8.1 Introduction
In Chapter 1 of this volume, Pham and Leuckert distinguish between two major conceptualisations of non-canonical syntax: constructions that represent a departure from ‘basic’ grammar, and constructions that represent a departure from typical or normal use. The first conceptualisation builds on a theoretical foundation of what constitutes ‘basic’ grammar in a given language. For example, the canonical or basic structure of English clauses is Subject-Verb-(Object or Complement). However, the second conceptualisation can provide a completely complementary perspective. For example, the typical/normal minimal response in English conversation consists of a single word like ok, yeah, oh. Such utterances are non-canonical in that they depart from ‘basic’ structures, but they are canonical in that they are the normal way for a speaker to respond to a previous utterance.
The present chapter documents a case where both perspectives are important: the use of Non-Canonical Reduced Structures (NCRSs) in TV news broadcasts (TVNBs). These are long, elaborated utterances with no main finite verb, and therefore they represent a striking departure from the rules of basic/canonical grammar. Such structures are also non-canonical in that they are rare or virtually unattested in most other registers – both spoken registers (including conversation) and written registers. But, at the same time, we show how a heavy reliance on NCRSs is becoming the norm in certain types of TV news broadcasts, and thus in that sense, these structures are becoming canonical in that register.
One of the major types of non-canonical grammar, in the sense of a departure from basic grammatical structure, involves structural reduction. Previous research has shown that such structures are especially prevalent in face-to-face conversation. The Grammar of spoken and written English (GSWE; Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021: 1031–120) groups these features into three major categories: Non-Syntactic Non-Clausal Units, Syntactic Non-Clausal Units, and Structurally Reduced Clausal Units.
The first of these categories – Non-Syntactic Non-Clausal Units – can be regarded as pragmatic rather than grammatical features. Non-Syntactic Non-Clausal Units do not have internal syntactic organisation, but they serve a wide range of pragmatic and interactional functions in conversation. Common types of Non-Syntactic Non-Clausal Units include vocatives (e.g., hey you), expletives (e.g., damn, shit), greetings (e.g., hi, bye), discourse markers (e.g., well, ok, right), or other utterance launchers (e.g., oh yeah, hey, there again) (see the survey in Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021: 1063–93).
In contrast, Syntactic Non-Clausal Units and Structurally Reduced Clausal Units have internal syntactic organisation. Syntactic Non-Clausal Units are phrases that can be described using the normal framework of syntactic analysis, but they are usually structurally incomplete because they lack a main verb (see Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021: 1093–9). These phrases can stand alone as a complete utterance, usually in response to some previous utterance as in examples (1) and (2).
Stand-alone noun phrase:
A: Is Nicki giving a lecture?
B: No, a training session
Stand-alone adjective phrase:
A: All of those houses have big square white things
B: Yeah, very solid inside
Structurally Reduced Clausal Units are main clauses where one or more of the semantically essential elements have been ellipted (see Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021: 1099–102). For example, subject noun phrases and auxiliary/copular verbs are commonly omitted in conversational utterances (marked by --), as in (3).
A: Are they happy?
B: -- depends on what you mean by that.
A: Well, -- -- at least content?
In practice, it is difficult to maintain the distinction between Syntactic Non-Clausal Units and Structurally Reduced Clausal Units in analyses of extended discourse. In the first place, dependent clause structures fail to fit tidily into one or the other category, because they are ‘clausal’ but often clearly not a main clause (cf. (4)).
A: I’m going to go back and get some more of those.
B: Yeah, because they’re a lot higher at the other store.
The bigger problem, though, is that identification of a reduced clausal unit requires speculative interpretation to try to reconstruct the underlying non-reduced form. For example, the stand-alone noun phrase in (1) and adjective phrase in (2) might have been interpreted as reduced clausal units with the subject noun phrase and copula be omitted. We show below that this problem is exacerbated in the analysis of TVNBs, which employs extremely complex sequences of reduced structures. For these reasons, the following analyses are based solely on the forms that are actually employed in discourse, with no attempt to reconstruct an underlying ‘original’ form. As a result, we do not distinguish between Syntactic Non-Clausal Units and Structurally Reduced Clausal Units; rather, these are both treated as instances of Non-Canonical Reduced Structures (NCRSs).
The reliance on NCRSs in conversation can be interpreted in functional terms, associated with the pressures of real-time language production, and the fact that speakers co-construct discourse and share the same time/place, resulting in a frequent use of context-dependent linguistic structures. NCRSs have also been described as a salient characteristic of ‘simplified registers’, which have been categorised according to their functional motivations as ‘handicap registers’ and ‘economy registers’. Ferguson (Reference Ferguson and Hymes1971), focusing on the ‘handicap’ functions, coined the term ‘simplified register’ to refer to the ‘modified varieties of speech typically addressed to listeners not believed to be fully competent in the language’, such as ‘baby talk’ or ‘foreigner talk’ (Ferguson Reference Ferguson, Obler and Menn1982: 49). Janda (Reference Janda1985) extended the analysis of simplified registers to include those that are functionally motivated by the need for efficiency and economy of expression (like note-taking), rather than a perceived listener handicap. Interestingly, though, Janda shows how simplified economy registers employ many of the same NCRSs as simplified handicap registers, such as the omission of subject noun phrases, omission of definite articles in noun phrases, and omission of copula/auxiliary be.
