History and methods of data collection

Collecting and preparing private correspondence at the Department of Czech Language 

The collection of correspondence began in the late 1990s at the Department of Czech Language, Faculty of Arts, Masaryk University, under the supervision of Z. Hladká. It initially focused on texts from around the turn of the millennium, when traditional handwritten letters were gradually giving way to email and SMS. The work resulted in the publication by Zdeňka Hladká et al., Čeština v současné soukromé korespondenci. Dopisy, e-maily, SMS [Czech in contemporary private correspondence: letters, e-mails, SMS.] (Masarykova univerzita, Brno, 2005), supported by Czech Science Foundation grant no. 405/03/0248. Two accompanying CDs contained approximately 5,000 texts, mostly by young writers, presented as images of the originals and transcriptions of all the documents. The letters and emails were also made available as corpora. Each text was annotated with sociodemographic and dialect information, the latter an innovation at the time. These annotations could be used to search both the database and the corpora. In 2006, part of the material, comprising 2,000 letters, was incorporated into the Czech National Corpus under the name KSK-dopisy. 

Collection subsequently focused exclusively on handwritten letters, while its chronological scope expanded to cover the entire twentieth century. Eight years after the first publication, another two CDs were released, presenting 2,000 letters from 1902–2012 using the same methodology: Zdeňka Hladká et al., 111 let českého dopisu v korpusovém zpracování [111 years of Czech letters from the viewpoint of corpus processing] (Masarykova univerzita, Brno, 2013), supported by Czech Science Foundation grant no. 406/10/0823. P. Rychlý of the Faculty of Informatics, Masaryk University, was a principal contributor to the corpus processing for both publications. Computer Aided Technologies, s. r. o., handled the technical implementation and graphic design. 

Both publications were produced in small print runs and distributed free of charge. They are now unavailable, and the CDs are also becoming unusable. A decision was therefore made to thoroughly clean and revise the material, add approximately 800 previously unpublished letters, and make it available through Digitalia MUNI ARTS. This will support its long-term preservation and access for a wider community of researchers and general readers. 

Collection and transcription methods 

The collection focused on previously unpublished private letters by ordinary writers: family members, partners and friends. To preserve the diversity of idiolects, only one letter per writer was included. Writers recur only in the SMS collection, which comprises complete sets of messages received over one week by several recipients, as well as a full year of messages received by a single recipient. 

Following consultation with a legal adviser, the legal and ethical issues surrounding publication were addressed by obtaining the letters directly from their recipients or, for older correspondence, from their heirs. This ensured the consent of at least one participant in the communication. Given the number of texts and the age of some of the material, it was not possible to obtain the writers’ consent. Donors gave written consent to the publication of the correspondence and its use for non-commercial purposes. They also removed all identifying details from the texts they supplied. Where this had not been done consistently, the remaining information was removed during transcription and archiving. Names of public figures could be retained. 

For each text, donors also completed an accompanying record card with standardised information about the writer and recipient. This information was used for sociodemographic and dialect annotation.

Transcribing handwritten letters 

The digital transcriptions of handwritten letters strictly preserved the original text, including spelling mistakes, typographical errors and other irregularities. Only identifying details, such as surnames, addresses and telephone numbers, were replaced with a standard @ symbol and a general description in curly brackets, for example {následuje příjmení} (“surname follows”). 

Texts were transcribed as continuous text, with the start of each paragraph marked by #. Separate lines in the opening and closing sections of letters were also treated as paragraphs. The start of a new page was marked as a paragraph only where this reflected the structure of the content. Illegible text was replaced with three hyphens, preceded by a note in curly brackets stating the number of illegible words, for example {následují 3 nečitelná slova} --- (“3 illegible words follow”). Curly brackets were also used for other transcriber’s comments, such as notes about inserted images or the style of lettering, which was not otherwise distinguished in the transcription. For example, {následující tři slova psána kurzivou} means “the following three words are written in italics”. 

Quoted passages of one sentence or more were also placed in curly brackets, followed by three hyphens indicating that the passage had been excluded from the main text, for example {citace: Dalekáť cesta má! Marné volání!!} ---. The label citace means “quotation”. Foreign-language passages and passages added by another writer were treated in the same way. These conventions were intended to prevent distortion of the writer’s idiolect in corpus processing: text in curly brackets can be located in the corpora but is excluded from frequency analysis. 

For corpus searches, curly brackets were also used to handle incorrect word boundaries, a frequent feature of older correspondence, although other errors were faithfully preserved. Incorrect word boundaries are harder for corpus users to anticipate than ordinary errors. They were therefore corrected in accordance with current standard spelling, with the original form recorded only in curly brackets. For example, {špatná delimitace následujícího slova: živo bytí} živobytí records the incorrectly separated form živo bytí in the note and supplies the standard form živobytí in the main text. The note means “incorrect word boundary in the following word: živo bytí”. 

Transcribing emails 

Missing diacritics in emails, often a technical limitation of the period, were handled in two ways. Where an email contained no diacritics at all, they were supplied according to standard spelling rules. This was indicated at the beginning of the transcription by the note {v následujícím e-mailu je doplněna diakritika} (“diacritics have been added to the following email”). Where diacritics were used only in part of an email, showing that technical conditions did not prevent their use, the original was transcribed faithfully, without adding or correcting diacritics. 

Transcribing SMS messages 

SMS messages were transcribed according to specific rules, with two versions produced for each message. One reproduced the original exactly, including its visual form and errors. The other made some changes to improve readability and comprehension: it adjusted the visual form, corrected spelling mistakes and expanded abbreviated words, while retaining features such as non-standard and emotionally expressive language. 

With minor variations, these rules were followed throughout the collection and preparation of the private correspondence. Before publication, all transcriptions were checked in detail and corrected at least twice. Unintentional departures from the original may nevertheless remain, particularly in transcriptions of older handwritten letters. If in doubt, consult the digital image of the original.