1 The aim of this paper is to discuss the results of a corpus-based investigation of the process that has transformed the very specific material meaning of the word “culture” into the extremely elusive, liquid (Bauman 2011) concept we are familiar with today. The analysis starts from the lexicogrammar profile of the world “culture” in contemporary syncronic corpus resources (Gatto 2011; 2014) and attempts further exploration of these findings on the basis of diachronic language resources. In particular, data from Google Books, accessed both via Ngram Viewer and through the tools available at BYU Corpora, have been used to test hypotheses for the behaviour of the world “culture” in the 19th and early 20th century, whereas data from EEBO (Early English Books Online) have been used to explore patterns of usage in the period of time from 15th to 18th century. The partial results of this research suggest that there is room for far reaching investigations into the (hi)story of this intriguing “complicated" word, as Williams (1985: 87) dubbed it, and that computational methods and language resources can well complement studies carried out in the context of the digital humanities, from the perspective of historical linguistics, sociolinguistics and cultural studies, when not providing the basis for fresh new insights and further explorations. Sketching the “historical peregrinations” of the concept of culture in his Culture in a Liquid Modern World, Bauman outlines the changing role of culture in society, from “an agent for
change”, to “a conservative force”, to an increasingly flexible and liquid concept “fashioned to fit individual freedom” (Bauman 2011: 1-17). It is against this background that this paper attempts an investigation of the multifaceted process that over the centuries has transformed the very specific material meaning of the word “culture” into the extremely elusive concept we are familiar with today. The basic assumption is that the process of semantic change which transformed a word originally referring to the concepts of tillage and husbandry (from the Latin colere) into a potentially polysemic word accommodating a far wider range of meanings is mirrored in changes in usage of the word, and in turn reflects changes in society. In the wake of a growing interest for the use of language resources for the investigation of cultural and social phenomena (e.g. Michel et al 2010) these changes can be observed through the quantitative and qualitative analysis of the lexicogrammar patterns the word “culture” has entered during its long history of existence. The very choice for the word “culture” originates in Raymond Williams’ famous statement that culture is “one of the two or three most complicated words in the English language”. By identifying “culture” as one of the key words of our times, Williams reminds us (1985: 87–93) that culture used to be, in its early uses, the noun of a process: the tending of something, basically crops or animals. This meaning provided a basis for the important next stage of metaphorization, when the tending of natural growth was extended to a process of human development so that the word “culture” came to be taken in absolute terms as signifying a process of refinement. After tracing the key moments in the development of this word, Williams distinguishes three categories in modern usage: (i) the noun which describes a process of intellectual, spiritual and aesthetic refinement; e.g. a man of culture; (ii) the noun which describes the products of intellectual and especially artistic activity; e.g. Ministry of Culture; (iii) the noun which indicates a particular way of life, whether of a people, a period, a group, or humanity in general; e.g. Jewish culture.
This paper takes Williams as a starting point to provide empirical evidence of the ways the word “culture” is used in the English language. Indeed Williams himself, in his introduction to Culture and Society, states that an enquiry into the development of this word should be carried out by examining “not a series of abstracted problems, but a series of statements by individuals” (1966: xvii), which bears striking similarities – though not intended - with the corpus-based approach adopted in the present research. 2
The starting point of the investigation carried out in the paper are the preliminary results of investigation into the lexicogrammar profile of the word “culture” using Sketch Engine, a corpus query tool that provides a one-page summary of the lexico-grammar patterns of a word from a given corpus, as reported in Gatto (2011; 2014). In the first part of this paper data from three synchronic corpora of English (BNC UkWaC and EnTenTen) will be compared. For a start, the table below reports the number of occurrences and the normalized frequency of the world CULTURE in each corpus: BNC 10,281 90,1 per million
UKWAC 200,663 129,70 per million
ENTENTEN 3,692,159 200,80 per million It should be noted, that these occurrences obviously include both those in which “culture” is used with its general meaning in the humanities, which is the primary concern of the present analysis, and those in which it is used as a scientific term (e.g. cell culture). Since the tools and resources used for the present research do not allow for a disambiguation between the two meanings, an attempt has been made – heuristically – to estimate the number of occurrences of “culture” in its scientific sense, by computing the number of occurrences of “culture” with the
2.1
Culture as object lemmas “cell” or “bacteria” in their co–text. This was done using the filter option and setting a broad co–text (15 words to the left and to the right of the node). The results seem to indicate that nearly 9,472 (2,46 per million) occurrences of “culture” can be related to its scientific meaning in the BNC, 2,499 (1,61 per million) in UkWaC, and 98,714 (3,47 per million) in EnTenTen. While the method used was not to be considered totally reliable, on the basis of the relative negligibility of the results this aspect has not been taken into account in the following commentary of the data (but this is certainly an aspect which needs to be handeld with care when pursuing further research on this topic).
Given the limited scope of the paper only three lexico-grammar patterns will be focused on in the subsections below.
