At a basic level, sonification design might look at:
One might use volume or pitch to represent quantity of data. The more of something, the higher the pitch or louder volume.
The tempo of a piece could be mapped to time in a piece. In the yield curve sonification earlier, the dates were filtered so only a few are sonified to allow the listener to understand the data.
Direction or progression might be shown using panning to move from one speaker to another.
These provide a basic form of sound design. We might take a more design approach and think about other aspects to make the sonification richer. We could use alter the sound, such as adding a lower tone to the sound to make it a bit deeper, or we can use the other types of sound.
Here you might think about the type of data to be sonified and the approaches, such as qualitative - such as categories - or quantitative - like discrete numbers or continuous values like time series data.
You might think about other approaches like earcons, audicons, or even samples. Here you are using more complex sounds and beginning to interpret sounds.
We could use audio graphs, such as line charts, to show relationships over time or more in depth sounds, such as using noise for sonifications of networks, like some earlier work on JANET. (Emsley, De Roure and Chamberlain, A, 2017). Two approaches are taken here: firstly, the noise of the traffic and secondly, a more audio graphic version with a voice over to contextualise the data.
Another way is to create layers of sound to contextualise data. In the UnheardCity work, sonifications of the data were overlaid with sound recordings to provide some context from the walk. https://unheardcity.org.uk/sounds/soundscape.html
A question that you might want to ask is how the listener will understand what you are sonifying. Do they have access to a description of the sounds?