How to Automatically Translate Text When Classifying Text Data

This article describes how to translate text into a different language when using Displayr's text categorization tools. For example, you can automatically translate Chinese text:  

...to another language such as English:

...and then classify the responses:

If you'd like to create new translated variables instead of classifying the translated text, please see: How to Automatically Translate Text Variables into Other Languages.

This article is broken out into the following sections:

• Method - Translate text containing a single language
• Method - Translating text containing multiple languages
• Method - Classify the translated text
• Technical Details

Requirements

• A data set with a text variable.
• If translating from multiple languages, you have the option to use a variable that stores the language for each text response. Click here for a list of the currently supported languages.
• Familiarity with Displayr's text categorization tool. For more information see: How to Classify Text Data

Method - Translate text containing a single language

The assumption is that all the text is in a single language.

1. Select the text variable from the Data Sources tree.
2. Hover and click + > Text Categorization.
3. Select one of the following options:
   1. If you want to classify responses into multiple themes, select Multiple themes.
   2. If you want to classify responses into a single theme, then select Only one theme.
4. Click Start.
5. To begin the translation, click Translate.
6. From the Source language dropdown, select the original language. You can select a specific language from the list, select Automatically detect language (note that this passes your data through Google to be translated), or select Specify with a variable. This last option allows you to use an existing variable in your data set to determine the language.
7. From the Output language dropdown select the language that you want to translate to.
   
8. Click Translate.
9. Click OK when you see a message confirming the language that was translated.
10. The translation will run and you will see "[Translated]" in front of each text response that was translated:

To see the original, untranslated text of a response, hover over the response. The original text will appear in a tooltip.

Method - Translate text containing multiple languages

This method allows you to translate text responses that come in a variety of languages all at once.

1. Select the text variable from the Data Sources tree.
2. Hover and click + > Text Categorization.
3. Select one of the following options:
   1. If you want to classify responses into multiple themes, select Multiple themes.
   2. If you want to classify responses into a single theme, then select Only one theme.
4. Click Start.
5. To begin the translation, click Translate.
6. From the Source language dropdown, select one of the following options:
   1. Automatically detect language (note that this passes your data through Google to be translated).
   2. Specify with a variable. This last option allows you to use an existing variable in your data set to determine the language.
7. From the Output language dropdown select the language that you want to translate to.
8. Click Translate.
9. Displayr will confirm the languages that were translated. Click OK.
10. The translation will run and you will see "[Translated]" in front of each text response that was translated:

To see the original, untranslated text of a response, hover over the response. The original text will appear in a tooltip.

Method - Classify the translated text

Once you have translated the original text, you can classify the text responses.

1. Click Create to create AI-generated themes.
2. Click Classify to classify the text responses into the themes.
3. Further refine the categorization using the steps found in How to Classify Text Data.
4. Click Save.

A new variable will appear in the Data Sources tree with "Categorized" in the name. You can use this variable for further analysis. For example, drag it onto a Page to create a summary table:

Technical Details

Translation

If Automatically detect language is selected, the source text is translated into the desired language by Google.

Automatic source language detection

If you choose to have Google automatically detect the source languages, the detected languages are summarized after translation. You may choose to cancel and re-translate with different settings if you are unhappy with the detection results.

Automatic translation for categorizations with translation settings

Once you decide on the translation settings in the translation dialog and save the categorization, in any future editing sessions of this categorization, all responses are automatically translated using the saved categorization settings.

Next

How to Classify Text Data

How to Automatically Translate Unstructured Text in Other Languages with Automatic Categorization

How to Automatically Translate Text Variables into Other Languages