Prof. Barbara J Mills — "Culture as Networks of Interaction and Transformation: Multiscalar Perspectives"
Duration: 24 mins 57 secs
Share this media item:
Embed this media item:
Embed this media item:
About this item
| Description: | A recording of Professor Barbara J Mills (University of Arizona) speaking on "Culture as Networks of Interaction and Transformation: Multiscalar Perspectives" as part of the "Cultures at the Macro Scale" seminar series. |
|---|
| Created: | 2021-02-18 13:30 |
|---|---|
| Collection: | Culture at the Macro-Scale |
| Publisher: | University of Cambridge |
| Copyright: | Barbara J Mills |
| Language: | eng (English) |
| Keywords: | Archaeology; Networks; Cultural Boundaries; |
Transcript
Transcript:
00:02
Hi, my name is Barbara Mills and I'm delighted to be participating in the seminar "Culture at the Macro-Scale: Boundaries, Barriers and Endogenous change."
00:13
Today I'm going to talk about culture as networks in interaction and transformation using multi-scalar perspectives. I'd like to acknowledge that this is really a collaborative project, and I especially want to thank my co-PIs, Jeffrey Clark, Matt Peeples and John Roberts, who have been instrumental in much of the work I'll be discussing today. I'd also like to give a shoutout to our latest iteration in the project "Cyber Southwest", where much of the data that I'm going to be talking about today, is now available along with some tools for analysis—many of which are still under development—but I hope that some of you can log on and take a look.
00:55
Today, I'm going to provide a couple of case studies of how network analysis can be useful for the seminar's goals. Before doing that, however, I will briefly frame why network analysis is important to use, and then talk about how archaeological networks are constructed. I'll end with some short responses to the seminar questions that I was provided.
01:24
One of the first questions that arises is, of course, "What is culture?" This composite definition would sit well with most anthropologists in the 20th century, and even into the 21st. Unfortunately, archaeologists use of the term culture was within a culture history paradigm that took the anthropological definition much too literally. "Cultures" were seen as spatially bounded (and that resulted in the use of culture areas) and temporarily bounded (resulting in phases). Our archaeologists were pretty good at recognising when and where architecture, artefacts, etc. changed, but that change was largely seen as created through external sources.
02:11
Gary Feinman and Jill Neitzel have neatly summarised this dilemma in their recent article—"Excising Cultural History from Contemporary Archaeology."—and especially in this table. Their programmatic statements emphasise the relational sociality of human life, the permeability of boundaries, the usefulness of social networks for modelling these webs of relationships, and the importance of taking multi-scalar approaches. And much of what they have to say it's not new, of course.
02:43
Kroeber famously depicted these two trees to illustrate biological evolution on the left and cultural evolution on the right. It is too bad that the reticulated model on the right was not more closely followed by archaeologists, for what we have inherited from our ancestors is a static view of culture. It is now becoming clear that even biologists are questioning the branching, non-reticulated structure, as evidenced in new analytical models for genetics that acknowledge processes of recombination.
03:17
The alternative is to use network models, including models that allow for recombination over time. However, I don't think this was the motivation of early adopters of network approaches in archaeology. For many of these early applications, it was the possibility of applying methods being productively used in other disciplines to better understand the structure of interactions that goes beyond spatial adjacency. As this citation network shows, there was an eclectic use of sources from complexity scientists, sociologists, anthropologists, geographers, actor-network theorists, computer scientists and physicists. The main sources that link these early adopters were I think the physicists, computer scientists and sociologists—all the fields that have been at the forefront of network science, and representing a convergence toward network applications.
04:12
Today, archaeological applications have grown exponentially, and there are considerably more archaeologists who use network approaches, as illustrated in a more recent citation analysis shown on the right. And important texts that many of us come back to is Carl Knappett's 2011 book "An Archaeology of Interaction", which was important for bringing up the issue of scale by looking at micro-, meso-, and macro- approaches to network formation; for connecting those scales to theories of communities of practice and regional exchange; and how at each scale the connections or ties are differently envisioned. His work is also significant for juxtaposing theories of relationality and materiality with formal network representations—a tension that is still present today.
