Preparing Your Files

Give Exempla a structure it can read

A corpus begins with source files, but Exempla also needs to know where each speaker turn begins and which passages are transcription comments rather than spoken words. Careful preparation produces reliable counts, accurate speaker attribution, and Concordance results that can be traced back to the correct file and line.

12 min
Why this matters

Formatting determines what Exempla counts and attributes

Reliable searches

Spoken words remain searchable while comments and speaker labels stay outside ordinary results.

Correct speakers

Each line remains connected to the speaker whose turn it belongs to.

Traceable evidence

Every Concordance occurrence can identify its corpus text, speaker code, and line number.

Choose the right file format

Exempla imports UTF-8 plain-text files and Microsoft Word documents. You can select one file or several files in the same import.

Plain text.txt

The file must use UTF-8 encoding. Exempla also accepts UTF-8 files that include a byte-order mark. Existing line breaks become Exempla transcript lines.

  • Best when you want direct control over line breaks.
  • Easy to inspect in any plain-text editor.
  • Invalid UTF-8 and empty files are rejected.
Microsoft Word.docx

Exempla reads Word paragraphs and manual line breaks as logical text blocks. Long blocks are wrapped near a natural space without splitting a word.

  • Useful when transcripts are prepared in Word.
  • The older .doc format is not supported.
  • Word styling is not part of the imported corpus text.
What UTF-8 means

UTF-8 is a text encoding that preserves accented characters and writing systems consistently. If a TXT file is rejected, open it in a text editor and save it again with UTF-8 encoding.

How the corpus name is created

The filename without its extension becomes the initial corpus text name. For example, `Sociolinguistic interview 01.docx` becomes `Sociolinguistic interview 01`. You can rename the corpus text inside Exempla at any time.

Use speaker turns, not a special table

A transcript can remain ordinary readable text. Add a speaker code in square brackets at the beginning of each new turn. The spoken text follows the code.

Sociolinguistic interview 01.txtUTF-8 plain text
[INT]Where did you grow up?
[MARINA]I grew up here in Ottawa. My whole family lived in the region.
[INT]How was it growing up in Ottawa?
[MARINA]It was different from today. We knew most of the families on our street.
People spent more time outside, and everyone looked after each other.
[DAVID]You could walk everywhere back then. (laugh)
[MARINA]Yes, and we were always visiting our neighbours.
1

Start a turn with a code. Write the code inside matching square brackets, then add a space before the spoken text.

2

Continue a long turn on another line if needed. A line without a new code inherits the most recently identified speaker.

3

Start the next speaker with a new code. Exempla then attributes the following line or lines to that speaker.

4

Keep the source readable. Blank lines and ordinary punctuation are acceptable. The essential rule is that speaker and comment markers are balanced.

Choose stable speaker codes

A speaker code is the identifier that connects a turn to a speaker in Exempla. It is not counted as a spoken word and does not appear as an ordinary search token.

Recommended[INT][MARINA][NARRATOR]

Use short, recognizable codes and keep the same spelling for the same person across files.

Avoid ambiguity[S1][S1]

Do not reuse one code for different people. If the same code appears in another imported transcript, Exempla asks you to confirm the identity.

The same person in several conversations

Use the same code whenever the same person speaks, even across different conversations and transcripts. Confirm Same Person during import so Exempla keeps the occurrences under one speaker code.

  • A code cannot be empty.
  • A code can contain letters, numbers, spaces, hyphens, underscores, and apostrophes, but it cannot contain square brackets or line breaks.
  • When Exempla asks you to type a new code, enter only the code itself. Do not type the surrounding square brackets.
  • Codes are compared without treating uppercase and lowercase as different identities. `[MARINA]` and `[marina]` therefore require an identity decision.
  • A code written inside a parenthesized comment is treated as part of the comment and is not registered as a speaker.

Put non-spoken comments in parentheses

Use matching parentheses for stage directions, transcription notes, contextual annotations, and other material that should not normally be counted as speech.

[MARINA] We met there every Friday. (brief pause) Everyone knew the building.
Counted as speechWe met there every Friday. Everyone knew the building.
Excluded comment(brief pause)
What is excluded

Parenthesized text is excluded from ordinary word counts and Word Lists. Concordance also excludes it by default.

When comments can be searched

Concordance includes an Include comments option. It can search inside parentheses without making bracketed speaker codes searchable. Word counts and Word Lists remain unchanged.

Compare a clean transcript with common problems

Correct
[INT] How was it growing up in Ottawa?[MARINA] It was a close community.[MARINA] (brief pause) We knew our neighbours.

Every speaker code and comment has matching markers.

Needs review
[INT How was it growing up in Ottawa?[] It was a close community.[MARINA] (brief pause We knew our neighbours.

The first line lacks `]`, the second has an empty code, and the third lacks `)`.

