Machine Translation of Public Health Materials From English to Chinese: A Feasibility Study

Area of Science:

• Public Health Communication
• Medical Translation
• Health Literacy

Background:

• Limited English proficiency individuals in the U.S. often speak Chinese, yet Chinese public health materials are scarce.
• Previous research suggested machine translation plus human postediting (MT+PE) could be faster and cheaper than human translation.
• This study evaluated MT+PE for translating public health documents into Chinese.

Purpose of the Study:

• To assess the feasibility of using machine translation (MT) tools followed by human postediting (PE) for Chinese public health materials.
• To identify potential barriers and challenges in the MT+PE process for this specific language pair and context.

Main Methods:

• Collected 60 English-to-Chinese public health documents, originally human-translated.
• Translated English documents into Chinese using Google Translate (MT).
• Analyzed MT errors, then had native Chinese speakers with health backgrounds postedit a subset.
• Measured postediting time and compared MT+PE versions against original human translations using blinded quality raters.

Main Results:

• Common MT errors included word sense (40%) and word order (22%).
• Posteditors worked at approximately 41 characters per minute.
• Blinded raters consistently preferred human translations over MT+PE versions.
• Poor MT quality, lack of posteditor expertise, and inadequate instructions were identified as barriers.

Conclusions:

• The MT+PE approach showed limitations in producing quality Chinese public health translations.
• Improvements in MT technology and careful design of postediting protocols are necessary.
• Further work is required before MT+PE can be routinely adopted in public health practice for diverse language pairs.