‘Call for Purge on the People Traffickers’: An Investigation into British Newspapers’ Representation of Transnational Human Trafficking, 2000–2016

Gregoriou and Ras draw on corpus linguistics and critical discourse analysis to examine a 61.5 million-word corpus of articles published by UK newspapers between 2000 and 2016, and on qualitative critical discourse analysis of a sixty-seven-article sample corpus in depth. Both approaches analyse the naming and describing of victims and traffickers, metaphors, transitivity, and speech and writing presentation, while the in-depth qualitative approach furthermore analyses the text (images) (multi)modally. Their findings conclude that trafficking for sexual exploitation is over-reported compared to other forms of trafficking, and that victims are generally presented as young, female, and vulnerable. As a result, non-stereotypical victims, of crimes like forced begging and domestic servitude, are not readily recognised as victims, and thereby are deprived of opportunities for assistance.


IntroductIon
To analyse UK media representations of human trafficking (henceforth HT), certain research questions must be answered. How are trafficking, traffickers, and HT victims represented, and to what extent are victims criminalised and victimised, 1 sex trafficking overrepresented, and smuggling and trafficking conflated by the UK press? Such overarching research questions are designed to shed light on underlying ideologies relating to agency, responsibility, and vulnerability, and will respectively guide the critical discourse analysis of the whole corpus 2 (quantitatively) and the spike sample corpus (qualitatively).

data collectIon
Given the 'present day' focus of this project, the time frame was restricted to 2000-2016; we focus on UK articles published after 2000 as this was the year of the ratification of the Palermo Protocol. 3 This 16-year period also covers the years running up to the UK Parliament passing the Modern Slavery Act 2015, which defines HT, slavery, and exploitation as applicable to the English and Welsh criminal justice system. The dataset was limited to UK national daily and Sunday newspapers alone.
To address corpus compilation considerations with regard to the relevance of the included texts and the exhaustiveness of the corpus, we drew on Gabrielatos' (2007) data collection method, which he developed as 1 We distinguish primary victimisation (which directly affects the victim) from secondary victimisation (such as in the form of victim blaming), re-victimisation (such as falling victim to re-trafficking), indirect victimisation (such as with reference to detrimental effects on one's family) and secondary exploitation (in the form of anyone benefitting from the problem's existence).
2 'A corpus is a collection of pieces of language text in electronic form, selected according to external criteria to represent, as far as possible, a language or language variety as a source of data for linguistic research' (Sinclair, 2005). 3 United Nation's Protocol to Prevent, Suppress and Punish Trafficking in Persons Especially Women and Children, supplementing the United Nations Convention against Transnational Organized Crime (UNODC, 2016). part of a research project examining UK media representations of refugees, asylum seekers, and (im)migrants. Our resultant search terms were: bonded labour, child labour, debt bondage, domestic servitude, exploitation, exploitative labour, forced criminality, forced labour, forced prostitution, human trafficker, human trafficking, labour trafficking, organ harvesting, raid-and-rescue, sex slave, sex slaves, sex trafficking, sexual exploitation, sexual servitude, slave, slave trade, slaves, trafficked, trafficked victims, traffickers, trafficking (in/of) persons, trafficking (in/of) human being/s, and woman trafficking. 4

The Corpus
Our method generated a resulting corpus of 80,608 articles. It contains 61,571,641 tokens (as in 'words') and 388,834 types (of words). As Table 2.1 shows, relatively large numbers of articles were published in The Guardian, The Times and The Independent, which suggests that results are skewed toward the representation of HT by British broadsheets. As such, the corpus is not representative of what the British public reads about HT, but of what the British press writes about HT. 4 Gabrielatos ' (2007) method works by first selecting a limited set of search terms that are then used to generate several sample corpora. Our initial search terms were human trafficking, human trafficker, trafficking (in/of) human being/s, slavery, sexual exploitation, sex trafficking, sexual trafficking, sex slave, and forced labour. We first collected sample corpora (i.e., a selection of the intended final, full corpus) from Lexis Nexis for the periods 1/1/2000-30/9/2000, 1/1/2008-30/9/2008, and 1/1/2016-30/9/2016, to ensure that our results would not be unduly skewed toward either end, or indeed the middle, of the overall time frame. Each initial sample corpus was then uploaded to Wmatrix (Rayson, 2009) and compared to a reference corpus (a corpus that serves as the benchmark against which the primary corpus is, or our sample corpora were, compared), which in our case was the BNC Written Sampler (a corpus intended to be representative of written British English), to generate three keyword (words used significantly less or more often in the primary corpus than in the reference corpus) lists, one for each sample corpus, which were then combined. The potential usefulness of each keyword was evaluated by calculating RQTR scores (a mathematical measure indicating the potential relevance of a keyword as a search term) following Gabrielatos (2007). Further potential search terms were selected introspectively by members of the project team, based on their reading of the relevant literature. These potential search terms were also evaluated using Gabrielatos' (2007) RQTR score. It was the resultant list of search terms that was used to collect articles from Lexis Nexis over the full time frame 1/1/2000-30/9/2016.

Diachronic Change and Spike Sample Corpus Construction
Figure 2.1 shows that the number of articles over time has steadily climbed, which is in line with Winterdyk, Reichel, and Perrin's (2012, p. 9) notion of an 'explosion of media coverage' of HT during this time. 5 Further to indicating that more texts were published in later years compared to the earlier ones, Fig. 2.1 also shows sizeable fluctuations in the number of articles published on HT. Even though dips in the graph are also worth examining, we here keep our focus on the graph's particularly 5 The figure's vertical, y-axis, shows the number of texts generated during the period, while the horizontal, x-axis, indicates the months of each of these 16 years, from January 2000 at the far left to September 2016 at the far right. All relevant data points are accounted for in the graph, though space constraints dictate that not all month-labels are visible on the x-axis. Quantitative methods of analysis are used to examine the full corpus, but qualitative methods, which are required to triangulate and complement the quantitative methods, cannot practically be used to examine a multi-million-word corpus. As such, we created another sample corpus, of prototypical 6 texts from each of these five spikes. The total number of articles in this spike sample corpus is sixty-seven.
