Discourse, Context & Media 44 (2021) 100539

Contents lists available at ScienceDirect

Discourse, Context & Media

journal homepage: www.elsevier.com/locate/dcm

Cut to the chase – How multimodal cohesion secures narrative

orientation in film trailers
Christian R. Hoffmann
Universität Augsburg, Philologisch-Historische Fakultät, Lst. Für Englische Sprachwissenschaft, Universitätsstraße 10, 86159 Augsburg, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Recent years have seen a growing number of studies explore the narrative, persuasive and multimodal
Received 1 March 2021 design of film trailers (Maier, 2009, 2011; Wildfeuer and Pollaroli, 2017). While such work has primarily
Received in revised form 23 July 2021 looked at the ways in which trailers realize persuasive functions (Hediger, 2001; Maier, 2011; Krebs,
Accepted 9 August 2021
2020), this paper examines how the (audio-visual) representational choices made by trailer editors can
Available online 6 October 2021
provide narrative orientation for the film audience. Inside a self-compiled corpus of 150 genre-
stratified US-American film trailers, recurring audio-visual, cohesive patterns are elicited. It will be
Keywords:
shown that the visual effects of this cohesive thrust can be tied to the cinematographic choices editors
Film trailer
Telecinematic discourse
make in the opening shots of the trailer, by using specific shot scales, (types of) settings and (non-)
Multimodality human represented participants. To capture the verbal dynamics of trailer cohesion, the study explores
Film the key semantic domains to which all lexical expressions in the trailer dialogues can be linked
Cohesion (Rayson et al., 2004). Results of this study confirm that both visual and verbal elements form recurring
Coherence genre-specific patterns. They unleash their multimodal potential of film trailers by tapping into the pre-
Semantic domains vious (genre) knowledge of trailer audiences, facilitating the cognitive uptake of audiovisual information
Corpus-assisted multimodal discourse for the audience as the trailer unfolds.
analysis
Ó 2021 Published by Elsevier Ltd.

1. Introduction Most importantly perhaps, trailer editors must balance trailers’ per-
suasive appeal against their narrative plausibility, so that the film
Film trailers (also called movie previews) play a decisive role for audience is enticed to go and watch the feature film the trailer pro-
today’s film industry in promoting feature films in movie theatres, motes without losing track of the film’s broad storyline and charac-
on home television or via Internet streaming services (Hediger, ters. In this study, it is argued that cohesion plays an instrumental
2001; Hixson, 2006; Karray and Debernitz, 2017; Redfern, 2020). role in balancing this structural ‘‘conflict”. To this end, trailers’ use
Despite their growing economic and cultural impact, linguistic of audio-visual cohesion will be analyzed systematically to show
research on film trailers is scarce, with few notable exceptions how recurring cohesive tracks enable film audiences to quickly rec-
(Maier, 2009; Tseng, 2013; Wildfeuer and Pollaroli, 2017). Despite ognize trailers as part of a specific film genre. This recognition, it is
this lack of research, film theorists have argued for a long time that shown, enables film audiences to understand trailer storylines and
film trailers constitute a veritable film genre in its own right that character types more quickly when they can integrate them into
deserves more scholarly attention (Kernan, 2009: 45, Hediger, their previous knowledge of film genres.
2001: 15).1 This study contributes to closing this research gap. To For the first time, the paper explores multimodal cohesive pat-
do so, it investigates the structural make-up of film trailers. As terns in a cross-generic corpus of 150 US-American standard film
Hediger (2001: 192ff) shows, the production of effective and persua- trailers (Augstrail corpus, 38,737 tokens). The study follows
sive film trailers is a complex and demanding task since editors Bednarek’s (2015:73) corpus-assisted multimodal discourse anal-
strive to make trailers comprehensible and entertaining while, at ysis (CAMD) which is applied to film trailers in this study. It com-
the same time, overcoming compositional obstacles along the way. bines the use of corpus linguistic methods (to analyze the verbal
(audio) track of the trailers) with a systematic investigation of
patterns of (visual) representation styles in trailers’ opening
E-mail address: [email protected]
1
Perhaps their strange nature as para-texts (Genette, 1997), their prominent use in shots, i.e. shot scales, settings (non-) human represented
mass-media entertainment, and trailer’s semblance of formal simplicity may have participants.
prevented further academic scrutiny in the past.

https://doi.org/10.1016/j.dcm.2021.100539
2211-6958/Ó 2021 Published by Elsevier Ltd.
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

