4 Results and discussionThis section consists of three subsections, where the results are presented and discussed from different angles. 4.1 is a presentation of the corpus in its entirety, with focus on the overall trends for the how the different coordination forms are distributed based on country. 4.2 is a breakdown based on phrase type, with a discussion of the relation between phrase type and coordination form. 4.3 is a brief discussion about whether the “tabloid” versus ”serious” classification seems to affect the use of coordination form.
4.1 By country
Table 2 [Table not shown]
Table 2 shows all the examined headlines and their distribution across the three forms (and, comma, &). Newspapers are ordered by country and there is also a total figure for each country. As mentioned in section 3.1, all newspapers have been examined through their RSS feeds, for the period stated under “period”.
When comparing the rows “Total number of hits” with “Number of headlines examined”, it can be seen that the constructions examined in this study were found in somewhere between 5,1% (50 exx in 972 headlines, the Telegraph) and 9,0% (50 exx in 556 headlines, Washington Post) of the headlines in the selected UK, US and Canadian newspapers. For the Australian newspapers the figure is much lower, only between 0,8% and 1,8%.
As mentioned, the &-form is only represented by a single newspaper, New York Post. Furthermore , all newspapers except Daily Mirror and the Guardian have at least one example of both comma and and forms. The UK newspapers show only 2 examples of the comma form spread over a total of 200 hits.
Fig. 1, Fig. 2, Fig. 3, Fig. 4 [Figures not shown]These pie charts (Figs. 1–4) show a country-by-country breakdown, showing the percentile distribution of the forms for each country. The US (Fig. 1) and Canada (Fig. 4) show a very similar distribution, with 73,9% and 75% comma form respectively. Australia follows a similar pattern, although with a stronger bias toward the comma form. In the UK newspapers, the dominance of the and form is, as mentioned, almost total.
The general trend in the US and UK results did not contradict the information gathered from internet sources, such as style guides and the BNC corpus search mentioned in the introduction to section 3. The extreme dominance of the and form in the UK newspapers indicates that the comma form is generally avoided in online UK newspaper headlines (although there might of course be newspapers that go against the grain). Against the statistical background , these two examples of the comma form could be viewed as more or less accidental:
(29) Van Morrison, Tom Jones top blues festival bill [TTNP1]
(30) Honolulu hiker drank sweat, rainwater wrung from clothes during 7-day ordeal in mountains [DMANP1]
As far as I can determine , the only thing (29–30 ) have in common is that they are both potentially ambiguous: (29) could be interpreted as either Van Morrison and Tom Jones performing together or in separate shows; in (30), it is not clear if the sweat was wrung from the clothes or not. It is possible that the comma is intended to somehow disambiguate these headlines, but as this is in the realm of speculation, I will not discuss this further.
Among the US newspapers, USA Today (see Table 1) is by far the most consistent in its use of the comma form, with 46 out of 50 commas. Although I have not counted the number of words in each headline in the corpus, USA Today seemed to have the shortest headlines of the US newspapers, with no headline exceeding 10 words. Since, as mentioned in section 2.2, the comma form seems to mainly be a space-saving device, one might expect a correlation between shorter headlines and more frequent use of the comma form in those newspapers that utilize it. The Australian results also indicate that there might be a correlation between shorter headlines and more frequent use of the comma form, as they have an even higher frequency of the comma form overall than the US newspapers, and also the shortest headlines of all countries examined. The longest headline among the Australian examples in this corpus is 8 words. It should also be noted that these headlines are, after all, the rare examples found with coordinated elements, and can thus be expected to be among the longer headlines found in these newspapers.
The striking similarity in distribution between the US (Fig. 1) and Canada (Fig. 4) is one of the more interesting results, since it indicates that US and Canadian newspapers may use a common North American headline style, at least when it comes to this feature. Unlike the UK newspapers, where the extreme scarcity of comma form examples made it hard to draw any conclusions, the and form examples in the US, Canadian and Australian newspapers constitute a category large enough to possibly show some trends regarding choice of form. Do the and form examples fall into some kind of special category, or are they simply cases where a comma would have been particularly farfetched, ambiguous or even unacceptable?