05:01
Formal network analyses are a structural approach, or perhaps a neo-structural approach, made up of nodes (or vertices) and ties (or edges). In social network analysis, the nodes may be actors, households, villages, etc., and the connections between them of kinship, friendship, work setting, religious affiliation and other social relationships. Translation to archaeology is required and most applications have taken settlements as nodes and the connections between them using material culture—as I do in this video. Data for constructing archaeological networks include historical, spatial and material sources. Each of these has been used in archaeology, with spatial and material networks the most common.
05:51
It is useful to think about attributes that represent different parts of the "chaine operatoire", such as production, distribution and consumption. Using pottery for example, archaeologists have used temper types, and construction techniques to represent shared communities of practice and learning; or decoration and special forms such as effigies, as representing shared ideologies, or social signalling.
06:16
Using assemblages of decorated pottery wares, as my team has done, we are looking at consumption: the cumulative choices of what kinds of containers to use to store prepare, and serve food and water. In the Southwest, these decorated wares are defined by different fabrics, so they are to some extent indicating zones of production, but they are also defined by their slip and paint colour combinations, which are more active indicators of social affiliation.
06:47
We use the percentages of all decorated pottery wares, and compare every possible pair of assemblages, to derive a similarity score. There are many different similarity indices, but the Brainerd-Robinson is most commonly used for nominal data. This produces a weighted tie strength ranging from zero to one, which is then used to derive network metrics. To illustrate or visualise, these have to be binarized. But weighted similarity scores are preferable over simple presence/absence data in calculations. I'm not going to discuss all of the steps in detail, but this slide summarises the protocols with references for those interested in the methodological details. Suffice to say that one needs a method of chronological control to divide assemblages into periods, and then a similarity index to construct the networks and calculate different network statistics at the node and whole network levels. For individual nodes, there are several different measures of centrality. And at the whole network level different measures of cohesion—such as density—all of which are available in packages, such as UCINET. Two other measures that we've applied, and which I will illustrate, are a community detection algorithm for identifying subgroups, and the identification of connections that are composed of weak in addition to strong ties.
08:16
Before getting into the case studies, I wanted to summarise why a network approach is so beneficial. First and foremost is to be able to identify similarities that are not tied to categories—such as culture areas—also called the "anti-categorical imperative". Second is to understand which nodes are more connected, or central, within a network based on their position, as well as to characterise overall network structures. Third, is to be able to look at networks dynamically over time. And finally, we can compare those networks to other dimensions of our archaeological cases, such as demography, environmental changes, and social inequality, to better understand culture change.
08:59
Now for our two case studies. Both build on a database of archaeological sites that included spatial locations and site size. The first one I'm going to talk about is the Chaco Social Networks database, shown in light red on the slide, which is composed of sites with monumental architecture called "great houses" and "great kivas" constructed between AD 800 and 1200. It covers the beginning, peak and end of the Chaco world, which is located in the San Juan Basin of northwestern New Mexico, and surrounding areas, including the Mesa Verde, Middle San Juan, Chaco and Upper Little Colorado areas. We wanted to know which great houses were interacting with each other more than others.
09:47
The second case study, called the Southwest Social Networks study dates to about A.D. 1200 to 1500, and includes sites with more than 12 rooms west of the Continental Divide, outlined here in dotted blue. Both of these case studies are punctuated by well-documented major droughts, migration, large-scale cooperative building programmes, and changes in degree of inequality.
10:19
As I mentioned for the Chaco World, we had a number of research questions, including "Which great houses were interacting with other great houses?" And also, "Was there evidence for subgroups within the networks?". We wanted to know how central were Chaco Canyon sites in the network, and when and how changes in network support or do not support different migration scenarios, especially for the origins and demise of the Chaco World.
10:52
Here's a map showing the location of the Chaco World in the southwest, with the locations of all great houses and great kivas. At the centre is Chaco Canyon, which includes Pueblo Bonito: One of the largest single buildings ever constructed in the pre-Hispanic Southwest with about 800 rooms, elite burials and dense ritual deposits. The canyon itself has the densest concentration of great houses and great kivas. (A close up of those locations is in the bottom right). We were able to compile ceramic data for about 75% of all of these structures.
