they did not participate; our party platform was chosen by the people. Normal word order, as in (27b), in the context of this discourse might be interpreted as being too blunt, almost condescending, even boastful, despite the ameliorative interpretation that the word "unfortunately" seems to be warranting. However, the it-cleft in (27a) clearly presents the softer interpretation. Thus, any confusion on the part of the reader in interpreting the intent behind the propositions embodied in (27) is avoided. We turn finally to an it-cleft (28a) from Message 0431 [Appendix L] that produces a generic reading of a focused noun phrase. (28)(a) It is an authoritarian ideology that does not question all types of states, but only the states of the South. (b) It's an ill wind that blows from the North. (c) An ill wind blows from the North. (d) An authoritarian ideology does not question all types of states, but only the states of the South. In the context of this message, (28a) clearly has the interpretation that it is a characteristic function of an ideology that if it does not question all types of states but only those of the South, then that ideology is to be characterized as being authoritarian. This interpretation is akin to the generic interpretation of (28b). There is no specific "ill wind" in mind in (28b). Contrast (28c). The truism is meant to be interpreted as applying to all objects of the type "ill wind." So too the it-cleft in (28a) is to be interpreted as applying to all "authoritarian ideologies." From the author's point of view, then, the it-cleft presents the idea clearly with no possible misinterpretation, inherent in the SVO paraphrase (28d). (28d) might be ambiguously interpreted as either generic or specific. The point here is that the it-cleft strongly favors the generic interpretation. Contrast (28a) and (28d). In the context of the entire message, (28a) might be interpreted as a sentence containing a noun phrase followed by a relative clause, and the construction has a definite referent in the previous discourse. If this is the case, the author's intent might be misunderstood. The author might be accused of pointing an accusatory finger at the referent. Thus, we see in (28a) an example of an it-cleft from a discourse that facilitates a particular interpretation. To our knowledge, very little if anything has been said in the literature about the generic reading of it-clefts, and while we have only uncovered this one instance in the data under investigation, we believe it is a fruitful area for future investigation. 6. Conclusion Information in a discourse can be obtained by analyzing several different linguistic properties of the discourse. Anaphoric pronouns, for example, provide referential information. Logical forms analyze the various propositions of the discourse into semantic predicates and arguments. Various syntactic and semantic triggers, therefore, provide clues to the complex informational structure of a discourse. For the purposes of this investigation into additional elements that provide information about a discourse, we selected several discourses taken from a series of wire service messages dealing with terrorist incidents that occurred in Central America from 1989 to 1991. Because of their property of invoking inferred information, we identified it-cleft constructions in the corpus. We have argued that, despite their rarity in the corpus under investigation, it-clefts in discourse provide additional information, and in some cases do not merely contrast or emphasize focused linguistic material, such as noun or prepositional phrases. By substituting so-called "normal" word order, or SVO paraphrases, for the it-clefts in the messages, we have determined what additional information it-clefts