20.1 Social, Sociohistorical, and Sociolinguistic Facts

With a population of 258 (April 2014),Tristan da Cunha is one of the smallest communities in which English is spoken natively. Its unusual social history has been researched thoroughly and is discussed elsewhere in detail (Schreier Reference Schreier2002, Reference Schreier2003; Schreier and Trudgill Reference Schreier2006; Schreier and Lavarello-Schreier Reference Schreier and Lavarello-Schreier2011), so the following account is kept brief. Originally discovered by the Portuguese in 1506, Tristan da Cunha was only settled in 1816, when the British admiralty formally annexed the island and dispatched a military garrison to the island. When it withdrew after one year only, some army personnel stayed behind and settled permanently: two stonemasons from Plymouth (Samuel Burnell and John Nankivel), a non-commissioned officer from Kelso, Scotland, named William Glass, as well as his wife, “the daughter of a Boer Dutchman” (Evans Reference Evans1994: 245), and their two children. The population increased when shipwrecked sailors and castaways arrived soon after, and in 1824, apart from the Glass family, the settlers included Richard “Old Dick” Riley (from Wapping, East London), Thomas Swain (born in Hastings, Sussex), and Alexander Cotton (from Hull/Yorkshire) (Earle Reference Earle1966 [1832]). The late 1820s and 1830s saw the arrival of a group of women from St Helena and three settlers from Denmark and the Netherlands, later joined by American whalers. The population then grew rapidly. By 1832, there was a total of thirty-four people on the island, twenty-two of whom were young children or adolescents.

The second half of the nineteenth century, on the other hand, was a period of growing isolation. There were hardly any new settlers, with the exception of a weaver from Yorkshire (who left after a few years only) and two Italian sailors (Crawford Reference Crawford1945). This state of isolation lasted well into the twentieth century. When visiting the island in 1937, the Norwegian sociologist Peter Munch found that the Tristanians basically lived in pre-industrial conditions (Munch Reference Munch1945), and Allen Crawford, the cartographer of the expedition, noted that only 6 out of a total of 170 Tristanians had ever left the island. This changed in April 1942, when the arrival of a British navy corps saw economic changes; a South African company employed the local workforce in the fishing industry, and the traditional subsistence economy was replaced by a paid labor-force economy. These social changes had sociolinguistic consequences, as the usage of local dialect features decreased somewhat (see Schreier and Lavarello Schreier Reference Schreier and Lavarello-Schreier2011).

Though sociodemographically and politically insignificant, TdCE has perhaps over-enthusiastically been referred to as the “sociolinguists’ Galapagos” (Chambers Reference Chambers and Kortmann2004: 134), and two reasons can be advanced why it is particularly suitable for studies on contact-induced language change and variationist sociolinguistics. First, it is a prime research site because the community's founders settled under tabula rasa conditions. There was no contact with indigenous varieties since the island was practically uninhabited when the garrison arrived (the population was precisely two: one of the people present, a boy from Menorca, left on the first occasion, the other one, an Italian sailor, drank himself to death in the soldiers’ canteen). As we can date its origins to the 1820s, TdCE is one of the youngest nativized Englishes around the world (approximately one generation or so older than New Zealand English). Second, the English input varieties to TdCE are well known, as is the development of the local population (there is an entire genealogical tree for the island community, reprinted in Crawford Reference Crawford1982). We know that the feature pool included dialects from the British Isles (the founders came from the Scottish Lowlands, East Yorkshire, East London, and Hastings), the United States (the most influential American resident being a whaling captain from Massachusetts), various second-language varieties (spoken by settlers with the mother tongues Danish, Dutch, Italian, and (perhaps) early nineteenth-century Afrikaans), and St Helenian English (StHE), a variety that has undergone some amount of restructuring and perhaps creolization (Schreier Reference Schreier2008). As a result, multiple contact processes were operative during the genesis and formation periods of TdCE. Schreier (Reference Schreier2002, Reference Schreier2003) and Schreier and Trudgill (Reference Schreier2006) argued that TdCE primarily derives from varieties of British/late eighteenth-century American English and StHE, and that it did not emerge in a context of prima facie language contact, excluding pidginization effects on Tristan da Cunha. Some features most likely had a vernacular British English origin, such as slight strut fronting, /t/ glottalling, /v ∼ w/ merger, thought in cloth, nucleus fronting in mouth, start backing, th fronting, fleece in fish, happY tensing, etc. On the other hand, L2 forms had some impact on TdCE as well, and a few non-native features were adopted also (th sibilization, i.e. dental fricatives realized as /s/, as in think, throw, etc.; cf. Schreier Reference Schreier2003: 211). The existence of creole-type features in TdCE (such as high rates of consonant cluster reduction and absence of -ed past tense marking, Schreier Reference Schreier2005: 152; /v/ realized as [b]; lack of word order inversion in questions; copula absence with locatives and adjectivals; etc.) served as a strong indication that a creolized form of English was transplanted via (at least some of) the women who cross-migrated in the 1820s.