2. Introducing film trailers mental process of meaning-making (Widdowson, 1980, Stubbs,

1983, Bublitz, 2006). With this renewed interest in matters of
Maier (2009: 159) defines film trailers as ‘‘multimodal texts in coherence also came a gradual shift of studies in the field. For-
which several semiotic modes are combined, and parts of texts cre- merly, studies had focused on the formal description of cohesive
ated for other purposes are transferred, rearranged and supple- links in text documents, but new research began to privilege ques-
mented in order to attain a promotional purpose.” Trailers have tions of comprehensibility, inference and interpretation informed
been classically used in the US film industry to promote future by new insights from psycholinguistics and cognitive studies (cf.
releases of feature films. Today they are still regularly shown in Hoey, 2003, 2005).
film theatres at the beginning of film screenings (see Krebs, Atthesametime,scholarschartedthecohesivelandscapeofdiffer-
2020). In their early days (1920’s and 1930’s), however, trailers ent text genres (Neumann and Fest, 2016) and adopted corpus-
would have been projected in movie theatres between the screen- linguistic methods to capture cohesive chains, collocations, intertex-
ing of what was known as ‘‘double features”, i.e. two short films tual patterns and discursive trends on a larger scale (see Flowerdew
screened in a row (Hediger, 2001: 61). and Mahlberg, 2009; Hoffmann, 2012; Tanskanen, 2006). More
Film editors usually distinguish different types of trailers, which recently, more attempts have been made to include visual resources
serve different promotional purposes. Teaser trailers, for instance, ofexpressionin the analysisof cohesion,elicitingmore complex,mul-
are the first trailers to be released in public (Kernan, 2009: 242) timodal patterns of cohesion, which work particularly well in digital
and they tend to be rather short (less than 2 min). Their primary discoursegenres(cf.Kress,2010;Janney,2010;Tseng,2012).Filmdis-
function is to arouse the interest of the viewing audience and coursecohesion,or in ourcase, cohesionin filmtrailers,canserveas an
evoke the general ‘‘feel”, mood or ambience or the feature film interestingcaseinpoint.Maier(2009),whowasoneofthefirstlinguists
being promoted. Standard theatrical trailers, exclusively used in this toexaminethemultimodalcompositionoffilmtrailers,hasshownhow
study, are typically longer than teaser trailers (approx. 2–3 min) various techniques, e.g. close-up shots, camera movements, captions,
and are commonly released in film theatres and on television chan- transitionsandspecialeffects,areco-deployedtoexpressvisualevalu-
nels. In contrast to teaser trailers, theatrical trailers commonly pro- ationsintrailers.Shealsoprovesthatshotscanbe‘‘strengthenedorsub-
vide more details about the general storyline of the feature film verted by evaluation provided by the non-diegetic voice-over
and its main characters. While some standard trailers are screened commentsor off-screen wordsuttered by characters [. . .]” (175).
exclusively in the USA (also called national trailers), international In a different paper, Maier (2011:153) reveals how the narrative
trailers are typically used in all regions around the globe. Other staging in film trailers can inform its persuasive shape and how the
forms of trailers include creative trailers, which are filmed and edi- functional relations that hold between the different narrative
ted by directors of the final feature films, video clips, i.e. short dia- stages of the trailer can trigger some of the persuasive effects the
logue shots, to be used for promotional purposes in television talk trailer genre is known for. In contrast, Janney (2012) describes
shows and featurettes, which provide rich insights into the (post-) how classic verbal cohesion can be fruitfully applied to capture
production of a given film. Before we return to trailers to introduce visual patterns of cohesion across film shots. Tseng (2013) provides
their cohesive structure in section five and six of this study, we a similar but more refined classification scheme to capture and
must first address the way in which linguists have explored cohe- classify a range of co- referential chains of cohesion across a series
sive patterns in the past. of film shots. While both Janney (2012) and Tseng (2013) have
been influential benchmarks for the study of multimodal cohesion
in films, they do not discuss the way cohesion is strategically
3. From verbal to visual cohesion deployed in film trailers. Wildfeuer and Pollaroli (2017) seem to
partially address this issue in their analysis of argumentative pat-
Cohesion in this study is regarded as ‘‘the texture that is con- terns which provide structure to the audio-visual progression of
structed by the ‘reappearance’ of film elements as a film unfolds, shots in the trailer of the feature film Gravity. Nevertheless, their
identifying, tracking and classifying such reappearances as they approach to studying cohesion by using argumentation theory is
occur” (Bateman, 2014: 321, my emphasis). Note that the term decidedly different from the one adopted in the other studies. In
‘‘film elements” does not relate to visual resources exclusively this study, I would therefore like to move beyond the existing work
but comprises other semiotic resources as well, e.g. spoken mono- in three major ways:
logue or dialogue, musical styles or chords, repetitive sounds. In
addition, filmic elements need not recur between different shots (1) The study does not investigate single trailers or reduced sets
of one and the same film trailer but can likewise re-appear in sim- of trailers but draws on a larger repository of film trailers to
ilar patterns across film trailers that belong to the same (sub)genre, examine their cohesive profile.
In this latter view, a ‘‘vertical” reading of cohesive patterns is tar- (2) The study examines the specific shape of audio-visual cohe-
geted to generate cohesive profiles of film genres over time sion relative to three different trailer (sub)genres, i.e. action
(Hoffmann, 2012: 155). trailers, comedy trailers and documentary trailers.
Note that such a broad reading of cohesion pays tribute to the (3) The study attempts to combine a corpus-driven analysis of
way the concept has evolved in discourse analysis. Firmly estab- verbal cohesion with a multimodal analysis of visual
lished by Halliday and Hasan’s seminal work (1976), studies on resources of representation, reflecting the audio-visual char-
(verbal) cohesion have shaped the linguistic subdiscipline of dis- acter of any film document.
course analysis for decades. While early work focused on discover-
ing the (semantic) bonds that hold between clauses, sentences or Before the methods used in this study will take center stage, I
utterances, later research (in the 19800 s and 19900 s) explored will spend a few lines first, reporting on the sampling methods
how (and to what extent) such cohesive ties could possibly consti- and the design of the corpus compiled for this study.
tute a text’s conceptual, general or social meaning (cf. Hasan, 1984;
Bublitz, 1994, 1999; Hoffmann, 2012). Scholars additionally
reflected on the intricate relationship that holds between concepts 4. Corpus design
such as cohesive texture or cohesion, which were classically con-
fined to the text’s internal structure and organisation, and coher- The corpus of film trailers (Augstrail) for this study draws on
ence, which was often seen as the individual, dynamic and three major youtube channels, that offer access to large online
2
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