provide in the discourse. Our investigation reveals that, as a subset of the information that can be obtained in a discourse, it-clefts can be used to avoid conflicting or differing interpretations inherent in their SVO paraphrases, thereby minimizing possible confusion regarding the interpretation of what the author is trying to communicate. In some cases the speaker/writer's point of view or attitude about the subject matter being discussed is also revealed. For example, we have shown how a focused generic noun phrase in an it-cleft sentence provides a clearer statement of the author's intended meaning than does its SVO paraphrase. Thus, while it is a well-known fact that it-clefts and their SVO paraphrases are not interchangeable, certain information becomes more readily apparent when these two constructions are contrasted in the discourse. By employing this method of investigation, we have simply uncovered some additional information, such as author's point of view, which is inherent in the discourse. Although we have not addressed how this information can be extracted algorithmically, the fact that such information as an author's point of view may be obtained by considering certain constructions, such as it-clefts, we hope leads others to investigating methods by which such information can be extracted. Appendix A MONTEVIDEO (URUGUAY), 22 AUG 89 (EL DIA) -- [TEXT] PRESIDENT JULIO MARIA SAGUINETTI YESTERDAY SENT A TELEGRAM TO COLOMBIAN PRESIDENT VIRGILIO BARCO VARGAS EXPRESSING HIS SOLIDARITY OVER THE DEATH OF SENATOR LUIS CARLOS GALAN. HE SAID THAT THE "CIVIC SPIRIT" OF THE PEOPLE WILL TRIUMPH "OVER ORGANIZED CRIME, ITS ACCOMPLICES, AND ITS PROTECTORS." SANGUINETTI ADDED THAT IT WAS "WITH PAIN AND WITH INDIGNATION" THAT HE RECEIVED THE NEWS OF THE DEATH OF THE COLOMBIAN SENATOR, "MURDERED BY CRIMINALS AT THE SERVICE OF SPURIOUS INTERESTS." . . . Appendix B . . . [ZAMORA] THERE IS A POSITIVE FACTOR THAT SHOULD BE BORNE IN MIND, NAMELY, THE CALL FROM THE FIVE PRESIDENTS FOR A RESUMPTION OF THE DIALOGUE IN EL SALVADOR AND THE APPEAL TO THE UN SECRETARY GENERAL TO CARRY OUT MEDIATION TO THIS END. BUT IN THE SUMMIT'S FINAL DOCUMENT, IT IS THE FMLN AND THE NICARAGUAN CONTRAS WHO PAY THE PRICE FOR THE AGREEMENT AMONG THE PRESIDENTS: THESE ARE DIPLOMATIC GAMES THAT DO NOT FACILITATE THE RESTORATION OF PEACE IN CENTRAL AMERICA.... Appendix C . . . [BEGIN RECORDING] [PEREZ BELTRAN] TO THIS MOMENT, AND DUE TO THE SERIOUSNESS OF THE CASE, I DO NOT DARE TO ADMIT THE EXISTANCE [sic] OF ANY RELATIONSHIP BETWEEN THESE TWO TERRORIST GROUPS. IF THE SHINING PATH IS OPERATING IN BOLIVIA, IT SHOULD ANNOUNCE THAT IT IS IN FACT SHINING PATH THAT IS COMMITTING THESE CRIMES. Appendix D . . . WE THINK ALL THESE ACTIVITIES OF THE DEATH SQUADS, THE 1ST INFANTRY BRIGADE, THE POLICE, AND THE NATIONAL GUARD ARE ONLY THE OPEN EXPRESSION OF THE ARMED FORCES' POSITION, WHICH IS TO PLAN A FULL-SCALE CONFRONTATION WITH THE POPULAR MOVEMENT. WE THINK THIS INDICATES [WORDS INDISTINCT] BEHIND THE BACKS OF THE MAJORITY OF ARMED FORCES MEMBERS. IT IS A MINORITY WHO HAVE DECIDED ON THESE PROVOCATIONS AGAINST THE POPULAR MOVEMENT. THIS MOVEMENT, HOWEVER, IS SUFFICIENTLY STRONG TO DISARM THESE PROVOCATEURS, PRESENT THEM TO THE PRESS, AND HAND THEM OVER TO SHOW THAT THE PEOPLE EXPRESS THEIR TREMENDOUS POWER WHENEVER THEY WANT.... Appendix E . . . DURING A RADIO PROGRAM ON MONDAY, 4 SEPTEMBER, JOURNALIST JOSE VICENTE RANGEL, FORMER MOVEMENT TOWARD SOCIALISM PRESIDENTIAL CANDIDATE, MAINTAINED THAT IT IS NOT AT THIS TIME THAT DRUG TRAFFICKERS ARE ATTEMPTING TO ENTER VENEZUELA BECAUSE THEY ENTERED A LONG TIME AGO. RANGEL MAINTAINED THAT VENEZUELA ANNUALLY SERVES AS A BRIDGE FOR DRUG TRAFFICKING WORTH SOME $20 BILLION. FURTHERMORE, HE SAID, THE EXPENSIVE VENEZUELAN ELECTORAL CAMPAIGNS HAVE ALSO RECEIVED RESOURCES OBTAINED FROM DRUG TRAFFICKING.... Appendix F . . . "THEY REASSERT THEIR PROFOUND CONVICTION THAT DEMOCRATIC SOCIETIES MAY EFFECTIVELY DEFEND THEMSELVES FROM THE SCOURGE OF TERRORISM WITHIN THE FRAMEWORK OF THE LAW AND THE EXERCISE OF POLITICAL PLURALISM." THE COMMUNIQUƒ ENDS BY STATING: "IT IS WITHIN THAT FRAMEWORK THAT THE ARGENTINE GOVERNMENT IS CONFRONTING EXTREMIST ACTIONS, AND WE THEREFORE WISH TO EXPRESS OUR RECOGNITION."