Meet some network scientists!
Interviews by Jia Yan
For this installment of our featured partner piece, we interviewed several CLS faculty who use network science in their own research: Michele Diaz, professor of psychology and linguistics, who studies language and aging; Chaleece Sandberg, associate professor of communication sciences and disorders, whose interests center around language disorders and bilingualism; and Roger Beaty, assistant professor of psychology, who investigates creativity. Their responses provide an interesting window into how using networks like the one above can help us learn about a wide range of topics in language science.
Michele Diaz
Chaleece Sandberg
Roger Beaty
We are all members of our own social networks. We’re connected to the people we know, who may or may not be connected to each other. And the people we know also know other people, who we ourselves are not connected to directly. This type of network of relationships can be a useful way to think about many aspects of the world, including language. The illustration to the right, provided by one of our featured partners Roger Beaty, shows how our knowledge of words can be understood as a sort of network in our minds.
1. In your research, you have used network science. Could you explain for our readers what
that is?
Michele: Network science is a set of tools that allows you to look at how things are related to one another. It can be applied to any number of fields and situations: looking at social relations among individuals, how planes fly between airports, or how words are connected via semantic (meaning), phonological (sound), orthographic (written form), or other linguistic links.
Chaleece: Network science is the study of any system that can be represented as a network of interconnected objects. For example, in a social network, each person is a node in the network and relationships between people serve as the links that connect the nodes into a network form. I mainly study functional brain networks in which brain regions serve as the nodes, and similarity in their activation patterns over time serve as the links that connect the nodes. Network science relies on graph theory, which is the area of mathematics that provides the method for analyzing the relationships within a network. For example, graph theory tells us that the node degree—which is the number of links any one node has—can provide useful information. In a social network, the node degree would be how many friends one person has, which can be interpreted as how popular a person is. In a functional brain network, the ‘popularity’ of a brain region has implications for how important it is in processing information within that network.
Roger: Our lab studies creative thinking: how people creatively combine concepts to form new ideas. We use network science to understand how creative thinking relates to the organization of concepts in semantic memory (our general knowledge about the word). Network science tools can be used to ‘map’ the relationships between concepts in memory, and to study how these maps differ in people who are more or less creative.
2. Why do you think network science is useful in studying language?
Michele: Network science can provide models for how words might be related or how words might be represented differently in different groups of individuals. It allows you to look at a number of dimensions that we know are important in language—like semantic or phonological relationships. You can also combine many of these features in one network
Chaleece: I think network science could be broadly useful in studying language because of how it can characterize complex systems—and language is definitely a complex system. For example, there is a long-standing hypothesis, that concepts within the semantic system are linked by semantic similarity and that brain activation spreads among concepts via these links. By characterizing these semantic networks, we can better understand how their structure may influence different types of behavior that rely on access to semantic knowledge, including how this structure may be affected by both developmental (such as specific language impairment) and acquired language difficulties (such as aphasia).
Roger: Network science gives researchers powerful tools to study how language is structured in our minds. Networks are also intuitive for people to understand, I think. And network visualizations—depicting the relationships between words in the lexicon (e.g., related words being more connected to each other, as in the graph above)—often help to facilitate that understanding.
3. Could you share an interesting result from your work in network science?
Michele: Abby Cosgrove (current fifth-year graduate student of psychology) and Amy Lebkuecher (current postdoctoral fellow at the University of Pennsylvania and Moss Rehabilitation Research Institute) have been working on several different network science projects. Abby’s findings suggest that the semantic networks in older adults differ from younger adults in that they’re less efficient and have more sub-groups. This is an important finding because typically, measures of older adults’ comprehension show little age-related decline. For example, vocabularies are often larger among older adults compared to younger adults and older adults have very good reading comprehension. What Abby’s results suggest, though, is that these increases in vocabulary affect semantic network structure and may have a cost.
Chaleece: One interesting result that I have found in my research comes from a study of brain networks. I compared the networks of people with aphasia, an acquired language difficulty due to stroke or other brain injury, with those of age-matched adults without aphasia. I expected that people with aphasia would have less connected networks because of lesions that eliminated brain regions from networks. However, something I found that I did not expect was that some brain regions that were spared by the lesion and would normally be connected to a network were also disconnected. This suggests that brain regions that do not appear to be structurally affected by the lesion are functionally affected by the lesion, and that lesions have far-reaching consequences for brain networks.
Roger: We’ve found that highly creative people, or those who do well on verbal creativity tests, have a unique semantic memory network: they have many connections between concepts, which might make it easier for them to connect concepts when thinking creatively. This finding has turned up across many different ways of conceptualizing verbal creativity, such as metaphor, which involves connecting concepts that are usually not related (e.g., ‘time is money’). People who can produce more creative metaphors tend to show this interconnected semantic memory network structure.
4. Can you tell us about your experience as a language instructor? What language(s) do you teach and what kinds of students are in your classroom? What kinds of things do your students struggle with? What do you think motivates them?
Michele: Network science is a useful set of tools that can be used to help us understand almost anything. Advancing our understanding of how we use language can benefit people across the lifespan, as well as individuals who experience language difficulties.
Chaleece: Network science in general connects to our daily lives through things like search engines and power grids. For example, Google uses an algorithm based on the graph theoretical metric PageRank to optimize searches. My hope is that my research using network science will eventually help people with aphasia by informing treatment practices. I assume that most researchers who use network science in their research are like me and hope that it will translate into a practical benefit for society.
Roger: A type of semantic network that people can use in their everyday lives is called a ‘mind map’—a diagram for drawing the relationships between parts of a whole. Mind maps are often used in educational settings to visualize a student’s knowledge of a concept, and to see how that knowledge map changes as students learn (e.g., about science concepts). Another way to explore your own semantic network is through free association, by saying the first word that comes to mind in response to any other word. The Small World of Words is a neat project based on free associations from thousands of research participants. Anyone can explore this semantic network to see how most people make associations to different words, just as in the example illustration above.