The PhD students looked at ways of incorporating rhetorical styles into their poster presentations. They were best at using the rule of three for repetition, but clearly need lots of practice in creating shorter, more powerful parallel phrases.

I demonstratrated the power of cutting out needless repetition through this correction (which is still not ideal):

  • To apply learning methods on our data sets we are looking for methods to group continuous data into discrete data. Such discretization methods are optimal if as little information as possible is lost and the discretized data still reflect the dependency structure.
  • Grouping continuous data into discrete data ideally requires methods that retain as much information as possible while still reflecting the dependency structure.

These were the phrases they came up with, which they practiced saying/ reading aloud:

rule of three

  • When I look for paleo-earthquakes in a certain area, I want to learn: Did big events happen, how big were they and how often did they occur?
  • Because there is such a deadlock in international climate negotiations, it is important to look at the levels below, namely the regional, the national and the local levels. (Use hands to scope from large to small, to express that region is larger than nation.)
  • Bayesian networks are a great tool since they help to discover dependency structures, to understand complex processes, and to communicate them to experts and non-experts.
  • Health depends on the fulfillment of physiological needs, the provision of adequate infrastructure, and the protection from disease exposure. (This nominal style needs rephrasing using verbs for spoken English: People can be considered healthy when their physiological needs are met, they are provided with an adequate infrastructure, and they are protected from exposure to disease.)
  • Finding alternatives to standard interpolation-based approaches allows us to stick with the original data, to retain the variance of the processes, and to adjust easily to different data qualities.
  • Interception cannot be measured; So we collect throughfall, we measure rainfall, and we subtract throughfall from rainfall.

parallel structure

  • Health is not simply the absence of disease, but in fact results from the presence of beneficial conditions. (This is contrast rather than parallel structure; a good example of how difficult it is to boil complex ideas down to simple phrases.)
  • There are two ways of looking at climate politics: One is the program, or policy; the other is its administration, or organization.

Next time I teach giving presentations, I’ll add logical shift: a change or movement in a piece resulting from an insight gained by the speaker. I’m just starting out, and so don’t have models and phrases from the students’ writing to work with yet. Work in progress.