Most languages use a subset of the sounds a person could produce. Different languages use different subsets. If a sound used by one language cannot be represented correctly in a non-native script, then it needs to be approximated. This approximation might be inaccurate if the sounds in the source and destination languages are significantly different.

The following table shows how transliteration can be inaccurate when the words are transliterated from English (Latin) to Japanese (Katakana) and back to English (Latin):

Original Latin (English)	Katakana (Japanese)	Transliterated Latin
Philippines	フィリピン	Firipin
Düsseldorf	デュッセルドルフ	Dyusserudorufu
Beethoven	ベートーベン	Betoben

Some Kanji characters in the first name of the contact in Japan addresses are incorrectly transliterated into Arabic numerals instead of Latin alphabets.

The following table shows the Kanji numerals that Informatica Address Verification might incorrectly translate into Arabic numerals instead of Latin alphabets:

Kanji Numeral	Arabic Equivalent	Latin Transliteration
一	1	ichi
二	2	ni
三	3	san
四	4	yon
五	5	go
六	6	roku
七	7	nana
八	8	hachi
九	9	kyū
十	10	jū