repositories of US-American standard film trailers, i.e. www.trai- tion of trailer dialogues. In the following, both methods will be dis-
ler-track.com, www.traileraddict.com and www.fandango.com cussed in due order.
(cf. Bednarek, 2018). The channels are searched for standard (the-
atrical) trailers, that exhibit a duration between two and three 5.1. Representational strategies in opening shots
minutes. Teaser trailers and other trailer types (see Section 2) are
precluded from the corpus for different reasons: First, standard The idea that film shots in movies do not simply represent an
trailers are more frequently used in movie theatres and on televi- eclectic, random and unrelated series of photographic images but
sion than any of the other trailer types. Second, standard trailers carefully crafted multimodal composites is not new. In 1919, the
also provide more substantial information of the film’s main plot Russian film director Oleg Kuleshov (cf. Bordwell, 1974: 9) noticed
and characters than teaser trailers. Understanding standard trailers that film viewers appraise a film sequence differently if selective
is thus more cognitively demanding than watching teaser trailers. shots of the sequence are added, deleted or replaced at different
Finally, standard trailers are more easily accessible and more temporal intervals. It follows that montage, or the lineal arrange-
widely distributed on online portals than the other types of trail- ment of shots in a film sequence can influence the way audiences
ers. Since more recent trailers likewise tend to be more accessible perceive characters, understand events or interpret storylines. Ever
than older ones, the corpus contains US-American film trailers since, lots of ‘‘schools” and traditions of film editing have held
released between the years 2017 and 2020. The final corpus con- quite distinct views on how film shots ought to be arranged but
tains 38,737 tokens, dispersed across 150 trailers (50 action trail- also how individual shots should be composed to induce visual
ers, 50 comedy trailers, 50 documentary trailers). The individual effects and gain control over the visual ‘‘reading” of images
trailers selected for the corpus were assigned to three main film (Kress and van Leeuwen, 1996: 214). To capture the representa-
trailer genres, defined in the following: tional choices that are involved in the visual composition of film
shots, Bateman et al. (2017: 331) provide a useful categorisation
(a) documentary trailers, i.e. feature films that report on histori- scheme of different types of shot scales (see Table 1); each of the
cal, social or political real-life personalities or events, scales in the scheme portray individuals, groups, objects or events
(b) action trailers, i.e. feature films that focus on one (or a group from a different depth and perspective. Some framings depict
of) fictional character(s) that must overcome a challenging objects from a vantage point far away (extreme long shots), some
or physically demanding obstacle to save (or help) others, from a closer range (simple long shots to medium shots) and some
(c) comedy trailers, i.e. feature films that typically revolve from a position very close to the object of interest (close-up shots).
around a funny or entertaining series of events, often includ- Shot scales cannot only involve the viewer in an event or distance
ing relationship conflicts that are resolved in humorous the audience from what is shown, they can also be used to provide
fashion more (spatial) information to the viewer or constrain their visual
field in close-up shots.
In addition to these genre descriptions, genre classifications Another visual parameter this study explores is Kress and van
suggested by the International Movie Database (www.imdb.com) Leeuwen’s distinction between represented and interactive partici-
were consulted, especially in cases in which genre classification pants (1996 [2005]: 119). Kress and van Leeuwen (1996) use the
remained unclear. In such cases, the most likely option of two term interactive participants to describe the participants engaged
likely genre categories was selected.2 The three genres of action, in the main act of a given communication. In the case of film trail-
comedy and documentary trailers were selected since their semantic ers, interactive participants refer to what Dynel (2014: 39) calls the
extension is much broader than other possible genre candidates, ‘‘collective sender”, i.e. the team of individuals that create the film
such as romance or crime. Main genres, such as action or comedy, trailer for a projected target audience. At the same time, a second
have also become what Bordwell calls ‘‘central Hollywood genres” (type of) interactive participant is constituted by the film audience,
which exert their influence ‘‘at all scales of [film] production” that watches and tries to understand the trailer.
(2012: 396). If genres are (conceptually) close, e.g. action & thriller In contrast, represented participants ‘‘are the subject of com-
films, recurring categorial overlaps seem unavoidable. To restrict munication, that is, the people, places and things (including
such effects of genre hybridization (Rodman, 2011: 35), genres were abstract ‘things’) represented in or by the film shot” (Kress and
chosen that are sufficiently different from each other in terms of van Leeuwen, 1996 [2005]: 119). Two types of represented partic-
their narrative organization and their default character types. ipants can be distinguished, i.e. human and non-human represented
The transcript function offered by youtube proved helpful in participants. The former refers to human individuals represented
obtaining verbal transcripts of spoken monologues and dialogues on screen, while non-human represented participants relates to
from the trailers, but these can be of poor quality. To meet the all other (non-human) sentient beings or objects, including ani-
quality requirements, all transcripts had to be manually checked mals, animate or inanimate objects, etc.
for accuracy, fluency and consistency and, if necessary, were for- Furthermore, different types of film settings are examined, i.e.
mally adapted. This included further mark-up, such as character/ different private or public places, urban or rural settings in light
speaker labels or spelling or punctuation changes. or dark shadings, pictured at different times of the day. All visual
features are systematically analysed and classified in all 150 open-
ing shots of the Augstrail corpus. Then, the visual analysis is
enhanced by a systematic look at the semantic patterns of frequent
5. Capturing audio-visual framing devices in film trailers lexical expressions in the trailer dialogues; a corpus-driven
approach which is explained in the next section.
This paper uses two different yet related methods to analyze
the audio-visual structure of film trailers: A visual analysis that 5.2. The key domain method - detecting cohesion in trailer dialogues
classifies the representational choices editors make in opening
shots, and a verbal analysis that centers on the semantic classifica- The trailer dialogues of each trailer are fed into three subcorpus
files, each representing one of the three trailer genres, introduced
2
The use of IMDB genre categorisations is standard practice in film corpus
before. Paul Rayson’s WMatrix tool (Rayson et al., 2004) is used
compilation (see Bednarek, 2018, URL: http://www.syd-tv.com/wp-content/uploads/ to automatically assign semantic domains to each lexical expres-
2019/04/Guide-to-SydTV_april19.pdf). sion. To this effect, it draws on a preset range of semantic tags,
3
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Table 1 ers (50% of all cases). In documentary trailers, extreme long shots
Shot Scale Classification (Bateman et al., 2017: 331). occur most rarely in opening scenes (40% of all cases). In addition,
Area Technical Brief description comedy and documentary trailers use (simple) long shots twice as
feature often as action trailers. The popular use of extreme long shots in
Shot Extreme long a shot taken from a considerable distance, giving the action subcorpus could be linked to the genre’s proclivity to
scales shot (ELS) an overall sense of the setting (establishing shot) begin trailers in the middle of a fast-paced action sequence, in
Long Shot (LS) a shot showing an entire action or entire which story characters must move swiftly from one place in time
character with some background visible
Medium Shot a shot typically framing a character from the
to the next. In such cases, extreme long shots can provide easy spa-
(CU) waist up, showing the character’s face and tial orientation for viewers and allow them to track and trace the
gestures spatial position of characters from one location to the next.
Close-Up (CU) a shot showing just the face of a character, In comedy trailers, extreme long shots are equally used to
drawing attention to expression and emotion
reveal story locations. Nevertheless, they are more likely to intro-
Extreme a shot showing a fine detail or part of a face,
Close-Up focusing specifically on what is shown duce a scene in a spatially constrained indoor setting such as a
(ECU) bar, kitchen or living room. Extreme long shots are also frequently
employed as aerial overhead shots of cities, towns, quarters or
external shots of houses, mansions or apartments. In the documen-
explained in Archer et al. (2005: 3).3 A reference corpus serves as tary subcorpus, trailers equally tend to begin in spatially con-
semantic benchmark to determine which semantic domains are strained indoor settings although documentary trailers have a
found to be more (or less) salient in each of the genre files, comput- clear preference for interview settings, whose dynamic interac-
ing the actual overuse or underuse of semantic domains in the cor- tions are better captured by long or medium shots.
pus at hand. For our purposes, the genre-stratified subsections of the We can see that the different trailer genres in the corpus appear
Augstrail corpus are used as suitable reference corpora for each to afford different opening shot scales that are conducive to the
other. This is feasible since the three corpus genre sections all consti- way editors wish to introduce the plotline of trailers. The deliber-
tute dialogues of the same type, are of similar size and origin, and ate, recurring choice of the same shot scale in opening shots can
differ only in regard to their individual genre profiles. To illustrate, thus serve an important cohesive force that frames the trailer at
the action and comedy genre files are collectively used as a reference a fairly early stage of the viewing experience; and it might induce
corpus for the semantic categorization of the documentary (sub)cor- audiences to activate different film genres (action: extreme long
pus. In contrast, the comedy and documentary subcorpus – in shots, comedy: mostly simple long shots, documentary: simple
combination- serve as reference corpus for the action corpus, etc. long or medium shots). As we shall see, this scaling effect is rein-
Positive key domains in the results are automatically highlighted forced by editors’ choice of settings and their strategies to repre-
by the WMatrix tool, and they indicate that a given semantic sent participants in the opening scene of trailers.
domain is used more frequently in the target corpus than in the Table 3 reveals the most frequent types of settings for the com-
reference corpus. Negative key domains suggest the opposite. When edy subcorpus. The most frequent category is PRIVATE HOMES, and it
positive key domains are found in a target corpus, the respective relates to external or internal shots of private family housings, flats
semantic domains can be assumed to be foregrounded in the or apartments. Comedy trailers use family home settings signifi-
respective dialogues, and thus can be expected to be particularly cantly more often than the other two trailer genres (p1 = 0,0061;
relevant for the trailers (or indeed genres) under study. Using these p2 = 0,03).This result may be explained by the traditional focus
frequency results as a springboard, the lexical expressions which of the comedy corpus on family members or close friends that live
frequently realize these semantic domains are examined in their or meet in the private settings presented. In contrast, action trailers
respective contexts of use. rely heavily on overhead shots of large cities, towns, featuring
skyscrapers, buildings, streets and cars at night-time; areas which
are often stylized as a somber, dark and dangerous places. While
6. First contact - visual representation in opening shots comedy trailers likewise use cities and towns in bright daylight
as a backdrop to their stories, the places are transformed into open,
Most film trailers in the Augstrail corpus comply with what friendly and inviting settings, using up-beat musical scores and
Bordwell calls ‘‘the premises of ‘classical continuity’ film editing lighting effects to sustain such an ambience. Very few action trail-
(2002: 16), so that trailers that begin with opening shots are ers rely on natural landscapes, including forests, lakes, cliffs or
designed to ‘‘cue us to where we are” (Thompson, 2003: 34). Ful- mountains; and when they do they typically act as a visual back-
filling this function, opening shots are typically ‘‘far-distance, long drop for ensuing action sequences (see Table 3).
shots that ‘‘establish place, often time, and sometimes other neces- Documentary trailers contain quite different settings in their
sary information” (Monaco and Lindroth, 2000: 210). However, opening sequences. The most frequent choice includes an array
film trailers can likewise start with medium or even close-up shots. of cultural venues, e.g. theatre stages, rock concerts or cinemas.
The next section reveals which shot scales are used as opening Political and social documentaries seem to prefer institutional set-
shots in the Augstrail corpus; it also discusses what visual (and tings, e.g. press conferences, political speeches, company
narrative) effects such choices may trigger. announcements. They also include workplace environments, such
as factories, corporate buildings, banks, etc. The latter are often
6.1. Shot scales and settings in film trailer genres used as metonymic references to the companies or institutions
which these buildings represent and whose potential (mis)conduct
This section explores trends in the representational strategies the film seeks to reflect or criticize. Clearly, nature documentaries
used by film editors in the first shots of trailers. Table 2 shows typically feature natural landscapes in their opening shots.
the respective choice of shot scales by genre. As can be seen, What we can see is that trailer genres do not only use different
two-thirds of the action trailers employ extreme long shots, while shot scales in their opening sequences, but also make different use
they are less frequently deployed in the first shots of comedy trail- of particular settings, times and lighting schemes to signal their
generic shape and design. In the following, we will continue this
3
For a very comprehensive description of the annotation process of key domains, analytical thread, exploring which humans, objects or events are
using the USAS semantic tagger, consult Leech (2013: 23). typically represented in openings shots. As can be seen, the choice
4
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Table 2 Table 4
Genre-Specific Dispersion of Shot Scales in trailer opening shots (Augstrail corpus; Genre-related use of human represented participants in the opening shots of trailer
150 trailers). genres.

