Scribd
Upload
12 views

TM Systems - Main Functions

Uploaded by

Оксана Малик
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
12 views

TM Systems - Main Functions

Uploaded by

Оксана Малик
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 38

Translation

Memory Systems:
main functions
Agenda of the lecture
01 Translation Memory (TM)

02 Matching, types of matches

03 Building and Maintaining a TM Database

04 Benefits and Limitations of TM Systems


Translation
Memory (TM)
a system that stores segments of previously
translated text for future use.
Historical Development
The concept of Translation Memory was first
proposed as early as the 1970s, but it didn’t
become widely available until the mid-1990s.

The original idea for Translation Memory


is often attributed to Martin Kay.
The P roper Plac e of M e n a nd M ac hines
in Languag e Tr an sla tion ’, 19 80

The original idea for Translation Memory


is often attributed to Martin Kay.
A translator’s
workstation
A translator’s
workstation
hardware and software
tools that enable
translators to work
efficiently and effectively
Kay’s vision included a
system where a translator
could command the
machine to display anything
in the stored database that
might be relevant to the text
being translated, allowing
them to review past and
future fragments containing
similar material.
t w o u ld
a m t h a
r o g r a ny
ap h e th e r
m b e r' w
're m e x t h a d
n e w te
rt o f a te d ,
pa tr a n s la
y b e e n
alre a d h t h e
s o , fe t c
and if t ra n s la tio n
on d in g
c o rr e s p o r y
it s m e m
from

Peter Arthern, a translator for the


European Economic Community (EEC)
Not only do they help you
work faster and more
consistently, but they also
keep you competitive in a
field where efficiency is key
Creative texts are not the
best match for a TM, but…
even for creative texts, TM systems can still offer
some value by ensuring consistency with
recurring phrases or terminology—like slogans
or disclaimers that show up multiple times.
Types of Translation Systems
Types of Translation Systems

Pros:
Offline Use
Independence from Server Issues
Reduced Distractions

Cons:
Storage Space
Performance Issues
Licensing Costs
Risk of Data Loss
Update and Bug Fixes
Types of Translation Systems

Pros:
Accessibility
Automatic Backups
Regular Updates
Collaboration
Cons:
Internet Dependency
Data Security
How does it work?
TM allows translators to consult a database of previous
translations, usually organized on a sentence-by-sentence basis.

When a translator encounters a new sentence to translate, they


can search this database for any similar sentences that have
been translated before.

If they find an exact match, they can simply copy and paste it into
their current project.
How does it work?

If the match is close but not exact, the TM system highlights the
differences, and the translator can then decide how to adjust the
translation accordingly.
Building the Database
building it as you go along

importing an existing
database

aligning a Parallel Text


building it as you go along
each sentence you translate is automatically added
to the database. Over time, as you work on similar
texts or projects, you can leverage this growing
database for future translations.
importing an existing database
TM databases created in one software can be imported
into another
Aligning a Parallel Text
the software matches segments of the source text
with their translations, creating translation pairs
Misconception #1: Translation
memories are unreliable
because they are a kind of
machine translation.
Translation memory vs. machine translation:
How do they differ?

hybrid translation workflow


Matching
the process where the TM system compares the
sentence you are currently translating with the
database of previously translated sentences.
01
Exact Match
the TM system finds a sentence in the database that perfectly
matches the sentence you need to translate
02
Fuzzy Match
The system can’t find a perfect match, so, instead, it may find
sentences that are similar but not identical.

+ Fuzzy Match Score


How does matching work?
Most TM systems calculate these matches based on
something called "sequence comparison" or "string-edit
distance." This method measures how many changes (like
insertions, deletions, or substitutions) are needed to turn
one sentence into another.

Example: Changing the word "waiter" to "waitress"


requires one deletion and three insertions.
Select 'Symbol' in the Insert
menu (Input sentence)

Select 'Paste' in the Edit menu


(Match found by the system)
More Sophisticated Matching
Some TM systems incorporate more sophisticated
methods, taking into account linguistic knowledge, like
inflection paradigms or synonyms.

"When the paper tray is empty, remove it and


refill it with paper of the appropriate size."

"When the tray is empty, remove it and fill it


with the appropriate paper."
Segment (Fragment) Matching
Most TM systems work on a sentence-by-sentence basis.
However, some systems are starting to offer "segment matching"
or "portion matching."

Example: If your input sentence is


"Select 'Symbol' in the Insert menu to enter a
character from the symbol set,"

and the TM system finds these two matches:


"Select 'Paste' in the Edit menu."
"To enter a symbol character, choose the Insert menu
and select 'Symbol.'"
Evaluation of TM systems
user-friendliness
productivity gains
the accuracy of the
matching algorithm
Who uses translation software?
Misconception #2: Translation memories
are not useful to my business because
our texts are not repetitive.
Benefits of Using Translation Memory Systems

+ Enhanced Collaboration
+ Simultaneous Shipping (Simship)
The best translation memory software in 2023
Thank you for your
attention!

You might also like