Use the translation memory

The translation memory (TM) stores approved translations for your whole organization. When a source text that’s already in the memory needs translating again, on any site and in any adapter, Ponyglot reuses the approved translation instead of calling an engine. These TM hits don’t count towards your characters.

Everyone in the organization can browse the memory. You need the editor role or higher to edit, delete, import or export entries.

Fill the memory

The memory fills itself:

  • when editors approve drafts in the CMS or admin (the connector reports it). If they changed the text, the memory stores their final version;

  • when a connector imports existing translations;

  • when you import a TMX file (see below);

  • when you edit an entry in the dashboard.

There is one entry per language pair and source text. The latest approval wins.

Browse and correct entries

  1. Open Memory in the dashboard.

  2. Filter by language pair, or search the source and translated text.

  3. Choose Edit to correct a translation or delete the entry.

A corrected entry is used for the next matching segments. Translations that were already delivered don’t change.

Export a TMX file

On the Memory page, choose Export TMX. The export uses the current filter, so select a language pair first to export only that pair. The file is TMX 1.4 and works with common translation tools.

Import a TMX file

  1. At the bottom of the Memory page, under Import TMX, choose the file (TMX 1.4, up to 10 MB).

  2. Choose the file’s source language.

  3. Check Overwrite if the file should replace existing translations of the same source text. Otherwise existing entries are kept and those units are skipped.

  4. Choose Import. The page reports how many entries were created, updated and skipped.

Keep in mind:

  • Language codes are matched to your sites’ languages: de-DE in the file becomes de if your sites use de. Units without text in the chosen source language are skipped.

  • Inline markup in the file is flattened to plain text. Files exported from Ponyglot keep the original format of each entry, so re-importing them works exactly. Entries from other tools count as plain text, so HTML segments rarely find exact matches from them.