Typography can display hierarchy in various ways. In Typography they are used to display relationships between text elements. We examine these presentation in the context of data visualization and how they can used. One could argue how these typographic style can be considered as graphics, but we propose for a wider understanding of visualizations with regard of the idea of invisible design and micro visualizations.
Most forms can be found in everyday life and therefor can be considered as learned. One point to keep in mind is that when text is also used for labeling the distinctions should always be applied with an eye on legibility.
Some of the most commonly used forms of displaying hierarchy in typography are even-handedly adapted from the encoding of other geometric shapes. Just like in bar charts the words or even single letters can be scaled according to the underlaying data. A recommended tool for harmonic text scaling is Type Scale by Jeremy Church.
Similar adoptions are the use of opacity, saturation or filling. Just like with classical elements these encodings transport a hierarchy, but fail to make the represented data precisely »readable«. Especially when using lighter fonts one should not use brighter gradations and not more than five shades to make them clearly distinguishable.
One example of filling is »The Cellphone Revolution« by GOOD and Fogelson-Lubliner.
Hierarchical classification based on the y-axis is actually quite common in scientific fields. For example chemist combine it with scaling for their nomenclature. This way the two layers clearly represent different hierarchal level and also allow a greater legibility for each element.
One aspect to consider when shifting on the y-axis is the undefined »bottom« of the letter, especially when working with letters with descenders. Therefor we would suggest to only work with capital letters.
The shift of words on the x-axis is also used frequently. Nearly every list makes use of this now learned encoding. But one must pay intention when the visualization is used globally due to the contrary reading direction in Arabic and Hebrew texts.
One similar approach but inappropriate for absolute values is the shift single letters on the x-axis, hence the kerning. Thus this form is often found in artistic fields like the »Concrete poetry«.
One example for this form where the visualization was used for text as data source is the book »Extremely Loud and Incredibly Close« by Jonathan Safran Foer.
There two unique forms of encoding data with text: fonts and the combination of capitals, small capitals and lowercase letters.
Even though one depends on the availability of different weights, it is a promising approach to create hierarchy. In contrast the use of different letter types is limited to three forms and can not as clearly be hierarchically ordered. This is an example where encoding should only be used complementary, because absolute can not be extracted from the representation.
Combining different typefaces may be tendering, but should not be used for a specific order, but for categorical data.
Overall we suggest to combine different distinctions to clarify the hierarchical order of the displayed text elements. Although we advocate the use of type to convey the underlaying data, we recommend to only support the already – and hopefully more precise – display of the data in a scientific environment.