The eastern Sepik river basin in Papua New Guinea is famous for its culturaland linguistic diversity. There are over 200 languages and 400 dialectswithin the region of the Sepik river basin. Although the populations interactsignificantly, the population density, river tributaries, and surrounding terrainact to restrict contact between groups.

In the CSULB collections are 53 carved wood masks and statues that werepurchased by May Company from the Abelam tribal area in the MaprikRegion in 1965. These wood objects display tremendous variability in bothform and design elements. By using stylistic classes to describe theseobjects, this enables us to generate hypotheses about the structure ofinteraction among this population.

Stylistic Analyses of Stylistic Analyses of AbelamAbelam Woodcarving from the Woodcarving from the SepikSepik River Papua New Guinea River Papua New GuineaJason A. Miller and Kristen Voss

Department of AnthropologyCollege of Liberal Arts, California State University Long Beach

The purpose of this analysis is twofold. The first purpose is to determine therelatedness of carving classes using statistical analysis tools of the carvingsbased on measurements made with paradigmatic classification structuredesigned to examine stylistic variability in the decorations and expressed onareas of head and face. The second purpose is to evaluate the possibleexistence of lineages of carvers. We expect that stylistic patterns of attributeclasses should form cohesive sets that are transmitted from carver to carver.Given that this collection consists of contemporary examples of wood carving,we believe that stylistic analysis should provide information as to carvers andtheir unique traditions.

We would like to thank Professor Carl Lipo for his help in getting the cluster and cladistic analyses to function and display properly in this poster.

Figure 5. The figure abovedemonstrates the location data andprovides and example of thedetermination of the traits, carved,painted, or carved and painted. Theun-carved and un-painted versionresembles an “egg” in shape withoutfeature beyond the wood itself.

Introduction Classification

Aim

Results

Future Research Avenues

For more information

Please contact Jason-A-Miller@hotmail.com orclipo@csulb.edu.

Figure 1. Map of Papua New Guinea,East Sepik Province. Maprik region.

Courtesy Papua New Guinea Department of Tourism & Mining.

HypothesisGiven the wide range of stylistic variability expressed in the wood carvings,we expect that patterns of lineages should produce a strong and consistentstructure using occurrence seriation and Phenetic analysis using hierarchicalcluster analysis and cladistic analysis. We expect the patterns to showsequences of stylistic classes that are equivalent to lineages.

Figure 3 Phenetic Analysis. Shown above is the result of ahierarchical cluster analysis. It shows the degree of overallsimilarity. We created the tree using binary data and Wardsmethod . Cluster Analyses order descriptions based on overallsimilarity rather than heritable continuity. Each row is a singleclass of carving and is placed adjacent to masks that mostresemble it on either side.

Figure 2 . Below is the results of a cladistic analysis. The cladogram is an un-rooted 50% consensus tree that we constructed using random sequencingand measures of parsimony. Cladistic analyses order taxa based on their degree of relatedness. The tree demonstrates strong lineage structure consistentwith the seriation and Phenetic analyses.

Figure 4 Occurrence Seriation. Occurrence seriation ordersclasses using the concept of heritable continuity. Each column listedabove is a single class attribute with the presence of that traitmarked in black and absence marked by white. Each row is onecarving. Continuous sets of traits represent lineages. Gaps in thesequence suggests that more than one lineage is present in thecarving traditions.

These results are promising. However, additional external information isrequired to evaluate the hypotheses presented. For example, moreinformation about the social structure of carvers would provide a means forus to distinguish whether we have identified lineages or if other kinds ofstylistic classes are represented (i.e., classes related to specific socialfunction). With the addition of this data we could determine the degree towhich distance, language, and geography influence the transmission ofculture and the emergence of cultural variation with scientificallyreplicable results. One promising avenue of research would be to measuretrace-elements from the wood using LA- ICP-MS. This analysis couldpotentially allow us to distinguish geological signals related to groundwateruptake by the trees. If trees were harvested in different parts of the valleyby different carver groups, this would enable to test the lineage hypothesespresented here.

First, we carefully photographed and examined each wood carving anddescribed overall sources of variation. Second, we created a paradigmaticclassification to describe the stylistic variation in carving form, expression anddecoration. Using these data, we constructed an occurrence seriation,performed a Phenetic analysis using hierarchical clustering algorithms, andgenerated cladogram. Each analysis provides a different examination of thestructure of stylistic variability among the carvings.

Seriation operates on events described in terms of stylistic classes. It ordersclasses based on the principle that stylistic variability will be patterned byheritable continuity. Since stylistic variability is not subject to performance -only replication, it’s patterning is determined by transmission and will varyover time across space. In this case the temporal component is limited to abrief time period during or immediately prior to 1965. Therefore, the patternwe should see is likely due to space and/or social structure.

Phenetic analysis measures overall similarity in carving form and expression.Here the general algorithm assumption is that similarity in overall form shouldbe a reflection of the degree of interaction between carvers. For this analysiswe used a hierarchical clustering algorithm to generate groups of similarcarvings.

Cladistic analysis measures the degree of relatedness based on the measure ofshared derived traits. By focusing on derived traits, cladistics produces a treethat represents how taxa (i.e., classes of carvings) are potentially related toone-another. The cladogram represents relatedness as a function of distancealong the tree.

Methods

A groupingwhose mostobvious trait isbased on thecheek designs

A grouping whosemost obvious traitis based on Tonguedecorations

Conclusions and Discussion

Figure 6. On the left are examples ofthe basic classification for designs andtheir two basic subdivisions geometricand effigy. Effigies are completeentities and have two classes Birds andOther which includes lizards, wings,flowers etc. that would otherwise bestatistically insignificant. Geometricshapes include ovals, dots, triangles,and lines. All geometric patterns wereplaced in five color categories; plain,black, white, red, and yellow.

There are consistent patterns to the three analyses. First, gaps in continuityof the occurrence seriation demonstrate that there is likely more than onelineage present among the carving data. The analysis show that there are atleast seven different groups present. Second, the cladistic and Pheneticanalysis also suggest that there are multiple discrete lineages. Theselineages include groups most obviously based on tongue shape and design,and cheek pattern and design. Additionally, there is a lineage that was notreadily apparent in the occurrence seriation most obviously based on thetrait of “Effigy.” Other lineages are consistent between the Phenetic andcladistic analyses. While we do not know exactly what these lineagesrepresent, they suggest that the patterns of design and expression formstructured groups that are consistent with transmission structuredpopulations.

A groupingwhose mostobvious trait isbased on theEffigy -otherdesigns