This paper is a presentation of a current work in progress, specifically the exploratory phase for determining a methodological framework, clear objectives, and establishing preliminary results to guide the future direction of the project. The paper sees the application of text analysis to a corpus body of texts with a focus on highlighting heritage and intersectional data present within these texts. The approach of text analysis allows for a quantitative analysis of modern perceptions of the past, narratives given to the past by modern people, and the resulting context elements of the past are placed in stemming from modern influences. With a focus on how prehistory is presented to modern people, in the specific context of museums, it is necessary to trace the contents of texts depicting the past in these museums. The overall goal of this paper is to have a deeper understanding of the impact modern narratives attributed to the past has on the prehistoric past in an educational context. Specifically, looking at narratives focused on the process of neolithization as discussed in museums. Additionally, preliminary explorations give insight into the benefits of the methodology and how to best establish next steps to propel future research.
Museums are integral tools through which the past is understood and interpreted using
tangible scientific evidence. These institutions provide modern people with both physical and
experiential elements for developing interpretations and relationships with the p3a3s]t a[nd
educational opportunities1[
The following paper details the exploratory stage for an in-progress research project utilizing quantitative textual analysis applied to a corpus of archaeological museum texts. This stage of the analysis was used to formulate the best methodological approach to carry out text mining for this corpus, set parameters, and solidify next steps in this project. This research is focused on the types of narratives linked with production societies within museums, specifically museums along the east of the Iberian Peninsula. The interest is in how prehistory-producing societies are taught in museums by analyzing associated texts.
As this is the first stage of the project all results are beneficial to evaluate the data, any need for additional data, and exploring the research questions through quantitative text analysis. The proposed questions are: 1) Through the texts produced in these museums is it possible to determine the key issues from those societies and periods? 2) Do the narratives fit with the current accepted research surrounding these societies and periods? 3) Are any problems from past societies related to problems of the present? and 4) How is the information treated and presented within these museums? These early results are indicative for next steps in this research project.
In the context of archaeological heritage, textual sources are an important source highlighting
shifting narratives, contexts, and interpretations linked to tangible and intangible
archaeological heritage – including landscapes, surviving sites, museums, etc. Discourse studies and
continually evolving quantitative methods continue to highlight the benefits of textual analysis
applied to archaeological heritage dat2a9[
In previous research, this method has been used to discern how archaeologists discuss
social issues over time 2[
Data from surveys and other research provides insight into how modern people cultivate
relationships with the past through perceptions, interpretations, and context in which people are
exposed to the past [
Looking at museums as agents facilitating that cycle, it is important to look to the
existing presentation of the past within these museums. In order to determine what constitutes the
narratives currently presented to people about the past, to see the efects of archaeological
education with this modality1[
Texts appearing within museums linked to the development of early farming societies and metal age societies– including discussions on the neolithization process – were utilized in this study. This project aims to gain a better understanding of the words and concepts linked to this part of the past. Specifically, the way in which narratives of neolithization and early producing societies are presented to the public in museums. Within the Iberian Peninsula, there are regional-specific variations and circumstances in which the overall development of these production societies occurred1][. There are diferences in which neolithization impacted specific regions and the people within, in conjunction with clear neolithic traits not dependent on a region (i.e., farming practices, economic structures, technological advancements, etc1.2)].[ Additionally, changes to social structures and interactions between others which impacted the role of reciprocity, growth of social inequalities, and the development of social netwo1]r,ks [ will be explored in future research.
Within the Iberian Peninsula, new research and discoveries focused on the regio1n4, [
The corpus used in this project was built from a collection of texts in Català, compiled from
permanent exhibitions appearing in eight museums over a period of thirty-years along the east
of the Iberian Peninsula. These texts – with an overall sum of twenty-five thousand words –
were extrapolated from panels that describe the diferent themes of the museum. They are
explanatory texts of general Neolithic culture characterizations (i.e., farming, work techniques,
religion, etc.). They exclude display cases as the sole interest is the main narratives presented
to visitors. The texts are general descriptors of Neolithic elements, with a select few descriptors
of specific sites. R Statistical Software (v4.2.2)[
The texts were all collected using the same process of OCR with manual and automatic
inference. The individual texts were photographed and converted using OCR, with the texts
manually verified when necessary. During this process, it was decided to record all texts as
a single document per museum. The reasoning is each museum splits the content diferently.
