While such set-level compression could be regarded as self-evident, we intend to complement this by averaging the visual complexity scores over the whole dataset for each generation. 

Part I: Analysis of historical data set

We have collected and tabulated individual Vai graphemes from 16 sources dated between 1834 and 2005. However, for the purposes of analysis our dataset is limited to the life histories of only the best-attested Vai graphemes. Thus, a Vai grapheme is included in the dataset if a) it is attested in more than 90 percent of the historical sources, and b) there are surviving examples in both the earliest sources (ca. 1834-1849) and the most recent source (2005). This constraint reduces our analysis to 62 graphemes. We derive a visual complexity score for each instantiation of the 62 graphemes across 16 time periods. 

Part II: The Bilerian experiment

The 'Bilerian' experiment involves several generations of participants who must learn to read and write six complex Bilerian symbols. These six symbols are selected from the full set of Vai graphemes as they were written in 1834. Those not attested in the 1834 manuscript are extracted instead from the 1849 sources. In a second part of the experiment the same six symbols are used, but are taken instead from the 2005 generation (Figure 5).