The use of NCRSs in note-taking can be attributed to the time pressure of recording language produced in a real-time situation. However, texts in other written simplified registers – such as newspaper headlines and classified advertisements – have been carefully pre-planned and edited. In these cases, the motivation for economy of expression relates to publication costs, resulting in linguistic forms that express maximal content in as few words as possible (see the detailed discussion in Bruthiaux Reference Bruthiaux1996). As a result, these written economy registers often employ extended sequences of NCRSs, assuming that the intended readers will have the background knowledge required for understanding. Thus consider example (5) from a classified advertisement (from Bruthiaux Reference Bruthiaux1996: 85):
88 ACURA LEGEND, 5 speed, 4 dr. Red ext w/grey leather int. Loaded, with car phone. Low mileage. Mint condition. Steal = $12,000 obo…
The use of extended sequences of phrasal NCRSs in written economy registers is perhaps not especially surprising, given the fact that writers can take as much time as they want to manipulate the final form of the language that they produce. In contrast, we would never expect a speaker in conversation to produce an extended sequence of phrasal NCRSs like in (5). Rather, corpus research has shown repeatedly that speakers in conversation rely heavily on clausal structures, even though there is also a dense use of reduced structures. That is, NCRSs in conversation tend to occur as responses to a previous utterance, rather than as a sequence of multiple reduced utterances. In addition, any single utterance in conversation is likely to include only a single simple NCRS, rather than a structure consisting of multiple phrases or reduced clauses.
In all of the registers discussed above, the use of NCRSs is motivated by functional considerations relating to the challenges of the production (or perceived comprehension) circumstances, or the desire to achieve more economical linguistic expression. However, Ferguson (Reference Ferguson1983) noticed that Sports Announcer Talk (SAT) is a spoken simplified register that employs many of these same types of NCRSs, but in some cases without time pressure, handicap, or economy motivations. In particular, Ferguson studied the discourse of baseball game radio broadcasts, a speech event directed to inform listeners, where fast-action events are usually separated by considerable amounts of time, permitting extensive opportunity for the announcer to plan their speech and produce as much language as they want. Despite those situational characteristics, the discourse of SAT frequently employs many of the same kinds of NCRSs as conversation and other simplified registers. Ferguson (Reference Ferguson1983: 159) interprets these as instances of clausal structures with a deleted subject pronoun, a deleted copula, or a deleted subject plus copula, as in examples (6)–(10).
-- pops it up.
Klutz – in close at third.
McCatty--in difficulty.
-- -- fastball. -- -- strike. -- -- one and one.
The functional motivation of such structures in SAT is less clear than with handicap or economy registers. One possibility suggested by Ferguson (Reference Ferguson1983: 168) is that these NCRSs might have been initially used in SAT modelled on the pattern of news headlines, and as ‘a way of sounding exciting’. Support for that interpretation comes from other studies that describe the characteristics of radio broadcasts of events in progress. For example, the Multi-Dimensional analysis of register variation in Biber (Reference Biber1988: 128) shows that radio broadcasts are surprisingly phrasal, in contrast to all other spoken registers. Detailed analysis of that pattern shows that radio broadcasts are actually structurally reduced: ‘there are relatively few verbs, because many of the verbs are deleted due to time constraints, or to give the impression of action that moves so fast that there is no time for a full description’ (Reference Biber1988: 135). As a result, this register has a higher density of nouns and prepositional phrases than other spoken registers. The radio broadcasts included in the 1988 study were recordings of events in progress (taken from the London-Lund Corpus). Some of these events were live sports competitions, while others were less dynamic events like a funeral procession or a state wedding. Surprisingly, though, radio broadcasts of all of these events relied on a dense use of NCRSs. For example, the excerpt in (11) reporting on a state funeral procession includes numerous phrases and non-finite (dependent) clauses, but no main finite verbs at all.
flanked [pause]
by its escort of the Royal Air Force [pause]
the gun carriage [pause]
bearing the coffin [longer pause]
draped with the Union Jack [longer pause]
on it [pause]
the gold [pause]
and enamel [pause]
of the insignia of the Garter
Similar to reportage of a baseball game, there is virtually no time pressure on language production in this case. But the broadcaster chooses to produce discourse that relies heavily on NCRSs, giving the impression of action-packed reportage.
The present study focuses on the use of NCRSs in another spoken register produced in circumstances with no time pressure: TVNBs. TVNBs differ from radio broadcasts of events in progress in that they usually report on past events. However, as we show below, TVNBs further differ from all simplified registers previously studied in three major respects:
They use NCRSs with much higher frequencies.
They employ a wider inventory of different grammatical types of NCRSs.
They combine NCRSs in much more complex structural combinations.
Similar to radio broadcasts, the use of NCRSs in this register is not motivated by either handicap or economy considerations. However, since TVNBs normally do not report on events in progress, the functional motivation for these patterns differs from radio broadcasts of live events. As we discuss in Section 8.4.3 below, one major clue to this functional motivation comes from a comparison of the TVNBs offered by different networks, which vary widely in their use of NCRSs associated with their differing emphases on ‘hard news’ versus human interest stories.
8.2 Situational Characteristics of Television Network News Broadcasts
TVNBs are composed mostly of scripted language, combined with some real-time speech produced by reporters in the field or people who are the subject of a news story. The fact that anchors and reporters read pre-scripted reports seems to be a major factor influencing the types and frequencies of NCRSs found in TVNBs (see below). Interestingly, though, the audience must comprehend the discourse in the spoken mode, in real time. We return to this consideration in the conclusion.
TVNBs are organised as a sequence of news stories or segments. The opening segment functions in a similar way to the headlines found in a newspaper, announcing the topics of the subsequent stories, and providing hints of why those stories are interesting. Then, a typical broadcast will include a series of stories/segments that present the news. Some of these will consist of the anchor in the studio presenting a story; some stories will consist of a reporter interviewing other people; and some stories will involve a reporter covering a live event outside of the studio. News stories can also be characterised for whether they cover ‘hard news’ (e.g., stories relating to national or international concerns) versus ‘soft news’ (also known as ‘human interest stories’, which focus more on entertainment purposes than informational purposes). In our analyses below, we compare the patterns of language use across four general types of news segments: the opening segment (referred to as the ‘headlines’), the lead story (i.e., the first major story covered in the broadcast), a later story that was presented in the studio, and a later story that was presented from a location out of the studio.