Shot Scale | Trailer Action/Thriller (50 Comedy (50 Documentary (50 Shot Scale | Trailer Genre Action/Thriller Comedy Documentary
Genre trailers) trailers) trailers) (50 trailers) (50 trailers) (50 trailers)
Extreme Long 33 25 20 Extreme Long Shots (HRPs) 1/33 4/25 5/20
Shots (total)
Long Shots (HRPs) 5/9 14/17 14/17
Long Shots (total) 9 17 17
Close-Ups (total) 8 8 12 Close-Ups (HRPs) 5/8 9/8 7/12
Other scales 0 0 1
(total)

ture two other presentation styles: One shows multiple film char-
Table 3 acters assembled at a social event, e.g. marriage, Christmas party,
Genre-related use of settings in the Augstrail corpus (2 shots in Documentary genre family reunion, typically located in a (semi-)private setting (see
show no discernible background). Fig. 2, shot 2). The other option presents story characters engaged
Settings | Trailer Genre Action/ Comedy Documentary in talk (see Fig. 2, shot 3). In the corpus, such dialogue shots were
Thriller one of the preferred choices in opening shots.
Private Home (Living Room, Bedroom, 3 15 5 While dialogue shots are equally used in documentary trailers
Kitchen, etc.) (see Fig. 3, shot 1), they come in a different shape, using footage
Natural Landscapes (Parks, Trees, 12 6 8 of prominent real-word film stars, politicians, scientists or artists
Lakes, Mountains, etc.) (Fig. 3, shot 2). Documentary trailers routinely film these HRPs in
Workplace Environments 3 3 5
public contexts, for instance at institutional meetings or events
Towns, Cities, Streets 19 12 5
Institutional Settings 1 2 5 in which larger crowds of people are co-present (see Fig. 3, shot
Futuristic Settings 2 0 0 1 and 3).
Famous Sites, Memorials 1 1 2 HRP numbers and choices have been found to stratify across the
Culinary Locations (Restaurants, Bars, 1 3 0
genre corpora in distinctive and meaningful ways. Similar to shot
Hotels)
Entertainment (Theatres, Cinemas, 2 1 10 scales and setting types, they constitute recurring cohesive tracks
etc.) for each trailer genre. More specifically, we can see that the action
Sports (Clubs, Training Grounds) 0 1 2 genre prefers to display its main characters either in private, reflec-
War Zones 2 0 1 tive settings or in fast-paced action sequences, explosions, or car
Vehicles (inside) 0 2 3
chases. HRPs in comedy trailers are shown to take part in private
Workplace Environments (Factories, 2 3 2
Stores, Shops) social events or talk sequences. Documentary trailers typically
Ruins, Castles & Churches 2 1 0 show lead characters in large crowds or social gatherings in corpo-
rate or political settings or in closed interviews. Section 6.3 will
of represented participants can enrich the cohesive profiles for our now shift our focus from human to the representation of non-
three trailer genres. human represented participants (henceforth: NHRPs) and the
way these are featured in the opening shots of the corpus.
6.2. Human represented participants in opening shots

