Transliteration vs Transcription: Definitions, Differences and Examples

When a word moves from one writing system to another, the resulting spelling may look simple. Москва becomes Moskva, 東京 becomes Tōkyō, and 한국 becomes Hanguk.

However, these conversions do not all follow the same method. Some systems preserve the written structure of the original word. Others represent how the word is pronounced. Some attempt to do both, while simplified systems may intentionally discard information to make the result easier to read or type.

The two most important concepts are transliteration and transcription.

In their strict technical senses:

• Transliteration converts written characters from one script into another.
• Transcription represents the sounds or pronunciation of spoken language.

The distinction becomes less clear in everyday usage because institutions, standards organizations, dictionaries, libraries, governments, and software platforms sometimes use the terms differently. Unicode, for example, uses transliteration as a broad term for script conversion, while also noting that non-reversible or pronunciation-oriented conversion is often called transcription.

Understanding the difference is essential when comparing romanization systems, choosing a conversion tool, cataloging names, building multilingual databases, or trying to pronounce a word written in an unfamiliar script.

Quick Answer

Transliteration represents the written form of a word in another script. Transcription represents how the word is pronounced.

For example, a strict transliteration may assign one Latin character or character sequence to every letter in the source script. Its purpose is to preserve spelling information and, where possible, make the conversion reversible.

A transcription instead begins with the spoken form. It may ignore spelling distinctions that do not affect pronunciation and add sound information that is not explicitly shown in the original writing.

The distinction is useful, but it is not absolute. Many widely used romanization systems combine orthographic and phonetic principles.

What Is Transliteration?

Transliteration is the systematic representation of text written in one script using the characters of another script.

A transliteration system primarily examines the source text as writing. It asks:

• Which source character is present?
• Which target character represents it?
• Is the character modified by a diacritic?
• Does its position affect the mapping?
• Can the original spelling be reconstructed?

Merriam-Webster defines transliterate as representing or spelling something with the characters of another alphabet. Unicode describes transliteration more broadly as converting characters from one script to another, normally producing a result that is at least approximately phonetic in the target script.

A Simple Transliteration Model

Imagine a fictional script with these characters:

The source sequence:

ΑΒΓΔ

would become:

abgd

The converter does not need to know how the word is pronounced. It only needs to follow the character-mapping table.

Real transliteration systems are more complicated because:

• One source character may require two or more Latin letters.
• Several source characters may sound alike.
• Diacritics may be needed to prevent ambiguity.
• Context may alter the output.
• A script may omit information that pronunciation requires.
• Characters may combine into larger units.

Transliteration and Reversibility

A major objective of strict transliteration is reversibility.

A reversible system allows the original source text to be recovered from the transliterated result. Unicode illustrates this with a Hangul conversion:

갗 → gach → 갗

The Latin form preserves enough information for the original Hangul sequence to be reconstructed. Unicode also explains that reversibility may work in only one direction: different Latin inputs can sometimes map to the same source-script result.

Reversibility is particularly valuable for:

• Bibliographic cataloging
• Archival records
• Scholarly editions
• Historical texts
• Authority files
• Database identifiers
• Automated data exchange
• Cross-script searching

A reversible system is not necessarily easy to pronounce. It may use diacritics, apostrophes, unusual letter combinations, or conventions that require specialist knowledge.

Transliteration May Use Diacritics

Diacritics help distinguish source characters that would otherwise collapse into the same Latin form.

A transliteration system may distinguish:

• s from š
• d from ḍ
• h from ḥ
• a from ā
• z from ž

These marks may represent different original letters rather than minor pronunciation variations.

Removing the diacritics can make the result easier to type, but it may also destroy reversibility. Unicode’s transliteration charts note that some Latin-script outputs use additional accents specifically to support round-trip conversion.

What Is Transcription?

Transcription is the written representation of speech sounds or pronunciation.