[MARINA I grew up in Ottawa.Opening [ has no closing ]
MARINA] I grew up in Ottawa.Closing ] has no opening [
[] I grew up in Ottawa.Speaker code is empty
[[MARINA]] I grew up in Ottawa.Nested [ inside a speaker code
The interview continued. (phone ringsOpening ( has no closing )
The interview continued.)Closing ) has no opening (

What happens after you choose Add Text Files

You can choose Add Text Files… or drag supported files into the Corpus tab in the left panel. Files are prepared one at a time, even when you select several together.

1
Read the document

Exempla confirms the file type, reads its text, rejects an empty file, and validates UTF-8 for TXT.

2
Create the corpus text name

The filename without `.txt` or `.docx` becomes the proposed name inside the project.

3
Resolve a duplicate

If the name already exists or the content is identical to another transcript, choose Replace Existing, Import with Different Name, or Cancel.

4
Check formatting

Exempla checks square brackets and parentheses before adding the file to the corpus.

5
Resolve speaker attribution

Exempla handles a transcript with no code, then reviews codes already used elsewhere in the corpus.

6
Finish and review the summary

Exempla adds every successfully prepared file to the project. After a multi-file import, the Import Summary shows how many files were imported, skipped, or had problems. Open its details to see the affected filenames and the reason for each skipped or problematic file.

Understand the formatting warning

If Exempla finds a possible problem, the warning names the source file and shows the first location. Show details reveals the excerpt and lets you move between problems.

Possible Formatting Errors
!
2 possible formatting errors

Sociolinguistic interview 01.docx

Word paragraph/break 14: opening [ has no closing ]

Show details ›

Illustration

Review and Fix

Opens a line-numbered editor on Exempla's import copy. Previous and Next move between problems. After editing, choose Check Formatting. Continue Import becomes available only when the revised text passes the check and is different from the original import copy.

Import Anyway

Adds the transcript with the possible formatting problems still present. Use this only when you understand the warning, because unmatched markers can change what is counted or which speaker receives a line.

Cancel

Stops the import for that file. Nothing from the file is added to the project, and the next selected file can still be processed.

Complete the speaker decisions

No speaker code was found

If the transcript contains no valid speaker code outside parentheses, Exempla asks how the text should be attributed.

Assign Speaker Code…

Enter one code for the entire transcript. Exempla adds it to the first attributable line of the imported project copy.

Import Anyway

Adds the transcript without speaker attribution. The text can still be searched by corpus or text-file scope.

Cancel Import

Stops the file so you can correct or reconsider it before adding it.

A speaker code already exists

If an incoming code is already used in the corpus, Exempla names the incoming text and the existing text or texts. This protects the integrity of speaker-based searches and statistics.

Speaker Code Already in Corpus
?
The speaker code “MARINA” appears in the incoming text “Sociolinguistic interview 02” and is already used in “Sociolinguistic interview 01”.

Does “MARINA” identify the same person in both texts?

Illustration

Same Person

Retains the established corpus code so occurrences from both texts remain connected.

Different Person — Change Code

Prompts for a unique replacement code and updates every matching occurrence in the incoming project copy.

Review Occurrences

Shows examples from existing texts and the incoming text, with line numbers, before returning to the same decision.

Cancel Import

Stops the file without making an uncertain identity choice.

Find passages with no speaker assigned

Text outside comments that is not attributed to a valid speaker code can remain unassigned, even when other passages in the same transcript have speakers. The Speakers list shows No speaker assigned when the corpus contains such text.

Double-click No speaker assigned to open a read-only list of unassigned passages, grouped by text. Select a passage to see its line number and excerpt, then choose Open Transcript (or double-click the passage) to jump to it. This does not assign a speaker or change the transcript.

To attribute one passage, use Edit Transcript from the Transcript view and add the appropriate [speaker code] at the start of that speaker's turn. Review the result before applying your change. Other unassigned passages are not assigned automatically.

Open the editor at the right place

In Transcript, highlight any amount of text—even one character—or simply place the cursor where you want to work. Choose Edit Transcript to open Review Transcript at that selection or cursor position. The line is shown from its beginning so you can see its speaker code and context.

Make the correction, use Check Formatting if needed, then choose Apply Changes. Editing changes the project copy, not your original TXT or Word file.

Know which text Exempla is using

Exempla reads the source document and stores a prepared copy inside the project. All analysis uses that project copy. The original file remains in its original location and is not rewritten.

Original file

Your `.txt` or `.docx` file remains unchanged on your computer.

Prepared project copy

May contain reviewed corrections, an assigned whole-text speaker code, or wrapped Word paragraphs.

Analysis and results

Concordance, Word List, Sequences, and statistics use the prepared copy.

TXT line references

Existing line breaks become Exempla lines, so Concordance line numbers follow the imported text structure.

Word line references

Paragraphs and manual breaks become logical blocks. Long blocks may be wrapped near 108 characters, so a Exempla line is not the same as a visual line or page in Word.

Text Information

For each corpus text, Exempla records the original filename, source type, import method, import time, whether long lines were wrapped, whether corrections were made, formatting status, line and word totals, and detected speaker codes.

Check again later

Use Check Formatting to examine selected text files or the entire corpus. You can review and correct the project copy while leaving the source file untouched.

Before importing

Final preparation checklist