... Appendix G . . . HERNANDEZ CONDEMNED THE PRESENCE OF U.S. TROOPS AND THE CONTRAS, AND HE SAID THAT "EFFORTS ARE BEING MADE TO MAKE US PAY FOR A FOREIGN DEBT WE DO NOT OWE." "THAT DEBT WAS PAID A LONG TIME AGO, AND WE MUST ASK WHETHER IT IS THE DEVELOPED CAPITALIST COUNTRIES THAT OWE US BECAUSE OF THEIR LOOTING OF OUR NATURAL RESOURCES," HE ADDED.... Appendix H SAN SALVADOR, 12 JAN 90 (CANAL SEIS TELEVISION) -- [REPORT] [NANCY PERLA] [TEXT] SALVADORAN VICE PRESIDENT FRANCISCO MERINO HAS DESCRIBED THE POSSIBLE RESUMPTION OF DIALOGUE BETWEEN THE GOVERNMENT AND THE REBELS AS POSITIVE. MERINO SAID THAT THE FMLN [FARABUNDO MARTI NATIONAL LIBERATION FRONT] SHOULD NOT CLAIM CREDIT FOR THE RESUMPTION OF A DIALOGUE BECAUSE IT WAS PRESIDENT ALFREDO CRISTIANI WHO TOOK THE INITIATIVE IN RESUMING THE TALKS. MERINO ADDED THAT THE REBELS SHOULD ABANDON THE DELAYING AND IRRESPONSIBLE TACTICS REGARDING THE DIALOGUE BECAUSE THE FMLN HAS LOST CREDIBILITY BOTH IN THE MILITARY SENSE AND IN INTERNATIONAL POLITICS.... Appendix I CLANDESTINE, 13 JAN 90 (RADIO VENCEREMOS) -- [STATION COMMENTARY] [TEXT] WHAT WE HAD PREDICTED HAS OCCURRED. CRISTIANI ANNOUNCED THIS AFTERNOON THAT A COLONEL, TWO LIEUTENANTS, ONE SECOND LIEUTENANT, TWO SECOND SERGEANTS, AND TWO SOLDIERS WERE INVOLVED IN THE MASSACRE OF THE JESUITS. THE ACCUSATION AGAINST COLONEL BENAVIDES IS THE FOLLOWING: THAT THE MULTIPLE CRIME WAS COMMITTED IN THE SECTOR UNDER HIS COMMAND. FROM THE LIEUTENANT ON DOWN, THE ACCUSATION IS THAT THEY DIRECTLY PARTICIPATED IN THE CRIME. [WORDS INDISTINCT], AS CRISTIANI SAID, THAT IT WAS COL BENAVIDES, BUT IT WAS NOT COL BENAVIDES WHO ISSUED THE ORDER. SO, THE QUESTION IS: WHO GAVE THE ORDER TO KILL THE JESUITS AND THEIR TWO MAIDS. IN OUR VIEW, COL BENAVIDES WAS IMPLICATED SOLELY AS A MEANS OF PROPAGANDA AIMED AT THE U.S. CONGRESS. IT IS CLEAR THAT COL BENAVIDES DID NOT GIVE THE ORDER. HE PROBABLY DID NOT EVEN KNOW ABOUT THE ACTION. THIS IS CONSISTENT WITH WHAT WE KNOW ABOUT HIS BACKGROUND. FIRST, HE IS NOT KNOWN TO HAVE CARRIED OUT ACTS OF REPRESSION OR CORRUPTION. OUR REPORT INDICATES THAT BENAVIDES HAS MAINTAINED A POSITION IN FAVOR OF INSTITUTIONS AND OPEN TO DISCUSSIONS. HE WAS DESCRIBED AS AN ACTIVE, STUDIOUS AND NOT REPRESSIVE CADET, ONE OF THE FEW EXCEPTIONS IN LA TANDONA [REFERENCE TO THE 1977 GRADUATING CLASS OF THE CAPTAIN GENERAL GERARDO BARRIOS MILITARY ACADEMY]. IN OTHER WORDS, FOR THE LA TANDONA, A GANG OF ASSASSINS, HE IS A PERFECT SCAPEGOAT.... Appendix J . . . REGARDING OUR FISCAL POLICY, WE MUST TRY TO DETERMINE WAYS TO MORE DIRECTLY MAKE UP FOR THE DIFFERENCE BETWEEN REVENUE AND DISBURSEMENTS IN THE COUNTRY. WE BELIEVE THAT WE CAN BRIDGE THIS FISCAL DEFICIT IN TWO WAYS -- WE MUST EITHER INCREASE OUR REVENUE OR LOWER OUR DISBURSEMENTS. THIS GOVERNMENT HAS ALREADY TAKEN AUSTERITY MEASURES REGARDING THE LATTER. THE GOVERNMENT HAS DETERMINED NOT ONLY TO MAINTAIN EXPENDITURES AT THE CURRENT LEVEL, BUT TO REDUCE THEM BY 15 PERCENT. WE WANT TO MAKE CLEAR THAT OUR GOVERNMENT'S AUSTERITY MEASURES MUST BE GENUINE AND SHOULD BE AIMED AT MAKING MORE EFFICIENT USE OF SCANT RESOURCES. THIS IS WHY WE ARE ASKING PUBLIC EMPLOYEES TO UNDERSTAND THAT IF THE GOVERNMENT IS TO BE MORE EFFICIENT, THE GOVERNMENT MUST PROVIDE BETTER SERVICES FOR THE SALVADORAN PEOPLE. AFTER