Instead of asking which characters appear in the source text, transcription asks:

• How is the word pronounced?
• Which sounds are actually produced?
• Which sound distinctions matter in the language?
• How detailed should the pronunciation record be?
• Which notation will the intended reader understand?

A transcription may be written with ordinary Latin letters, a language-specific pronunciation system, or a specialized notation such as the International Phonetic Alphabet.

The International Phonetic Association maintains the IPA as a standardized system for representing the sounds of languages. Its purpose is phonetic representation rather than conversion between ordinary writing systems.

Transcription Begins with Sound

Suppose a language writes a word with five letters, but only four sounds are pronounced. A transliteration may preserve all five written elements. A transcription may represent only the four sounds.

The reverse can also happen. A writing system may omit vowels or other pronunciation information. A transcription may supply those sounds even though they are not directly represented in the written source.

This is common in writing systems where:

• Short vowels are normally omitted
• Letters have several contextual pronunciations
• Historical spelling differs from modern speech
• Tone is not fully marked
• Consonants change at word boundaries
• The same character has different readings

Broad and Narrow Phonetic Transcription

Phonetic transcription can be broad or narrow.

A broad transcription records the major sound contrasts needed to identify a word. A narrow transcription adds finer pronunciation details, such as aspiration, nasalization, vowel quality, or other phonetic features.

In linguistic notation:

• Angle brackets may indicate ordinary spelling: ⟨word⟩
• Slashes commonly indicate a phonemic representation: /word/
• Square brackets indicate a phonetic realization: [word]

UCLA linguistics materials describe slashes as marking phonemes and square brackets as marking phonetic transcription, including more detailed surface pronunciations.

A pronunciation-oriented romanization written with ordinary Latin letters is not necessarily IPA. It may be a practical transcription designed for travelers, students, journalists, or general readers.

The Core Difference: Written Form vs Spoken Form

The clearest way to distinguish transliteration and transcription is to identify the source of the conversion.

Transliteration pathway

Written source → character-mapping rules → target script

The system starts with visible letters or characters.

Transcription pathway

Spoken form or pronunciation analysis → sound-mapping rules → written notation

The system starts with sounds, phonemes, or a pronunciation dictionary.

This difference affects almost every aspect of the result.

Transliteration and Transcription Are Not Translation

Neither transliteration nor transcription normally changes the language or meaning of a word.

Translation replaces the original expression with a word or phrase carrying approximately the same meaning in another language.

Consider the Japanese place name しんばし:

• Hepburn romanization: shimbashi
• Kunrei-style romanization: sinbasi
• Literal translation of the name: new bridge

The first two represent the Japanese name using Latin letters. The third communicates its lexical meaning in English. Unicode uses this example to demonstrate that script conversion and translation are separate operations.

A similar distinction applies to names:

• A person’s name is generally transliterated or transcribed.
• A descriptive title may be translated.
• A place may have both a romanized local name and a conventional translated or foreign-language name.

Transliteration, Transcription and Romanization

Romanization is the conversion of a non-Latin writing system into the Roman or Latin script.

Romanization describes the target script, not necessarily the conversion method.

A romanization system may therefore be:

• A strict transliteration
• A phonemic transcription
• A pronunciation-oriented transcription
• A hybrid system
• A conventional spelling system
• A simplified version of a more precise standard

For example:

The Library of Congress calls its ALA-LC resources Romanization Tables: Transliteration Schemes for Non-Roman Scripts. This wording illustrates how romanization and transliteration can describe overlapping aspects of the same system.

Why the Terminology Is Sometimes Confusing

There is no universal boundary that every institution follows in exactly the same way.

In a narrow technical definition:

• Transliteration is character-oriented and reversible.
• Transcription is pronunciation-oriented and not reversible.

In broader usage, transliteration may refer to almost any predictable script conversion, including systems influenced by pronunciation. Unicode explicitly states that a non-reversible transliteration is often called transcription, a lossy transliteration, or an ambiguous transcription.

Formal standards sometimes combine the terms. ISO 843, for example, establishes a system for the transliteration and/or transcription of Greek characters into Latin characters.