Therefore, a side-by-side comparison of each text (i.e., panel) would not be useful and would
include strong biases (i.e., museums where text is split in several panels would have a higher
weight that other ones where text is concentrated on a few large panels). Following the text
collection, Topic modelling was applied as a type of ‘distant reading’ to trace linguistic patterns
regarding the storytelling process between diferent museums presentation of archaeological
knowledge. R was used to explore content, word appearance, word frequency, and context. All
texts were loaded into, processed, and the corpus built within R using a number of packages:
tidytext (v0.3.4)[
Topic modelling was selected for its machine learning prowess in the sorting and
classification of documents in a corpus, assigning topics, and highlighting temporal distribution of topics
[
Previous research was consulted to determine a preprocessing chain to help remove bias and
noise as much as possible. A metrics test was run to determine the appropriate number of topics
per the corpus, parameters of utilizing stopwords and applying lemmatization, and training an
LDA model was decided upon using previous research as guidelin2e6s, [
This first exploration of the data used text mining techniques using R to analyze word frequency and occurrence. The focus was to find the overall most frequently appearing terms throughout the corpus and then the most frequently occurring terms for each text per museum. Looking ifrst at the frequency of term appearance overall, as seen in Figure 1, the entirety of the terms present are of importance. The terms all are indicative and relative to the overall process of neolithization. As the terms are in Català, translations include: neo(lnietioclithic), pedra(stone), poblats(villages), cova/coves(cave(s)), restes(remains), edat(age), silex(flint) , objectes(objects).
Now looking at the frequency of word appearance per text per museum, diferences in the narratives per museum are discernible within Figure 2. For starters, in the entries for M BBAA Castelló(Museo de Bellas Artes de Castellón), key terms include: cour(ecopper), vida(life), canvis(changes), and all centered around the province Castelló – also a term – located in Valencia. Whereas, in the entries for MAC Barcelon(aMuseu d’Arqueologia de Catalunya, Barcelona), key terms are: sílex(flint) , cabanes(huts), and fossa(moats).The last example are the entries for Museu Lleida( Museu de Lleida) with the terms: restes(remains), eines(tools), lloc(site) all pertaining to Minferri – an additional term – which is an Early Bronze Age settlement located in Lleida. Just from this intervention, there is a delineation between narratives in museums – linked to location and elements of value in each location.
The next step was running topic modelling. In this iteration, the model is made up of nine topics composed of the ten most frequently occurring terms per topic. These preliminary explorations made visible the types of topics within the corpus and the trends in each topic, as illustrated in Figure 3.
Inference based on the terms give insight to the general theme of the topic itself. For example, in Topic 4 terms include: llo(pclace), bronz(bronze), vid(life), cult(worship), and art – which indicates the context of art and culture within the bronze age. Whereas in topic 8 key terms are: cov(cave), cultur(culture), décor(decoration), and bronz(bronze) – leading to infer this topic pertains to cave art during this period. Additionally, it is through interpreting these topics that names are developed as a result. Proposed names per topic are presented in Table 1.
Focusing on the themes within each of the topics highlight contexts within the text, of what terms were occurring in proximity together. These early results provided the researchers of the project the chance to consider potential changes to the dataset, adjustments to the methodological approach, and other ways of analyzing the contents of the texts.