It is hard to see any clear pattern in the and form examples from the US, Australia and Canada. There are examples where replacing the and with a comma would seem forced, or would even make the headline difficult to comprehend:
(31) Astral buy means Bell owns HBO and a lot more [TSNP26]
(32) In Puerto Rico, Romney and statehood inextricably linked [WPNP25]
(33) In Alabama and Mississippi, Romney aims for one-third of delegates [WPNP31]
However , examples such as (31–33) comprise only about a dozen of the 73 and form examples from the US, Australian and Canadian newspapers in the corpus. The majority of the and form examples found in the newspapers that clearly prefer the comma form are cases where a comma could just as well have been used, such as:
(34) Southern sweep: Rick Santorum takes Mississippi and Alabama [LTNP33]
(35) Santorum wins in Alabama, Mississippi reshape GOP race [USTNP33]In other words, even when a comma form seems entirely possible, sometimes the newspapers that prefer the comma form will use the and form for no obvious reason, while the opposite is not true (at least nowhere near as much) of the and-preferring UK newspapers.
This seems to validate – or at least not argue against – the idea presented in section 2.1, that the and form is the unmarked form in this type of construction, as it seems likely that it is easier to consistently avoid a marked form than to consistently use it. Even though it has been established in section 2.2 that newspaper headlines have specific and unique traits and “rules”, I would argue that the sense of what is marked or unmarked in formal writing probably still has some effect on a writer of headlines. That headline writers in UK newspapers simply follow the company style guides much more strictly than their US, Australian or Canadian counterparts seems far-fetched. At this point, it is also interesting to revisit Washington Post’s reply to the email survey, referenced in section 2.2. The multiplatform editor claims that “We do use the word ‘and’ and ‘says’ when space allows”. That would indicate that the and form could be used with shorter headlines, but neither the corpus in general nor the Washington Post part of it seems to support this explanation, as indicated by these two examples:
(36) Sean Parker and Al Gore discuss ‘Occupy Democracy’ and the ‘hacking’ of U.S. politics at South by Southwest [WPNP32]
(37) Rebels, security forces battle in Damascus [WPNP4]
Of course, it is not possible to draw definitive conclusions from examples (36–37). Different company guidelines for headline length may apply for different newspaper sections or article placements, and this is beyond the scope of this thesis to investigate. I would nevertheless argue that, based on the data gathered for this thesis, there exists a general headline style, followed by all US, Australian and Canadian newspapers in this corpus, where the comma form is clearly preferred whenever possible. Sometimes, however , the and form seems to look better to the headline writer, for one reason or another. I will not speculate further about what those reasons may be.
4.2 By phrase type
Table 3 [Table not shown]
Table 3 shows the distribution of the four phrase types (NP, VP, AP, MP) for all the newspapers across the two main forms (&-form is excluded). The newspapers are also ordered by country, with the total sum for each country.
Fig. 5 Fig. 6 Fig. 7 [Figures not shown]These pie charts (Figs. 5–7) show the percentile distribution of the four phrase types for each coordination form, as well as the distribution of all the hits. The similarity of distribution for the phrase types among the different forms is striking; the distribution of NPs , VPs and MPs is more or less the same in both of the main forms and the corpus as a whole (Fig. 7). In Fig 7, the unusual &- form (which only had 6 hits: 4 NP, 1 AP and 1 MP) is also included.
The only notable difference is that APs are more than twice as common among the and form examples, but this is also the least common form, with only 29 hits, so here too the statistical ground becomes somewhat shakier. Sixteen of the 20 APs in the and form column are found among the UK newspapers, which means that they make up more than half of all the APs in the corpus .
No particular features or trends among the APs that would explain why APs are more common with the and form were found. One might suspect a link between long headlines and coordinated APs , since it is easy to imagine that longer headlines would be more likely to contain many adjectives, but the papers with the longest headlines in this corpus, Daily Mail and Daily Mirror, did not have more APs than the US papers. Instead, it was the Telegraph and the Guardian who provided most of the APs, with 6 each. Judging from these figures, there does not seem to be any significant differences between the coordination forms based on phrase type, but it makes sense to exclude the UK results from such a discussion because of the almost complete dominance of the and form in those newspapers . I will therefore now proceed to look at differences between phrase types in US, Canadian and Australian newspapers. and comma total
Table 4 [Table not shown]
In Table 4 , not only the UK newspapers, but also the anomalous &-form and the few multiple phrase (MP) examples have been removed. What is shown is the distribution of the preferred, but marked, comma form and the dispreferred but unmarked and form for the three main phrase types.