The question thus is how mixing between the inputs proceeded and why TdCE selected the features it did, and it is the earliest set of recordings of Tristanians (born in the late nineteenth century) that is of utmost significance here.

20.2 Researching Tristan da Cunha English: Taking Stock

There are several corpora currently available for a historical analysis of TdCE. These include the 1961–1962 BBC/UCL corpus, the Svensson/Munch corpus, the 1999/2010 Schreier corpus and, as the latest substantial addition, a set of recordings made by Scottish oral historians in 2006. All these collections have their advantages and shortcomings, mostly due to the context(s) in which they were complied, and these are briefly described in the following.

The first ever linguistic description of TdCE comes from Zettersten (Reference Zettersten1969). Zettersten analyzed recordings made in 1961–1962 by the British Broadcasting Corporation (BBC) in cooperation with the University College of London (UCL). At the time, the entire Tristan population was residing in Calshot, a disused army camp near Southampton, UK, where they were forced to live after volcanic activities endangered the settlement and enforced a full-scale evacuation. Following a major eruption on 9 October 1961, the Tristanians had to leave their island within twenty-four hours and were first shipped to Cape Town, where many of them for the first time saw modern civilization. However, since South Africa was no longer a member of the Commonwealth (it had been expelled after the National Party instated and implemented the Apartheid regime in 1948), the Tristanians were brought to England, arriving in November 1961. Though settling in remarkably well (many of them had jobs in the local industry), the majority voted for a return to the South Atlantic the following year, when the island was declared safe.

The data compiled by UCL, in collaboration with the BBC, amount to a total of about four hours, or about twenty-five recordings made with approximately twenty Tristanians. Most interviewers were well-known radio journalists (e.g. René Cutforth) or linguists working for UCL (among them Jan Svartvik), all of whom were highly educated and had RP accents (the exception being the technicians, who at times contributed to the conversation but were not in charge of leading the interviews). Since they were not familiar with the Tristanians, they tended to ask rather general (at times banal or rather insensitive) questions about island life and history, occasionally revealing their ignorance about the island and its past. The context of these recordings was very formal indeed: interviewees were asked to join the interviewers in a separate room that was specially equipped for the purpose, and where they had to speak into a microphone placed in front of their mouth. Other factors contributed to the stiff environment of the recordings: each interviewer wore a suit and tie, contrasting with the islanders’ traditional knitwear, often there were several people present at interviews (one or two interviewers plus a technician), and it was conducted in an upper-class accent, which contrasted notably with TdCE and reinforced the social distance between interviewer and interviewee. All this bore little resemblance with the principles of sociolinguistic fieldwork developed to elicit vernacular speech in a relaxed and informal setting (Labov Reference Labov1982; Schilling Reference Schilling2013; Schreier Reference Schreier2014). As a result, the recordings made by the BBC tended to be short, not yielding a lot of data: most were between six and fifteen minutes long.

The Svensson/Munch Corpus was compiled by the Swedish Painter Roland Svensson (1910–2003) and the Norwegian sociologist Peter A. Munch (1908–1984). Svensson was a passionate painter of islands who travelled extensively in the Atlantic and also in the Pacific. He developed a special liking for Tristan da Cunha and visited the island several times. He had a strong ethnographic interest and collected all sorts of material on domestic life, shared the everyday experiences of the islanders and also carried out recordings (some in Calshot in 1962, much more when visiting the island in the 1960s and 1970s). Though he gave much room to his own reflections (at times there are lengthy monologues and musings in his recordings, to which the islanders listened politely), he was well liked and popular with the population. Since the recordings were carried out in the Tristanians’ homes, often over a cup of tea or something stronger, the context was very relaxed and informal. Moreover, Svensson occasionally recorded couples or even asked some of the “old hands” for a group interview (a real treasure chest for everybody interested in the social history of the island and the historical sociolinguistics of TdCE).