Table 4 shows the different use of human represented partici- 6.3. Non-human represented participants in opening shots
pants across the trailer genres of the corpus. In the corpus, comedy
and documentary trailers use human (represented) participants In the Augstrail corpus, action trailers frequently feature NHRPs
(henceforth: HRP) more frequently than action trailers. In all three related to the semantic domain of WAREFARE, DEFENSE & THE ARMY. Apart
subcorpora, HRPs appear more prominently in simple long and from the regular display of soldiers or mercenaries, we can find
close-up shots than in extreme long shots. This difference seems military gear and vehicles, naval warships, army vessels, bombers,
particularly pronounced in the action subcorpus, which features rockets and weapons of all kinds (NHRPs). In fact, action trailers
only one HRP in an extreme long shot. Following this rationale, that did not use either weapons, military equipment or aggressive
one could infer that the smaller the shot scale, the more likely fight scenes were exceedingly rare. In extreme long shots, typical
the use of HRPs in opening shots. This sounds plausible since NHRPs are skyscrapers, apartment buildings as well as commercial
extreme long shots are primarily used by editors to provide spatial or industrial buildings that can be classified under the rubric of AR-
orientation by using aerial shots and wide-angle lenses. Instead, CHITECTURE& BUILDINGS. Additional, yet less frequent NHRPs include
simple long and close-up shots tend to zone in on the (non)verbal tables, chairs, counters, fridges, cupboards that can mainly be sub-
(responsive) behaviour of film characters and are thus more likely sumed to the domain of FURNITURE. The least frequent category NATURE
to include and highlight HRPs. contained objects, such as trees, lakes, mountains, rocks or waves.
There are three typical ways HRPs are represented in the action Comedy trailers frequently depict both HRPs and NHRPs
trailers of the corpus: the first option shows the main character(s) assigned to the semantic domain of HOME AND HOUSE. As such, many
in a relaxed or reflective mood (see Fig. 1, shot 1). The second shots include film characters lying on beds, couches, sitting on
option presents a story character engaged in some form of (mate- chairs or resting in swimming pools. When they do, characters
rial) action, e.g. walking, moving, fighting, etc. (see Fig. 1, shot 2). are usually not alone but accompanied by family members, rela-
The third option presents a random group of individuals in a casual tives or friends. Since private residences, houses, manors, and
or professional setting, whose daily routine is suddenly shattered apartments are frequently used in extreme long shots, they can
by a disruptive event (see Fig. 1, shot 3). While option one and likewise be aligned with the category of ARCHITECTURE & BUILDINGS. A
two are dominant in the action subcorpus, option three is clearly few establishing shots use visual elements from the semantic
used more rarely. domains of NATURE or WORK & ENVIRONMENT, picturing gardens, lawns,
While comedy trailers also repeatedly draw on the second parks, lakes or exterior shots of shops, such as pawn shops, karate
option described before (see Fig. 2, shot 1), they additionally fea- schools.
5
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Fig. 1. Action Trailer Opening Shots including human represented participants.

Fig. 2. Opening Shots in Comedy Trailers, including human represented participants.

Fig. 3. Opening Shots in Documentary Trailers, including human represented participants.

In documentary trailers HRPs and NHRPs are more varied than generate a cohesive effect that affects not only individual trailers
in the two other trailer genres, and perhaps this is based on a but works to define the prototypical thematic core of trailer genres.
greater range of topics covered in the promoted documentary Patterns of this nature can be likened to the cohesive category Hal-
films. HRPs can involve a range of different human individuals liday (1976: 286) originally labelled collocation. It clusters expres-
from the pop icon down to the ‘‘ordinary person”. The diversity sions that ‘‘share some recognisable lexico-semantic relation”. This
is equally corroborated by NHRPs that include objects as diverse time, however, the concept does not only capture cohesive rela-
as cars, engines, sunglasses, chairs, trees, shelves, books, moon tions within single texts but lexical links which span across differ-
shuttles, microphones, buses or rocks. Despite such variability, ent film trailers and can even spread beyond different semiotic
NHRPs are most frequently assigned to the domains of VEHICLES & resources. So, in this study, the original scope of collocation will
TRANSPORTATION for external shots while FURNITURE as well as POLITICS & be expanded considerably to include both verbal and visual modes
of expression.
GOVERNMENT serve as semantic designators for objects displayed in
internal shots. ECONOMY and SCIENCE are appropriate candidates for
objects such as banks, dollar notes, stock exchange, space shuttles
7.1. Key semantic domains in the action trailers
or scientists, which appear more rarely in the corpus.
Table 5 reveals the top ten most frequent semantic domains in
7. Key domains and their lexical expressions the Augstrail corpus as expressed by the lexical expressions found
in the entire dialogue corpus of all trailers explored in this study.
While we have so far looked at the cohesive patterns that Results were filtered by assessing the statistical significance of lex-
emerge from the visual design of opening shots in different trailer ical overuse and underuse in the target corpus with the help of the
genres, this section examines the lexical expressions which realize log-likelihood test (for methodological details, see Section 5.2).
these semantic domains in the accompanying trailer dialogues of Table five shows that the most frequent semantic domain in the
the corpus. It will be investigated how the prominent use of lexical action corpus is WARFARE, DEFENSE & THE ARMY. The most frequent lexical
expressions link up with the genre-bound semantic domains item is war (see Fig. 4), and it is also used in compound expres-
explored so far (cf. Rayson and Smith, 2006). Key semantic patterns sions, such as war criminal. Other tokens include various word
6
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Table 5
Semantic Keyness of lexical expressions in the action/thriller trailer genre subcorpus (spoken language, Augstrail corpus, statistical measure: Log-Likelihood).

