Align the material first
If one result is 1,980 words and another 2,015, first compare saved version, unit and included parts. A title, note or bibliography present on one side can explain the gap without involving tokenisation. Keep a common text copy. Inspect its first and last lines in both environments, including copied labels. If the gap disappears, investigate collection or selection before changing global settings.
Understand Writer’s documented rule
LibreOffice describes strings between separators and explicitly treats visible internal hyphens and slashes, such as those in technical identifiers, differently from ordinary spaces. Additional separators are configurable. Do not extend that description to every similar-looking mark. Copy the actual token: retyping it with another dash can remove the very distinction you need to diagnose. Added spaces also change its structure.
Use samples that isolate the cause
Place blu-verde, blu/verde, blu verde and the actual troublesome separator in a small comparison table. Record observed results in both programs rather than claiming a test not performed. Start with cases common in the document, such as slash codes or URLs. Change one feature at a time so a different result identifies a plausible rule rather than an uncontrolled collection of changes.
Record additional separators
Check Writer’s Tools > Options > LibreOffice Writer > General, under Additional separators; macOS may use LibreOffice Preferences. Record existing settings before editing. A required counting convention can justify a change, but seeking a lower total alone does not. Settings may affect later documents too. Keep a control sample and describe the configuration alongside the selection and file version.
Reconcile the whole difference
Suppose twelve verified hyphen cases contribute one extra unit each in a comparison. They explain 12 units. For a total gap of 35, 35-12=23 remain unexplained. This is a conditional arithmetic scenario, not a reported app test. Add causes only after verifying their sample contribution; a token containing both slash and hyphen must not be counted twice in the explanation.
Check the transferred characters too
Different character totals on a simple shared extract can suggest missing text, added spaces or omitted lines. Equal totals still do not prove identical strings: different text can have the same length. Inspect boundary words and the relevant marks too. The Writer counting guide locates statistics; an external character counter is useful only on the same string and stated unit.
Choose the acceptance method
When instructions name a program, it is the operational acceptance reference. Otherwise request clarification if the ambiguity changes eligibility, retaining both results and a concrete explanation. For internal work, agree the document version and recurrent-token treatment before revision begins. That project-specific convention need not become a universal definition. Apply it to the final text instead of repeatedly comparing incompatible totals.
Keep a short explanation
A concise report can say “Writer count on selected body text; external difference concentrated in slash codes.” Retain the sample table if needed. Do not assign an unverified cause or claim tests in unopened applications. The word counter can explore an extract, but the delivered number follows the agreement. Recheck after changing text or settings, preserving meaning while explaining the remaining gap.
Reconcile a hypothetical difference
| Item | Calculation | Contribution |
|---|---|---|
| Verified cases differing by one | 12 × 1 | 12 units |
| Total gap in the scenario | 35 | Explain all of it |
| Still unattributed | 35 − 12 | 23 units |
| Next step | The first cause is insufficient | Check other tokens or scope |
Sources and checks
- LibreOffice: word-count rules · Reconcile text scope, hyphens, slashes and additional separators before attributing a Writer word-count difference to an error.
- LibreOffice: word-count rules · Reconcile text scope, hyphens, slashes and additional separators before attributing a Writer word-count difference to an error.