Software products may also use transliteration for phonetic input. Google Input Tools describes transliteration as typing an approximate pronunciation in Latin letters and converting it into another script, such as typing namaste to generate नमस्ते. This is useful input behavior, but it is not necessarily a strict reversible character transliteration.

For that reason, a reliable reference should explain the behavior of a system rather than relying on its label alone.

Detailed Comparison

1. Source Material

Transliteration works from written text.

A transliterator may process:

• Unicode characters
• Letters
• Diacritics
• Syllabic signs
• Character sequences
• Orthographic boundaries

Transcription works from pronunciation.

A transcriber may rely on:

• Recorded speech
• Native-speaker pronunciation
• A pronunciation dictionary
• Phonological rules
• Regional accent information
• Contextual sound changes

2. Unit of Conversion

In transliteration, the unit is usually a grapheme, character, sign, or defined character sequence.

In transcription, the unit may be:

• A phoneme
• An allophone
• A syllable
• A mora
• A tone
• A stress pattern
• A phonetic feature

3. Relationship to the Original Spelling

Transliteration normally preserves spelling distinctions even when they are not pronounced.

Transcription may remove:

• Silent letters
• Historical spellings
• Morphological distinctions
• Unpronounced vowel markers
• Character differences that produce the same sound

Conversely, transcription may insert pronunciation information absent from the source spelling.

4. Reversibility

Strict transliteration aims to support reverse conversion.

Transcription usually cannot reconstruct the exact original spelling because several spellings may share the same pronunciation.

For example, if two source characters are pronounced identically and both are transcribed as s, the transcription does not reveal which original character was used.

5. Reader Accessibility

Transliteration may require knowledge of a formal standard.

Unicode notes that Japanese Kunrei-style ti is intended to be read with a sound comparable to English chee. A reader who assumes ordinary English spelling rules may mispronounce it.

A reader-oriented transcription may choose chi because it is more immediately pronounceable for an English-speaking audience. That convenience can come at the cost of structural regularity or reversibility.

6. Language Dependence

A script-oriented transliteration can sometimes be applied without fully analyzing pronunciation, although the source language may still matter.

Transcription is inherently language-dependent. The same written character can represent different sounds in different languages, historical periods, dialects, or phonetic environments.

Unicode notes that modern Greek beta is pronounced like v, although traditional or etymological transliterations may represent it with b.

7. Stability

A formal transliteration system can remain stable even when pronunciation varies.

A transcription may change according to:

• Region
• Accent
• Historical period
• Speech style
• Speaker
• Surrounding sounds
• Intended audience

Examples by Language and Script

Japanese: Hepburn and Kunrei

Japanese provides one of the clearest examples of competing priorities.

Hepburn romanization generally produces forms that are more intuitive to many English-speaking readers. Kunrei-style romanization maintains a more regular relationship with Japanese sound patterns and kana structure.

Unicode gives the example:

Unicode also explains that Hepburn can be ambiguous. Both ジ and ヂ may appear as ji, while ズ and ヅ may both appear as zu. A modified system can preserve the distinction with spellings such as dji and dzu, but those forms are less familiar to general readers.

This creates a typical trade-off:

• Hepburn: more reader-oriented
• Kunrei: more structurally regular
• Modified reversible form: more precise but less familiar

Neither goal is inherently superior. The appropriate system depends on whether the user needs pronunciation help, linguistic consistency, cataloging, or reverse conversion.

Korean: Pronunciation-Based Romanization

The Republic of Korea’s Revised Romanization is explicitly based on standard Korean pronunciation and avoids symbols other than Roman letters as far as possible.

Some Korean consonants receive different Latin spellings depending on their phonetic position. The official rules state, for example, that ㄱ, ㄷ and ㅂ may be written g, d and b before a vowel but k, t and p before another consonant or at the end of a word.

This makes the system relatively accessible, but it means the result is not a simple one-character-to-one-character transliteration.