Rank Freq SC %1 Freq RC %2 Keyness LL LogRatio Semantic Domain

1 58 0.60 17 0.10 + 53.32 2.60 Warfare, defense and the army; weapons
2 42 0.43 28 0.16 + 16.54 1.41 In power
3 32 0.33 18 0.10 + 16.07 1.66 Crime
4 10 0.10 1 0.01 + 14.61 4.15 Electricity and electrical equipments
5 137 1.41 157 0.91 + 13.77 0.63 Time: Future
6 50 0.52 41 0.24 + 13.45 1.11 Dead
7 15 0.15 5 0.03 + 12.60 2.41 Flying and aircraft
8 19 0.20 11 0.06 + 9.19 1.62 Sailing, swimming, etc.
9 30 0.31 24 0.14 + 8.50 1.15 Vehicles & Transportation
10 42 0.43 39 0.23 + 8.41 0.93 Law & Order

Fig. 4. Concordance sample of the lexical expressions which realise the [WARFARE, DEFENSE & ARMY] domain.

forms of the lemma SHOOT as well as firearms, ammunition or weap- crime-dominated themes, characters and objects, involved in
ons. Military-related NHRPs refer to lexical expressions that repre- dynamic fight or flight scenes, using technical equipment of vari-
sent military vehicles or related concepts, e.g. pilot, chopper, ous sorts.
helicopter, parachute, plane or airship. Interestingly, this result Note that in Table 5 the semantic category TIME: FUTURE emerges
seems to corroborate research by McIntyre (2012: 19) whose key- as another prominent semantic category in the action genre. The
word analysis of thirteen blockbuster screenplays (mostly action dominance of the domain is primarily due to the regular use of
films) also listed expressions like ship as abbreviation for starship) the modal auxiliary will in action trailer dialogues. While some
or cockpit among the 20 most frequent keywords in the corpus use of the modal expression is volitional, other uses clearly express
dialogues. futurity, pointing out coming states or events (see Fig. 5). This is
Similar HRPs include tokens like thief, gangster, smuggler, kid- sustained by the frequent use of the be going to construction in
napper, robbers or felons in the semantic domain of CRIME. The the corpus. Frequent references to future events in verbal dialogues
domain IN POWER is commonly realized by nominal groups related might be used by trailer editors as narrative ‘‘shortcuts” of key
to HRPs assuming a position of rank or power, e.g. king, queens. transitions or future developments in the film’s narrative. The
Some of these units pay tribute the frequent presence of characters use of such expressions can thus be a useful strategy to compress
in action films that are responsible for the planning and execution story events.
of military missions, criminal raids or spy extractions, e.g. leader,
controller or administrator. They commonly use a range of technical 7.2. Key semantic domains in the comedy trailers
tools to break into safe rooms or to escape from prisons listed in
the domain of ELECTRICITY AND ELECTRICAL EQUIPMENTS. Table 6 offers an overview of key semantic domains used in the
To this end, we can see how the recurring lexical expressions comedy trailer subcorpus. The most frequent semantic category in
can be subsumed to semantic domains that, when combined, gen- the comedy trailer genre is labelled DISCOURSE BIN (see Table 6). A clo-
erate cohesive patterns of genre setting, protagonists and semantic ser look at the lexical expressions that constitute this category
‘‘landscapes” conducive to the trailer genre under study. In the case reveals a range of lexical features, traditionally linked to the use
of action genres, such a cohesive profile amounts to military-or of spontaneous, casual face-to-face conversation (cf. Biber and
7
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Fig. 5. Concordance sample of the lexical expressions which realise the [TIME: FUTURE] domain.

Table 6
Semantic Keyness of lexical expressions in the comedy trailer genre subcorpus (spoken language, Augstrail corpus, statistical measure: Log-Likelihood).

Rank Freq SC %1 Freq RC %2 Keyness LL LogRatio Semantic Domain

1 456 3.13 199 1.45 + 89.15 1.11 Discourse Bin
2 52 0.36 9 0.07 + 31.06 2.45 Education in General
3 67 0.46 20 0.15 + 24.13 1.66 Relationship, Intimacy and Sex
4 3001 20.58 2495 18.14 + 21.79 0.18 Pronouns
5 147 1.01 79 0.57 + 17.00 0.81 Kin
6 30 0.21 7 0.05 + 14.09 2.02 Drinks & Alcohol
7 27 0.19 6 0.04 + 13.26 2.09 Language, Speech & Grammar
8 323 2.22 239 1.74 + 8.18 0.35 Negative
9 113 0.77 72 0.52 + 6.93 0.57 Evaluation: Good
10 9 0.06 1 0.01 + 6.90 3.09 Mathematics

Conrad, 2019: 86f). Such expressions include inter alia interjections equally embedded in the talk exchanges and point to the often con-
(Ah; Oh), greetings (Hey; Hello; Good morning), expressives (Oh my flictive and emotive thrust of comedy dialogues (see Fig. 7).
God), thanking responses (Thank you), backchannel signals, affir- The category of DRINKS & ALCOHOL might show that comedy char-
mation markers (Yeah; Yes; Yep; Right), interrogatives (Do you acters tend to consume (alcoholic) beverages to distract them-
think), hesitation signs (um), politeness markers (please) or infor- selves from interpersonal conflicts with their peers. It might also
mal intensifiers (see Fig. 6). simply show that they typically frequent places, like bars or restau-
The semantic domain EDUCATION IN GENERAL contains frequent lexi- rants where alcoholic beverages are ordered or served. The fre-
cal expressions, such as class, viva, SATs, students, professors or quent use of kinship terms, such as father, mother, brother,
teachers. Another popular plotline in the dataset revolves around daughter, uncle, in turn, establishes the comedy genre as one that
love Affairs or coming-to-life stories of teenagers. This finding ties centers on family-related concerns and events.
in with the third most frequent semantic domain OF RELATIONSHIP, INTI- The frequent use of negations in the comedy genre is notable.
whose lexical expressions emerge in scenes of romantic
MACY AND SEX, The many contracted negations are often coupled with first person
encounters, dates or talk about story characters’ interpersonal rela- pronouns as well as modal auxiliaries, i.e. I can’t do X, I don’t
tions. The domain also contains different forms of verbal aggres- want X, I haven’t thought about X. In this shape, they epitomize
sion, simple profanities (fuck) or intensifiers (fucking), which are the conversational conflicts that often dominate the spoken inter-

8
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Fig. 6. Concordance sample of the lexical expressions which realise the [Discourse Bin] domain.

Fig. 7. Concordance sample of the lexical expressions realising the [RELATIONSHIP, INTIMACY AND SEX] domain.

action between comedy characters. Equally frequent is the use of 7.3. Key semantic domains in the documentary trailers
first and second person pronouns, that is typical in casual conver-
sations (cf. Biber and Conrad, 2019: 7). We should also mention the The visual analysis of opening shots in documentary trailers
frequent use of positive evaluations, e.g. super, well, great, good, already revealed a trend to portray protagonists from the world
fantastic, that comedy characters routinely employ to negotiate of politics, the media or individuals with a corporate background.
and secure interpersonal rapport. The semantic domains reappear in the list of frequent lexical
Most of the semantic domains are conducive to narrative con- expressions in documentary trailers (see table 7). The domains of
ventions in the comedy film genre, such as the relatively young POLITICS & GOVERNMENT are primarily represented by morphological
age of story characters, their typical settings in schools, colleges, variants of the base politic*, e.g. political, politicians or political par-
restaurants, bars or family homes. The lexical expressions are also ties, or in the case of GOVERNMENT, by expressions such as president,
reminiscent of the fact that conflict often serves as a key plot ele- nation, government regime, state or country. Frequent expressions
ment in comedy films. for the domain of MEDIA are reporters, journalism, press or papers.