Similarly, Peter Munch was highly familiar with life on Tristan da Cunha. He was a member of the Norwegian expedition that visited the island in 1937–1938, when he collected information to produce the most detailed account of social organization in the community. Munch spent World War II in a concentration camp and emigrated to North America in 1946, where he had a remarkable academic career at Southern Illinois University in Carbondale and held a chair in sociology. He published several books on the sociology of Tristan da Cunha and had an engagement with the community until he passed away in 1984. His recordings and notes are priceless, since he studied the island for almost half a century and began studying the community when it lived in practically pre-industrialized conditions (Munch Reference Munch1945). Though the observer's paradox (Labov Reference Labov1982) can never be fully discounted, the audio data at hand come rather close to unmonitored casual speech. The recordings collected by Svensson and Munch comprise a total of twenty-three hours with about fifty individuals, recorded alone or in groups. Both Svensson and Munch were “informed outsiders,” so to speak, they both had close connections with the Tristanians, and were well liked and respected on the island. In their interviews, they asked about local life, family histories, all sorts of incidents in the late nineteenth and early twentieth century (including ghost stories, mysterious sightings of missing ships, information on some of the earliest settlers on Tristan, etc.), and collected all sorts of reminiscences of islanders, immersed in the habitat of the people they spoke to. The fact that these were fruitful topics for discussion is attested both in the length of responses given and stories elicited and also in the total size of the corpus.

The Schreier Corpus consists of recordings made in 1999, 2002, and 2010. The total length of the recordings is about fifty hours with eighty Tristanians. For his PhD project, Schreier initially spent six months on Tristan in 1999. Adopting a “friend of a friend” approach, he expanded his acquaintances in snowball fashion. He spent the first three months getting to know the island and carried out very few recordings, with elderly members only (see the methodological description in Schreier Reference Schreier2003). Only later were younger Tristanians recorded. Great care was taken to bring up local topics and stories (ghost stories, past experiences, e.g. during the volcano years in England), and all the individuals were recorded in places where they felt comfortable. The aim was to analyze a representative sample of the population, so Tristanians of all age groups were recorded, either in group interviews (up to four speakers) or individually. More recordings were carried out in follow-up visits in 2002 and in 2010. On his last fieldwork trip, Schreier was joined by a local fieldworker, the Head of the Tourist Division, who developed a strong interest in the project and helped during the fieldwork activities with the aim of collecting a body of oral history, to be housed in the local museum. The 1999 subcorpus was analyzed in Schreier Reference Schreier2003, but the recordings made in 2010 have not been studied so far.

The Scottish Corpus, finally, consists of recordings made with some twenty-five individuals, both locals and expatriates alike. The interviewers were Scottish historians who visited the island on the SA Agulhas in September 2006 and spent three weeks on the island. They had no prior first-hand information and singled out individuals with the help of a Scottish dentist, who had a long-term relationship with Tristan as he was sent to Tristan on an annual basis starting in the late 1990s. They mostly worked with a semi-structured questionnaire and asked general questions about island life. Within merely three weeks, they managed to conduct twenty-five interviews (mostly single interviews) with a total of twenty-five residents, thirteen of which were locals. Interviews were carried out in local homes and mostly lasted between forty-five and sixty minutes.

When comparing the various corpora, we find that there are rather substantial differences relating to context, length of interview and interview styles, informant selection, and also quality of the recordings. With an interviewer to interviewee rate of at times three to one, the 1961 BBC corpus is formal, and interviews rarely last more than twenty-five minutes. As for the principle of accountability (Labov Reference Labov1982), sufficient material for a quantitative analysis of sociolinguistic variables is available for few speakers only. On the upside, recordings were made with a representative sample of the population, with children and also elderly members (the oldest speakers were born in 1887, 1894, and 1895) and the overall quality is good. The Svensson/Munch corpus, by contrast, is rather informal. Svensson was well acquainted with the Tristanians and brought up topics of interest to them. Interviews vary in length and were at times led in groups, as a result of which they are highly informal. The disadvantage is that there are only a few recordings of younger speakers and that Svensson and Munch concentrated on the “old hands” (in fact, the oldest Tristanian whose speech is preserved, Granny Mary, born in 1874, was recorded in 1963; others were born in 1889, 1900, and 1902); moreover, the quality is not always good (there is a constant humming from the background noise of the tape recorder, and some tapes are worn out because they were not stored properly). The Schreier corpus (Schreier Reference Schreier2003) is rather informal as well (interviews on the island, in the informants’ homes, ethnographic approach, and six-month residence on the island, local topics, etc.). Some speakers were recorded twice (1999 and 2010), allowing for a (albeit modest) real-time study. The Scottish corpus, again, is somewhat more formal in that the interviewers worked in pairs and only had three weeks on the island. However, they were briefed by a Scottish dentist (see above) who was familiar with the community and made arrangements for them beforehand.