Examples in the official rules include:

The system is commonly called romanization, while its official English documentation describes its principles as transcription. This is another illustration of overlapping terminology.

Thai: Transliteration and Simplified Transcription

ISO treats transliteration and transcription as distinct but related approaches for Thai.

ISO 11940 provides a formal transliteration framework, while ISO 11940-2 defines a simplified transcription system based on pronunciation rules and dictionary forms.

The simplified transcription:

• Uses pronunciation rules
• Does not distinguish short and long vowels
• Does not normally include tones
• Sacrifices some information for simpler output

ISO states that vowel-length and tone information can be added when fuller pronunciation representation is needed.

This is a particularly clear demonstration of the difference:

• A transliteration attempts to preserve written distinctions.
• A simplified transcription gives a more practical pronunciation-oriented form.
• Information omitted from the transcription cannot necessarily be recovered.

Greek: More Than One Valid Conversion Goal

ISO 843 establishes rules for both transliteration and transcription of Greek characters into Latin characters. The standard applies to Greek script from classical and modern periods.

The distinction matters because Greek spelling and pronunciation have changed over time.

For example:

• A historical or character-based system may represent β as b.
• A modern pronunciation-oriented system may represent it as v.

Unicode notes that traditional transliteration can preserve the association between beta and b, even though modern Greek pronunciation uses a v sound.

The best form therefore depends on whether the text is:

• Ancient Greek
• Medieval Greek
• Modern Greek
• A scholarly citation
• A modern name
• A pronunciation guide

Library Cataloging

Libraries need stable forms for names, titles, subjects, shelving and record exchange.

The ALA-LC Romanization Tables provide approved schemes for converting numerous non-Roman scripts. The Library of Congress describes the tables as a definitive guide for transliterating names, words, titles and text into Roman script.

Romanized cataloging supports:

• Searching
• Alphabetical arrangement
• Shelving
• Circulation
• Acquisitions
• Authority control
• Reference work
• Metadata exchange

The Library of Congress increasingly records original-script data alongside romanized access forms. Romanization remains useful for indexing, sorting and systems with incomplete support for non-Latin scripts.

A library-oriented transliteration may be less intuitive than a popular spelling because preservation and consistency take priority over immediate pronunciation.

Transliteration and Transcription in Digital Tools

A script-conversion tool should state what kind of output it produces.

Character transliterator

A character transliterator applies a defined mapping to written input.

Best for:

• Data conversion
• Cataloging
• Scholarly work
• Standard comparisons
• Round-trip testing

Pronunciation-based converter

A pronunciation converter analyzes the word or looks it up in a lexical database.

Best for:

• Language learners
• Travel
• Speech interfaces
• Public-facing names
• Pronunciation assistance

Phonetic transcriber

A phonetic transcriber produces IPA or another sound notation.

Best for:

• Linguistic analysis
• Dictionaries
• Speech therapy
• Pronunciation research
• Accent comparison
• Speech technology

Input-method transliterator

An input-method tool allows users to type approximate Latin spellings and receive candidates in the destination script.

For example, a user may type a Latin approximation of a Hindi word and select the intended Devanagari result. Because several input spellings may produce the same candidate, this process is often fuzzy and not reversible.

Why Automated Conversion Can Be Difficult

A character table alone is not sufficient for every language.

Unwritten vowels

Some writing systems do not normally show every vowel. A pronunciation-oriented transcription must infer or look up missing sounds.

Context-dependent pronunciation

A character may be pronounced differently according to:

• Its position in a word
• Neighboring characters
• Morphological boundaries
• Stress
• Dialect
• Historical period

Multiple readings

A written sign may have several possible readings. The converter may need lexical or grammatical context to choose the correct one.

Sound changes

Speech may contain assimilation, consonant weakening, vowel reduction or other changes not visible in spelling.

Ambiguous target spellings

Latin letters themselves are pronounced differently across languages. Unicode points out that a letter such as j can suggest very different sounds to English, German, French or Spanish readers.

