People and places: towards an understanding and categorisation of reasons for place attachment – case studies from the north of England

Abstract People develop a sense of place, belonging and identity when a place affords tangible and intangible benefits like security, familiarity, shelter, food, work opportunities, and social interaction. Places form landscapes individually valued by people for these reasons. This paper describes Topic Modelling as a new grounded approach to assessing people’s sense of place in a rural landscape affording special qualities for everyday working and living situations – the Peak District National Park, UK. This novel approach is applicable and scalable to any landscape, rural or urban, iconic, or everyday. Results of this study show that significant themes and phenomena not hypothesised at the initial research design stage can emerge from interview data. Examples include pro-environmental behaviours resulting from traditional farming practices, environmental benefits of the drystone-walling tradition, and attitudes towards rewilding initiatives. We argue that such phenomena arise from people’s attachment to place and influence their behaviours.


Introduction
People connect to places for different reasons and can develop a sense of place, belonging and identity through those connections (e.g.Cresswell, 2015;Feld & Basso, 1996;Jones & Leech, 2015;Seamon, 2020).This occurs when a place affords tangible and intangible benefits like security and familiarity, shelter, food, work opportunities, social interaction, and a space for well-being.The human geographer Yi-Fu Tuan developed the 'sense of place' concept to theorise connections between people and places (Tuan, 1977).Such connections were subsequently developed as the concept of place attachment.Over the past 50 years, place attachment studies have provided theory and methods for assessing the connection between people and places (e.g.Altman & Low, 1992;Lewicka, 2011).The following will elaborate on (1) place and (2) reasons for attachment.
(1) Place, as a concept, is variously defined across disciplines (e.g.Cresswell, 2015;Ingold, 1993;Tilley, 1994).Places are elements of rich and dynamic historic and contemporary landscape that require various forms of heritage management and planning.Tools, such as Historic Landscape Characterisation (HLC, e.g.Aldred & Fairclough, 2003) and Landscape Character Assessment (LCA, e.g.Swanwick, 2004), were developed specifically for such planning and management purposes.The focus on integrating people-centred approaches to decision-making for public benefit has only recently entered into management and policymaking but is still dominated by '"objective" outside experts' (Butler, 2016, p. 1).But there are examples indicating a change.For example, in Scotland research conducted by Community Land Scotland with Inherit involved interviewing practitioners and members of the public to explore ways of integrating public opinion meaningfully into the planning and decision-making process (Dalglish, 2018, pp. 54-55, see also e.g.Koblet & Purves, 2020;Wartmann, Acheson, & Purves, 2018).Nevertheless, landscapes -and places -are subject to change and development and there is a significant research gap in identifying the impact of change on place attachment and people's perception of the quality of familiar places (Hedblom et al., 2020, p. 58; see also Hunziker, Buchecker, & Hartig, 2007).This gap is due mainly to the inherently challenging and time-consuming qualitative data analysis methods that are typically used in such investigations.
(2) We have previously identified the significance of social values held by individual people or communities as reasons for a connection to places and landscapes (Tenzer, 2022;Tenzer & Schofield, 2023;Tenzer, 2023).Social value is created when the meaning of the qualities of a place are weighed and signified (Williams & Patterson, 1999, p. 142) and can emerge from family or local history, memory, traditions, myths, legends, and beliefs (S.Jones, 2017).Stefaniak, Bilewicz, and Lewicka (2017) showed the development of place identity as the relation of people's personal life stories with the history and past of a place based on ancestry, memories, and traditions, emphasising that people who wish to 'actively engage with place … come to feel a part of place' (Seamon, 2020, p. 37).In this holistic approach to landscapes, we follow Tilley and Cameron (2017) who recognise the agency of both human and nonhuman actors and the material world and an emotional, social bond between people and places.
In this paper, we explore what insights can be gained into the meaning and significance of social values through in-depth interviewing of people living and working in the Peak District National Park (PDNP) (UK) with a view to tangible and intangible elements in the landscape that afford strong connections.We categorise these elements and ask how a strong place attachment influences people's behaviours and approaches to the place or landscape?These insights can provide vital background for proactive planning and development, adapting to the needs and visions of people with a strong place attachment to be socially sustainable.
We propose the application of Topic Modelling to interview data as a qualitative assessment methodology, which is time-and resource-efficient and allows the exploration of qualitative data.Artificial Intelligence tools, such as Natural Language Processing and Topic Modelling (TM), have been developed since the late 20th century.The Council of Europe actively encourages the use of these new tools within heritage activities (e.g.Traviglia, 2022).However, the deployment of these methods and tools in heritage studies is a recent development and they have yet to be fully integrated into the field of heritage and landscape studies (but see Fiorucci et al., 2020;Matrone et al., 2020;Purves, Koblet, & Adams, 2022;Bordoni, Mele, & Sorgente, 2016;Wartmann, Koblet, & Purves, 2021).We apply TM, which we have previously used to analyse survey data (see Tenzer & Schofield, 2023; see also Abram, Mancini, & Parker, 2020;Cai, Siebert-Evenstone, Eagan, & Shaffer, 2021;Franzosi, Dong, & Dong, 2022).TM allows the exploration of qualitative unstructured data to reveal themes latent within or emerging from the empirical data without preconceived assumptions.This principle adheres to the Grounded Theory elements underpinning this research (e.g.Charmaz, 2006;Odacioglu & Zhang, 2022).While the case study focuses on rural areas, this approach is applicable and scalable to any landscape.