Consequently, the four corpora differ on a number of accounts: place of recording (Calshot/Southampton or South Atlantic Ocean/Tristan da Cunha), degree of formality, degree of familiarity with the population (ranging from rather high in the case of Svensson/Munch and Schreier to low in the case of the BBC reporters), length of recordings (mostly forty-five to sixty minutes in the case of the Scottish oral historians to about five minutes in the BBC corpus), and overall quality of the recordings (high, BBC, as opposed to varying, sometimes very low, Svensson, decreasing towards the end of an interview since some tapes are worn out). Considering these differences, it is clear that they should not be compared without qualification, and the advantages and limitations of each corpus must be taken into consideration at all times (this is the reason why I myself have not worked with the extremely formal BBC corpus as of yet, because I felt it contrasted too much with the informal character of my own 1999 recordings). Analyses of change in real time, for instance, based on a comparison of speakers recorded by the BBC (1962) and Schreier (1999) would confuse the degree of formality and compare data from different styles, which obviously influences the results. This must be kept in mind at all times.

Bearing in mind these methodological concerns, the potential of the four corpora for research on variation and change or new-dialect formation is certainly great. Together, they comprise some 140 hours of tape-recorded speech with more than 150 Tristanians born between 1874 and 1990. Moreover, two of the corpora (Svensson/Munch, Scottish) have not been subject to linguistic analysis, neither has Schreier (Reference Schreier, Auer and Schmidt2010), so the potential for future research of these recordings for research on historical phonology, contact linguistics, and variation and change is immense.

20.3 Analyzing Earlier Tristan da Cunha English: General Findings and a Case Study

Despite the caveats outlined above, it is clear that we must take care when comparing data from different corpora. However, the material available lends itself to studies on language variation and change. For one, a replication study can be carried out since individual speakers were interviewed several times (of course, one would have to take great care to select recordings taken in similar contexts of formality so as not to confuse real change with variation along the informal-formal axis; see above). Individuals recorded by Svensson in the 1970s and Schreier (Reference Schreier2003, Reference Schreier, Auer and Schmidt2010) would lend themselves really well for such purposes. This would enable researchers to study the speech of single individuals over a period of time (see Rickford and McKnair Knox Reference Rickford, McKnair-Knox, Biber and Finegan1994 for a longitudinal case study). Alternatively, in a sample study, one can select pre-specified samples of individual speakers in an apparent-time approach, a practice adopted by Schreier (Reference Schreier2003). This is ideal for an analysis of mobility-related language change (difficult to travel to and away from, Tristan da Cunha is sociodemographically stable with a small population).

The Svensson corpus allows us to extend the degree of time depth well into the nineteenth century and to identify features in the earliest-born Tristanians recorded. The oldest islander whose speech is preserved is that of Granny Jane, born in 1874, recorded by the South African Broadcasting Corporation in 1954.Footnote ² Unfortunately, the recording with her lasted merely 9.06 minutes and is thus too short for a quantitative analysis. Notwithstanding, it is noteworthy that Granny Jane uses several local features that have been recorded in Zettersten (Reference Zettersten1969), Schreier and Trudgill (Reference Schreier2006), etc. and is in fact strong evidence that these forms were present in generation three of the Tristan population already (and not innovations of the twentieth century), which is most useful for the historical reconstruction of TdCE with reference to current models on English as a world language at large or koinéization in particular. A descriptive profile of the data includes the features shown in Table 20.1.

Therefore, a recording such as this one, as short as it may be, pushes back the time frame by 30–40 years (or a generation or two) and allows precious glimpses into mid-nineteenth century TdCE. Some of these variables were subject to a quantitative analysis in twentieth century TdCE (past be leveling in Schreier Reference Schreier2003, consonant cluster reduction (CCR) in Schreier Reference Schreier2005).