11:32
For each 50-year interval, we constructed networks based on the protocols discussed earlier, with visualisations based on strong ties of 75% to 100% similarity. I'll be showing a series of these figures, and in each the size of the nodes indicates how central in the network each building is based on eigenvector centrality (which is similar to Google PageRank algorithms). The colours of the nodes are based on a community detection algorithm. The lightest blue ties connect nodes that are spatially closer and darker are those more distant. For reference, known Chaco roads are shown in dotted yellow. These roads are not well dated, and we wanted to see when they coincide with network ties between the sites that are along them.
12:25
The resulting series of 50-year interval snapshots is shown here. And it resulted in a number of new insights into how the Chaco World changed over time. This includes the early centrality and density of sites in Chaco Canyon, the presence of strong ties between Chaco Canyon sites with sites to the south and the southwest, and the early and continued differentiation of sub areas. In particular, the Mesa Verde area of Southwestern Colorado was largely disconnected from the rest of the network, but highly cohesive on its own.
13:06
Sites in the Middle San Juan are shown with yellow dots. (Here you can see the yellow dots kind of go in and out of the graph, and they come back towards the end of the sequence as that particular area becomes more important.) The Middle San Juan is the area where the longest road segment connects it with Chaco Canyon, called the North Road, including many of the settlements in the area called the Aztec Complex.
13:40
Here's a map showing the distribution of some of the monumental architecture from the Aztec complex. And here's the Great North Road in an aerial photograph, and then also showing the connectivity of the road with the sites along it in this other map of Chaco routes.
14:05
So getting back to the snapshots, the Great North Road is completely covered by network ties in the A.D. 1000 and 1100s, when archaeologists proposed that a migration occurred out of Chaco Canyon, joining settlements in the Middle San Juan. Moreover, sites in Chaco Canyon become more heterogeneous, with some sites being more closely tied to sites in the Middle San Juan than to their neighbours. This also fits with a hypothesised migration of people from the Northern San Juan (Mesa Verde) area to Chaco. A major drought in the early 1100s is often brought up to explain this movement.
14:48
But why move to Chaco during this time? We now know that violence was a big part of the Northern San Juan settlement life and likely a major factor in this realignment of alliances among great house residences. The entire four corners, including all the sub areas shown in the Chaco network plots—with the exception of the southernmost area—was depopulated in the last half of the 1200s.
15:17
Our next case study looks at networks in the post-Chacoan Southwest at an even larger scale than the Chaco World, and encompasses most of that area with the exception of the Mesa Verde area. Site locations and site sizes for all known sites greater than 12 rooms had already been compiled by some of our team at Archaeology Southwest, showing how population densities changed across the Southwest. The depopulation out of the four corners by A.D. 1300 clearly shows up in these panels. And one of our major questions was how is population movement affecting networks of interaction in the western Southwest, covering an area that has traditionally been divided into Ancestral Pueblo, Mogollon, and Hohokam areas. This is the four corners area up here. And you can see that this dense occupation is pretty much disappeared by 1300.
16:22
The resulting snapshots for the A.D. 1200 to 1500 case study illustrates the rewiring of connectivity with population movement. The first two 50-year intervals show several subgroups' disconnected components, including the post-Chacoan sites that still remained in the highlands surrounding Chaco Canyon. As I mentioned, Chaco Canyon itself was largely depopulated by this time. The largest connected component or subgroup was in the Upper Little Colorado area, which had already seen a major movement of people from the Chaco area—so this is the largest connected component in these illustrations.
17:03
And by 1300, the population of the Four Corners region was complete and there was a rapid transformation, the settlements in the southern Southwest becoming more densely connected. A second major event occurred around 1400, as sites in the Transition Zone between the Colorado Plateau and the basin-and-range country to the south would be populated. (So this is the transition zone. This is the Colorado Plateau. This is the basin-and-range. And these sites are in a transition zone here.)