ALL, IT IS THE PEOPLE WHO PAY OUR SALARIES AND PROVIDE THE REVENUE THAT ALLOWS THE GOVERNMENT TO OPERATE.... Appendix K . . . [CRISTIANI] I BELIEVE THE WORD NEGOTIATION HERE IN OUR COUNTRY NO LONGER HAS THAT GENERIC MEANING GIVEN BY THE DICTIONARY. THIS WORD HAS BEEN MANIPULATED BY LEFTIST FORCES IN THE SENSE THAT TO NEGOTIATE IS TO ACCEPT WHAT THE FMLN WANTS. ACCORDING TO THE LEFT, THAT IS WHAT NEGOTIATING MEANS. I BELIEVE THE WORD NEGOTIATION HAS A GENERIC MEANING. A NEGOTIATION IS TO TRY TO FIND JOINT AGREEMENTS TO SOLVE A PROBLEM AND THAT IS WHAT WE HAVE BEEN PROPOSING. WE HAVE SOUGHT AGREEMENTS ON THE LARGEST NUMBER OF POINTS POSSIBLE. LOGICALLY, NOT ALL POINTS WILL BE ACCEPTED BY THE OTHER SIDE AND WE BELIEVE THE FMLN IS MAKING A MISTAKE WHEN IT THINKS IT WILL NEGOTIATE ITS ECONOMIC, POLITICAL, OR SOCIAL PLATFORM WITH THE GOVERNMENT. THE FMLN MUST UNDERSTAND THAT THEY HAVE TO NEGOTIATE THEIR PLANS FOR THE COUNTRY WITH THE PEOPLE. THAT IS DONE BY POLITICAL CAMPAIGNS AND BY SHOWING THEIR PROPOSALS TO THE PUBLIC. IF THE PEOPLE ARE CONVINCED THAT THEIR PROPOSALS ARE BEST, THEY WILL GET THE VOTES AND THEY WILL HAVE THE RIGHT TO IMPLEMENT THEIR PROGRAMS AND THEIR POLITICAL, ECONOMIC, AND SOCIAL PROGRAMS. IN THIS CASE, UNFORTUNATELY THEY DID NOT PARTICIPATE; IT WAS OUR PARTY PLATFORM THAT WAS CHOSEN BY THE PEOPLE. THIS GIVES THE GOVERNMENT THE RIGHT TO IMPLEMENT WHAT THE MAJORITY OF THE PEOPLE ASK OF THEM.... Appendix L LIMA, NO DATE (TELEVISION PERUANA) -- [SPEECH] [GARCIA] [TEXT] [CONTINUED] IT IS CLEAR THAT NEO-LIBERALISM IS THE IDEOLOGY OF THE NORTH. IF IN THE PAST LIBERALISM WAS A FACTOR THAT PROMOTED PROGRESS AND INDEPENDENCE IN LATIN AMERICA, NEO- LIBERALISM IS NOW THE STATE IDEOLOGY OF THE MORE POWERFUL NATIONS. IT IS AN AUTHORITARIAN IDEOLOGY THAT DOES NOT QUESTION ALL TYPES OF STATES, BUT ONLY THE STATES OF THE SOUTH. UNDER THE GUISE OF ELIMINATING BUREAUCRACY AND PROMOTING EFFICIENCY, NEO-LIBERALISM WANTS TO DESTROY THE STATE TO ELIMINATE ANY RESISTANCE TO THE HEGEMONY OF THE WORLD MARKET. WE SHOULD NOT PLAY THAT GAME. IDEOLOGICAL NONALIGNMENT SHOULD THEREFORE REDEFINE THE STATE AND DEFEND ITS STRENGTH IN OUR COUNTRIES.... Notes *The following research was conducted at the Naval Research Laboratory in Washington, DC with funding from both the Office of Naval Research and the Army Research Laboratory. 1 In the following investigation, we do not ultimately distinguish between the spoken and written language, although initially the distinction is made. The corpus which we used and upon which we base our claims contained the written forms of both textual communiquŽs and oral dialogs. Therefore, we collapse this identity into one role, since we only used the written forms of both types of communication. 2 Underlining words or phrases and the use of italics are common typographical or stylistic features to indicate stress and various intonational features, such as emphasis and contrast, in text. However, for the purposes of this study, we exclude these alternative methods of indicating such features in text analysis and concentrate on a specific syntactic construction. 3 For our study, we downloaded the raw data from the Naval Ocean Systems Center (NOSC) database and processed the messages from the third in a series of evaluations of English text analysis systems held in San Diego, CA from May 21-23, 1991. The Defense Advanced Research Projects Agency, Software and Intelligent Systems Technology Office (DARPA SISTO), sponsors the yearly conference. For this analysis we utilized some in-house UNIX utility programs at the Navy Center for Research in Artificial Intelligence (NCARAI) and some text formatting utilities obtained from New York University. The latter were obtained from The Linguistic String Project of New