Overview of the method
Typically, place attachment is measured using Likert scales, which risks missing the fine nuances shaping attachment (Boley et al., 2021;Brown, Raymond, & Corcoran, 2015).Social values as a basis for place attachment can be captured in different ways.Surveys have been used elsewhere and provide a broader view on the reasons behind attachment (e.g.Wartmann et al., 2021).For this study, we apply the approach we developed by Tenzer & Schofield (2023) and apply Topic Modelling to interviews with people living and working in the Peak District National Park as a first pre-assessment stage of the data analysis, followed by a second stage of manual observation.
Figure 1 demonstrates the steps involved in the combined approach of Topic Modelling and direct observation used in this research.Interviews were collected in person at the places to which interviewees described as having the strongest bonds.Single paragraphs were then treated as text documents and fed into the topic modelling algorithm.This approach allowed the discovery of themes within the data that might otherwise stay undetected or disguised by the researcher's assumptions and predefined codes.

Study area
The Peak District was designated as the first National Park in the UK in 1951 (Figure 2).The PDNP has a history of at least 10 000 years of human occupation and covers an area of 1438 km 2 .It has 38 000 residents and receives more than 13 million visitors per year.The PDNPA contains a large number of designated sites, including 2900 listed buildings, 109 conservation areas and 450 scheduled monuments, including prehistoric burial mounds, stone circles, medieval field systems, castles and country houses (PDNPA, n.d.).One third of the PDNP are Nature Protected Areas, designated as Sites of Specific Scientific Interest (SSSI), Ancient Woodland, National Nature Reserves (NNR) and Local Nature Reserves (LNR).A specific character feature of the National Park are the drystone walls with a total length of 26 000 miles. 1 The palimpsest of human impact on the landscape includes traces of industrial action including remnants of the millstone industry and rakes from lead extraction.This quality of open landscape across the PDNP stands in stark contrast to the adjacent industrial cities of Sheffield and Manchester.