17:37
The demographic effects on social networks are shown in the series of close-ups by 50-year intervals, starting with the A.D. 1200 to 1250 intervals shown here. So this is again—these are the same snapshots. This is the area surrounding Chaco Canyon, this is where Chaco Canyon is you can see that it's completely depopulated by this time. This is a mountain range, the Chuskas that are to the west of Chaco, and then middle San Juan. And here's our big connected component. The area to the south, where many people we think from Chaco migrated to at 1250 to 1300 interval shows also that Chaco had completely emptied. And the remaining sites in the far western area form another separate subgroup or connected component. There's one down in the Phoenix area and then one in the Phoenix basin, excuse me, the Tucson basin.
18:47
For the interval of 1300 to 1350 population declined and networks realigned to the south. This realignment continues into the 1350 to 1400 period. And it's pretty much a maximum build-up of population—and also the greatest connectivity, in which the Hopi area is connected all the way down to the Hohokam area. This is the Zuni area. And it's still connected to areas just north of the Transition Zone. So this is actually Petrified Forest region here and some very long connections to the Verde Valley.
19:33
And finally, in the last period, after 1400, most of the Transition Zone was completely depopulated, another migration happened—and population sizes were plummeting throughout the Southwest at this time. Only the Hopi and Zuni areas are left at the end.
19:55
So we were curious as to the impact of spatial propinquity on network connectivity. For the earliest period of 1200 to 1250, when settlements were most widely distributed across the landscape, there was an unexpectedly higher correlation. But the northern and southern Southwest showed very different correlations over time. With increasing coalescence in the north, the correlations predictably decreased over time, and on the whole spatial and social propinquity were positively correlated. In the southern Southwest, however, increasing coalescence resulted in a slight increase in connectivity at 1300 when migration from the northern areas have been well documented. But on the whole, spatial and social propinquity were negatively correlated. In other words, more settlements that were far away from each other, were more well connected.
20:48
So the explanation for this divergence comes from looking at the pottery that drives the similarity coefficients for the southern Southwest. The major decorated ware in the southern Southwest, was the ideologically charged "Roosevelt Red" ware shown here. These were initially made by northern migrants who moved into the Transition Zone and southern Southwest. Patricia Crown's work established that these ceramics were part of a new religious ideology introduced in the early 1300s, which included designs such as this plumed serpent. She also showed that they were made in most areas that they were found. So it was the knowledge of how to make them that moved, not the pots themselves. That is, people moved, and were connected with each other and shared the information.
21:38
All of what I've talked about so far, has been based on strong ties. To wrap up, I'd like to take a minute or two to talk about what weak tie analyses have shown, as I think they provide important insights on the topics of boundaries, and permeability of boundaries. The sociologist Mark Granovetter is famous for his insights on the strength of weak ties. Weak ties are those on the low end of similarity indices in our work. According to his theory, they are strong because they can provide access to people and resources outside of their regular strong tie groups, which tend to be those based on residence, kinship, and friendship.
22:23
When we look at where weak ties tend to occur in the Southwest, they are in those areas with lower population densities, and in physiographically transitional zones. Materially they co-occur with high architectural diversity, including multiple kinds of ritual spaces, high ceramic ware diversity, and a high degree of innovation in ceramic technology, such as new paints, colours and decorative styles.
22:51
These are also areas in which small group migration and intermarriage have been proposed. Migration into these areas fits with a model of the internal frontier—but with an important corrective to Kopytoff's original formulation that was based on a world-systems core/periphery model in which frontiers were diluted representations of metropoles. Instead, we follow new interpretations, such as by Ogundiran, who sees these zones as areas of population diversity and innovation. The strength of weak ties for us is in pointing out how boundary or edge regions are important for encouraging culture change.
23:39
In closing, I offer these very short responses to the questions that were posed by the seminar organisers. I think that cultural groups should be distinguished from one another relationally, through their connections. Archaeologically, this is based on similarities in material culture, whether using production, distribution, consumption, and or symbolic attributes, which are even more powerful when combined with spatial data.
24:06
"How can cultural groups be identified from data?" Through clusters of relations, such as community detection algorithms on networks and identification of different network components.