TVNBs can be regarded as a register with mixed characteristics of speech and writing. Similar to conversation, TVNBs are produced in the spoken mode. However, most language produced by the announcers and reporters has been previously scripted in the written mode and was then read out loud during the broadcast. In contrast, recorded language spoken by other people (either interviewees or people involved in events/situations outside the news studio) has usually been produced in real time as the speaker is deciding what to say.
In the United States, there are several different types of TVNBs, including broadcasts offered by major commercial networks like NBC and CBS; broadcasts offered by non-profit or semi-governmental agencies like PBS; and broadcasts offered by 24-hour news channels like CNN and Fox News. The present study focuses on the evening news broadcasts offered by the three major commercial networks: ABC, CBS, and NBC.
It is easy to assume that the main communicative purpose of TVNBs is to present information about current and past events, similar to newspaper reportage. However, this turns out to be a simplification. In fact, media researchers describe the main communicative purpose of TVNBs as presenting the news stories that will attract the largest audiences possible. As a result, TVNBs have evolved to focus more on entertainment than information (see the extended discussions in Postman and Powers Reference Postman and Powers2008 and Montgomery Reference Montgomery2007). This shift began in the 1970s and 1980s, as TV networks realised that TVNBs could be extremely profitable (because of the advertising revenue) if they succeeded in attracting large audiences. Unlike a newspaper, a TVNB has a limited time duration and can therefore cover only a limited number of stories. For these reasons, TVNBs have evolved to be different from newspapers in their focus on stories that have high audience appeal/interest.
It turns out that the three networks included in the present study differ in this regard. ABC has widely publicised its attempts to humanise the news and thus mostly broadcasts ‘soft news’. In contrast, CBS (and to some extent NBC) focuses much more on the coverage of ‘hard news’ (see Moos Reference Moos2011; Stelter Reference Stelter2012). These characterisations are based on a survey of the topics covered in the respective broadcasts. However, as we show below, these differences also have a major impact on the linguistic style of the broadcasts, especially in relation to their use of NCRSs.
8.3 A Taxonomy of Non-Canonical Reduced Structures Found in TV News Broadcasts
In the simplest cases, NCRSs in news broadcasts occur as an utterance consisting of a single phrase or partial clause. These may be, for example, a noun phrase (NP; e.g., A terrifying situation), an adjective phrase (AdjP; e.g., Yes, very close), a prepositional phrase (PP; e.g., In my career, yes), an ‑ing-clause (e.g., And videotaping their crimes), an ‑ed-clause (e.g., Taken down), a wh-word (e.g., Why not?), or a wh-clause (e.g., What you will never see again).
It turns out, though, that such examples are quite rare in TVNBs. Instead, what we normally find are utterances consisting of a NCRS that consists of multiple constituents. In some cases, these structures could clearly be analysed as a top-level constituent with an embedded constituent, as in (12)–(13).
[NP [PP]]: A sharp fall on Wall Street
[AdjP [PP]]: So, close to a ten percent correction
However, as we tried to apply such analyses to coding extended texts, we found that clear-cut examples like those above were rare. Rather, NCRSs in TVNBs normally consist of multiple constituents that have an unclear syntactic relationship to one another. For example, the NCRS in (14) consists of a prepositional phrase followed by a noun phrase.
In Portland, a hidden world
We interpreted this structure as consisting primarily of a noun phrase, with the prepositional phrase serving an adverbial function. However, there are no explicit signals of syntactic function here, and so many structures of this type are open to multiple interpretations.
Noun phrases followed by non-finite clauses also usually have unclear syntactic structure. For example, the title of our chapter illustrates an NCRS with an embedded -ing-clause (cf. (15)).
The President wide awake at 3:14 AM tweeting about CNN
In this case, the utterance can be analysed as a reduced finite main clause with two coordinated main verbs (i.e., ‘The President was awake and was tweeting …’). In other cases, like (16), though, similar structures are more plausibly interpreted as a noun phrase modified by a non-finite relative clause.
The major change tonight involving the Miss America pageant.
[compare: the major change which involved …, versus *the major change was involving …]
However, many structures of this type are not readily interpretable as deriving from either a reduced finite main clause or from a non-finite relative clause. For example, consider the second utterance in the announcement in (17).
We also have new reporting coming in right now involving a troubling security breach at a US Air Force base. A driver breaching the main gate at Travis Air Force Base, then crashing, the SUV exploding.
In this NCRS, the -ing-clauses do not function to specify the identity of head nouns, and thus a non-finite relative clause interpretation is not plausible. That is, the goal is not to identify a particular ‘driver’ or a particular ‘SUV’. Rather, the -ing-clauses function to tell us what the ‘driver’ did, and what happened to the ‘SUV’. At the same time, though, these structures cannot be readily interpreted as reduced progressive aspect clauses. That is, the intended meaning is not that ‘the driver was breaching’ or that ‘the SUV was exploding’. Rather, the more likely canonical forms would have been ‘A driver breached …’ and ‘the SUV exploded’.
Reflecting such uncertainties, our analysis here focuses on the sequences of grammatical structures employed in NCRSs, rather than trying to force a specific analysis of syntactic relations. For the quantitative analyses in Section 8.4 below, we made an interpretation of the top-level constituent in each sequence of structures, and counted each NCRS as a token of that category. As a result, the overwhelming majority of tokens in our corpus analysis are categorised as top-level noun phrase structures combined with other secondary structures. However, many of these instances are like example (17), discussed above, which could have been analysed as top-level clausal structures with a secondary noun phrase.
It turns out that the specific classification of NCRS tokens has little bearing on the overall patterns found in our corpus analysis. Rather, the main finding about the nature of NCRSs in TVNBs is that they are incredibly long and complex, in contrast to the NCRSs documented in previous research. As a result, any attempt at classification is a very poor reflection of the huge range of structural and syntactic variation and complexity actually found in these utterances. Thus, consider examples (18)–(21), which are typical of the NCRSs found in this register.