Museums are one example of a center for archaeological heritage discourse in which
individuals can experience material and immaterial culture which helps develop one’s relationship
with the past [
Museums are integral with cultivating memory of the past seen through the interpretations
used to create narratives presented to visitor1s5[]. Yet, narratives surrounding the past are not
stagnant and as new information becomes available, new technologies allow for more complex
study [
As far as assessing how text analysis can illuminate key issues from the past as it is connected to early production societies, the present themes and word frequency illuminate the focal point of the narratives and interpretations presented to visitors. If museums are determined to provide visitors with insight into all facets of the past, terms and themes related to said issues should be visible through text mining. At this stage of the project, any answers to research questions are preliminary and are more valuable in determining next steps. Again, the primary focus has been on the efÏcacy of the methods, potential changes to the data, and what adjustments need to be made for parameters.
Originally, during the preprocessing of the texts before building the corpus, there was a limit to
the cleaning of the text. In part, this is due to the text in Català and the need to see how the
language is recognized within R. There was no general list of stopwords for Català, though there
is one in English. Nor a personalized list of stopwords, something that would aid in removing
noise [
The preliminary figures illustrate how with the strengthening of the methodological approach, the research questions can be answered through a quantitative analysis. With the goal of exploring the extent current exhibitions ofering this archaeological knowledge consider all advances of knowledge over the last three decades.
With Figure 1, focusing on early production societies and neolithization, the terms we can expect about these peoples and societies in general are present (i.e., ped(srtaone), poblats(villages), cerámica(ceramics), silex(flint) . For Figure 2, it is further possible to discern the relationship between topics and museums. For example, looking at M BBAA Castelló, in this area neolithization is in relation to animals, with habitat more often in caves which is reflective of the mountain region. Where, MAC Barcelona sees terms that fit more describing the neolithic cultures in general within Catalunya, with more specialized concepts used by archaeologists. This is reflective of the museum having an academic and technological discourse.
Regarding these texts reflecting narratives in accepted research, there are discrepancies. For example, within these texts there is no mention of Mesolithic peoples, nor any mention of the transformations of neolithic landscapes. There is also a focus on specific neolithic activities that take precedence over others. There is a larger appearance of farming compared to other neolithic activities, such as herding. Additionally, there is no mention of the impact of neolithic on prehistoric landscapes. This reflects that some issues are the importance of purely economic activities and technology beyond other types of traits that define neolithic (i.e., social structures, sexual division of labor, environmental dynamics, etc.). Furthermore, with the goal of being able to interpret and understand the treatment and presentation of these texts, additional analysis can consult linked metadata or comparing the evolution or stagnation of the information.
Regarding the selected texts, at the present moment, these are the only texts and only museums utilized – but there is a potential to bring in texts from an additional four museums in the specified region. As far as narratives present within museums remaining up to date with current research, the use of a qualitative approach – ‘close reading’ – of these texts and recent articles could provide more details. Another option could be adding an additional corpus made up of recent research to analyze separately and compare results. Comparing corpus to corpus, to see potential similarities and diferences between the narratives presented in museums and those produced in peer-reviewed journals. Based on the early explorations of the corpus, the final analysis filtered the corpus for any term not present in at least 2 museums and the number of topics was set at nine following metrics testing. The number of term appearances can also be adjusted, along with additional metrics testing to further evaluate the number of topics. The end result of this stage is a future direction for how this project can further develop, solidify the code and methodological framework.
Our ability to interpret, perceive, and relate to the past is only possible resulting from the value and significance of surviving material and immaterial culture. The increasing accessibility of this material and immaterial culture present within museums is linked to what modern people find valuable and significant enough from the past, to hold this level of presence within modern culture – on display in the modern collective memo1r3y].[Much like value and significance assigned to remnants of the past can change, so does the knowledge and interpretation of the surviving past change. Text analysis is a valuable tool to trace shifts in the knowledge and interpretation of archaeological heritage – including texts surrounding archaeological heritage currently residing in museums. This project will continue applying the methodological approach to better understand and disseminate current patterns of archaeological knowledge, and their respective narratives and contexts as they exist within museums.