York University under the direction of Naomi Sager, and were used to format the texts into a regularized machine-readable form. In this process, elements of the messages are separated into lexical items with punctuation being treated as lexemes. Each paragraph and sentence is given a unique identifier, making for easier processing, but the specifics need not concern us here. 4 By "normal" or SVO word order in English, we mean that the sentence can be analyzed as a declarative sentence in which the subject precedes the verb and its complement. Furthermore, the subject of the sentence is a semantically significant phrase. Thus, something like (i) exhibits "normal" or SVO word order for English, while (ii), its interrogative paraphrase, and (iii), a sentence containing a so-called semantically "empty" subject, do not. (i) The president met with the advisors. (ii) Did the president meet with the advisors? (iii) It is raining. 5 When we looked at the 12 occurrences of it-cleft sentences from the wire services, we noticed that 8 of the 12 sentences were embedded in some sort of negative context; that is, some sort of overt negative or contradictory marker, such as the word "no," "not," "but," "however," or the like, was immediately available in the surrounding context. We return to this point shortly. 6 For the purposes of this investigation, we will consider such adverbial phrases as "yesterday" in (i) in some sense a prepositional phrase (PP). (i) It was yesterday that I saw the president. Restrictions on the seventh term of the it-cleft construction, i.e. NP or PP, have been noted throughout the literature on the subject of it-clefts. The it-cleft construction does not permit adjectives or various types of verbal constructions, as in (ii-iv), at least in colloquial English. (ii) *It was black that the sky turned. (iii) *It was rotten that the eggs became. (iv) *It was quickly that the boy ran. In terms of X-bar Theory (Chomsky 1981) the generalization that this particular constituent is restricted to phrases whose heads can be characterized as [+/-N, -V] has not been fully investigated. Further research on this particular point might lead to greater understanding about the nature of clefting in grammar. Furthermore, the restriction on the range of type of adverbs in it-cleft constructions is still not clearly defined. 7 The canonical form for it-clefts that we consider here rules out such demonstrative uses as (i). (i) That was the man that shot the president. Such clefts as (i), as well as other types of clefts, are discussed in Gurney and Perzanowski (forthcoming). 8 We do not discuss the lexical, syntactic, or semantic content of this item here. We leave the discussion of whether it is a categorial NP or some sort of filler to cover up an empty subject in certain languages, such as English, for others. 9 Here we loosely define "cataphoric" usage of pronouns as the referential use of "it" in an it-cleft, since a grammatical antecedent for this particular pronoun is absent in the text. Furthermore, there seems to be some semantic link between the "it" in an it-cleft sentence and the focused element in the complement of the it-cleft sentence. For a much more detailed explanation of the cataphoric usage of pronouns in general and the cataphoric usage of the "it" in it-clefts, the reader is directed to (Quirk et al. 1985). 10 The original texts for the messages under consideration are included in the appendices to this report. All the words of the messages were capitalized and only terminal punctuation and spaces between words and sentences distinguished these elements one from the other. 