For each spike, Table 2.3 indicates the total number of articles, the number and percentage of articles sampled, the numbers of tokens per spike, and the number and percentage of the number of words in the sample generated. As the table shows, even with our focus on spikes, and hence periods in which media reported on HT the most, our qualitative analysis was limited to 18,000 words in total, out of the 61.6 m words available. For the multimodal part of the analysis, which explores the images 6 ProtAnt (Anthony, 2015) was first used to generate a list of the most prototypical articles in each spike, after which random articles were selected using Excel, to reduce each list to 1%. This percentage generated a set of sixty-seven articles, and hence a sample corpus manageable for manual analysis. For a final check, these articles were manually evaluated to ensure relevance. accompanying the spike sample corpus, we obtained facsimiles and online newspaper website article versions of 53, 7 (i.e., 80%), of these 67 articles. 7 Facsimiles were required to facilitate a multimodal analysis of the sample corpus texts' accompanying images. Most articles were available from the British Library Newsroom in digital form, on microfilm, or in the original newspaper, although about half a dozen were This study examines the linguistic characteristics of UK newspaper reporting on HT that are normally also examined by critical discourse analysts/ stylisticians in relation to media texts (Fairclough, 1992;Gregoriou & Paterson, 2017;Jeffries, 2010;Mayr & Machin, 2012). Given our research questions, we prioritise the investigation of metaphors, the naming and describing of HT, its victims, traffickers, and other HT perpetrators, but also examine transitivity, 9 modality, 10 and the reporting of speech and writing. We adopted a 'mixed methods' approach by manually analysing our spike sample corpus of sixty-seven articles, supported by a corpus linguistic approach to the whole 80,608 text corpus. The qualitative analysis was conducted independently of the quantitative analysis to pre-empt the criticism of 'cherry-picking' (see Poole, 2010;Widdowson, 1995Widdowson, , p. 169, 1998Widdowson, , pp. 143-146, 2004) that critical discourse analysts often face.
In metaphor, one cognitive domain is mapped against another (see Lakoff & Johnson, 1980). We adopt the traditional linguistic means of referring to metaphors using small capitals. We first focused on the spike only available as online articles from the newspaper websites. A further fifteen, including nine from The Mirror, were simply unavailable for reasons including the fact that only one edition of a newspaper, of multiple, is actually stored and digitised by the British Library. 8 See Baker et al. (2008, p. 295). 9 Answering the question of who is doing what to whom/what, how and with what? (Mayr & Machin, 2012, p. 52;Richardson, 2007, p. 54  sample corpus to find sustained, or mega, metaphors, and explored these further across the whole of our data set via corpus analysis, using a method similar to the one advocated by Koller, Hardie, Rayson, and Semino (2008). 11 The corpus linguistic methods that were used to examine which topics were most important in the corpus and how various participants and concepts were (metaphorically, at times) described included the generation of a key word list. 12 These key words were subsequently categorised with the help of concordances. Categories include 'Participants', 'Acts of Abuse and Exploitation', and 'Movement'.
Various methods were used to examine the modification of the target nouns/lemmas included in this key word list. Concordance displays, c-collocate 13 lists (derived using the method set out by Gabrielatos & Baker, 2008 14 ), and 2-gram lists (an n-gram is a word group of n size, and in this case we refer to word pairs) were particularly useful. 11 Koller et al. (2008) take (unexpected) semantic categories, that is, groupings of broadly related words, a list of which is generated through Wmatrix, as indicative of source domains. They advocate that when source domains of interest are established prior to the examination of the corpus through Wmatrix, its 'domain push' function is used to 'push' to the foreground secondary semantic domain tags. 12 To generate a key word list of the human trafficking news corpus, the British National Corpus Written Sampler that forms part of Wmatrix was used as a reference corpus. Two cut-off points were selected to shrink the list to a manageable length. Gabrielatos and Baker (2008) set their log-likelihood (LL; a statistical measure often used to indicate keyness) cutoff at p < 0.000000000000001, which they indicate as an 'extremely low' value, or an extremely high threshold. The threshold for the current key word list was set at LL > 15.13, p < 0.001. A further threshold of relative frequency was set, to ensure that all interpreted key words are used relatively often. Given the fact that the BNC was, at the time of the present study, 20 years old, and includes texts that are older, it is possible that certain words that are used by current-day newspapers simply did not appear in the time that many BNC texts were written. 13 A concordance displays a word of interest in its co-text, usually 50 characters to the left and the right, for every instance of this word of interest in the corpus. A collocate is a word that occurs within a certain number of words to the left or right (usually 4 or 5) of the word of interest. Collocates of interest are those that occur at a certain (relative) frequency. A c-collocate, short for 'constant collocate', re-occurs above a certain threshold over a predefined number of years.
14 The main issue in defining a c-collocate list is defining 'consistency'. Gabrielatos and Baker (2008) indicate that a collocate is consistent when it occurs in 7 years, at a minimum frequency of 5. As their corpus spans a decade, whereas ours spans 17 years, the minimum number of years was raised to 75%. As this, however, would mean that a collocate has to occur at least 65 times (compared to Gabrielatos and Baker's (2008) minimum overall fre-As Jeffries (2010, p. 47) points out, through transitivity choices, the reader is presented with 'clear notions of who is in control, who is a victim and so on'. Most researchers only examine transitivity in small corpora, semi-manually. 15 As the key word analysis already distinguished between pertinent processes (e.g., acts of abuse and exploitation) and participants (e.g., survivors/victims and traffickers/offenders), this transitivity analysis focuses on a limited set of words. We took key words referring to acts of trafficking and exploitation and noted whether they are in nominalised form ('trafficking'), are past participles ('trafficked'), or have another form.
We work with a more traditional understanding of transitivity (see Jeffries, 2010, p. 38), rather than Halliday's (1994) transitivity model. Our approach is, in some ways, a throwback to critical linguistics, as proposed by Fowler, Hodge, Kress, and Trew (1979), and driven primarily by the limitations of SketchEngine as experienced by us. We used SketchEngine's word sketch functionality, which shows how often a target lemma is used as a subject, as an object, and is preceded by the prepositions 'by' (for an approximation of how often lemmas were used as agents in agentive passives). For each target lemma, these frequencies were simply recorded in an Excel file. Unfortunately, word sketches cannot show how often the target lemma is used as the subject of a passive sentence. We used CQL to examine how often a target lemma was the subject of a passive sentence. In The query produced above was run a second time for each lemma, with the following sequence attached, to examine how often these passives included an agent 17 : quency of 35), the minimum frequency was dropped to 3 (resulting in a minimum overall frequency of 39). Like Gabrielatos and Baker, collocates were collected from the span 5L-5R, with the target word at position 0. 15 For instance, Rodrigues Jr. (2005), de Lima-Lopes (2014), Tabbert (2015), Bartley and Hidalgo-Tenorio (2015), and Lee (2016) manually annotate concordance lines. 16 We acknowledge that this query only produces results for regular past participles and passive constructions created with a version of 'to be'. The best approximation of the number of agentless passives was reached by calculating the number of passive constructions that include an agent. 17  Frequencies were recorded on a year-by-year basis to facilitate future diachronic analyses, complementary to the current synchronic analysis.