As for the early twentieth century, Zettersten (Reference Zettersten1969) provided a detailed qualitative description and made some general assessments on differences between older and younger speakers, and Schreier and Trudgill (Reference Schreier2006) reported their impressionistic findings from some of the oldest speakers available, comparing their findings with observations on the speech of younger members of the community. They further discussed them in order to retrace koinéization and language-contact processes in the formation period of TdCE. Schreier (Reference Schreier2003) used the apparent-time construct to study change in the twentieth century, reporting that some local variables were on the decline (e.g. third person singular present tense zero or present be-leveling), presumably as a result of increased contact with the outside world during the years 1961–1963 in England following the volcano eruption on the island.

Research has also been carried out in segmental phonology, and Figure 20.1 provides a vowel plot for a male TdCE speaker, born in 1902 (one of two speakers analyzed so far).

Figure 20.1 Vowel plot for a male TdCE speaker, born 1902.

Figure 20.1 indicates some prominent features, such as the near–square merger, kit backing and centralization, goose fronting, and the merger of north and force; further research needs to be carried out with regards to Tristanians born in the second half of the twentieth century.

Quantitative research is possible as well, and in what follows the focus is on shifting by interview context (which the corpora lend themselves to ideally, see above). The question is whether there are differences between single speakers in the corpora, i.e. whether or not there is evidence of context sensitivity that gives rise to differential usage rates of sociolinguistic variables that can be explained by the setting of the interview. The variable selected is /h/ insertion, which is analyzed quantitatively for some selected speakers here.

The history of /h/ variation in English has been studied intensely, by historical linguists, philologists, and sociolinguists alike. As is well known, /h/ underwent a long process of weakening in the history of English (Lass Reference Lass, Hogg and Denison2006) and has now become a minority feature in some regions, particularly in the English southeast and in vernacular London English. On the other hand, there have also been reports of /h/ insertion, i.e. usage in words where it is not found etymologically, for instance in the fourteenth-century Norfolk Guilds and the Paston Letters (Wyld Reference Wyld1925). The Linguistic Atlas of Late Medieval English (McIntosh et al. Reference McIntosh, Samuels, Benskin, Laing and Williamson1986) indicates that it is mainly found in texts from the East Midlands, East Anglia, and the south. In the period from c. 1190 to 1320 the texts range from Lincolnshire or Norfolk to the southern counties but “the instability seems to be greatest in the East Midlands” (Milroy Reference Milroy1992: 140). This has been noted frequently in the eighteenth and nineteenth centuries, for instance by Batchelor (Reference Batchelor1809: 29), “the aspirate h … is often used improperly, and is as frequently omitted where it should be used. Give my orse some hoats has been given as an example of these opposite errors from the Cockney dialect,” and by Walker (Reference Walker1791), who speaks of a “fault of the Londoners: not sounding h where it ought to be sounded, and inversely.”

As has been reported elsewhere (Zettersten Reference Zettersten1969; Schreier Reference Schreier2003; Schreier and Trudgill Reference Schreier2006), TdCE is quite remarkable in that it is almost certainly the only variety in the world that still has frequent /h/ insertion. It is found elsewhere, e.g. in the Caribbean (“on the island of Abaco in the Bahamas ear is what you do with your hear (or vice versa),” Holm Reference Holm1988: 76, quoted in Childs, Reaser, and Wolfram Reference Childs, Reaser, Wolfram, Aceto and Williams2003: 15), and the Appalachians, but there it is restricted to very few speakers, often elderly members of the community, and a small set of lexical items (such as the letter <h>). On Tristan da Cunha, this is much more widespread as Allen Crawford noted in the 1930s, “the tendency to add an ‘h’ before vowels makes all islanders ‘highlanders’!” (reported in Crawford Reference Crawford1982: 49), and a previous study (Schreier Reference Schreier2006) provided quantitative evidence that one speaker had a total of approximately 30% of /h/ insertion in words such as oil, I, egg, etc. The same source (a case study of individual variation) showed that /h/ was inserted with prepositions (out, on, after, up, over), pronouns (it, I), verbs (ask, offer, open), adjectives (other, every, old), adverbs (outside), nouns (army, area, officer), and proper names (Ernie), making this not only a frequent but also a widespread and productive process that did not seem to be restricted to certain word types, which was counterevidence to historical reports (where examples of /h/ insertion were mostly nouns). /h/ was thus inserted in function and content words, with an overall insertion rate of 64.7% for lexical words and 24.7% for function words.