Loss of distinctions

A simplified output may merge:

• Two source letters
• Long and short vowels
• Tones
• Stress patterns
• Aspirated and unaspirated consonants
• Retroflex and dental consonants

Once merged, those distinctions cannot reliably be restored.

When Should You Use Transliteration?

Choose transliteration when preserving the written form is more important than giving an immediately intuitive pronunciation.

Common situations include:

• Academic citations
• Manuscript studies
• Library catalogs
• Historical research
• Archival databases
• Legal record matching
• Authority control
• Linguistic comparison
• Script-conversion software
• Digital humanities projects

A strict transliteration is also preferable when users must compare several standards or convert the result back to the original script.

When Should You Use Transcription?

Choose transcription when the primary question is how the word is pronounced.

Common situations include:

• Language-learning materials
• Pronunciation dictionaries
• Travel guides
• Speech research
• Voice interfaces
• Broadcasting
• Linguistic fieldwork
• Accent documentation
• Speech therapy
• Public pronunciation guides

The level of detail should match the audience. A traveler may need a simplified spelling, while a phonetician may require detailed IPA notation.

How to Identify Which Method a System Uses

Do not rely only on the name of the standard. Examine its rules.

Signs of transliteration

A system is probably transliteration-oriented when:

• Every source character has a defined equivalent.
• Diacritics distinguish similar source letters.
• Silent or historical letters remain visible.
• Reverse conversion is possible.
• Pronunciation changes do not alter the mapping.
• The documentation emphasizes characters or script.

Signs of transcription

A system is probably transcription-oriented when:

• Output follows actual pronunciation.
• Silent letters disappear.
• Unwritten vowels are supplied.
• Sound changes affect the result.
• Dialect or standard pronunciation must be specified.
• Several source spellings can produce the same output.
• The documentation relies on a pronunciation dictionary.

Signs of a hybrid system

Many practical systems are hybrids.

A hybrid may:

• Preserve most source characters
• Apply some pronunciation rules
• Simplify rare distinctions
• Retain conventional spellings
• Omit selected diacritics
• Use context-dependent mappings

Calling such a system either transliteration or transcription may be defensible, provided its behavior is explained clearly.

Accuracy, Readability and Reversibility

Most script-conversion systems balance three competing goals.

Accuracy

Accuracy means preserving relevant information from the original.

This may include:

• Character identity
• Vowel length
• Tone
• Stress
• Consonant distinctions
• Word boundaries
• Capitalization
• Punctuation

Readability

Readability means producing a form that the intended audience can understand or pronounce.

A spelling familiar to an English speaker may not be intuitive to an Indonesian, French, German or Spanish speaker.

Reversibility

Reversibility means allowing the original written form to be recovered.

Unicode’s guidance emphasizes that these requirements are rarely satisfied simultaneously. Designing a useful transliteration therefore involves balancing competing needs.

A highly reversible system may be difficult to read. A highly readable transcription may lose the exact original spelling.

Best Practices for Romanization.org Tools and References

A trustworthy conversion platform should make the method transparent.

Preserve the original script

The native form should always remain visible. A transliteration or transcription is an additional representation, not a replacement for the source.

Name the standard

The result should identify the system used, such as:

• ISO 9
• ISO 843
• ALA-LC
• Hepburn
• Kunrei
• Revised Romanization
• ISO 11940
• ISO 11940-2

Label the conversion type

Where possible, identify whether the result is:

• Transliteration
• Simplified transliteration
• Phonemic transcription
• Phonetic transcription
• Reader-oriented pronunciation
• Conventional romanization
• ASCII fallback

Preserve diacritics in the standard form

If a standard requires diacritics, display the complete version first. A simplified output without diacritics can be provided separately.

Show alternative systems

Different outputs may all be valid for different purposes.