As indicated, we also examine modality, the linguistic area of analysis that investigates the attitude toward the propositions that speakers/writers express. Features of interest include modal verbs (like 'could'), lexical verbs ('try'), but also nouns ('possibility'), adjectives ('possible') and adverbs ('likely'). The modal systems that are of particular interest in relation to the HT corpus are the deontic, concerned with one's 'attitude to the degree of obligation attaching to the performance of certain actions', the epistemic, concerned with one's 'confidence or lack of confidence in the truth of an expressed proposition', and the perception, also communicated through a degree of commitment to the truth of a proposition, but a degree given specifically by some reference to (mostly visual) perception (Simpson, 1993, p. 47, 48, 50).
Finally, we also analyse some noteworthy images of the manageable, yet representative, set of fifty-three media texts on HT in either its original print or online form multimodally, in an approach consistent with the social semiotic multimodal analytical paradigm, covering aspects relating to the relevant photographs' representational, interactional, and compositional meaning (Kress & Van Leeuwen, 2006). Besides, much like the surrounding verbal text, photographs make a significant contribution to a news story's meaning, and are deserving of the same scrutiny as the verbal text (Caple, 2013, p. 3).

Metaphors
One frequent conceptual mega metaphor is that of TraffIckIng Is A Trade. See, for instance, references to people as 'cargo', 'commodities', or 'investment' to be 'bought' and 'sold' in a 'trade', 'market' or 'business'. In '"well stocked" orphanages were seen as a successful business venture', people are conceptualised along the lines of goods with 'worth', or as livestock worth exploiting. This qualitative finding is supported by the categorisation of key words, as at least 0.5% of corpus tokens 19 have been categorised as Trade-related. Given that at least 0.85% of tokens are categorised as Abuse/Exploitation-related, and 0.63% as Movement-related, the Trade-related percentage is high; Trade is clearly an important semantic category. Wmatrix similarly shows that a full 2.4% of all items in a 100,000-word sample corpus are in the semantic domain of 'Business'. When considering pushed domains, following Koller et al. (2008), this number rises to 3.3%. The majority of key words categorised as Traderelated are to do with acts of trading, such as buying and selling. This Trade metaphor is clearly grounded in the everyday practices of human traffickers; since trafficking literally involves the illegal trade of people as stock, one may question whether TraffIckIng Is A Trade is metaphorical at all.
The following excerpt illustrates the difficulty in distinguishing between the metaphorical and actual trade in human beings: Anyone with the slightest degree of worldliness would have thrown the deal back in the Calcutta Wallah's face but, partly through ignorance, partly through short-term greed, and partly because he persuaded himself it might be in the girls' best interests, Chandra signed the contract of indentures. By doing so, he sold his daughters into seven years of slavery and almost certain sexual degradation. (Darbyshire, 2000, emphasis added) The full article describes how these girls' father, Chandra, in a time of grief and in a very difficult economic position, was persuaded to give up two of his daughters, in exchange for an up-front sum, future monthly payments, and the promise that they would be trained and cared for. This complex situation is then simplistically described as a trade of money for girls.
Regardless of whether this use of Trade is metaphorical or literal, the press adopting this metaphor invites an acceptance of the ideology whereby people are commodities. News texts focusing on the trade-element of trafficking, rather than, for instance, survivors' suffering or how to recognise trafficking, perpetuates the idea that human beings can be treated as goods. Though highly conventional (see Kövecses, 2010, p. 211), this mega metaphor still requires a critical response; it contributes to the ongoing objectification of victims/survivors, and to the erasure of HT victims' agency.
A second cluster of conceptual metaphors revolves around the nature of trafficking and how to respond to it. TraffIckIng is frequently conceptualised as A SpreadIng Unwanted Substance and as one which Can Be Broken, and indeed common responses to trafficking are in line with the metaphor RespondIng To TraffIckIng Is War. These metaphors are related in the sense that the representation of trafficking as a threatening physical matter creates a matter that can physically be fought. They also oversimplify responding mechanisms and suggest that HT is something to deal with in a reactionary rather than preventative manner.
The TraffIckIng Is A SpreadIng Unwanted Substance metaphor is found in references to trafficking as 'epidemic', 'widespread', or 'rampant'. What spreads seems to be both the victims and the perpetrators of this crime; there is an 'influx' of those prone to trafficking 'pour[ing]' into the UK, all the while trafficking 'parasites' need to be 'curb[ed]' or 'root[ed] out'. The overwhelming and indeed growing matter of trafficking is emphasised throughout the corpus. Of collocates to trafficking, 14.56% indicate either growth (more, rise) or imply largeness (most, scale). Wmatrix shows that 0.37% of items in the Wmatrix sample corpus are tagged N3.2+, indicating largeness and growth. Furthermore, numbers, such as billion, millions, thousands, and hundreds form another 1.69% of the Wmatrix sample corpus. These words can refer to numbers of victims as well as profits made through trafficking, but nonetheless underscore the scale of this problem.
The metaphor TraffIckIng Is A Substance which Can Be Broken is found in references to 'strong' units to 'crackdown', 'break up', 'rap', 'stamp out', 'smash', or 'crush' the gangs or HT problem. RespondIng To TraffIckIng Is War is found in the texts' reference to a need for an 'international onslaught' to 'tackle ', 'combat', 'fight', 'target', 'confront', 'attack', 'purge' or 'spearhead' against  We most often conceptualise war as a violent conflict that engages societies as large as nations. Though the War metaphor suggests there being different sides, who features on each is unclear/hidden. War-related references also problematically conceptualise trafficking as an issue specific nations cause and other nations have to deal with, when the realities of trafficking suggest otherwise. Even more so, metaphorical constructions, such as the one that refers to a telephone hotline as 'the latest weapon in the fight against human trafficking', portray responses to trafficking as a battle against an overwhelming matter, and allow state actors to portray themselves as saviours, with survivors as acted upon, in other words, saved. Such references to trafficking also establish links between trafficking, on the one hand, and masculinity, power and violence on the other. Violence, such metaphorical constructions suggest, legitimises police violent reactions to HT, not to mention reducing HT victims to war collateral damage. By focusing on the Fight, survivors' responses to trafficking are, lastly, limited. They are either 'useful' in this fight, by contributing statements and evidence, or they are not. This limitation on options further erases survivors' agency and can actually contribute to their secondary victimisation. Not unlike the notoriously flawed War on Drugs, the trafficking media metaphors fail to focus on structural factors, and instead only respond to symptoms.