That controversial tweet one of eighteen in five days about Russian meddling following the indictments of thirteen Russian officials by Special Counsel Robert Mueller, including blaming his predecessor.
Also tonight, after President Trump told millions that the government is fully prepared for this hurricane, the President tweeting today, questioning how many really died in Puerto Rico, saying 3,000 people didn’t die after hurricanes hit Puerto Rico.
On the eve of Michael Flynn’s sentencing, former FBI director James Comey pushing back on Flynn’s claim the FBI never warned him of the consequences about lying about his contacts with the Russian ambassador.
New tonight, what the judge is now saying involving Paul Manafort after Robert Mueller’s team accused him of tampering with witnesses while he was under house arrest, allegedly trying to get them to lie.
These utterances all lack a finite main verb, making them instances of NCRSs. The utterances also share the characteristic that they are composed of multiple grammatical structures (phrases and dependent clauses) that have complex (and often indeterminate) syntactic relations to one another. Beyond those similarities, however, the four examples are strikingly different in terms of the particular structures that are combined and in terms of the syntactic relations among those structures. As such, they clearly illustrate the inadequacies of any attempt to exhaustively classify tokens of NCRSs in TVNBs. Rather, the main pattern that emerges from the corpus-based analysis of these structures is their extreme diversity and complexity, in contrast to the types of NCRSs previously described in other registers. Those corpus findings are discussed in the following section.
8.4 Distribution of NCRS Types in TV News Broadcasts
8.4.1 Corpus and Methods
Our quantitative-linguistic description of TVNBs is based on analysis of a corpus of 144 texts sampled from the three US television network news broadcasts for the year 2018. We began by selecting one broadcast for each month of the year, and then sampled four specific segments/stories from each broadcast: the ‘headlines’, the lead story, a story presented in the studio, and a story presented outside of the studio (see Table 8.1). All texts were collected from the major evening news shows of the three networks (i.e., ABC’s World News Tonight, CBS’s Evening News, and NBC’s Nightly News). The entire 30-minute broadcast was downloaded for each week. Most of the language in these broadcasts had been previously scripted and then read out loud by the announcers or reporters. However, broadcasts also include some spontaneous spoken language produced by interviewees or by other people who appear in video clips.
Table 8.1Long description
The table is divided into 3 columns. The labels for the columns are network or broadcast segment, number of texts, and number of words. The three major network broadcasts are A B C, C B S, and N B C. A last row is provided for total. Each broadcast channel is subdivided into 4 sections: headlines, lead story, in-studio, and on-scene.
The data in the table provided are as follows:
1. A B C broadcast:
The values for headlines are 12 and 2569.
The values for lead story are 12 and 7940.
The values for in-studio are 12 and 3930.
The values for on-scene are 12 and 4120.
2. For C B S broadcast:
The values for headlines are 12 and 2504.
The values for lead story are 12 and 4842.
The values for in-studio are 12 and 4396.
The values for on-scene are 12 and 4266.
3. For N B C broadcast:
The values for headlines are 12 and 2183.
The values for lead story are 12 and 6021.
The values for in-studio are 12 and 4159.
The values for on-scene are 12 and 4122.
4. The values for the total are 144 and 51052.
Texts were downloaded from the Access World News database. Transcription was taken at face value. We relied on the punctuation in the transcripts to determine utterance boundaries. An informal comparison of actual video broadcasts to the transcribed texts indicates that the transcriptions are accurate representations of the words produced in speech, and that the clausal punctuation conventions generally corresponded to our own perceptions of utterance boundaries. As noted in Section 8.2, we distinguish among the four major segments of a TVNB (headlines, lead story, in-studio reportage, on-scene reportage) for our analyses of NCRSs.
The texts included in our corpus were hand-coded to identify all tokens of NCRSs, operationally defined as an utterance that does not include a finite main verb. NCRSs were then further hand-coded for their structural characteristics, including the top-level structure (e.g., noun phrase, prepositional phrase, -ing-clause), the structural type of any adjacent secondary structure, the presence of multiple secondary structures, and the presence of a deictic time/place adverbial. The frequency of each major combination of structures was counted in each text, and then converted to rates of occurrence (per 1,000 words).
8.4.2 Comparison of NCRSs in TV News Broadcasts versus Conversation
To establish a baseline for comparison, we carried out the same analyses on a small corpus of 10 AmE conversations (from the LSWE Corpus), totalling roughly 11,100 words. The results, summarised in Figure 8.1, confirm the qualitative descriptions in the GSWE (Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021: chapter 14), showing that most NCRSs in conversation are simple structures. Three structural types predominate in conversation: simple noun phrases, wh-questions with no verb, and simple verb phrases (i.e., with no subject and/or with an omitted auxiliary verb).
Figure 8.1 NCRS types in TV news broadcasts versus conversation (rate per 1,000 words)
Figure 8.1Long description
The vertical axis marks rate per 1000 words and ranges from 0 to 30, in increments of 5. The horizontal axis marks the non-canonical reduced structures or N C R S, including total N P, simple P P, simple V P, simple w h-structure, simple N P, N P plus e d, N P plus i n g, N P plus P P, N P plus other, and total complex. The vertical bar graph includes two types: a shaded one indicates news broadcasts, and a nonshaded one indicates conversation. The trend observed in price from left to right for a set of vertical bar graphs are 25, 4, 1, 0, 0, 1, 1, 1, 2, 3, 2, 0, 8, 0, 7, 0, 3, 0, and 16, 0.
In most cases, all three of these structural types occur in adjacency pairs with a preceding utterance, where the NCRS provides a response, asks clarification, or provides active engagement (often with a direct repetition of part of the preceding utterance). Noun phrases occur most frequently with these functions, as in examples (22)–(27):
A: And her, her brother once played with uh.
B: Oh. Duke Ellington?
A: Yeah Duke Ellington.
A: Were they like throwing things at you? Like shoes and stuff?
A: The oysters were this big, I’ve never seen oysters that big.
