Style

The Machine Translate style guide


The Machine Translate Foundation was founded as open information and community for people who want to use machine translation in their work, products or services. They are diverse and busy, so the content should be high-level and understandable.

Less is more.

Table of contents
  1. 📂 Content
    1. Default to Wikipedia style
    2. Minimal content
    3. Consistency
    4. No marketing
    5. No empty phrases
    6. No prescriptions
    7. Not too academic
    8. Machine translation ready
    9. Chronological order
    10. Names list
    11. Terminology
    12. The Machine Translate Foundation and machinetranslate.org
  2. 🖌 Formatting
    1. Capitalisation
    2. Spacing
    3. Dates
    4. Times
    5. Oxford comma
    6. Apostrophes and quotation marks
    7. Acronyms and abbreviations
    8. Lists
    9. Examples
    10. File names
    11. Images
  3. 🛠 Markdown
    1. Front Matter
    2. Headers
    3. Body
    4. Links
      1. Paths
    5. Mathematical notation

📂 Content

Default to Wikipedia style

Default to the Wikipedia’s Manual of Style and the European Commission English Style Guide:

  • Create content that is neutral, factual and evergreen.
  • Follow their guidelines in titles, capitalisation, punctuation, and formatting.
  • Link to other articles with relative paths.

Default to UK English orthography and to International English vocabulary.

Minimal content

Unlike Wikipedia, the goal is to provide a helpful overview that is easy to read in full and easy to maintain.

Keep pages short, paragraphs short and sentences short and simple.

Add lists for clarity at a glance.

Avoid parentheses, except for defining common abbreviations.

… translation management system (TMS) …

Consistency

Use sister articles as reference and keep consistency with their structure. Similar articles should have similar sections and focus on the same kind of information.

No marketing

Do not promote businesses.

No empty phrases

Do not use cliché or meaningless terms.

  • language barriers
  • multilingual audiences

No prescriptions

Write about facts that are always true.

  • Post-editing effort may be is…

Not too academic

Talk straight and make complex topics simple.

Machine translation ready

Machine Translate will be machine translated. Write articles that are ready for machine translation:

  • Follow the Minimal content guidelines.
  • Be concise and consistent.
  • Use active voice.
  • Check spelling.
  • Use nouns instead of pronouns.

Tags format, structure and annotate texts. They Tags can also be used as placeholders.

Chronological order

Arrange events, publications, and any information related to time in chronological order.

Names list

In lists, separate names with commas. Do not add “and” before the last name.

Philipp Koehn, Franz Josef Och, Daniel Marcu

Terminology

  • API or system engine
  • machine translation automatic translation AI translation

For terms with multiple spellings, check Machine Translate to find the preferred choice.

  • Rules-based machine translation Rule-based machine translation

The Machine Translate Foundation and machinetranslate.org

Refer to this project and community as the Machine Translate Foundation.

🖌 Formatting

Capitalisation

In headers, capitalise only the first letter.

How to contribute

Spacing

Avoid trailing spaces and double spaces inside sentences.

Separate sentences with a single enter. Separate paragraphs with two enters.

Dates

Put dates before months. Omit the leading 0 in single-digit dates.

2 June, 2019

Times

Include the leading 0 in single-digit hours.

09:00

Use - to separate time ranges.

09:00 - 09:30

Oxford comma

Use the Oxford comma where appropriate.

Apostrophes and quotation marks

Use curly apostrophes and quotation marks.

‘…’

“…”

Acronyms and abbreviations

Only use very widely understood acronyms, like URL and API.

NMT neural machine translation …

If a concept like CAT or TMS is often referred to by its acronym, then include the acronym in parentheses the first time it is mentioned.

… computer-aided translation (CAT) tool …

Neural machine translation (NMT) …

Otherwise, do not use abbreviations.

Do not use punctuation marks as abbreviations.

USD/EUR USD or EUR

Lists

Introduce lists with full sentences.

For list items without a main verb, use initial lowercase and no end punctuation.

For list items without verbs:

- item 1
- item 2
- item 3

For list items with verbs, use initial uppercase and a period.

For list items with verbs:

- Item 1 starts with uppercase and ends with a period.
- Item 2 starts with uppercase and ends with a period.
- Item 3 starts with uppercase and ends with a period.

Examples

Write examples between backticks.

Italian: `Il corso è lungo.`
Translation into English: `The course is long.`

File names

When creating a new article, give the file a name that does not exist in Machine Translate.

Images

Upload image files to the corresponding article directory.

- workflows
  - hybrid-translation.md
  - hybrid-translation-workflow.png

Embed images in articles using the <img> tag.

<img title='Hybrid translation workflow' src='/workflows/hybrid-translation-workflow.png' width='700' style='padding: 1em;' />

🛠 Markdown

Machine Translate articles are written in Markdown.

Follow the structure from published Machine Translate articles.

Front Matter

Add Front Matter variables at the beginning of each article.

---
parent: Customisation
title: Training data
description: Training data for machine translation
---
  • parent is the name of the section.
  • title is the article header.
  • description is a phrase that summarizes the content.

Headers

The Front Matter title will automatically be displayed as an H1.

For sections header, use H3.

### Challenges

Leave a space between # and the title.

Body

Leave an empty line between sub-headers and paragraphs.

Leave an empty line between text and tables.

### Schedule

|      |      |
| ---- | ---- |
| 8:00 | **Opening** |

Only link the same word or phrase once per article.

Avoid external links, except for databases, important papers and events.

Paths

For articles in the same directory, use a relative path.

She researches quality estimation with [Kevin Johnson](kevin-johnson.md) at Microsoft.

For articles in any other directory, use an absolute path.

She researches [quality estimation](/quality/quality-estimation.md) with Kevin Johnson at Microsoft.

For a specific section of an article, use the fragment identifier.

She researches quality estimation with Kevin Johnson at [Microsoft](/companies#microsoft).

Avoid trailing slashes at the end of links.

jekyllrb.com/docs/front-matter/

Mathematical notation

To render short mathematical formulas, like , use LaTeX notation and insert the LaTeX code after = in the below URL.

<img src="https://render.githubusercontent.com/render/math?math=[LaTeX Code]">

Include the complete URL to add the mathematical formula in the article.

<img src="https://render.githubusercontent.com/render/math?math=p(t_i|s_{1\ldots |s|}, t_{1\ldots (i-1)})">

Want to learn more?


Edit this article →

Machine Translate is created and edited by contributors like you!

Learn more about contributing →

Licensed under CC-BY-SA-4.0.

Cite this article →