Less prominent and hence micro metaphorical spike sample corpus excerpts draw on various other source domains. These include the TraffIckIng Is Drama/Spectacle metaphor 20 ('Children forced into Oliver twist thieving', 'Oliver Twist Scenarios happening in our streets and cities', a trafficking case 'could have ended in tragedy', a raid is 'dramatic'); indeed, in the Wmatrix sample corpus, the semantic domain K4, 'drama, the theatre', has a relative frequency of 0.15%. This rises to 0.41% when K4 is a pushed domain. Another is the TraffIckIng Is HIdden/Not VIsIble metaphor ('turned a blind eye to trafficking', 'A report entitled Young Invisible Slaves', 'They are effectively invisible, isolated behind the closed doors of private households'), the related semantic domain, A10 ('hiding, showing') making up 0.54% of items in the Wmatrix sample 20 Exploring the gender and age dimensions of this metaphor would be interesting tasks in themselves. Are women and children victims more prone to be referred to via drama metaphors compared to men? Are men more prone to animal metaphors, perhaps? Such questions merit investigation through the use of concordances of Wmatrix examples, though doing so is beyond the scope of the present chapter. corpus, 1.23% as a pushed domain. A final metaphor, the TraffIckIng Is AnImated and Beastly ('Halt this evil trade', 'we all thought that [slavery] was gone and in the past, but actually it's alive and kicking in London', 'the trade's dark underbelly', 'Sex trafficker known as The Snake') and, in Pope Francis' terms, even VampIrIc metaphor ('Bloodsucker [trafficking] bosses', '[l]iving on the blood of the people'), is less clearly present in the Wmatrix sample corpus, as there are no directly corresponding semantic tags. Related tags, however, include L1+ ('alive'), L2 ('animals'), and S9 ('religion and the supernatural'), which occur at relative frequencies of respectively 0.12%, 0.38%, and 0.36%. Note, however, that the first semantic domain, L1+, also includes instances in which someone is literally alive after life-threatening events, and the use of 'life' to indicate someone's daily routines and habits, whilst the last, S9, also includes many instances in which someone's religion is identified. Though each of these metaphors is not prominent individually in the sample corpora, this micro metaphorical set can altogether be read along the lines of the TraffIckIng Is Horror Story metaphor, which itself suggests an audience in need of entertainment of something mysterious, disturbing and fictional. By relating the problem to narratives of horror, stories with binary portrayals of goodness and badness, it is, once, again, simplified.
Metaphorical portrayals of trafficking signal state actors as heroes and reduce victims to the measure of value/usefulness; as such, the media trafficking metaphors prove ultimately unhelpful in responding to the HT problem.

Naming and Describing HT
HT tends to be described as a large, imported, serious problem, which must be responded to urgently. People's suffering tends to be foregrounded, as the section on 'Agency and Focus' also shows, and this suffering is made particularly salient through the use of (negative) sensory-related words.
Quantitative results are largely similar, as shown in Table 2.4, in which we categorise c-collocates to human trafficking: C-collocates categorised as 'other' (articles, pronouns, verbs such as 'said' and 'make'), 61% of all c-collocates to trafficking, have been excluded from this table. The percentages in this table represent the share each category has of the remaining c-collocates. Certain categories are newly introduced or replace others compared to the categories found through the qualitative analysis. These include 'foreign', replacing 'migrant', which includes words like 'migrant' as well as 'Europe' and 'Britain', indicating trafficking to be an imported issue. New categories relate to those involved, 'people' and 'responders', including words such as 'woman' and 'police'. Other new categories relate to official responses, categorised as either 'fight' ('crackdown') or 'law', indicating legal responses ('court', 'evidence').
Overall, some terms and descriptors vary from neutral, suggesting that there is no problem whatsoever (at least not when looking at them in isolation), to those that, like the metaphor-related analysis showed, relate to a business or trade, to those that suggest there is something wrong (but not necessarily criminal), to those that suggest there is indeed something criminal going on and to those that strongly negatively (and metaphorically) evaluate the issue. The fact that some of these are perception (sightand taste-) related is interesting. And as the quantitative analysis also suggests, most characterisations of trafficking mark it as a large, modern, imported problem, which is very bad and must be fought. Much like the horror-related metaphor previously touched on, these findings support O'Connell Davidson's (2010) argument as to the modern slavery narrative being fairy-tale-like; a narrative of innocence and evil, its figures problematically 'stripped of ambivalence'.

Trafficking and Smuggling
As earlier research shows, trafficking and smuggling are, to a certain extent, conflated. An example of this conflation is found in The Daily Mail: '[NCIS director general Abbott] said the smuggled immigrants were often victims misled by traffickers about conditions [in Britain]' (Hinckley, 2000). In line with such research, an examination of our spike sample corpus also reveals a close correlation, even conflation, between (potentially illegal) immigration/smuggling and trafficking. See, for instance references to 'a crackdown on people traffickers [following the Home Secretary's promise to] crack down on illegal immigrants', to '[t]raffickers hid[ing] among migrants', to Cameron wanting to 'disrupt the trafficking and smuggling gangs', and to the London Olympics as a 'magnet for people traffickers smuggling sex slaves and illegal workers into Britain'. Similarly, the press reports a man confessing to 'trying to have his  Table 2.5 shows, smuggling is used in almost the exact same way as trafficking in the wider corpus, with three exceptions; 'routes', 'immigrants' and, perhaps most surprisingly, 'traffickers', are not c-collocates to trafficking.
When considering the c-collocates to keywords categorised as 'participants', a similar picture is drawn. The majority of participants with either 'trafficked' or 'smuggled' as a c-collocate are described as having been trafficked. Intriguingly, this also includes 'immigrants' and 'workers'. Only 'women' and 'people' are consistently described as having been smuggled, which is contrary to Musto's (2009) and Shalit, Heynen, and Van der Meulen's (2014) findings showing that men are generally represented as having been smuggled, whereas women are normally considered having been trafficked. An optimistic interpretation of the use of these descriptors would be that newspaper writers recognise the scope for exploitation in economic migration, and are offering these people the 'traffickingvictimhood' status in order to make sure they receive assistance and support. However, given the rhetoric of tabloids like The Daily Mail and The Sun, this seems an unlikely interpretation. For instance, in an article in The Daily Mail about NCIS warnings about the trafficking and smuggling of people into the UK, the following, only tenuously related, was added: Asylum seekers are putting overwhelming strain on some GPs' surgeries, doctors complained yesterday. Practices in Gateshead said they are having to organise extra care without extra resources after the arrival of 600 asylum seekers. And a practice in Leicester has threatened to close its list after 430 asylum seekers were moved into its catchment area. (Hinckley, 2000) One explanation for this discrepancy is that newspapers do not refer to cases of smuggling and trafficking following legal definitions, but instead use trafficking as simply the weightier, more loaded term. The loaded trafficking, to refer to all movement of people into the UK, is instead used to communicate newspapers' ideology that economic migration is a very serious problem; trafficking evidently has more serious connotations than the more accurate but less sensational smuggling.

