Translation tools process a text not as a whole but in units called segments. This unit is a sentence most of the time, but not always. Where the split is made directly affects both the comfort of working and the match rate from the memory.
Definition
A segment is the smallest unit in which the tool processes the text. Translation proceeds through these units. Memory records are also kept at segment level.
Relationship with the sentence
A segment generally corresponds to a sentence, but it does not have to. Headings, list items and cells are separate segments too. This is why the two are not used as synonyms.
Segmentation rules
The tool recognises the end of a sentence from the punctuation marks, and these rules can be adjusted. The default rules do not suit every text. The setting is made to suit the job.
Wrong splits caused by abbreviations
Abbreviations containing a full stop can be mistaken for the end of a sentence, and the segment is split in the middle. This is the most frequently seen splitting error. The list of rules is extended.
The numbering problem
The full stops in article numbers and dates can also lead to wrong splits. In legal texts this problem is concentrated. The settings are tried out in advance.
Effect on the match rate
A wrongly split segment does not match the record in the memory. In this way the accumulated past work becomes unusable. Correct splitting protects the gain.
Merging segments
The translator can merge two segments where necessary. This operation affects the memory record as well. It is used with care.
Splitting segments
A long segment can be split, but this too changes the match. The operation is carried out for a reason. Arbitrary splitting is avoided.
Empty segments
Untranslated segments remain flagged in the tool. Before delivery, a search is made for empty segments. A skipped segment means missing text.
Locked segments
Sections that must not be translated can be locked. The lock prevents accidental changes. It is applied during preparation.
Segment statuses
Every segment carries a status such as draft, translated or approved. The statuses allow progress to be tracked. Tracking is done with the tool.
Table cells
Every cell becomes a separate segment, and the context may be lost. In tables with a single word per cell this problem grows. The context is checked separately.
Headings
Headings are separate segments and appear without context. The same heading may carry a different meaning in different sections. Care is needed.
Position of the tags
Formatting tags sit inside the segment and are preserved. A tag that shifts breaks the format on delivery. A check is made.
Numeric segments
Segments consisting of numbers only can be carried over automatically. This automatic operation is verified as well. It is not taken on trust.
Visibility of context
Good tools show the section in which the segment sits. Working without context raises the risk of error. Visibility is looked for.
Relationship with the file format
Segmentation works differently depending on the file format. The same text may be split differently in two formats. The difference is known.
Common mistake
The most common mistake is to think that a segment is the same thing as a sentence. A heading, a cell or a sentence left half finished can also be a segment. This distinction makes a context check compulsory.