**3. Materials and Methods**

Exemplar student concept maps (here constructed by students during school science lessons) are translated into commentaries on the types of knowledge depicted by converting the linking phrases between concepts into descriptions of their semantic density and semantic gravity (see Figure 2). In terms of semantic gravity (SG), each proposition is considered in relation to the way in which the student has articulated the degree to which the knowledge is either tied to a particular context (SG+) or o ffers a more generalizable view (SG-) (see Table 1). We distinguish here between knowledge that is very context bound (SG++) and that which is less tightly bound (SG+) to o ffer a more nuanced description of the knowledge quality. Propositions are also evaluated according to the semantic density that is depicted (SD), where students may be using simple, everyday descriptions in their explanations (SD-) or may be offering much more technical summaries that exhibit considerable condensation of meaning (SD+). Again, the degree of condensation is considered by using SD+, SD++, SD- and SD– (see Table 1). In this way, each of the quadrants of the semantic plane itself has four sub-quadrants into which propositions may be plotted, giving up to 16 variants across the semantic plane. Once each proposition has been translated to indicate its semantic profile (SD±SG±), this is then plotted on the semantic plane (Figure 2) to indicate the semantic range depicted within the map. When using this method, researchers may need to establish the degree of inter-rater reliability to decide on ++ or + and on – or -. In this study we had three authors who were familiar with the content and agreed upon the level of density and gravity within each proposition.

**Figure 2.** A three step process of map construction (1), translation (2) and plotting (3) on the semantic plane.

### **Table 1.** Proposition analysis translation device. Modified from [44].


**Table 1.** *Cont.*