Description of Victims/Survivors
Victims tend to be described in a stereotypical way: as female, young, and coerced, creating an ideal victim.
In the sample of articles that were qualitatively analysed, survivors are mostly unnamed but, when named ('Ms Lin'), tend to be first named ('Abou', 'Anna', 'Favor', 'Han'), which creates a close, intimate relationship with them as opposed to a distant/official one. The spike sample corpus also refers to HT victims being numerous. There is very much a focus on their high volume with either reference to them in terms of groups of a specific number ('eleven', 'thirty', 'fifty-eight') or indeed large numbers ('dozens', 'hundreds', 'thousands', 'million'). The fact that victims are generally perceived as numerous is tied with the notion of trafficking as a large, and growing, problem. In fact, Dijk (2013) notes that the Polish legislature debated whether HT necessarily involved a plurality of victims, as the Polish translation of this phrase, 'handel ludzḿi', refers to people as plural. Of the full corpus key words referring to victims, 65.58% are indeed plural (e.g., girls). This multitude of victims is also evident in concordance lines, which show results such as '[t]he International Labour Organisation estimates that 20.9 million people are victims of forced labour globally, including victims of human trafficking for sexual exploitation. "While it is not known how many of these victims were trafficked, the estimate implies that there are millions of trafficking victims in the world," said Mr Fedotov' (The Morning Star, 2013).
Several of the key words applied to victims throughout the full corpus indeed indicate a non-British origin: these are immigrant, migrant, and refugee. Other victims, too, are often identified as non-British: 26% of c-collocates indicating nationality or origin of victim keywords refer to Britain (e.g. British, London, here), whereas 58% indicate a non-British location, most often Eastern Europe and Africa. These c-collocates also include words such as illegal, across, and foreign, and support the notion that trafficking is framed as an 'imported' problem (see Johnston, Friedman, & Sobel, 2015). It should also be noted that immigrants, migrants, refugees, and workers are often young (57% of age-related c-collocates), and workers are explicitly identified as vulnerable. One possible interpretation is that these modifiers are used to negotiate the negative evaluation of migrants in general by British newspapers (see  and the fact that these people may be considered victims of trafficking. However, as the qualitative analysis shows, British newspapers can nonetheless be ruthless about these victims. The corpus linguistic findings also refer to victims as female. Of the keywords that are used to refer to victims of trafficking and that inherently indicate gender, 60% are female (e.g., girls, daughter, woman). Having said that, those victim-labelling related keywords that do not inherently indicate gender (e.g., victim) have 54% masculine gender-related c-collocates, a result that can be explained by the fact that gender collocates can indicate deviations from a perceived norm, in the sense that 'lady doctor' and 'male nurse' indicate that doctors and nurses are expected to be respectively male and female. If the majority of c-collocates indicate masculinity, it is possible that survivors are inherently presumed female and their occasional male-ness is noteworthy.
Another trafficking finding from the qualitative analysis is that victims are young. The quantitative analysis supports this finding: 24.58% of all key words applied to victims in the full corpus inherently indicate youth (e.g., girl, boy, baby, children) while, for the remaining three-quarters of key words that do not inherently indicate youth, 46% of gender-related c-collocates imply youth (e.g., young, teenage). A further 29% of age-related c-collocates are generic indicators of age, and even those c-collocates that appear to imply a lack of youth (e.g., old) may also be taken as generic, as in the phrase 'X years old'. Concordances show that this is, in fact, how old is to be interpreted. According to concordances, victims are generally under the age of twenty-five, although more so in the case of those trafficked for sexual exploitation than those trafficked for labour exploitation. One notable exception is the case of Elizabeth Fritzl, who was forty-two at the time of her release (see for instance Hall, 2009;Milmo & Peachey, 2013;Nicks, 2008;The Mirror, 2008).
These results are unsurprising given that the stereotypical victim of HT is, as noted in this collection's introduction, female, young, unwilling to perform the work she is doing, but coerced, portrayals not just common in news media and pop culture, but also shared by policy makers and the public (Buckley, 2009;Dando, Walsh, & Brierley, 2016;Gould, 2010;Musto, 2009). In short, victims of HT are often described in line with Christie's (1986) theory of the ideal victim; she is young and she is female. Furthermore, in line with the representation of HT as a large and growing problem, victims are represented as numerous. Finally, in accordance with the idea that HT is an imported problem and a threat to national security, victims are described as foreign. The stereotypical victim of trafficking is therefore perhaps best described in the following excerpt from The Daily Mail: '[a]ccording to information given by the traffickers, many of the victims smuggled into Britain are "vulnerable" young women from poor or broken families' (Borland, 2009).

Description of Traffickers and Related HT-Perpetrators
Traffickers and other perpetrators are less clearly defined. Newspapers appear to focus on victims and their suffering, neglecting to discuss traffickers in depth.
Traffickers are sometimes identified as 'families forced to sell their children to survive' ('Romanian Fana Moses, 32, had given birth and sold the boy' to a 'French' 'childless couple'), but mostly as 'recruitment agencies solely for overseas workers', 'firms', 'businesses', 'traders', 'businessmen looking for cheap and expendable labour', 'salon' owners, 'employers who exploit migrant workers and pay less than the minimum wage', and 'gangs who sell false dreams and trade on the free borders within the EU'. They are 'masters', 'owners', 'captors', 'abusers', and 'slavers of our day'; in other words, they are defined by their ownership of the victims they have 'control and power over'.
Quantitative analysis also suggests that, like victims, traffickers are numerous. In fact, traffickers are more often plural than victims (at 69.32% compared to 58%). This use of plurals may help to signify the 'large threat' of trafficking, justifying Fight-responses and newspapers' ongoing attention to migration/trafficking. The use of this strategy is illustrated by the following quote from The Daily Express, discussing migrants choosing to camp at Zeebrugge rather than Calais: 'Belgian authorities have stopped 890 migrants without residency papers since January-450 of whom travelled to Zeebrugge which is seven miles from Bruges. Dozens of traffickers have also been detained' (Sykes, 2016, emphasis added). This excerpt also displays a conflation of trafficking and smuggling, as it is unclear whether these supposed traffickers simply worked to transport people across borders, or did so with the intent of exploiting them.