B: I know you said that.
A: Six huge, huge oysters.
A: They didn’t even bring us forks.
B: Yeah no forks, no, no plates.
A: Also our friend is very interested because he’s the one that started the, the little uh, the saints.
B: The Young Saints?
A: It says, what?
B: Dinner at five-thirty, reception to follow
However, wh-questions with no verb (cf. (28)–(30)) and simple verb phrases (with omitted subject and/or omitted auxiliary verb; cf. (31)–(32)) are also found.
A: Of course, if she’s late, then
B: Who?
A: I think you’ll need a fork for this.
C: Mm.
A: Nancy, what about you?
A: Can we take out the middle part?
B: What middle part?
C: Of what?
A: The table.
A: We saw pictures of you doing that. -- Doing your act.
A: Well then I’ll go at four.
B: -- you going to take your car?
In contrast, Figure 8.1 shows that NCRSs in TVNBs (1) are much more frequent overall, but (2) are rarely simple structures. Simple verb phrase structures rarely occur in news broadcasts, but simple noun phrases and wh-structures do occur, with roughly the same frequencies as in conversation. However, because the discourse of TVNBs is usually not dialogic, these structures are usually not part of an adjacency pair. Rather, simple noun phrases and wh-structures are usually used as topicalisation devices, emphasising the topic of the following discourse. Such structures are especially common in the headline segments of a TVNB, as in (33).
And, a holiday jobs bonanza. Companies so desperate for workers, they’re offering job perks and higher pay. How you can cash in.
In example (33), the first NCRS is a simple noun phrase (a holiday jobs bonanza), introducing the topic. The last NCRS is a wh-clause – a dependent clause with no accompanying main clause. The middle NCRS, which is actually much more typical of TVNBs, is a highly complex noun phrase, followed by a postnominal comparative construction, which itself consists of multiple phrases and a dependent clause.
Figure 8.1 also shows that simple prepositional phrases occasionally occur in TVNBs. Similar to the simple NP and wh-NCRSs, these simple PPs serve topicalisation functions; as in (34).
First, to the Mueller interview itself.
The most dramatic pattern shown in Figure 8.1, however, is the much higher frequencies of complex NCRSs in TVNBs compared to conversation. In fact, complex NCRSs are much more frequent than simple NCRSs in this register. The overwhelming majority of these are noun phrases combined in a sequence with other phrases and dependent clauses. Three major types of combination, illustrated in (35)–(38), are especially prevalent: NP + PP (+PP), NP + -ing-clause, and NP + -ed-clause.
NP + PP (+PP)
A sharp fall on Wall Street.
<(Bell tolling)>
The Dow drops more than eleven hundred points as a sell-off continues.
Also, tonight, the security plans for an Olympics just fifty miles from North Korea. […]
And the leader of the pack of underdogs.
NP + -ing-clause
The lead personal attorney John Dowd resigning. This leaves a big vacancy. The other headline breaking at this hour, General H.R. McMaster now out, a new national security adviser coming.
NP + -ed-clause
In central Pennsylvania, this morning, a seven-year-old killed in a hit and run.
NP + -ed-clause
Our team on Syria tonight, the horror, new airstrikes. Families, young children seen screaming. More than 200 killed.
It actually turns out that it is difficult to find relatively non-complex examples like those above. Rather, the overwhelming majority of NCRSs in these TVNBs are extremely complex constructions, involving multiple phrases and dependent clauses. The result shown in Figure 8.1 for ‘Total complex’ (i.e., NCRSs that include more than three structures) shows that these long, complex structures are extremely common in news broadcasts, occurring over 15 times per 1,000 words. In contrast, such structures are virtually non-existent in conversation. We have already presented several examples of such structures in TVNBs; other examples include (39)–(41):
A new tropical threat closing in tonight with fears of mudslides and landslides and life threatening flash floods.
Plus, NBC News obtaining this photo showing Kavanaugh and his second accuser, amid a new effort to get information to the FBI and the third accuser tonight.
With a potentially dramatic and contentious showdown set for Monday, Kavanaugh spotted this morning en route to the White House, said to be eager to testify.
These examples illustrate a major functional characteristic of NCRSs in TVNBs: that they use phrases and non-finite clauses to compress a lot of information into relatively few words. Phrasal modifiers with similar functions are common in the prose of academic research writing and other informational written registers (see Biber & Gray Reference Biber and Gray2016; Biber et al. Reference Biber, Gray, Staples and Egbert2022), but they are rare in most spoken registers. Thus, the extremely frequent use of such structures in TVNBs is especially noteworthy. We return to a fuller discussion of this pattern in the conclusion.
8.4.3 Use of NCRSs across the Segments of TV News Broadcasts
As discussed in Section 8.2, the segments of TVNBs have quite different situational characteristics. Therefore, it is likely that NCRSs would be used differently in the four discourse types.
Figure 8.2 shows that this expectation is realised, with NCRSs being much more frequent in the headline segment than in any other segment of TVNBs. The overwhelming majority of those structures are headed by a noun phrase.
Figure 8.2 NCRS types across segments of TV news broadcasts (rate per 1,000 words)
Figure 8.2Long description
The vertical axis marks rate per 1000 words and ranges from 0 to 60, in increments of 10. The horizontal axis marks the non-canonical reduced structures or N C R S, including total N P, P P, w h-structure, i n g clauses, and others. The vertical bar graph includes four types, which are shaded from dark to light and indicate headlines, lead, scene, and studio. The total N P is the maximum for the segments of T V news which declines gradually over P P, and other segments, and rises slightly at others around 4.