Fig. 8 Fig. 9 Fig. 10 [Figures not shown]
These pie charts (Figs. 8–10) show the percentile distribution of the two main forms (and, comma) for the three main phrase types.
Figs. 8–10 show that the dispreferred and form is distinctly more common in the NP category (25%) than in the VP category (10%), while it is most common in the AP category (33%). Since there are only 12 examples in the AP category, and based on the above discussion of the AP examples, it is difficult to draw any conclusions about the AP category, but the difference between VPs and NPs is interesting and warrants a closer look.
Some of the VPs in the US, Canadian and Australian newspapers suggest a connection between the more ample use of the comma form in the VP category and the fact that coordination becomes more semantically complex when verb phrases are involved. More specifically, a comma sometimes looks better than simply using and, because of the relation between the coordinated elements. Although an and may not change the meaning of the headline, it may have the effect of narrowing or “flattening” the meaning. As in this example:
(38) Terri-Lynne McClintic breaks down, describes sex assault at Stafford trial [TSVP6]
Here, replacing the comma with an and could give more of an impression of a simple, non-causal and non-temporal coordination: She broke down, and she also described the assault. This comma, on the other hand, seems more open to an interpretation such as and then or and as a result. Or to put it another way: The reader may be more likely to interpret the second verb phrase to follow the first, temporally or causally, if the comma is used instead of an and. Since the present tense is very common in headlines, the comma may also be a way out of the “flattened” temporal context that the often somewhat unnatural present tense of the headline forces upon the text. There are several examples in this vein:
(39) Pope arrives in Mexico, denounces violence [USTVP1]
This comma could be replaced with an and (and certainly not with an or or a but) but if we were to re-interpret the events behind this headline in normal, descriptive language, it would look something like this:
(40) The Pope arrived in Mexico, where he (then) denounced violence
Again, an and would certainly work, but it would arguably make the sentence look more “flat” than the comma does. One might argue (as perhaps someone already has ) that the most prototypical meaning of and is that of a simple coordinator, without temporal or causal meaning, whereas a comma is more open to different interpretations. Such a semantic discussion – interesting though it is – is beyond the scope of this thesis, however.
Regardless, the commas in (38–39) show a distinctly different semantic potentiality compared to this following example, where it is hard to imagine any temporal or causal charge in the comma, and replacing it with an and does not seem to change it at all:
(41) Gingrich stays on, sets sights on state votes to come [WPVP4]
There are some examples in the corpus of other reasons for preferring a comma, such as there being another and in the headline, but these “semantically versatile” commas are both more common and more interesting. This phenomenon may, to some extent, explain why the comma form dominance among the newspaper that prefer this form was significantly stronger for VPs than NPs, since coordinated NPs cannot express temporality or causality in this manner.
4.3 “Tabloid” vs. “serious”
Nothing has been found in the corpus to indicate that the classification of a newspaper as “tabloid” or “serious” has any bearing on the choice of coordination form. It seems to be more relevant what kind of headline style a newspaper uses. Shorter headlines, like USA Today, for example, seem to lead to more use of the comma form and less use of the and form. USA Today would by most be considered more serious than New York Post, but less serious than Los Angeles Times, but it was the most consistent in its use of the comma, while New York Post and Los Angeles Times had very similar results.
The one thing worth mentioning in this context is the New York Post’s use of the &-form. However , without any background or information about the possible use of the &-form by other newspapers, there is not much to comment on. The six examples of this form did not have anything particular in common, except for two instances where the &-form is used with first names, giving perhaps a stronger impression of the two persons comprising some kind of team or unit, as in this example:
(42) x-NAACP big rips Al & Jesse for handling of Trayvon Martin shooting [NYPNP34]
It seems unlikely that any newspaper that tried to uphold a “serious” image would refer to Al Sharpton and Jesse Jackson in this manner. Perhaps the &-form can be viewed as part of this informal tone.