In the sample corpora, traffickers' foreign nationality is often highlighted ('four Libyans and one Tunisian', 'ten other Hungarians', 'Ivorian', 'Dutchman') or indicated merely by their full, and most often foreign, name ('Wei Yu-a naturalised Irish citizen who lives in Limerick'; 'Razvan Ursu'; 'Zoltan Raffael'; 'Toi Van Le', and 'Perry Wacker', who was actually smuggling rather than trafficking immigrants into the UK) or surname ('Blaga'), the latter suggesting formality and also distance. Indeed, only 26% of c-collocates referring to nationality or origin to keywords identifying traffickers refer to Britain. This is equal to the proportion of 'local' c-collocates associated with victims. 58% of c-collocates referring to nationality or origin indicates that traffickers are associated with other countries or are foreign. This is also similar to the proportion of 'foreign' c-collocates associated with victims. This finding again underlines the interpretation that trafficking is framed as an imported problem. Reporting on the rise in prostitution, for instance, The Daily Star claims: 'Large numbers of women are also being trafficked by foreign gangs into prostitution' (Malley, 2006).
Other than these, however, there are few characteristics that traffickers have in common. They include gangs, family members (father, mother), men and women. However, most traffickers in the spike sample corpus data are indeed male ('four men', '42-year-old man'), some are female (a 'Nigerian woman', a '19/38-year-old woman'), with their age ranging from nineteen to fifty-seven when specified. In terms of their evaluation, such individuals are 'dubious' and 'unscrupulous' at best, and '(true) bloodsucker bosses', 'callous', 'evil', 'brutal', and 'heartless' 'monsters'/' parasites'/'villains' at worst.
In short, it is difficult to establish whether newspapers actually create a 'big, bad' trafficker, as the characteristics of the represented individuals actually vary. Likewise, in her 6-month corpus containing 354 trafficking cases in news texts, Denton (2010) found that all possible gender-and role-combinations were present. Sobel (2016) also found that while victims are primarily presented as female, traffickers are either mixed, or with their gender unmentioned, something that, for Sobel, is indicative of the gender of traffickers being male, by default. Instead, newspapers appear to focus on the experiences and characteristics of victims. By doing so, they fail to address the structural factors that cause people be involved in trafficking, either as victims or as traffickers.

Agency and Focus
This section argues that there is a general obscuring of agency in acts of trafficking. Furthermore, victims are presented as having little agency. As 5.39% of keywords relate to an act of trafficking and/or abuse, this overall lack of agency suggests that newspapers tend to focus on the suffering of victims, rather than on those parties who cause such suffering in the first place. Having said that, traffickers are generally identified as actors and agents; they are not acted upon either.
This focus on suffering, that does not specify who caused this suffering, is clear when considering the form of those keywords categorised as 'acts of trafficking and abuse'. They are generally nouns (abuse, death, exploitation, issue, labour, marriage, prostitution, sex, slavery, threat, violence), intransitive verbs (died, worked), or adjectives relating to these nouns (hard labour, modern slavery, sexual exploitation, sexual violence). Those transitive verbs that have been included (forced, trafficked, killed) are all in the past participle form, meaning that they can also be used to form (agentless) passives, in which the focus is on the subject undergoing the act, rather than on the object performing the act (if at all included).
Qualitative analysis of corpus data revealed many nominalisations. Some are found in the reference to 'economic migration having led to "conditions in which […] enforced labour could occur"'. Here, the active voice is employed; the nominalised migration causes conditions that could lead to enforced labour, rather than identifying the global socio-economic and political factors related to migration. Also notice the nominalisation of 'enforced labour', which leaves it up to the reader to determine who forces labour. Similarly, there is 'abuse' ('cocoa farm abuse', 'suffered'/'subjected to (terrible) abuse'), 'exploitation' ('the exploitation of migrant workers'), violence ('subject to threats of violence') and a 'demand for these crimes', again leaving it unclear who performs these acts. Lastly, in '[s]ex trafficking and illegal workers threaten Olympics', illegal workers are presumably victims of trafficking, but are presented as agents of threat, and as something to be dealt with. They are presented as problem-causing individuals that threaten the major institution that is the Olympics. As these workers are illegal, they are simply stripped of their victim-status.
The sample dataset is also full of agentless passives that do not require the agent to be referred to. Vehicles carrying victims are 'destined' and 'being used' with no indication as to who is behind the act. As taken in by deception, victims are 'promised', 'told' lies, 'cruelly snared' and 'duped'. As goods, victims are 'handpicked', 'scooped up', 'taken'/'separated from their families', 'being b(r)ought', 'sent out' to engage in forced criminality, 'delivered', 'sold', 'trafficked' (sometimes conflated with 'smuggled') and 'sent' to live elsewhere. As kept against their will, they are 'kidnapped', 'held hostage', 'locked (up)', 'held', 'tied (up)', 'hidden', 'imprisoned', and 'crammed' into small spaces. They are also vulnerable when 'exposed to hazards and illnesses or left alone on city streets' and come to be 'used ', 'groomed', 'exploited', 'coerced', 'mistreated', 'enslaved', 'given no money', 'denied'/'not afforded'/'robbed'/'deprived of' 'legal'/'basic rights' and 'discriminated against'. As violently acted upon, they are 'forced to work […] not being given food or paid for their work', 'attacked', '(sexually) abused', 'molested', 'raped', 'beaten' and end up 'traumatised', their lives 'ruined'. Victims are 'stranded', cases are sometimes eventually 'uncovered' and the victims 'found'/'discovered' and 'freed'/'rescued'/'taken to a place of safety'. In other words, victims are always 'done to', whether by the perpetrators or the police. They lack agency. As these forms allow the actor to be discarded (Jeffries, 2010), newspapers do indeed appear to focus more on the acts of trafficking and abuse than on those performing these acts.
Some of the keywords in this category have an ambiguous form, that is, it is unclear whether they should be read as adjective, noun, or verb. Domestic, for instance, can refer to a person working as a domestic servant, but can also be used to refer to domestic slavery. More striking is the set of words that can technically be read either as a noun or a verb. These are attack, force, murder, rape, work, services, hit, and trafficking. If taken to be verbs, they would mostly be present tense first person ones, a form not commonly used in newspaper writing; the most straightforward interpretation is that these are actually nouns.
When considering the grammar of those target words that specifically refer to traffickers, another pattern emerges. Despite a general lack of focus on traffickers (traffickers are also only explicitly referred to half as much as victims), traffickers are assigned much more agency than victims. They are the grammatical subject more often than victims are, at 39.58% against 31.77%. Victims, on the other hand, are more often the grammatical object. More importantly, traffickers are identified as agents (through the collocating preposition 'by') at 2.46%, against victims' 1.46%. These tendencies are especially pronounced in the target nouns trafficker and victims. At 7.52%, trafficker is particularly often indicated as an agent in passive sentences. Furthermore, a trafficker is often the grammatical subject of an active, rather than a passive, sentence.
