This figure shows a cloud with the following text content:

Dictionaries Model (additions to the base layer provided by target manufacturer/prov. or 3rd parties)

Presentation-specific and other UI components, and alternative texts:

• Text translations (for labels, help texts)
• Alternative labels, help texts (language-spec.)
• Additional semantic tagging Images, icons, video clips, sound files (may be language-spec.)
• Pronunciation information (language-spec.)
• Concrete UI component implementation (e.g. JAR file, flash code)
• Synonyms for natural language interaction (language-specific)
• Grammar for voice input (language-specific)
• Additional semantic tagging

Other considerations

• Metadata tagged to each dict entry describes applicability, media dependencies, etc.