This was used as sociohistorical evidence as to who was responsible for bringing this feature to Tristan da Cunha. Since /h/ was reported extensively in London English (Walker Reference Walker1791; Batchelor Reference Batchelor1809) and Cockney in particular (Jespersen Reference Jespersen1909), i.e. from precisely the area of origin of one of the most influential founders of the community (Richard Riley), Schreier (Reference Schreier, Auer and Schmidt2010) speculated that it was a legacy of early nineteenth-century London English.

Given what was said above about the different methods of data collection in the various corpora, the question addressed here is whether speakers vary in their usage of local variables when recorded in diverse settings. Of course, this in no way challenges Labov, who claimed that

Indeed, it would be unexpected and counter-intuitive to find no variation, so the question is rather when, how often, and to what extent the Tristanians vary between zero (0) and inserted /h/. Notwithstanding, one should bear in mind that the speakers, born between 1895 and 1910, were socialized in a hyper-isolated environment. These speakers grew up in a TdCE-speaking community with practically no exposure to other dialects (the exception being the missionaries sent out by the Society for the Propagation of the Gospel). Peter Munch reported that the island community was encapsulated and remarkably endocentric. The question then is to what extent selected speakers vary in their usage of a hyper-isolated local variant in the two interview modes, which allows us to address issues such as the effects of isolation on “shifting potential,” the potential for shifting of individuals in dense and highly complex social network systems, etc.

A total of four speakers met the criteria and were selected:

1. 1. MS (female, 1895–1974), recorded by: BBC 1961 (Southampton), Peter Munch 1964 (TdC), Roland Svensson 1970 (TdC).

2. 2. WR (male, 1889–1963), recorded by: BBC 1961 (Southampton), Roland Svensson 1963 (en route to TdC on a sailing boat).

3. 3. BGG (male, 1899–1995), recorded by: BBC 1961 (Southampton), Roland Svensson 1963 (TdC), Peter Munch 1964 (TdC).

4. 4. SG (male, 1910–1994), recorded by: BBC 1961 (Southampton), Roland Svensson 1970 (TdC).

In order to minimize the potential of type-token relations, only five tokens were collected for each lexical item. Moreover, as is standard in sociolinguistic interviews, the first minutes of interviews were not considered (three minutes for shorter interviews from the Svensson corpus (lasting between fourteen and twenty minutes)) with the exception of interviews shorter than ten minutes. It should also be pointed out that one islander, BGG, was born in the Cape and came to Tristan with his family in 1908, when he was aged eight. Another Tristanian, WR, was interviewed on board the sailing ship back to Tristan, shortly before arriving on the island (where he passed away three months later). If possible, a total of 100 tokens was extracted for each individual in each interview setting; based on the principle of accountability (“all occurrences of a given variant are noted, and where it has been possible to define the variable as a closed set of variants, all non-occurrences of the variant in the relative environments”; Labov Reference Labov1982: 30), all instances of lexical items with initial vowels and alpha stress were selected and coded, no matter whether they had /h/ insertion or not. In some of the shorter recordings, the total could not be reached (here the minimum was forty-eight).

A first look at /h/ variability indicates that the speakers vary, at times extensively, between zero (0) and /h/ insertion, as the following three examples show (all from the Svensson corpus), where /h/ is found before prepositions, nouns, and verbs:

1. 1. We give 0 each one so many. And then they (h)ordered potatoes, see, into the 0 island, and then they give 0 every cellar like a (h)onion size bag. (LR_1970)

2. 2. Well 0 I got six drivers and sometime 0 I am (h)on my (h)own. (LG_1970)

3. 3. It's the best to place a postal (h)order when you go down to the post 0 office. (MS_1964)

With regard to variation in different recording contexts, Table 20.2 and Figure 20.2 provide the results for the four speakers who were interviewed twice (in Southampton (BBC) and the South Atlantic (TdC)).

Figure 20.2 /h/ insertion in four TdCE speakers by place of interview (Southampton vs. TdC).