In other words, nominalisations and transitivity analysis reveal that traffickers act, victims are powerless and acted upon, and victims' suffering remains the focus of these texts. Traffickers' direct responsibility for this suffering lacks exposure.

Modality
This section turns to the attitudinal study of the language used in the spike sample corpus texts specifically. It shows that the use of deontic modality suggests that HT is presented as something that must urgently be dealt with, whereas the use of epistemic modality is used by journalists to overstate the scale of the problem and to present the problem as (excessively) serious, without having to write untruths. The deontic modality system, that of duty, is expressed through verbs like 'should' ('members states should step up', 'The harrowing testimony of a woman forced into sex slavery as a teen in Ireland should be a wake-up call'), 'let' ('Police let them walk away'), 'must' ('We must make sure they are aware there are women held against their will', 'they must prostitute themselves', 'A FIRM deadline must be put in place'), 'need' ('Consumers also need to be aware of potential exploitation of nail bar workers', 'People really need to open their eyes', 'a fuller package of proposals is needed'), 'commit' ('the European Convention on Human trafficking commits signatories to tackling the traffickers'), 'allow' ('In future domestic workers will only be allowed in on non-renewable business visas'; 'freed victims of trafficking will be allowed to stay', 'allowing her to be raped by a string of men', 'allowed employees to be paid appropriately') and 'oblige' ('Britain […] is not obliged to take refugees'). Overall, deontic expressions are very often to do with what victims and potential victims are and are not 'allowed' to do, what 'we'/the politicians/police/states should be doing, and what laws commit/allow people to do things. The news texts suggest that there is urgency, commitment, and necessity to respond to the HT problem.
The epistemic system deals with (un)certainty. Certainty is often communicated through the verb 'will' ('[trafficking] will continue to exist'; 'Illegal immigrants will continue to pour into the UK', 'Britain will remain the number one destination for sex-slave traffickers'). Uncertainty, on the other hand, is expressed through verbs like 'may' ('The slave traders may dump [their cargo of children] overboard'; 'Criminals may make promises', 'what may be hundreds of people'), 'can' ('child labour can involve youngsters', '[t]he type of work children can be involved in differs greatly', 'slavery can involve sexual exploitation', 'someone can make £1 million a year out of 10 women'), 'could' ('She could then earn her buyer £800 a day', 'Migrant workers […] could become virtual slaves', 'It could have ended in tragedy/be abused by illegal immigrants'), 'would' ('patrolling the African sea-lanes would be a far wiser policy than […]', 'the convention would help Britain', 'it is just inconceivable they would give evidence to trial', 'They would leave migrant workers […] destitute and homeless'; 'Changing employers would not be appropriate', 'Home office policy reversal would strip them of their right […]') and 'might' ('might have been involved in the human trafficking syndicates', 'might have become a watery grave'). This first set of epistemic phrases presents human trafficking as a very serious, large, problem, without the journalist having to present precise numbers. This also allows them to increase the threshold value (see Galtung & Ruge, 1965) of this news.
Uncertainty is also communicated through nouns ('a possibility for sexual exploitation', 'allegations that they treat their workers [… like] slaves'), adjectives ('possible/alleged victim/abuse/people trafficking', 'suspected traffickers/victims of human trafficking'), and adverbs like 'allegedly' ('allegedly held as slaves') and 'likely' ('Gangs are likely to target', 'the migrant crisis was likely to make people-trafficking globally easier', 'more likely to be in unpaid family work'). Such uncertainty is also expressed through (visual) perception modality ('If you see a car wash, and it is clear that people look like potentially being exploited', 'workers look hungry', 'who doesn't look like he works in that world', 'clearly the problem is most serious in lower income countries', 'she seemed uncertain'). Uncertainty can also be communicated through cognition modality, that is, uncertainty through belief ('officers believe women are frequently raped', 'cages believed to have been used'). As epistemically modalised expressions allow the writer to say things that might not be true and are often (legally) necessary, their overuse exaggerates the nature/ extent of the problem, again increasing threshold value, and can even generate panic.

Speech and Writing Presentation
This section turns to the voices behind such urgency and uncertainty. Other than the reporting voice, who/what else is quoted? It shows that primarily powerful people are quoted. This is problematic, as by reporting certain opinions over others, reporters lend legitimacy to these viewpoints over others (Louw, 2005). This may not be intentional, but journalists often report the viewpoints of people who are perceived as authoritative sources, who also tend to be socially powerful (Cottle, 2003;Kuhn, 2007;Machin & Niblock, 2006). 21 When exploring the spike sample corpus data in terms of who does and does not talk and write, we found that the reported tend to be either police officers (such as Detective Superintendents, Deputy Chiefs and Inspectors), and also various senior government members and ministers. The latter range from the Prime Minister (or their spokesperson/advisor), the (Shadow) Home Secretary, various (foreign office/immigration/ defence) ministers, Home office/UN reports, chief council executives, party (candidate) leaders/(European) Parliament Members, advisers to the parliamentary committee on HT, the anti-slavery commissioner, and even the Governor of New York. Also reported are those parties whose interests lie in protecting victims (including the Pope, judges, community support workers, charity/Union spokespeople, International Labour Organisation specialists/data, 'Save the Children', European Commission reports/working documents), with little said by those parties whose interests lie in protecting those accused of HT (there is just one instance of 'their lawyer said').
Readers also hear nothing from the HT perpetrators and very little from the HT victims themselves (quotes attributed to 'the boy' found in a suitcase and 'a child sex abuse victim' reporting on a UK intranational compared to stories without. Incidentally, many news values are also those aspects of HT news that are problematically overreported: unambiguity (Galtung & Ruge, 1965)/simplification (Jewkes, 2011), or the binary representation of victims and offenders; sex; violence or conflict; visuals, and the inclusion of/a focus on children (Jewkes, 2011). In other words, problematic representations of HT are perpetuated as they are perceived to resonate with readers, with the result that global inequalities are maintained (Gregoriou & Ras,Chap. 1 in this book). Churnalism is the increasing tendency for journalists to publish any story, 'whether real event or PR artifice, important or trivial, true or false' (Davies, 2008, p. 59). This trend is due to the increased commercialisation of the British press, which led to staff cuts despite growing revenues, and an increased pressure on remaining staff to 'churn out' as many stories as possible, reducing the ability to (fact) check stories and actually do investigative journalism (Davies, 2008). As a result of these pressures, much of modern journalism is little more than the recycling of PR, press office, and wire agency outputs (ibid.). The problem with PR and press offices, however, is that they work for institutions and individuals with their own agenda, such as the police, political parties, the government, and corporations (ibid.). When churnalism and news values are considered together, what becomes clear is that newspapers are given stories that represent events in a way that benefits the establishmentparticularly politics and business; have no time to fact check these stories and uncover different perspectives, and are, due to an increasing demand for revenue and readers, pushed to highlight reader-friendly aspects. This becomes more sinister when considering that highlighting these aspects also benefits those organisations whose press offices put out these stories in the first place. trafficking case are exceptions). In fact, there is reference to the latter being unable to provide necessary evidence altogether ('these women are so traumatised that it is just inconceivable they would give evidence in a trial'). Even though language/communication often proves problematic for victims, and legal consequences may arise when victims do away with their anonymity, the tendency is to silence them completely and regardless.