Figure 8.3 (which contrasts headlines with all other segments combined) provides further details about the NP NCRS constructions, showing that most of these NP-headed structures are complex rather than simple noun phrases. More detailed consideration of these results shows that an extremely high proportion of all discourse in the TVNB headline segment consists of complex NCRSs. That is, Figure 8.3 presents the rates of occurrence of NCRSs per 1,000 words of text. However, a typical complex NCRS is longer than 20 words. As a result, the finding for complex NCRSs in Figure 8.3 – that there are approximately 34 complex NCRSs per 1,000 words in headlines – means that almost 70% of the words in headlines consist of complex NCRSs (i.e., 34 NCRSs × 20 words each). The extended example in (42) illustrates the dense concentration of complex NCRSs in a headline segment.
Opening segment with the ‘headlines’
<ANCHOR> Tonight, the fiery hearing on Capitol Hill.
<GRAPHICS: BREAKING> The FBI agent in the hot seat after that agent sent text messages about then-candidate Donald Trump, what he wrote. And, tonight, he’s now firing back at allegations of personal bias within the FBI as they investigated Russia, the Trump campaign and Hillary Clinton’s emails. At times, today, the hearing turning ugly and personal.
<VIDEO with CONGRESSMAN talking> Mr. Chairman, this is outrageous.
<GRAPHICS: TRUMP DECLARES> Also tonight, President Trump declaring victory at NATO, saying other world leaders had just agreed to pay more. But what really happened? The French president, Emmanuel Macron, then saying there was no such agreement.
<GRAPHICS: DEADLY RIP> Deadly rip currents up and down the east coast tonight. The body of a swimmer pulled from the water. And, tonight, the teenager surviving nearly 10 hours after being pulled out to sea.
<GRAPHICS: DEADLY BOULDER> The tragedy on an American roadway, the driver under arrest. The 800-pound boulder falling out of the back of his truck, killing a mother and daughter.
<GRAPHICS: PAPA JOHN’S> The fallout tonight involving Papa John’s and its founder after using a racial slur during a conference call. What’s happened now?
<GRAPHICS: TOY STORE> The American toy store and its promotion growing out of control. After telling parents they could pay their child’s age, they could not handle the lines forming across America.
<GRAPHICS: MALL> And the shopping mall collapsing tonight. Worried shoppers had just been evacuated.
In some ways, these structures are reminiscent of the titles of newspaper articles. For example, consider the titles of articles from the New York Times given in (43)–(47), which all illustrate NCRSs constructed with noun phrases, prepositional phrases, or wh-clauses.
The Man Who May Challenge Putin for Power
The Revolutionary Power of a Skein of Yarn
On Trump’s Social Network: Ads for Miracle Cures, Scams and Fake Merchandise
Why the New Obesity Guidelines for Kids Terrify Me
How Barr’s Quest to Find Flaws in the Russia Inquiry Unraveled
Such examples show that the NCRSs in TVNBs versus newspapers are similar in their basic grammatical building blocks. However, NCRSs in TVNBs are noteworthy in three other respects: (1) the extreme complexity of individual NCRSs; (2) the extreme density of NCRSs over an extended stretch of discourse; and (3) the fact that these NCRSs are intended to be comprehended in real time by listeners. Newspaper titles are usually between 5 and 10 words long with a few embedded structures, in contrast to the NCRSs in TVNBs, which are frequently longer than 20 words with numerous embedded structures. The two are further different in that the newspaper title is immediately followed by the actual prose story, using standard canonical structures. The title functions to announce the general topic. In contrast, the opening segment of a TVNB consists of a barrage of long, complex NCRSs in sequence. These function as mini-synopses of the content of news stories, rather than as simple announcements of topics. The fact that listeners are able to comprehend discourse of this type in real time requires explanation; we return to that issue in the conclusion.
Figure 8.3 NP NCRS types in headlines versus other segments (rate per 1,000 words)
Figure 8.3Long description
The vertical axis marks rate per 1000 words and ranges from 0 to 40, in increments of 5. The horizontal axis marks the non-canonical reduced structures or N C R S, including simple N P, N P plus e d, N P plus i n g, N P plus P P, N P plus other, total complex. The vertical bar graph includes two types: a dark shaded one for headlines and a light shaded for other segments. The rate for headlines across the various N C R S begins from 5, remain constant, rises to 15, 18, drops to 4, and then peaks to 34. The rate for other segments across the various N C R S starts from 2, remain constant, rises to 6, diminishes to around 4 to 3, and then peaks to 11.
The extremely dense use of complex NCRSs in the ‘headlines’ opening segment might overshadow their use in other segments of TVNBs. However, Figures 8.2 and 8.3 show that NCRSs – including complex structures – are also quite common in the other segments of news broadcasts. This pattern shows that NCRSs are not restricted to ‘headline’ functions. Rather, they are frequently integrated into the normal discourse of regular news stories, as in the story in (48) about then-President Trump’s recent actions. Notice in particular the contrast between discourse produced in real time (by the people talking in the recorded videos) versus the discourse produced by the news anchor: all utterances in this story produced by the news anchor incorporate complex NCRSs, reflecting the fact that the language was probably pre-scripted. In contrast, recorded discourse produced by either Trump or Sanders is probably not pre-scripted, and shows no instances of NCRSs.
In-studio news story:
<ANCHOR> Tonight, President Trump vowing to take action after the Florida school massacre, which White House sources say impacted him personally.
<VIDEO with TRUMP> We must do more to protect our children. We have to do more to protect our children.
<ANCHOR> The President asking for regulations banning bump stocks, the devices allowing some guns to shoot hundreds of rounds per minute. The President had already called for a review of bump stocks after they were used by the Las Vegas shooter.
<VIDEO with TRUMP> I expect that these critical regulations will be finalized, Jeff, very soon.
<ANCHOR> Under fire from victims to do more--[incomplete]
<VIDEO with CROWD > Shame on you. Shame on you.
<ANCHOR> The White House now considering legislation that would strengthen background checks, but neither of those efforts would have stopped the Florida shooter.
<VIDEO with SANDERS> The President is trying to do everything that he can.
<ANCHOR> White House Press Secretary Sarah Sanders also facing questions tonight about the President’s tweet blaming the FBI for missing warning signs about the shooter because, quote, “They are spending too much time trying to prove Russian collusion with the Trump campaign. There is no collusion.”