The results show that all four speakers tend to have higher /h/ insertion rates when interviewed on Tristan da Cunha (or en route to the island), thus on or close to their home ground. While varying between /h/ insertion or 0 in both settings, all speakers have higher /h/ rates when interviewed by Peter Munch or Roland Svensson. This is even more remarkable since the questions and topics raised were not much different, mostly focusing on life on Tristan and the South Atlantic, reminiscences of the time before the volcanic eruption, the evacuation and also the experiences in England, which was of particular interest since the islanders decided to return home in a democratic vote held in 1962 (see Schreier and Lavarello-Schreier Reference Schreier and Lavarello-Schreier2011). The total difference is remarkable: in three speakers (MS, BGG, and WR), it amounts to an increase of about 12–15%. The one exception is SG, where the increase, while still manifesting itself, is slight (1.6%). Here it is striking that /h/ insertion in the Southampton recording is very high to start with (37.3%), so this may well be one of the strongest “/h/ inserters” in the entire corpus. The difference is statistically significant in all four cases (p < 0.05).

20.4 Conclusion

This chapter has given a short introduction to the sociolinguistic value of different corpora on late nineteenth- and twentieth-century TdCE, all of which will be subject to further research. A major dialect study based on several corpora carries immense potential (real vs. apparent time, style shifting, individual variation), though doubts remain as to what extent data obtained under different conditions can be compared strictu sensu. Criteria related to validity, formality, and authenticity have to be assessed and taken into consideration at all times. The potential of corpus comparison in historical linguistics has been dealt with in recent handbooks (in Krug and Schlüter Reference Krug and Schlüter2013), particularly the usefulness of comparative research (qualitative vs. quantitative approaches, which depends on the amount of data available via the length of recordings, as for instance in group vs. individual interviews). The data can be used for apparent-time studies (cf. Schreier Reference Schreier2003), for descriptive purposes or a reconstruction of earlier forms (Schreier and Trudgill Reference Schreier2006), or for research on historical phonology (cf. Figure 20.1 above).

The stance taken here is that the existing corpora are most suitable for an analysis of effects of setting and interviewer. This confirms Labov's maxim “there are no single-style speakers,” yet it is noteworthy that all the speakers studied show sensitivity to social context. All four speakers are more /h/-ful (inserted before words such as island, on, I, Albert, etc.) when they are talking to someone they are familiar with (a Norwegian sociologist, a Swedish painter, informed outsiders both) and in a place where they feel comfortable (in their own homes or on a ship in the South Atlantic Ocean), despite the fact that the topics raised are very similar. This shows that elderly speakers of hyper-isolated dialects such as TdCE are certainly aware of social constraints on variability, perhaps even more so than one would have expected. Though three of them had never left the island or spent time outside the local community (the fourth, born in the Cape to a Tristanian father and Irish/South African mother, moving to Tristan aged eight), they all have the same context-related effects. At the risk of slight exaggeration, this is some indication that speakers in hyper-isolated communities do show remarkable awareness as to when and with whom to use local features and when not. Hyper-isolated speakers style-shift with interviewer and context, and local features are used more frequently in informal settings. Some problems for analysis remain to be solved, such as assessing the impact of individual parameters (topic, interlocutor, etc.), the individual orientation of speakers toward island life and the outside world (which has been shown to be a crucial factor in variation, or of course “proactive” variation – enregisterment, identity formation, etc.). In this sense, the findings presented here can be no more than a first beginning.

At the same time, the potential for future research is immense. The next steps would be to look into variation within speakers recorded on Tristan under similar contexts (by Svensson in the 1970s and Schreier in 1999/Reference Schreier, Auer and Schmidt2010), which would allow for change in real-time, to look (perhaps in a more qualitative approach) at the overall presence of features in speakers born between 1876 and 1992, which would allow one to investigate the overall usage (or perhaps the disappearance) of features or to reconstruct earlier forms (perhaps even the inputs). More research on different variables is definitely needed, and the corpora, if handled with care, may certainly hold the key both to change in twentieth-century TdCE and the origins of the dialect in the nineteenth century.

Acknowledgments

I wish to thank Erik Thomas, North Carolina State University, for his help with producing the vowel plot, Nicole Studer-Joho, for her help with the history of /h/ in English, and Danae Perez-Inofuentes and Nicole Eberle, University of Zurich, for most valuable comments on an earlier draft version of this chapter. I also wish to acknowledge the input of the editor of the volume, Raymond Hickey, whose input was most helpful, as always.