Multimodality
Lastly, we turn to look at those fifty-three texts we were able to trace the facsimiles (45)/online (8) newspaper article versions of, questioning first the placing of the text among others on the page/relevant newspaper, and second the choice and nature of any accompanying images only some of which are printed, or at least made available to us, in colour.
We were able to locate forty-five of the spike sample corpus articles in the form of facsimiles, which enabled an examination of the journalistic context surrounding the HT articles as originally printed, telling as it might be to consider how the relevant stories relate 'to other stories on the page which are also competing for the reader's attention' (Caple, 2013, p. 160). Surrounding texts mostly dealt with people movement in general (smuggling, refugees and asylum seekers, migrants and deportation), politics, finance, national crises, international relations/conflict, crime (including abuse, murder, and fraud), various other unfortunate events (such as scandals, disasters, and suicides) and, lastly, sex and celebrity. None of these HT-related articles appeared on newspaper front pages. On average, these featured around page 15 in each newspaper and were about 269 words long; these articles' placing far into the paper and short length suggests they are not flagged up as of the utmost importance (see Chermak, 1994), however newsworthy they may be.
Only twenty of the spike sample corpus articles located (fourteen facsimiles, and six online) had any accompanying images. Several of these twenty merely contained photographs of the officials reporting or reported, the victims/perpetrators/officers involved in the cases reported, or displayed HT-related means (lorries, boats, a suitcase) or stereotypical circumstantial scenes (cages, houses, farms, stations and camps, and stock images of workplaces such as nail salons and car washes). Most noteworthy in terms of their relatively large size and enactment of temporal relations/provision of narrative progression (see Caple, 2013, p. 177), are pairs/series of images: that depicting a capsized boat of migrants, that of two women being questioned when found 'crammed' in the back of a van by bight-yellowed servicemen (the attire hence emphasising the latter's official role, see Caple, 2013, p. 49), and that with red-circled supposed traffickers 'hiding' in a group of migrants on a Navy ship. Perhaps most striking, however, are those images depicting (ideal victim) children. One pair of shots depicts a child physically seen inside a suitcase (with one of these shots being an airport scanner image); in another, a colourfully dressed isolated child breaks tiles outdoor in Mumbai; in a third, a group of four 'unaccompanied', the caption says) children walk next to a pond in a refugee camp in Greece at dusk, their reflection seen across the water. This last one adopts the 'iterating', 'matching', and 'mirroring' compositional configuration (Caple, 2013, p. 99) and can be read along metaphorical lines-the life of these children can take several forms, and it is up to us (Guardian) readers perhaps, to help them find the right one. Similarly, an image accompanying a Guardian article of domestic workers depicts two Filipino women wiping a wooden floor on their knees, and hence also adopts the 'iterating' and 'matching' configuration (though not the 'mirroring' one). Unlike all other images in the sample the perspective of which is horizontal, the camera angle is here high, the two participants are photographed from above, and the viewer is hence placed in a dominant position (Caple, 2013, p. 39); the viewing perspective here suggests a master/employer literally looking down on, and hence metaphorically taking advantage of, the women. In one last image, a man, head cropped off the stock image, is shown washing a car, the sky's slightly cloudy blueness suggesting optimism for one's future.
Such HT-related photojournalism depicts mostly groups rather than individuals, which suggests mid/low compositional salience, construes the news values of 'Impact, emphasising the consequences of an event, but also Superlativeness [as in the maximised or intensified aspects of an event]' (Caple, 2013, p. 41) while, even where subjects are seen in isolation or in pairs, there is a distinct lack of focus on actors' faces or emotional responses. Such images being mostly long shot, with actor faces being obscured and facial features mostly indiscernible or with 'neutral and negative facial affect' (Caple, 2013, p. 79), is suggestive of these foreign-looking migrants (possibly involved in trafficking) being a faceless and nameless group, and one hard for the viewer to relate to or sympathise with; the relationship between the viewer and those viewed is impersonal, detached. There are no frontal angles and no direct gaze at, or acknowledgment of, the viewer, no 'demand' (Kress & Van Leeuwen 2006, p. 118) for the viewer to do anything in response to the image. Represented participants are instead in oblique angles, and merely on 'offer' (Kress & Van Leeuwen, 2006, p. 119) for the invisible viewer's contemplation. The depicted actors are shown mostly engaged in what Kress and Van Leeuwen (2006, p. 79) describe as 'Narrative Structure', that is, 'unfolding actions and events, processes of change, transitory spatial arrangements'. Here, actors are working, walking, falling into the sea, or swimming. Though depicted as agents, the image cropping/framing suggests that it is others who are imposing these actions on those depicted, in fact. Where masses of immigrants are seen crammed in a fishing boat, they are passively being carried, with little indication of who the agent of this potentially criminal action is. Similarly, in those few images where the depicted actors are shown as standing or sitting, these individuals are portrayed as waiting for others to make life-altering decisions as to their future. In sum, the images suggest that all those (potential victims) depicted are ultimately all agentless. conclusIon Our quantitative and qualitative critical discourse analysis of 2000-2016 British news media reveals a close correlation/conflation between transnational HT and smuggling (concepts also blurred with those of immigration and asylum seeking), which contributes to HT victim criminalisation. Where female, victims are sexualised, which supports the notion of sexual exploitation being over-reported. Both HT perpetrators and victims are seen as numerous, illegal and non-British. Where the victims are young, vulnerable, coerced, agentless, silenced, unrelatable, and offered merely for reflection, perpetrators of the HT problem vary, are not focused on/ identified, and are not acted upon. HT itself is represented as an imported, perhaps exaggerated problem, which is very bad and must be fought, urgently. Such journalism foregrounding official responses is the result of vested interests, while state actors are shown to be heroes, its victims reduced to the measure of mere usefulness or value. The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.