Table 7
Semantic Keyness of lexical expressions in the documentary trailer subgenre (spoken language, Augstrail corpus, statistical measure: log-likelihood).

Rank Freq SC %1 Freq RC %2 Keyness LL LogRatio Semantic Domain

1 27 0.37 10 0.04 + 40.73 787.32 Politics
2 1797 24.33 4983 20.53 + 37.08 18.51 Grammatical Bin
3 67 0.91 104 0.43 + 21.30 111.72 Geographical names
4 12 0.16 4 0.02 + 19.06 885.91 The Media: Newspapers, etc.
5 37 0.50 47 0.19 + 17.42 158.71 Work & employment: Generally
6 30 0.41 36 0.15 + 15.50 173.86 Government
7 12 0.16 6 0.02 + 15.20 557.27 Degree: Approximators
8 22 0.30 27 0.11 + 10.97 167.78 The Media. TV, Radio & Cinema
9 66 0.89 130 0.54 + 10.76 66.85 People
10 8 0.11 4 0.02 + 10.14 557.27 Quantities: many, much

9
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Fig. 8. Concordance sample of the lexical expressions realising the [GRAMMATICAL BIN] domain.

WORK & ENVIRONMENT is typically reified by word forms of the the middle of larger social gatherings. In a similar vein, NHRPs in
lemma work*, e.g. works, working, worked, etc. as well as lexical action trailers are assigned to the semantic domains of WARFARE, MIL-
items related to professions, e.g. journalist or staff. The category ITARYor CRIME, while NHRPS in Comedy trailers primarily feature
GRAMMATICAL BINcontains mostly prepositions, such as for, about, up, objects from the semantic domains of HOME, HOUSE or ARCHITECTURE.
on as well as the infinitive marker to. The frequent use of preposi- documentary trailers, instead, used objects from the domains of VE-
tions in documentaries might imply a necessity for spatial descrip- HICLES & TRANSPORTATIONor ARCHITECTURE & BUILDINGS.
tions; a common feature of stories that interviewees or off-screen In line with this genre-defining use of visual elements, the key
narrators relay to the viewers (see Fig. 8). Such descriptions also domain analysis of trailer dialogues has uncovered the key seman-
align with the frequent use of approximators (nearly, almost, pretty) tic domains of the three genre subcorpora. When combined, the
and quantifiers (most, many) in the dataset, deployed by intervie- verbal and the visual analysis yield decisive multimodal patterns
wees to mitigate the probability of their assertions. of cohesion which differ in significant ways from one genre to
As can be seen, semantic domains are relatively stable within the next. We can therefore contend that trailer genres define the
genre boundaries but they differ from one genre corpus to the next. way multimodal cohesion is construed in film trailers. The cohe-
While the dispersion of lexical expressions that realise the seman- sive means explored in this study reinforce each other, creating a
tic domains is limited between trailers of the same genre subcor- simultaneous audio-visual stream that stems from the same
pus, their frequent lexical expressions largely correspond to the semantic lineage. The resulting cohesive force makes its mark on
same, limited range of semantic domains described before. Trailer the early dialogue and imagery of trailers and thereby projects
dialogues thus repeatedly use particular semantic domains in dif- the (proto)typical narrative hallmarks each film genre is known for.
ferent trailer genres and they frequently use particular lexical The study hopes to have shown how multimodal cohesion can
expressions or patterns of lexis which sustain the general plot be measured and tracked within and across larger sets of film trail-
and character orientation of each genre in question. We could elicit ers and trailer genres. It has likewise provided a new corpus-
distinct patterns of (verbal) cohesion which provides a recurring assisted framework which captures both verbal and visual means
semantic prosody for each of the three trailer genres investigated of expression to explore multimodal cohesive profiles across genre
in this study. These visual patterns described in section seven boundaries.
strike a chord with the visual patterns of cohesion in opening sce-
nes explored in section six. Both coincide to enable movie audience Declaration of Competing Interest
to assign genre characteristics to each movie trailer as quickly as
possible. Genre categorization, it is believed, can contribute The authors declare that they have no known competing finan-
tremendously to integrating the viewing experience into the gen- cial interests or personal relationships that could have appeared
eric frame knowledge of the movie audience. As a result, under- to influence the work reported in this paper.
standing a trailer will be much easier for viewers who succeed in
intergrating the (often subtle) cohesive, audio-visual clues and
References
clusters into their previous cinematographic knowledge of film
genres. Archer, D., Culpeper, J. and Rayson, P., 2005. Love - a familiar or a devil? An
exploration of key domains in Shakespeare’s Comedies and Tragedies.
Presented at the AHRC ICT Methods Network Expert Seminar on Linguistics.
8. Conclusion Lancaster University, 8 September 2005.
Bateman, J., 2014. Looking for what counts in film analysis: a programme of
empirical research. Visual Commun. 4, 301–330.
The study has shown what choices editors make to stage the Bateman, J., Wildfeuer, J., Hiippala, T., 2017. Multimodality: Foundations, research
opening scenes of film trailers. Such choices yield cohesive clusters and analysis - A problem-oriented introduction. Berlin, Walter de Gruyter.
that comprise specific types of shot scales, settings or (non-)human Bednarek, M., 2015. Corpus-assisted multimodal discourse analysis of television
and film narratives. In: McEnery, A., Baker, P. (Eds.), Corpora and discourse
represented participants. For instance, in action trailers main char-
studies: Integrating discourse and corpora. Palgrave Macmillan, London, pp. 63–
acters are primarily shown in a private, reflective mood or depicted 87.
in everyday activities which are suddenly disrupted by unforeseen Bednarek, M., 2018. Guide to the Sydney Corpus of Television Dialogue (SydTV).
events. In contrast, HRPs in comedy trailers are usually assembled Corpus manual vailable at http://www.syd-tv.com/publications/. (accessed 07
October 2020).
in private social venues, e.g. parties, meetings, and in documentary Bordwell, D., 1974. The idea of montage in Soviet art and film. Cinema J. 11 (2), 9–
trailers, characters are typically placed in interview settings or in 17.

