Exploring semantic networks of words and morphemes, particularly, the way in which they are built, and how they expand as new words and morphemes are added to them, is a crucial step in analyzing the development of the language and, therefore, identifying the current tendencies in its evolution. As a result, the foundation for predicting its future development and carrying out a comprehensive analysis can be built. It is the openness of the semantic networks, however, combined with their radiality that makes it possible for semantic networks to evolve and incorporate an increasingly large number of words and morphemes, therefore, allowing or an unceasing evolution of the language.
By definition, semantic networks incorporate the words and morphemes that are distinctly related to each other and, therefore, create a semantic unit together. It should be borne in mind that the identified definition is rather loose, which allows for stretching the concept of similarity and incorporate several domains under the umbrella of a single semantic network (e.g., the morphemes that include people, phenomena, ideas, etc., such as “teacher-student,” “challenge – accomplishment,” etc.). Indeed, since the very concept of similarity cannot be viewed as an invariant construct, the opportunities for building large semantic nets are truly ample.
Furthermore, seeing that the relationships between the identified words and morphemes are represented in the form of a radial structure, an array of opportunities for including new lexemes and morphemes into the network remain open. Consequently, the foundation for an unceasing increase of the semantic network is created.
The concept of radiality in linguistics is fairly simple. In semantic networks, the relationships between the words and morphemes are represented as a labeled graph – or, to be more specific, a multigraph – where the relationships between the construals may vary based on the scale of the net. For instance, the radial structure may imply a single-level relationship (e.g., a part-of one, such as “a door – a doorknob,” etc.). However, expanding it further, one may add new layers of meaning to the links between the elements included in the network.
The semantic indexing used to determine the correlation between the elements of the network, in turn, suggests that concepts should be true as opposed to terms (e.g., “weather” instead of blizzard“). The identified approach toward marking the correlation between the construals of a single network helps handle the problems associated with synonymy, at the same time, simplifying the overly complicated relationships between the contents of the network.
Therefore, using the identified method, one may make the existing semantic networks blend and integrate into a single entity by changing the proximity between the construals and determining the semantic similarity to the rest of the elements of the net (e.g., “parent-child,” “adult – child,” etc.). As a result, the classification framework rooted in the concept of perceived similarity can be built, allowing one to place the construals into different categories based on their proximity to the center of the semantic net.
Seeing that the opportunities for locating similarities between different construals based on their semantic proximity are practically endless, the idea of a semantic network as the entity into which an increasingly large number of new phenomena can be incorporated based on their perceived similarity is quite sensible. Indeed, with the addition of new semantic links to the semantic net, one is likely to have a continuously expanding entity that engulfs new construals as it grows and develops new connections between the ones that already exist in it. Taking the relationships between the elements included in the net on a new level, one will create the premises for a consistent and unceasing expansion of the semantic network, therefore, maximizing its potential.
One must admit that there are problems with creating a clear and accurate taxonomy of the relationships between the construals within a specific semantic net. For instance, the calculation of the degree of similarity between the construals based on the “is-a” framework may lack accuracy. Thus, the similarity levels are often identified based on the existing examples as opposed to considering words and morphemes on a case-by-case basis. Consequently, there is a significant degree of approximation in the outcomes of developing a semantic network.
Furthermore, some parts of the net where the construals are linked closer to each other than in other areas may have a higher density, which may lead to a disruption in the links between the elements of the semantic network. Nevertheless, the identified issue creates leeway for building increasingly large semantic nets. Furthermore, the very concept of a semantic network is finite can be questioned once the principles of radiality and semantic similarity are deployed into the process of constructing semantic nets.
Therefore, it can be assumed that perceived similarities can be deemed as the tool for assimilating semantic networks into a single entity. As a result, a large and comprehensive system can ostensibly be built with the help of the said framework. In the latter, all construals will be categorized based on their relation to the focus of the net and, thus, have strong ties with the rest of the words and morphemes. In other words, the phenomenon of perceived similarities helps build connections between any semantic units and, therefore, create large networks.