A comparison table should explain:

• The governing authority
• The intended use
• Whether the system is reversible
• Whether it follows spelling or pronunciation
• Whether diacritics are required
• Whether an ASCII form exists

Do not invent pronunciation from spelling alone

When pronunciation cannot be derived reliably, the tool should:

• Request the language
• Use a verified dictionary
• Show multiple possibilities
• Mark the result as uncertain
• Avoid presenting a guessed transcription as authoritative

Distinguish computed forms from established names

A mechanically generated conversion may differ from:

• A passport spelling
• A person’s preferred name
• An official place name
• A historical conventional form
• A brand name
• A recognized English exonym

Established forms should be preserved and labeled separately from calculated outputs.

Common Misconceptions

Transliteration always shows pronunciation

Not necessarily. A strict transliteration may preserve spelling at the expense of intuitive pronunciation.

Transcription is always written in IPA

No. IPA is a specialized phonetic system, but practical transcriptions can use ordinary Latin letters.

Every romanization is a transliteration

Romanization only specifies that the target is Latin script. The process may be transliteration, transcription or a hybrid.

Transliteration is always reversible

Reversibility is a common technical goal, but the term transliteration is also used more broadly. Unicode acknowledges both usages.

Transcription gives one universal pronunciation

Pronunciation may vary by accent, dialect, speaker and historical period. A transcription should identify the pronunciation variety when that difference matters.

Removing diacritics does not change anything

Diacritics may distinguish separate source characters, tones, long vowels or consonant types. Removing them can make the result ambiguous.

Translation and transliteration are interchangeable

Translation changes meaning into another language. Transliteration changes the script used to write the original expression.

Frequently Asked Questions

What is the simplest difference between transliteration and transcription?

Transliteration follows written characters. Transcription follows spoken sounds.

Is romanization transliteration or transcription?

It can be either. Romanization means conversion into Latin script. Some romanization systems are character-based, some are pronunciation-based, and many are hybrids.

Is transliteration more accurate than transcription?

It depends on what must be accurate. Transliteration can be more accurate about spelling, while transcription can be more accurate about pronunciation.

Can a transcription be converted back into the original script?

Usually not reliably. Multiple original spellings may have the same pronunciation, and the transcription may omit orthographic distinctions.

Can transliteration be pronounced directly?

Sometimes, but not always. Formal transliteration systems may use letter values that differ from ordinary English pronunciation.

Is IPA a transliteration system?

No. The International Phonetic Alphabet is primarily a notation system for representing speech sounds. It can be used to transcribe words from any language, regardless of their original writing system.

Why do standards use different spellings for the same word?

Different standards prioritize different goals, such as reversibility, pronunciation, typing simplicity, national policy, bibliographic consistency or reader familiarity.

Which method is better for personal names?

Use the person’s established or officially documented spelling when available. Generated transliterations can be shown as alternatives but should not replace a chosen identity form.

Which method is better for language learners?

A pronunciation-oriented transcription is often easier for beginners. However, it should be labeled clearly and used alongside the original script rather than as a permanent substitute for learning it.

Which method is better for databases?

A reversible transliteration is useful for indexing and structured data, but the original Unicode text should also be stored whenever possible.

Conclusion

Transliteration and transcription solve different problems.

Transliteration transfers writing from one script to another. It focuses on characters, spelling structure and, in stricter systems, reversibility.

Transcription represents pronunciation. It focuses on phonemes, speech sounds and the way a word is actually spoken.

The distinction can be summarized as:

Transliteration represents how a word is written. Transcription represents how a word is pronounced.

Real-world systems often fall somewhere between those two ideals. A romanization may preserve most written distinctions while applying pronunciation rules. A library standard may favor reversibility, while a national road-sign system may favor readability. A language textbook may simplify the result further for beginners.

For this reason, no Latin-script form should be presented without context. A reliable reference should identify:

• The original script
• The source language
• The conversion standard
• Whether the output follows spelling or pronunciation
• Whether it is reversible
• Which distinctions have been omitted
• Whether alternative forms are widely used

The goal is not merely to convert characters. It is to explain what the converted form represents, what information it preserves, and what information it leaves behind.