24:19
"How should cultural barriers and borders be understood?" I think they should be understood as dynamic zones of interaction and innovation, not within core-periphery models, but as internal frontiers that are demographically and culturally diverse.
24:37
As I mentioned, we have a large team of collaborators and I'd like to acknowledge all of those who have participated in this work. Thank you so much for listening and I look forward to the seminar.
Hi, my name is Barbara Mills and I'm delighted to be participating in the seminar "Culture at the Macro-Scale: Boundaries, Barriers and Endogenous change."
00:13
Today I'm going to talk about culture as networks in interaction and transformation using multi-scalar perspectives. I'd like to acknowledge that this is really a collaborative project, and I especially want to thank my co-PIs, Jeffrey Clark, Matt Peeples and John Roberts, who have been instrumental in much of the work I'll be discussing today. I'd also like to give a shoutout to our latest iteration in the project "Cyber Southwest", where much of the data that I'm going to be talking about today, is now available along with some tools for analysis—many of which are still under development—but I hope that some of you can log on and take a look.
00:55
Today, I'm going to provide a couple of case studies of how network analysis can be useful for the seminar's goals. Before doing that, however, I will briefly frame why network analysis is important to use, and then talk about how archaeological networks are constructed. I'll end with some short responses to the seminar questions that I was provided.
01:24
One of the first questions that arises is, of course, "What is culture?" This composite definition would sit well with most anthropologists in the 20th century, and even into the 21st. Unfortunately, archaeologists use of the term culture was within a culture history paradigm that took the anthropological definition much too literally. "Cultures" were seen as spatially bounded (and that resulted in the use of culture areas) and temporarily bounded (resulting in phases). Our archaeologists were pretty good at recognising when and where architecture, artefacts, etc. changed, but that change was largely seen as created through external sources.
02:11
Gary Feinman and Jill Neitzel have neatly summarised this dilemma in their recent article—"Excising Cultural History from Contemporary Archaeology."—and especially in this table. Their programmatic statements emphasise the relational sociality of human life, the permeability of boundaries, the usefulness of social networks for modelling these webs of relationships, and the importance of taking multi-scalar approaches. And much of what they have to say it's not new, of course.
02:43
Kroeber famously depicted these two trees to illustrate biological evolution on the left and cultural evolution on the right. It is too bad that the reticulated model on the right was not more closely followed by archaeologists, for what we have inherited from our ancestors is a static view of culture. It is now becoming clear that even biologists are questioning the branching, non-reticulated structure, as evidenced in new analytical models for genetics that acknowledge processes of recombination.
03:17
The alternative is to use network models, including models that allow for recombination over time. However, I don't think this was the motivation of early adopters of network approaches in archaeology. For many of these early applications, it was the possibility of applying methods being productively used in other disciplines to better understand the structure of interactions that goes beyond spatial adjacency. As this citation network shows, there was an eclectic use of sources from complexity scientists, sociologists, anthropologists, geographers, actor-network theorists, computer scientists and physicists. The main sources that link these early adopters were I think the physicists, computer scientists and sociologists—all the fields that have been at the forefront of network science, and representing a convergence toward network applications.
04:12
Today, archaeological applications have grown exponentially, and there are considerably more archaeologists who use network approaches, as illustrated in a more recent citation analysis shown on the right. And important texts that many of us come back to is Carl Knappett's 2011 book "An Archaeology of Interaction", which was important for bringing up the issue of scale by looking at micro-, meso-, and macro- approaches to network formation; for connecting those scales to theories of communities of practice and regional exchange; and how at each scale the connections or ties are differently envisioned. His work is also significant for juxtaposing theories of relationality and materiality with formal network representations—a tension that is still present today.
05:01
Formal network analyses are a structural approach, or perhaps a neo-structural approach, made up of nodes (or vertices) and ties (or edges). In social network analysis, the nodes may be actors, households, villages, etc., and the connections between them of kinship, friendship, work setting, religious affiliation and other social relationships. Translation to archaeology is required and most applications have taken settlements as nodes and the connections between them using material culture—as I do in this video. Data for constructing archaeological networks include historical, spatial and material sources. Each of these has been used in archaeology, with spatial and material networks the most common.
