Two people can look at the same translation and one may say "very good", the other "unacceptable". This debate is settled only when the findings are sorted into types and counted. An error typology sets up this sorting and moves the quality conversation from impression to measurement.
Definition of a typology
It is the classification scheme that sorts findings by type. Every finding goes into a category. The structure is defined.
Why it is needed
Without classification the quality debate rests on impression. The two sides look at the same text differently. Measurement ends the debate.
The main division
Findings are first divided into those that concern meaning and those that concern language. This first division is the most decisive. The structure is built from here.
The meaning level
The content of the source text being carried over wrongly or incompletely belongs to this group. Severity is high. Priority is given.
The language level
Findings relating to the rules of the target language form a separate group. The content may be correct. The severity is different.
The format level
Page layout and tag problems are the third group. Even if the text is correct, the document may be unusable. The group is kept apart.
The compliance level
Failure to follow the terminology list and the guide belongs to this group. The basis is the client's document. The criterion comes from outside.
Number of categories
A small number of distinguishable categories is chosen. A very fine division gets confused in practice. Simplicity is preferred.
Categories must not overlap
A finding should go into one category only. Overlapping definitions spoil the count. The boundaries are made clear.
Writing the definitions
Each category is defined with examples. An example teaches more than a definition. A document is prepared.
Borderline cases
A rule is set for findings that could go into either of two categories. The decision is not left arbitrary. The priority is written down.
Relationship with severity
Type and severity are separate concepts. The same type can hold both a minor and a major finding. The two axes are kept apart.
Variation by document type
A typology is not applied in the same way to every type of text. Priorities differ in legal and promotional texts. It is adapted.
Relationship with industry standards
Established typology models exist. There are also organisations that set up their own scheme. The choice is justified.
Sharing it with the client
The typology in use is explained to the client. A common language is established. The assessment becomes understandable.
Use in feedback
Corrections from the client are entered under these categories too. This is how the trend shows. Data accumulates.
Use in training
The typology shows the translator what counts as an error and why. Learning speeds up. The benefit is direct.
Recording scheme
Every finding is recorded with its type, location and severity. The record is analysed afterwards. A table is kept.
The real gain from classification
The value of a typology lies less in assessing a single job than in showing which error type recurs over the months; a recurring type points to a training or process problem, and the root cause is reached from there. The view is on the trend. The gain is cumulative.
Limits of measurement
Classification does not remove human assessment; which finding goes into which category is still an expert decision. Measurement makes that decision visible and consistent. The limit is known.