When considering the list of verbs having “culture” as their object, it seems that according to data computed by the Sketch Engine for all the three corpora, the word “culture” has a consistent tendency to occur as the object of such verbs as foster, promote, create, reflect, understand, shape change. Respect does not appear only in the list for EnTenTen, as it is n.16, just out of the first 15 positions chosen as a sample. While these data are encouraging in showing that corpora built in different ways yield consistent results for the collocates of “culture”, something interesting can be observed with reference to the changing position of some collocates. The collocates foster, promote, create, and change seem to gain prominence in the two more recent web corpora, but it is also interesting to consider how experience, which did not appear in the top 15 list from the BNC, is one of the most significant collocates for “culture” in the other two corpora. By contrast, such patterns as assimilate, absorb transmit, which indirectly refer to power relations, appear to be unique to the BNC.
Previous research has already discussed how concordance lines for such pattern as create+ culture or foster + culture have a frequent co– occurrence with words relating to the socio– economic domain, such as staff, enterprise, job, work; this, in turn, reveals that CULTURE, in this context, has partially lost its original meaning of a process/product of refinement, as in the famous Arnoldian sense of “a pursuit of our total perfection, …the best which has been thought and said in the world” (Arnold, 1869, p. viii) and rather concerns a set of ideas/behaviours relating to a specific group in a specific context, like a workplace, company or organization – a new restricted meaning of “culture” (Gatto 2011; 2014).
This view is also supported by the prominence of the pattern culture + within computed by the Sketch Engine (542 occurrences, 4.2 per million) as an interesting pattern especially in UkWaC, which can well be interpreted in terms of the concept of “small culture” (Holliday 1999) As to the more recent collocation of CULTURE with such verbs as experience and explore, we notice the emergence of the word cultures in the plural in their immediate co-text. These concordance lines clearly reflect the anthropological/ethnographic meaning of culture inaugurated between late nineteenth and early twentieth century by Franz Boas and other scholars. This pattern has a quite consistent collocation with adjectives like new, different, origin, other. It must be acknowledged, however, that – as far as UkWaC in particular is concerned, these concordance lines often originate in specialized sites dealing with the typically British experience of the gap year, a datum which relates to the choices made by the developers of ukWaC, who included academic sites (i.e. ac.uk sites) extensively in the crawl. Nonetheless, these occurrences testify to a radical shift in the meaning of culture, whereby culture is something to be experienced, rather than to be found in books (as Arnold would have argued), again something quite distant from its more traditional meaning. 2.2
As for the list of nouns modified by culture, this is opened, in the three corpora, by shock, a collocation which relates to a distinctively modern experience defined as culture shock, “the feelings of isolation, rejection, etc., experienced when one culture is brought into sudden contact with another, as when a primitive tribe is confronted by modern civilization” (Collins Cobuild Dictionary). Significantly, one of the most prominent collocates for culture shock in ukWaC is reverse, which originates in the phrase reverse culture shock, a form not yet attested in the BNC, probably because the experience itself had not yet been fully conceptualized. In this way the recent web corpora ukWaC and EnTenTen do not only provide evidence of a relatively new linguistic formation, but in doing so they point to the emergence of a new social and psychological condition, resulting from a change in society itself. The very existence of a reverse culture shock is related to novel ways of experiencing mobility and migration, which entail continuous dislocations and relocations. 2.3
Finally a particularly significant pattern emerging from the word sketch for “culture” is the pattern culture + of, which seems to turn the word “culture” into an extraordinarily capacious and inclusive category that can be used for anything. And while some collocates might seem to confirm Stubbs’ intuition that the pattern has a relatively negative semantic prosody (1996, p. X), owing to collocations with words bearing negative connotations, there is ample evidence that the pattern can equally accommodate positive notions, like secrecy/openness, blame/impunity, entrepreneurship/dependency, etc. as the list of collocates reported below suggests: Furthermore, the concordance lines for the pattern a culture of include many phrases in inverted commas, which seem to create a culture virtually ex–nihilo, as Barker (2003, xix) would argue, reducing culture to little more than an attitude, as in the examples reported below: Here culture seems to have become a sort of neutral term that can keep the company of many different words: a culture of corruption, and a culture of accountability, a culture of violence, a culture of peace, and even a culture of ‘buy now pay later’. As suggested by the various collocates for the pattern a culture of shown above, this lexico–grammar pattern really has the power of turning culture into a sort of vox media, a liquid modern word that can be used for anything. 3
Starting from these preliminary observations on the lexico-grammar behaviour of the word “culture” in corpora representative of contemporary usage, a new research question emerged. To what extent can these patterns be considered as characteristic of contemporary usage? And if they are, how and when did they emerge? Is there any other information that could be gathered from the investigation of diachronic resources?