05:51
It is useful to think about attributes that represent different parts of the "chaine operatoire", such as production, distribution and consumption. Using pottery for example, archaeologists have used temper types, and construction techniques to represent shared communities of practice and learning; or decoration and special forms such as effigies, as representing shared ideologies, or social signalling.
06:16
Using assemblages of decorated pottery wares, as my team has done, we are looking at consumption: the cumulative choices of what kinds of containers to use to store prepare, and serve food and water. In the Southwest, these decorated wares are defined by different fabrics, so they are to some extent indicating zones of production, but they are also defined by their slip and paint colour combinations, which are more active indicators of social affiliation.
06:47
We use the percentages of all decorated pottery wares, and compare every possible pair of assemblages, to derive a similarity score. There are many different similarity indices, but the Brainerd-Robinson is most commonly used for nominal data. This produces a weighted tie strength ranging from zero to one, which is then used to derive network metrics. To illustrate or visualise, these have to be binarized. But weighted similarity scores are preferable over simple presence/absence data in calculations. I'm not going to discuss all of the steps in detail, but this slide summarises the protocols with references for those interested in the methodological details. Suffice to say that one needs a method of chronological control to divide assemblages into periods, and then a similarity index to construct the networks and calculate different network statistics at the node and whole network levels. For individual nodes, there are several different measures of centrality. And at the whole network level different measures of cohesion—such as density—all of which are available in packages, such as UCINET. Two other measures that we've applied, and which I will illustrate, are a community detection algorithm for identifying subgroups, and the identification of connections that are composed of weak in addition to strong ties.
08:16
Before getting into the case studies, I wanted to summarise why a network approach is so beneficial. First and foremost is to be able to identify similarities that are not tied to categories—such as culture areas—also called the "anti-categorical imperative". Second is to understand which nodes are more connected, or central, within a network based on their position, as well as to characterise overall network structures. Third, is to be able to look at networks dynamically over time. And finally, we can compare those networks to other dimensions of our archaeological cases, such as demography, environmental changes, and social inequality, to better understand culture change.
08:59
Now for our two case studies. Both build on a database of archaeological sites that included spatial locations and site size. The first one I'm going to talk about is the Chaco Social Networks database, shown in light red on the slide, which is composed of sites with monumental architecture called "great houses" and "great kivas" constructed between AD 800 and 1200. It covers the beginning, peak and end of the Chaco world, which is located in the San Juan Basin of northwestern New Mexico, and surrounding areas, including the Mesa Verde, Middle San Juan, Chaco and Upper Little Colorado areas. We wanted to know which great houses were interacting with each other more than others.
09:47
The second case study, called the Southwest Social Networks study dates to about A.D. 1200 to 1500, and includes sites with more than 12 rooms west of the Continental Divide, outlined here in dotted blue. Both of these case studies are punctuated by well-documented major droughts, migration, large-scale cooperative building programmes, and changes in degree of inequality.
10:19
As I mentioned for the Chaco World, we had a number of research questions, including "Which great houses were interacting with other great houses?" And also, "Was there evidence for subgroups within the networks?". We wanted to know how central were Chaco Canyon sites in the network, and when and how changes in network support or do not support different migration scenarios, especially for the origins and demise of the Chaco World.
10:52
Here's a map showing the location of the Chaco World in the southwest, with the locations of all great houses and great kivas. At the centre is Chaco Canyon, which includes Pueblo Bonito: One of the largest single buildings ever constructed in the pre-Hispanic Southwest with about 800 rooms, elite burials and dense ritual deposits. The canyon itself has the densest concentration of great houses and great kivas. (A close up of those locations is in the bottom right). We were able to compile ceramic data for about 75% of all of these structures.