10
C.R. Hoffmann Discourse, Context & Media 44 (2021) 100539

Bordwell, D., 2002. Intensified continuity visual style in contemporary American Kress, G., 2010. Multimodality: A social semiotic approach to contemporary
film. Film Quart. 55 (3), 16–28. communication. Taylor & Francis, London.
Bordwell, D., 2012. Poetics of cinema. Routledge, London. Kress, G., Van Leeuwen, T., 1996. Reading images: The grammar of visual design.
Biber, D., Conrad, S., 2019. Register, genre, and style. Cambridge University Press, Routledge, London.
Cambridge. Leech, G., 2013. Virginia Woolf meets Wmatrix. Etudes de Stylistique Anglaise No. 4,
Bublitz, W., 1994. In the eye of the beholder: the rather mystical notion of 15-26.
coherence. In: Carlon. Maier, C., 2009. Visual evaluation in film trailers. Visual Commun. 8 (2), 159–180.
Bublitz. W., 2006. It utterly boggles the mind: knowledge, common ground and Maier, C., 2011. Structure and function in the generic staging of film trailers: A
coherence. In: Pishwa, H. (Ed.), Language and Memory: Aspects of Knowledge multimodal analysis. In: Piazza, R., Bednarek, M., Rossi, F. (Eds.), Telecinematic
Representations. (Trends in Linguistics, Studies and Monographs 173). Mouton Discourse. Benjamins, Amsterdam, pp. 141–158.
de Gruyter, Berlin: De Gruyter, pp. 359–86. McIntyre, Dan, 2012. Prototypical characteristics of blockbuster movie dialogue: A
Bublitz, W., 1999. Introduction: views of coherence. In: Bublitz, W. (Ed.), Coherence corpus stylistic analysis. Texas Stud. Lit. Lang. 54 (3), 402–425.
in spoken and written discourse: How to create it and how to describe it. Monaco, J., Lindroth, D., 2000. How to read film. The World of Movies, Media and
Benjamins, Amsterdam, pp. 1–7. Multimedia. Language, History, Theory. Oxford University Press, Oxford.
Dynel, M., 2014. Participation framework underlying YouTube interaction. J. Neumann, S., Fest, J., 2016. Cohesive devices across registers and varieties: The role
Pragmat. 73, 37–52. of medium in English. In: Schubert, C., Sanchez-Stockhammer, C. (Eds.),
Flowerdew, J., Mahlberg, M. (Eds.), 2009. Lexical cohesion and corpus linguistics Variational text linguistics: Revisiting register in English. Mouton de Gruyter,
Amsterdam: Benjamins. Berlin, pp. 195–220.
Genette, G., 1997. Paratexts: Thresholds of interpretation, vol. 20. Cambridge Rayson, P., Archer, D., Piao, S.L., McEnery, T., 2004. The UCREL semantic analysis
University Press, Cambridge. System, Proceedings of the workshop on Beyond Named Entity Recognition.
Halliday, M.A.K., Hasan, R., 1976. Cohesion in English. Routledge, London. Semantic labelling for NLP tasks in association with 4th International
Hasan, R., 1984. Coherence and cohesive harmony. In: Flood, J. (Ed.), Understanding Conference on Language Resources and Evaluation (LREC 2004), 25th May
Reading Comprehension: Cognition, Language, and the Structure of Prose, 2004, Lisbon, Portugal, 7-12.
International Reading Association, New York/Delaware. International Reading Rayson, P., Smith, N., 2006. The key domain method for the study of language
Association, Newark, Delaware, pp. 181–219. varieties. The Third Inter-Varietal Applied Corpus Studies (IVACS) group
Hediger, V., 2001. Verführung zum Film. Der amerikanische Kinotrailer seit 1912, International Conference on ‘‘LANGUAGE AT THE INTERFACE”. University of
vol. 5. Marburg, Schuren. Nottingham, UK, 2324 June 2006; URL: http://ucrel.lancs.ac.uk/publications/
Hixson, T.K., 2006. Mission possible: Targeting trailers to movie audiences. J. Target. IVACS06_raysonsmith_6UP.pdf (accessed 07 October 2020).
Meas. Anal. Market. 14 (3), 210–224. Redfern, N., 2020. Sound in horror film trailers. Music Sound Moving Image 14 (1),
Hoey, M., 2003. Why grammar is beyond belief. Belgian Essays Lang. Lit., 183–196 47–71.
Hoey, M., 2005. Lexical Priming. A new theory of words and language. Routledge, Rodman, Ron, 2011. ‘‘Coperettas””, Detecterns”, and Space Operas: Music and Genre
London. Hybridization in American Television. In: Deaville, J. (Ed.), Music in television:
Hoffmann, Christian, 2012. Cohesive Profiling. Meaning and Interaction in Personal Channels of listening. Routledge, London, pp. 47–68.
Weblogs. Benjamins, Amsterdam. Stubbs, M., 1983. Discourse Analysis. The sociolinguistic analysis of natural
Janney, R., 2010. Film discourse cohesion. In: Hoffmann, C. (Ed.), Narrative language. University of Chicago Press, Chicago.
Revisited: Telling a Story in the Age of New Media. Benjamins, Amsterdam, Tanskanen, S.K., 2006. Collaborating towards coherence. Benjamins, Amsterdam.
pp. 245–266. Thompson, K., 2003. Storytelling in film and television. Harvard University Press,
Janney, R., 2012. Pragmatics and cinematic discourse. Lodz Pap. Pragmat. 8 (1), 85– Cambridge, M.A..
113. Tseng, C., 2012. Audiovisual texture in scene transition. Semiotica 192, 123–160.
Karray, S., Debernitz, L., 2017. The effectiveness of movie trailer advertising. Int. J. Tseng, C., 2013. Cohesion in film: Tracking film elements. Springer, Berlin.
Advert. 36 (2), 368–392. Widdowson, H.G., 1980. Discourse Analysis. Oxford University Press, Oxford.
Kernan, L., 2009. Coming attractions: Reading American movie trailers. University of Wildfeuer, J., Pollaroli, C., 2017. Seeing the untold: Multimodal argumentation in
Texas Press, Austin. movie trailers. In: Tseronis, A., Forceville, C. (Eds.), Multimodal Argumentation
Krebs, H., 2020. Effectful advertising? Film trailers and their relevance for and Rhetoric in Media Genres. Benjamins, Amsterdam, pp. 190–216.
prospective audiences. In: Hoffmann, C., Kirner-Ludwig, M. (Eds.),
Telecinematic Stylistics. Bloomsbury, London, pp. 205–223.

11