Researching & Analysing Olive The Epistolarium in Social ... · each letter, postcard and so on – this is the basic information for working ... DailyNews/1 B. Meta-data

Researching & Analysing Olive Schreiner’s Letters: The Epistolarium in Social Science Perspective Universities of Edinburgh, Leeds Metropolitan, Sheffield

So what does ‘researching and analysing Olive Schreiner’s letters in social science perspective’ involve? Some detail about what this entails follows.

• An ESRC funded project • The biggest letters project since Thomas & Znaniecki • Multi-disciplinary within a sociological frame

o Sociology o Cultural history o Literary studies o Social geography

• Computer/software technologies

• A combination of focuses o The research project o Publication of ‘the Olive Schreiner Letters’

o Knowledge transfer – international users with

world-wide free access via the user interface o Knowledge exchange – VRE-use workshops

Schreiner’s are fascinating letters which open up for reading and analysis some key things:

• Aspects of the social history of Britain & South Africa as seen by a keen social commentator

• Epistolary insights into a period of

great change & momentous events

• The emergent features of Schreiner’s social theory, as shown by her letters

• The ‘great and the good’ & the now

obscure in a network of epistolary exchanges

• Development of theoretical concept of

the epistolarium using this very large letter collection

British Library

Harry Ransom Center University of Texas at Austin

Manuscripts & Archives University of Cape Town

There are c4800 Schreiner letters, in 16 archives on 3 continents

William Cullen Library University of Witwatersrand

National Library of South Africa

What they look like... like this, but thousands more of them, and many of them much more difficult to read than these

First we transcribe the letters, postcards, telegrams, notes and so on – and this means transcribing every insertion, deletion, omission, mistake. Then we check this once, twice, three times. Then, some weeks or months later, another researcher checks the transcriptions against the originals and edits any mistakes made by the person who first transcribed them. What results can be seen on the next slide – mark-up is turned on in Word to show the number of things spotted by the checker: being very very accurate regarding the ‘bird in flight’ aspects of how letters are written is surprisingly difficult and needs pains-taking attention to the detail

But this is the start, and not the end, of what we do in preparing our transcriptions for research and analysis – and also publication – purposes. The next stage is to ensure there is fully accurate consistent ‘meta-data’ for each letter, postcard and so on – this is the basic information for working with the letters ‘as data’ using computer-assisted methods of ‘managing’ them all as a data-set, and also for analysing them. An example is shown below.

A. CLASSIFYING INFORMATION Letter date [20 April 1899] Address from [2 Primrose Terrace, Berea, Johannesburg] Address to [Lyndall, Newlands, Cape Town] Who to [William Philip (‘Will’, ‘WP’) Schreiner] B. EDITED COLLECTION Editor [Rive 1987: 348-9] C. ARCHIVE COLLECTION REFERENCE Archive name [University of Cape Town, Manuscripts & Archives, Cape Town] Archive ref [Olive Schreiner BC16/Box2/Fold1/Jan-June1899/16]

We also then write Notations. These are a bit like Footnotes, but with a different philosophy concerning editorial intervention – they are mainly concerned with explaining uncertainties and oddities in the meta-data, commenting on differences between our transcription and any other version that has been published, and providing information about references to Schreiner’s unpublished or published writings in the letter, as follows:

[NOTATION: The date is derived from the postmark on an attached envelope, which also provides the address this letter was sent to..] [NOTATION: A version of this letter appears in Rive (1987). Rive’s version omits part of the letter and is also in a number of respects incorrect. A short extract also appears in Cronwright-Schreiner (1924) which is incorrect in multiple ways.]

[NOTATION: The ‘Bushman paper’ refers to one of Schreiner’s ‘Returned South African’ essays, now in Thoughts on South Africa; ‘The Boer Woman’ also appears in the same posthumous collection.]

But it doesn’t stop there! In order to be able to analyse the letters within a project specific Virtual Research Environment or VRE, they have to be prepared in what is called an XML format – this is a way of coding, for example, underlining, deletions, paragraphing & so on. And because of the way the kind of XML we use has been developed, it is ‘future-proofed’, enabling letters to be shown in successive versions of HTML (HyperText Markup Language) and so seen ‘as letters’ are meant to look via a web-browser.

XML = a set of rules for encoding text-based documents electronically and a textual data format frequently used as a basic language that can be converted into HTML HTML is the most common markup language used for writing and preparing webpages ‘jEdit’ is used as an XML processor (that is, we write our xml using jEdit) and it involves a number of complexities, including ‘tagging’, for example to indicate whether something is a letter, a postcard, a telegram and so on. An example follows on the next two slides:

A. Meta-data in XML:

<letter> <!-- defaults to type 'letter' -->

<date>28 December 1885</date>

<addressfrom>London</addressfrom>

<!--<addressto></addressto>-->

<to>Daily News</to>

<editor>Rive 1987: 70-1</editor>

<archive>Harry Ransom Center,

University of Texas, Austin</archive>

<ref1>HRC/OliveSchreinerLetters/OS-

DailyNews/1</ref1>

B. Meta-data in HTML:

O. TRANSCRIPTION INFORMATION

Letter type [letter]

A. CLASSIFYING INFORMATION

Letter date [28 December 1885]

Address From [London]

Who to [Daily News]

B. EDITED COLLECTION

Editor [Rive 1987: 70-1]