11:32
For each 50-year interval, we constructed networks based on the protocols discussed earlier, with visualisations based on strong ties of 75% to 100% similarity. I'll be showing a series of these figures, and in each the size of the nodes indicates how central in the network each building is based on eigenvector centrality (which is similar to Google PageRank algorithms). The colours of the nodes are based on a community detection algorithm. The lightest blue ties connect nodes that are spatially closer and darker are those more distant. For reference, known Chaco roads are shown in dotted yellow. These roads are not well dated, and we wanted to see when they coincide with network ties between the sites that are along them.
12:25
The resulting series of 50-year interval snapshots is shown here. And it resulted in a number of new insights into how the Chaco World changed over time. This includes the early centrality and density of sites in Chaco Canyon, the presence of strong ties between Chaco Canyon sites with sites to the south and the southwest, and the early and continued differentiation of sub areas. In particular, the Mesa Verde area of Southwestern Colorado was largely disconnected from the rest of the network, but highly cohesive on its own.
13:06
Sites in the Middle San Juan are shown with yellow dots. (Here you can see the yellow dots kind of go in and out of the graph, and they come back towards the end of the sequence as that particular area becomes more important.) The Middle San Juan is the area where the longest road segment connects it with Chaco Canyon, called the North Road, including many of the settlements in the area called the Aztec Complex.
13:40
Here's a map showing the distribution of some of the monumental architecture from the Aztec complex. And here's the Great North Road in an aerial photograph, and then also showing the connectivity of the road with the sites along it in this other map of Chaco routes.
14:05
So getting back to the snapshots, the Great North Road is completely covered by network ties in the A.D. 1000 and 1100s, when archaeologists proposed that a migration occurred out of Chaco Canyon, joining settlements in the Middle San Juan. Moreover, sites in Chaco Canyon become more heterogeneous, with some sites being more closely tied to sites in the Middle San Juan than to their neighbours. This also fits with a hypothesised migration of people from the Northern San Juan (Mesa Verde) area to Chaco. A major drought in the early 1100s is often brought up to explain this movement.
14:48
But why move to Chaco during this time? We now know that violence was a big part of the Northern San Juan settlement life and likely a major factor in this realignment of alliances among great house residences. The entire four corners, including all the sub areas shown in the Chaco network plots—with the exception of the southernmost area—was depopulated in the last half of the 1200s.
15:17
Our next case study looks at networks in the post-Chacoan Southwest at an even larger scale than the Chaco World, and encompasses most of that area with the exception of the Mesa Verde area. Site locations and site sizes for all known sites greater than 12 rooms had already been compiled by some of our team at Archaeology Southwest, showing how population densities changed across the Southwest. The depopulation out of the four corners by A.D. 1300 clearly shows up in these panels. And one of our major questions was how is population movement affecting networks of interaction in the western Southwest, covering an area that has traditionally been divided into Ancestral Pueblo, Mogollon, and Hohokam areas. This is the four corners area up here. And you can see that this dense occupation is pretty much disappeared by 1300.
16:22
The resulting snapshots for the A.D. 1200 to 1500 case study illustrates the rewiring of connectivity with population movement. The first two 50-year intervals show several subgroups' disconnected components, including the post-Chacoan sites that still remained in the highlands surrounding Chaco Canyon. As I mentioned, Chaco Canyon itself was largely depopulated by this time. The largest connected component or subgroup was in the Upper Little Colorado area, which had already seen a major movement of people from the Chaco area—so this is the largest connected component in these illustrations.
17:03
And by 1300, the population of the Four Corners region was complete and there was a rapid transformation, the settlements in the southern Southwest becoming more densely connected. A second major event occurred around 1400, as sites in the Transition Zone between the Colorado Plateau and the basin-and-range country to the south would be populated. (So this is the transition zone. This is the Colorado Plateau. This is the basin-and-range. And these sites are in a transition zone here.)
17:37
The demographic effects on social networks are shown in the series of close-ups by 50-year intervals, starting with the A.D. 1200 to 1250 intervals shown here. So this is again—these are the same snapshots. This is the area surrounding Chaco Canyon, this is where Chaco Canyon is you can see that it's completely depopulated by this time. This is a mountain range, the Chuskas that are to the west of Chaco, and then middle San Juan. And here's our big connected component. The area to the south, where many people we think from Chaco migrated to at 1250 to 1300 interval shows also that Chaco had completely emptied. And the remaining sites in the far western area form another separate subgroup or connected component. There's one down in the Phoenix area and then one in the Phoenix basin, excuse me, the Tucson basin.