A first attempt at answering these questions was to query the Google Books database, through an extremely limited and yet fascinating tool like Ngram Viewer, a tool which allows to read line charts representing n-grams i.e. continuous sequences from text, from the digitized books in the Google Book collections, in terms of frequency per year. The data can be accessed through a free web-based interface that enables relatively complex queries which support wildcards, POS-tagged search, case-sensitive queries, etc. For this reason, this is a tool commonly used in what has come to be known as “culturomics”, a research trend which aims “to observe cultural trends and subject them to quantitative investigation” on the basis of data obtained from Google books (Michel et al 2016). This approach is definitely controversial, especially from the perspective of corpus linguistics which is engaged in more theoretically sound and qualitatively reliable empirical research, and a very convincing overview of its limitations is found in McEnery and Baker (2016: 11-17). Nonetheless, in this specific case, information retrieved form such imperfect resources and limited tools could be still be used as indications, as fingers pointed to some interesting phenomena that might be worth being investigated in more detail with more appropriate resources. For instance, faced with the prominence and significance of the pattern “a culture of” as typical of the lexico-grammar profile of the word as described on the basis of data on contemporary usage, Ngram viewrs was used to try and see whether the pattern had always been there, or had it somehow emerged at a certain point in time.
Indeed a search in Google Books using Ngram Viewer apparently suggests that the pattern emerged in the late 19th century, with most occurrences in the biological field, when it was referred to the recent discovery of bacteria. Anyway around the 1990s there was a dramatic surge in usage for this pattern, possibly connected with the growing tendency to use culture as a vox media devoid of any specific meaning as in the examples reported in the previous section. Similarly, the emergence of the phrase “culture shock”, which seems to be prominent in contemporary corpora, can be located, with the help of NGram Viewer, in a specific moment in history, in the late 50s: Whereas “reverse culture shock” seems to have appeared in the 1960s: Also interesting is the possibility to have a big picture in terms of changing behaviour of different lexico-grammar patterns. This is the case of the diverging fortunes of the two patterns experience + culture and understand + culture which seem to provide evidence of the fact that culture is more and more seen as something to be experienced than to be learnt or, as the verb suggests, cognitively appreciated and understood: Besides using Ngram Viewer, the present research has also tried to profit from the interface for Google Books made available through Mark Davies well-known Corpora website to search the One Million Books and Fiction datasets. Based on the same Google Books data the interface was created by Mark Davies, Professor of Linguistics at Brigham Young University, and it is related to other large corpora made available through the same service. The system allows more refined queries than Google Books interface, and supports the comparison of the data in two different sections of the corpus.
The interface available at Corpora BYU confirms at a glance that the collocation “culture shock” appeared between the 1950s and the 1970s and has dramatically grown in frequency since the 1980s. The same tools provide evidence of the emergence and decline of the collocation between culture and refinement around the publication of Arnold’s seminal Culture and Anarchy in the late 19th century.
A further attempt at casting a backward look to envisage the changing face of the world “culture” over time has been finally made by using data from the EEBO corpus available through Sketch Engine.
While limited in scoped, these data provide clear evidence of the fact that the noun “culture” was not a particularly frequent in Written Early Modern English, as the EEBO corpus has only 2283 occurrences for this word (2.31 per million). Indeed, when CULTURE was used only in its original ‘agricultural’ meaning it was probably something which was not to be written about. Anyway data from EEBO makes us see firsthand the origins of its subsequent metaphorical meanings. Especially in the dataset for the period 16001699 the analysis of collocates for “culture” shows the emerging coexistence of the literal agricultural meaning and of a spiritual metaphorical meaning. It is at this stage that cultivation emerges as a meaningful collocate for “culture”, often in the such sentences as “cultivation of the minde”. However we have to wait until 17001799 for civilization to appear among the most salient collocates for “culture”. Which brings us back to the beginning of this story… 4
Using different resources to map such a complex research field, so as to obtain a general picture of significant patterns of usage in the evolution of language is certainly fascinating, but this is not enough. And it goes without saying that it is necessary to be extremely cautious before drawing conclusions, if any, from investigations like these. Anyway, the data analyzed confirm that that there are resources and tools that can support the investigation of huge amount of data, pointing to interesting research areas to be analyzed with more refined ad hoc tools. In any case a rewarding exploration of these data from a cultural perspective can perhaps only come as the result of teamwork in the context of a multidisciplinary approach in the growing research field of the Digital Humanities.
C. Barker (2003). Cultural Studies: Theory and practice. London: SAGE.
Z. Bauman. (2011). Culture in a Liquid Modern World. Cambridge: Polity Press.
M. Gatto (2014). Web as Corpus. Theory and Practice. London: Bloomsbury.
M. Gatto (2011). “Sketches of CULTURE from the web. A preliminary study”. In G. Dimartino, L. Lombardo, S. Nuccorini (eds.). Challanges for the 21st Century. Dilemmas, Ambiguities, Directions. vol. II, Roma: Edizioni Q, pp. 275-283.
A. Holliday (1999). “Small cultures”. Applied Linguistics 20/2, pp.237-264.
A. McEnery, H. Baker (2016). Corpus Linguistics and 17th-Century Prostitution: Computational Linguistics and History. London: Bloomsbury.
J.B. Michel et al. (2011). “Quantitative analysis of culture using millions of digitized books”. Science, 331(6014):176–182.
R. Williams (1985). Keywords. A Vocabulary of Culture and Society. Oxford: Oxford University Press. Resources and tools: Early English Books Online, https://eebo.chadwyck.com/home Google Books Ngram Viewer, https://books.google.com/ngrams Google Books (British English), http://googlebooks.byu.edu/x.asp Sketch Engine, https://www.sketchengine.co.uk/ [reference stub]