C. ARCHIVE COLLECTION REFERENCE

Archive name [Harry Ransom Center, University of Texas, Austin]

Archive Ref 1 [HRC/ OliveSchreinerLetters/OS-DailyNews/1]

From XML to HTML: An example Schreiner letter

A. Letter Content in XML:

<text>

<p>Sir,</p>

<p></p>

<p>...in London no Englishwoman was safe from the

hands of the police.</p>

<p></p>

<p>I regarded this statement with the cool scorn with

we are apt to <add>regard</add> <del><ur/> those

<add>statements</add> which we regard</del>

<add>consider</add> <add>as</add> <add>who

make</add> uncritical <del>,</del>

<add>statements</add> <del> &amp; nothing we <ur/>

?away</del> A few miserable &amp; for-lorn women

with out money or friends might suffer; but the mass of

English women armed with friends &amp; intellectual

power were safe from insult</ p>

<p></p>

<p>...... a wellknown ^wordspace

ph^wordspace ... offered to

<del>accompany</del> <add>conduct</add> me home.

...</p>

<p></p>

<p>London Dec: 28th 1885.</p>

<p></p>

<p><not>This is one of very few draft letters in the

Schreiner collections. ......</not></p>

</text>

B. Letter Content in HTML:

Sir,

A short time back the remark was made in my pres-

ence, that in London no Englishwoman was safe from

the hands of the police.

I regarded this statement with the cool scorn with we are

apt to ^regard^ Unreadable text (1 word) those

^statements^ which we regard ^consider^^as^^who

make^ uncritical , ^statements^& nothing we

Unreadable text (1 word) ?away A few miserable & for-

lorn women with out money or friends might suffer; but

the mass of English women armed with friends &

intellectual power were safe from insult. ...

...a well known medical man at the West End; on

leaving...a wellknown [wordspace] ph

[wordspace]...offered to accompany ^conduct^ me

home. ...

London Dec: 28th 1885.

[NOT: This is one of very few draft letters in the

Schreiner collections. ...]

XML and its Issues •‘Tags’ have to be carefully defined; putting them in using jEdit is time-consuming – but they are project-defined & developed & eventually will aid the analytic & publishing processes; • XML tagged versions of the letters are very hard to work on analytically – but XML is easily converted into HTML, which looks like MS Word; • XML files are small compared with MS Word & resist corruption; • jEdit is unforgiving to use – but it does enforce the ‘wellformedness’ of letters by showing up errors, something which it is very important we as editors and researchers should know about as soon as possible.

So what are all these technicalities for, then? Working with c4800 letters, ranging from 300 or 400 words to 40 densely written pages, is not easy in itself. However, analysing this number of letters, and letters of such size and magnitude, is even less easy. Large-scale datasets (c4800 ‘cases’ is large-scale in anyone’s terms) are often analysed using a ‘social science standard’ commercial package called SPSS – but this is designed to analyse very simple information (yes, no; age 30-39; single, partnered) for a lot of cases, not immensely and intensely detailed ones like the Schreiner letters. The analysis of qualitative text-based datasets often makes use of CAQDAS (computer assisted qualitative data analysis software) packages – but c4800 cases, many of them a huge size, and containing different ‘genres’ of data, is beyond their capacity. The Schreiner letters are too many and too large. At basis, we transcribe and xml the Schreiner letters as we do to enable us to devise our own project specific analytical tools, tools suitable for analysing a very large, but also qualitative, dataset. This process is driven by our analytical needs, for software which can both help in ‘project management’ aspects of this large qualitative dataset and also the analytical interests we have. This involves our Virtual Research Environment or VRE.

THE OSLP VIRTUAL RESEARCH ENVIRONMENT (VRE) A. Project Management

• Data packaging – helps us work with the letters as ‘a set’ from contents which are both very large in number, and also internally diverse

• Data organisation – constrains us to devise basic organisational principles (for the OSLP = around the collection, the archive, the date)

• Data identification & retrieval – search & find on both meta-data headings & full-text indexing can be provided

B. Computer-Assisted Analysis Tools

• Enables us to avoid variable analysis & hierarchical linked indexing of contents, constrained if not determined by most CAQDAS ‘off the peg’ software packages

• We can write in project-specific analytical tools, to work the letters in exactly the ways we want.

PUBLICATION & THE USER INTERFACE This slightly strange phrase, ‘the user interface’, refers to one of the products of our analysis of the Schreiner letter transcriptions. Generating HTML versions of our transcriptions is crucial to aid & support our analytic work – and it also enables electronic publication in a form in which readers can access & read the letters. The letter transcriptions will be published in early 2012 & made available in electronic format to readers world-wide. But alongside being able to ‘just read’ the letters, readers will face the same kind of issues we do in working with such a very large number of letters – where do you start, how can it be made manageable, how can you find out the things you’ll be particularly interested in, what other important aspects are there that you’ve never thought about? The role of the user interface is to support readers in asking – and answering – these questions and related ones that occur to them. The user interface will provide a high level of sophisticated ‘search & find’ tools, enabling readers to navigate their own routes through the letters.

WATCH THIS SPACE! From time to time, we will rewrite the Schreiner Letters website and re-launch it with new pages and information, to keep visitors up-to-date with progress. Any comments? Any information you’d like about the project that isn’t yet on the website? Please contact us! Thank you. Please press ‘back’ on browser to return to home page