11 Prince (1978) and Declerck (1984) discuss two types of it-clefts: CF or contrastive focus and IPR or informative presupposition clefts. As noted in Gurney and Perzanowski (forthcoming), CFs receive contrastive stress, but the stress of IPRs is different. One might ask two questions: first, how does anyone then know which stress pattern to assign the it-cleft construction. And secondly, one might ask how anyone reading a cleft knows what intonation to assign the cleft mentally, as it is being read. If, as we note, CFs are frequently surrounded by negative contexts, then the negative context itself can be used as a precondition under which contrastive stress is elicited in such responses that can be used to categorize CFs. IPRs do not occur in these negative contexts, and their stress patterns are different. While the constructions are the same: CFs and IPRs both have the syntactic pattern as indicated in (4) above, their environments may differ significantly. 12 Transcripts of the various messages from the MUC-3 corpus used in this study are provided in the appendices. 13 It is not altogether clear that the reporter is quoting a sentence that was originally an it-cleft; i.e., Sanguinetti's original quote may or may not have been an it-cleft sentence. Our inclination is to believe that it was. However, regardless of our inclination, the fact is that the sentence in our text is an it-cleft, and in the context of the wire service message, certain information is being communicated, which is our concern here. 14 Gurney and Perzanowski (forthcoming) offer a more comprehensive position on it-clefts, although the conclusions reached there do not contradict the ideas contained here. 15 In actuality, this sentence exhibits a more complex contrastive pattern than has been traditionally referred to in the literature on it-clefts. They discuss this contrast at some length. 16 Such a strong claim based on suppositional argumentation is immediately testable under the theory of scientific falsifiability and, for this very reason, is included here. 17 Hopefully, the translators of the original wire service messages were translating clefted original sentences into clefted constructions in English. Note further, then, that if we are correct in our claim that the particular type of syntactic construction used to convey an idea can evoke some meaning or interpretation, it is incumbent upon anyone dealing with translations or transcripts of proceedings to ensure that the same or comparable syntactic type of construction is used in the translations, as we are sure many translators are already aware. As we argue in the case of it-clefts, the particular syntactic construction employed can indicate speaker's intentions, leanings toward particular interpretations, and the like. Works Cited Chomsky, N. Lectures on Government and Binding. Dordrecht: Foris Publications, 1981. Declerck, R. "The Pragmatics of IT-Clefts and WH-Clefts." Lingua 64 (1984): 251-289. Delin, J. "Properties of It-Cleft Presupposition." Journal of Semantics 9 (1992): 289-306. Gazdar, G. Pragmatics: Implicature, Presupposition and Logical Form. New York: Academic Press, 1979. Gurney, J. & Perzanowski, D. "Understanding Clefts in Discourse," unpublished manuscript, (forthcoming). Heggie, L. "The Range of Null Operators: Evidence from Clefting." Natural Language and Linguistic Theory. 11 (1993): 45-84. Jones, L.B., & Jones, L.K. "Discourse Functions of Five English Sentence Types." Word. 36.1 (1985): 1-21. McLuhan, M. The Medium is the Message. New York: Random House, (1967). Prince, E. "A Comparison of WH-clefts and IT-clefts in Discourse." Language. 54.4 (1978): 883-906. Quirk, R., et al. A Comprehensive Grammar of the English Language. New York: Longman, 1985. Sundheim, B. Third Message Understanding Conference (MUC-3): Proceedings of a Conference. San Diego, CA: Morgan Kaufmann, 1991. Medium/Message 37