<VIDEO with SANDERS> We would like our FBI agencies to -- to not be focused on something that is clearly a hoax in terms of investigating the Trump campaign and its involvement.
<ANCHOR> That controversial tweet one of eighteen in five days about Russian meddling following the indictments of thirteen Russian officials by Special Counsel Robert Mueller, including blaming his predecessor.
Examples like (48) show that complex NCRSs can actually be the normal strategy for constructing certain types of news stories, rather than a specialised device restricted to the opening segment that introduces the stories included in the news broadcast.
8.4.4 Use of NCRSs across TV Networks
The style of discourse illustrated in (42) and (48) above is a recent historical innovation, in addition to being peculiar in comparison to other spoken registers. In fact, some present-day news stories in TVNBs make minimal use of NCRSs, as in (49).
<ANCHOR> We begin tonight with a deadly collision on an interstate highway in New Mexico. A tractor-trailer truck hit a Greyhound bus head-on. Hospitals report at least four people were killed and at least thirty-five injured, at least three critically. The bus was carrying forty-seven people from Albuquerque to Phoenix. Mireya Villarreal has late details of this developing story.
<REPORTER with VIDEO> A horrific scene sprawled out over New Mexico’s Interstate 40 near the Arizona state line. This Greyhound bus’s front end ripped off, debris scattered across the highway as emergency crews desperately tried to reach victims.
<WOMAN talking > Oh, God.
< REPORTER with VIDEO> New Mexico state police confirm multiple deaths. The bus was carrying forty-seven passengers and heading to Phoenix, Arizona, from St. Louis. Many of the seriously injured were taken to area hospitals. One trauma center says it has received six patients, three in critical condition. Investigators believe a tractor-trailer crossed the median and hit the bus head-on. The tractor-trailer lost most of its haul, and another vehicle was left a mangled mess.
<REPORTER> Witnesses say it took a while to get some of the passengers out of the bus. Bystanders were actually wrapping children in blankets as they sat along the highway. And at least two hospitals sent their choppers in to pick up the most critical patients. John, clearly these victims are the top priority, but troopers will quickly move to the investigation to try and figure out exactly how this happened.
Our casual observations of news broadcasts indicated that their differential reliance on NCRSs is strongly associated with the preferred discourse styles of different networks (e.g., commercial versus public; all-news networks versus general networks). The present study focuses only on news broadcasts offered by the three major commercial networks in the United States. However, Figure 8.4 shows that there are major differences in preferred style even within that restricted sample. ABC exhibits a much greater reliance on complex NCRSs than the other networks, while CBS exhibits the least frequent use of these structures. Excerpts (42) and (48) above illustrate the style of discourse typical in ABC broadcasts, while (49) illustrates the more traditional and conservative style found in CBS broadcasts. NBC is intermediate between the two. In fact, (48) above is a story from an NBC broadcast, exhibiting an extremely frequent use of complex NCRSs, but many other NBC stories are much more similar in style to (49).
Figure 8.4 NCRS types across networks (rate per 1,000 words)
Figure 8.4Long description
The vertical axis marks rate per 1000 words and ranges from 0 to 45, in increments of 5. The horizontal axis marks the non-canonical reduced structures or N C R S, including total N P, w h-structure, i n g clause, other, simple N P, N P plus e d, N P plus i n g, N P plus P P, N P plus other, total complex, and deictic adverbs. and total complex. The vertical bar graph includes three types: a dark shaded one for A B C, a mild shaded one for N B C, and a light shaded one for C B S. The rate is maximum for A B C news in all domains and peaks for total N P. The C B S and N B C have approximately lower costs in all domains.
It is not merely a coincidence that these linguistic differences correspond to the advertised emphases of the news networks, with ABC emphasising ‘soft news’ and humanising the news, versus CBS emphasising coverage of ‘hard news’. Previous descriptions of these differences have focused on differences in the topics of news stories. For example, hard news stories tend to cover major national and international events and issues, while a soft news story might be about an alligator wandering into a swimming pool in Florida. However, the results here suggest a much more basic and pervasive difference between the emphases, with stories on essentially the same topic being presented with fundamentally different linguistic styles. Apparently, the extreme reliance on complex NCRSs in ABC broadcasts is intended to convey high human interest, perhaps by implying a sense of urgency that precludes production of complete canonical structures. In contrast, the reliance on canonical structures in CBS broadcasts conveys a no-nonsense reporting of ‘just-the-facts’, associated with the emphasis on hard news. Future research is required to explore the differing motivations and effects of these contrasting discourse styles. However, the results here clearly show that the dense use of complex NCRSs is not only unique to TVNBs, but that it is further restricted to particular kinds of broadcasts hoping to achieve specific audience effects.
8.5 Conclusion
NCRSs in TVNBs are interesting because they are ‘non-canonical’ in both of the two major senses introduced in Chapter 1 of this volume: they fail to conform to the minimally grammatical clause structures recognised by grammatical theory, and they represent types of grammatical structures that are extremely rare in most other registers. NCRSs are especially interesting because they are rare in face-to-face conversation, making it unlikely that they have a functional motivation relating to constraints of the production circumstances. Rather, it turns out that these are often extremely complex constructions, both structurally and syntactically, which could not normally be produced in spontaneous speech. Thus, their functional motivation seems related to creating a perception of urgency and excitement, coupled with compressing a lot of information into a single utterance.
NCRSs are further interesting because they are arguably becoming the ‘canonical’ form in certain types of TVNBs. In particular, news broadcasts from ABC make extensive use of NCRSs, in all segments of the broadcast, to the extent that reliance on these structures has become the unmarked style of discourse. Thus, from the perspective of language use, NCRSs could be portrayed as the canonical form of utterances in ABC news broadcasts.