18:47
For the interval of 1300 to 1350 population declined and networks realigned to the south. This realignment continues into the 1350 to 1400 period. And it's pretty much a maximum build-up of population—and also the greatest connectivity, in which the Hopi area is connected all the way down to the Hohokam area. This is the Zuni area. And it's still connected to areas just north of the Transition Zone. So this is actually Petrified Forest region here and some very long connections to the Verde Valley.
19:33
And finally, in the last period, after 1400, most of the Transition Zone was completely depopulated, another migration happened—and population sizes were plummeting throughout the Southwest at this time. Only the Hopi and Zuni areas are left at the end.
19:55
So we were curious as to the impact of spatial propinquity on network connectivity. For the earliest period of 1200 to 1250, when settlements were most widely distributed across the landscape, there was an unexpectedly higher correlation. But the northern and southern Southwest showed very different correlations over time. With increasing coalescence in the north, the correlations predictably decreased over time, and on the whole spatial and social propinquity were positively correlated. In the southern Southwest, however, increasing coalescence resulted in a slight increase in connectivity at 1300 when migration from the northern areas have been well documented. But on the whole, spatial and social propinquity were negatively correlated. In other words, more settlements that were far away from each other, were more well connected.
20:48
So the explanation for this divergence comes from looking at the pottery that drives the similarity coefficients for the southern Southwest. The major decorated ware in the southern Southwest, was the ideologically charged "Roosevelt Red" ware shown here. These were initially made by northern migrants who moved into the Transition Zone and southern Southwest. Patricia Crown's work established that these ceramics were part of a new religious ideology introduced in the early 1300s, which included designs such as this plumed serpent. She also showed that they were made in most areas that they were found. So it was the knowledge of how to make them that moved, not the pots themselves. That is, people moved, and were connected with each other and shared the information.
21:38
All of what I've talked about so far, has been based on strong ties. To wrap up, I'd like to take a minute or two to talk about what weak tie analyses have shown, as I think they provide important insights on the topics of boundaries, and permeability of boundaries. The sociologist Mark Granovetter is famous for his insights on the strength of weak ties. Weak ties are those on the low end of similarity indices in our work. According to his theory, they are strong because they can provide access to people and resources outside of their regular strong tie groups, which tend to be those based on residence, kinship, and friendship.
22:23
When we look at where weak ties tend to occur in the Southwest, they are in those areas with lower population densities, and in physiographically transitional zones. Materially they co-occur with high architectural diversity, including multiple kinds of ritual spaces, high ceramic ware diversity, and a high degree of innovation in ceramic technology, such as new paints, colours and decorative styles.
22:51
These are also areas in which small group migration and intermarriage have been proposed. Migration into these areas fits with a model of the internal frontier—but with an important corrective to Kopytoff's original formulation that was based on a world-systems core/periphery model in which frontiers were diluted representations of metropoles. Instead, we follow new interpretations, such as by Ogundiran, who sees these zones as areas of population diversity and innovation. The strength of weak ties for us is in pointing out how boundary or edge regions are important for encouraging culture change.
23:39
In closing, I offer these very short responses to the questions that were posed by the seminar organisers. I think that cultural groups should be distinguished from one another relationally, through their connections. Archaeologically, this is based on similarities in material culture, whether using production, distribution, consumption, and or symbolic attributes, which are even more powerful when combined with spatial data.
24:06
"How can cultural groups be identified from data?" Through clusters of relations, such as community detection algorithms on networks and identification of different network components.
24:19
"How should cultural barriers and borders be understood?" I think they should be understood as dynamic zones of interaction and innovation, not within core-periphery models, but as internal frontiers that are demographically and culturally diverse.
24:37
As I mentioned, we have a large team of collaborators and I'd like to acknowledge all of those who have participated in this work. Thank you so much for listening and I look forward to the seminar.