The findings here are also interesting because they complement previous studies of grammatical complexity carried out from a Register-Functional perspective (see especially Biber et al. Reference Biber, Gray, Staples and Egbert2022). In particular, the use of NCRSs in TVNBs is noteworthy because:
(1) These non-canonical structures are more grammatically complex than those found in most other registers, and they occur with much higher frequencies.
(2) These non-canonical structures result in a type of text complexity that has not been found in other spoken registers – that is, a discourse style relying primarily on phrasal rather than clausal structures.
In the preceding sections, we have focused on the first of these two considerations. The corpus findings presented above document both the grammatical characteristics of NCRS types that had not been previously described and the characteristics of a register and discourse style that had not been previously noticed.
However, the theoretical implications of the second consideration are equally important. The findings here appear to contradict one of the major generalisations of previous research on text complexity carried out from a register-functional perspective: that all spoken registers, regardless of their communicative purpose, rely on clauses to construct text (including an extensive use of dependent clauses), in contrast to certain written registers, which rely on phrasal complexity (see the detailed descriptions of complexity variation in Biber Reference Biber1992, Biber & Gray Reference Biber and Gray2016, Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan2021, Biber et al. Reference Biber, Gray, Staples and Egbert2022, and Biber et al. Reference Biber, Larsson and Hancock2023, Reference Biber, Larsson and Hancock2024). This generalisation dates back to the Biber (Reference Biber1988) multi-dimensional study of spoken/written register variation, which concluded that the patterns of register variation within the spoken mode are fundamentally different from those within the written mode:
There is a difference between speech and writing in the range of forms that are produced in each mode. That is, there seems to be a cognitive ceiling on the frequency of certain syntactic constructions in speech, so that there is a difference in the potential forms of the two modes. … [This difference …] seems to be related primarily to the processing constraints of speech – to the fact that even the most carefully planned and informational spoken [registers] are produced and comprehended in real-time, setting a cognitive ceiling for the syntactic and lexical complexity ….
In short, spoken discourse is produced in real time, while written discourse is produced in situations that allow for extensive planning and revision. Writers have the possibility of taking as much time as they want to plan exactly what they want to write, and if they write something unintended, they can delete, add, revise, or edit the language of the text.Footnote 1 As a result, other factors – especially communicative purpose – can have a major influence on the linguistic characteristics of written texts. In contrast, all spontaneous spoken registers are produced in real time, which constrains the extent to which the speaker can vary linguistic characteristics, regardless of the communicative purpose.
This difference in production circumstances is especially relevant for the kinds of complexity features used in speech versus writing. Popular written registers (like fiction or blogs) employ many clauses, including finite dependent clauses. But informational written registers (like academic research articles) use relatively few verbs, and instead employ many phrases (noun phrases, prepositional phrases, adjective phrases), functioning mostly as modifiers of other noun phrases. In contrast, all spoken registers – regardless of their informational or interpersonal focus – have been found to rely on a clausal style of discourse (see Biber et al. Reference Biber, Gray, Staples and Egbert2022). Thus, the extreme reliance on phrasal text complexity is found only in written informational registers, apparently enabled by the extended production opportunities provided by the written mode.
One previously noted exception to this general pattern is the complexity profile of radio broadcasts documented in Biber (Reference Biber1992), which was intermediate between the characteristics of other spoken registers versus informational written registers in its use of phrasal complexity features (see Biber Reference Biber1992: 155, figure 3). The findings presented in the preceding sections of the present study indicate that TVNBs are an even more dramatic exception to this general pattern, representing a spoken register with an extremely frequent use of phrases and non-finite clauses, contrasted with a comparatively rare use of main verbs and finite clauses.
While we do not have space for a full discussion of this finding here, there are three major considerations that should be noted. First, the phrasal NCRSs in TVNBs are for the most part produced in writing. That is, they are originally scripted in writing, even though they are delivered and comprehended in the spoken mode. Thus, they do not represent an exception to the general claim that the dense use of phrasal complexity features is not normally feasible in spoken production. In this regard, it is interesting to contrast the previously scripted discourse spoken by the anchor in (48) – with a dense use of phrasal NCRSs – versus the recorded spoken utterances produced spontaneously by others, which never employ NCRSs.
This finding indicates that there is a major difference between the production constraints versus the comprehension constraints of the spoken mode. That is, the characteristics of TVNBs documented here indicate that the dense use of phrasal NCRSs poses little problem for the comprehension of discourse in the spoken mode, even though the findings here are consistent with the claim that such structures are difficult to produce in the spoken mode.
However, there are other differences between the phrasal complexity features in informational written registers versus the phrasal NCRSs common in TVNBs. The first difference is that informational written registers rarely use reduced non-canonical structures. Rather, written discourse relies on complete ‘canonical’ main clauses. Phrasal complexity features are common because information is packaged primarily in noun phrases that are modified by other phrases; but the discourse overall consists of complete main clauses with finite verbs.
This difference relates to the differing syntactic functions of phrases and non-finite clauses in TVNBs versus informational writing. Phrases and non-finite clauses usually function as modifiers of other phrases in informational written registers. In contrast, the NCRS phrases and non-finite clauses illustrated in the TVNB excerpts above are more readily interpreted as clause constituents rather than phrasal modifiers. We have deliberately avoided such syntactic interpretations in our analysis, choosing instead to limit ourselves to the structures found on the surface. However, consideration of the likely syntactic functions of constituents in these examples indicates that they are fundamentally different from the typical syntactic functions of phrases in informational written texts.
Finally, it is important to remember that the spoken discourse of TVNBs is almost always supported by videos and other visual images. While we do not have direct evidence of the importance of this characteristic for comprehension, it represents a major difference from most other spoken registers. Thus, it is likely that this characteristic is a major factor enabling the ready comprehension of NCRSs in TVNBs. In future research, we plan to explore these considerations in much more detail, and also undertake historical research on the emergence of complex NCRSs and the evolution of the register of TVNBs.
Open access