Data sources
To assess the landscape factors dominating the perception of people living/working in PDNP, ten in-depth in-person interviews were conducted over 16 months between late 2021 (after COVID-19 restrictions were lifted) and early 2023 (locations in Figure 2) (Supplementary Material 1).The case study approach was chosen for in-depth data on perception of individuals within a bounded system (Creswell, 2017, p. 96;Flybjerg, 2011).This approach completes a triangulation of methodologies developed to extract information on perception (Altman & Low, 1992;Low, 2002), meaning-making and value creation in the study area based on three different data sources: social media data (Tenzer, 2022), online survey data (Tenzer & Schofield, 2023), and in-depth interview case studies, as presented in this paper.The sampling of participants was based on the Typical Case sampling strategy advised by the PDNPA and from research into place history, providing typical, information-rich examples for the exploration of landscape perception (Creswell, 2017; see also Koblet & Purves, 2020;Tilley & Cameron, 2017;Wit, 2013).
The semi-structured interviews were held at the place of residence or work of the interviewees and comprised three parts.Part 1 involved general questions about people's connection to the PDNP and general landscape and local heritage perception.Part 2 of the interviews focused on the specific aspects of living or working in each particular case and were dynamic.Part 3 focused on tangible and intangible elements of the landscape.The interview transcripts ranged in length between 4351 and 12 773 words.The answers were separated into paragraphs, resulting in a total of 298 separate paragraphs (documents) and forming the dataset for the analysis using TM tools and manual assessment.

Topic Modelling and direct observation
We apply Topic Modelling to interview data as a qualitative assessment methodology, which is time and resource-efficient and allows the exploration of qualitative data as developed and described in Tenzer and Schofield (2023).This algorithm allows an insight into the empirical data that conforms to fundamental elements of Grounded Theory (Charmaz, 2006), where data are clustered by emerging themes.This approach allows the discovery of themes within the data that researchers might not have anticipated or discovered (objectifying researcher bias).For the data analysis, we used the R package textmineR to pre-process the data (including data cleaning and lemmatisation) for TM, following an innovative method developed by Jones (2021; see also Jones, Doane, & Attbom, 2021).The method implements Latent Dirichlet Allocation (LDA), a statistical method to identify themes based on keywords in the documents (Blei, Ng, & Jordan, 2003;tqx94, 2022).This statistical analysis method clusters text according to themes based on keywords.For a detailed description of the TM methodology, see Tenzer & Schofield (2023).Here we give only a short outline.
LDA clusters the data into a predefined number of topics.The optimum number of topics can only be tested but not determined from the start.Therefore, 60 models were created and assessed using the topic coherence factor (Figure 3).The higher this factor, the more association between the words in each topic cluster.Figure 4 shows the relationship between number of topics and coherence of terms within topics.We choose k ¼ 37 topics since the curve flattens out beyond, meaning coherence does not increase significantly with a subdivision into more clusters.TM forms themed clusters based on the probabilistic distribution of words over topics and topics over documents, labelled as bi-grams (two closely associated words in a text) (tqx94, 2022). 2 These labels are not meaningful as such but give a good indication of the latent themes (Supplementary Material 2) for manual labelling.
While the application of TM can support the first thematic insight into the data, it cannot and should not replace the direct observation of content and themes (Chang, Gerrish, Wang, Boyd-Graber, & Blei, 2009).Therefore, the subsequent manual evaluation and annotation included an assessment and refinement of the topic labels, as described in the next section.

Direct observation and categorisation
A first observation of the model labels gave insights into the themes latent in the data (Supplementary Material 2).For instance: 'stewardship_scheme' and 'tree_plant' pointed towards a connection of management in the PDNP and the subsidies for pro-environmental action; 'plan_permission' and 'farm_building' hinted at a trend towards diversification of farms in the PDNP; 'national_park' and 'people_visit' contained the theme of tourism and challenges with visitors; while 'climate_change' and 'blanket_bog' offered itself for a theme of climate change that also impacts the PDNP.However, the approach of lemmatisation (grouping of inflected word forms, presenting the basic dictionary form) and bi-gram creation (two words closely associated and positioned in the document) also showed problems with the text-based analysis.For example, 'good_dress' (lemma of 'well dressing') pointed to the tradition of well dressing typical for the PDNP.Only with previous knowledge about the questions and answers can such a label be meaningfully interpreted.

Examples for categories
The following section will provide descriptive examples for the seven overarching categories that were defined based on the Topic Modelling and the subsequent manual categorisation.Figure 5 demonstrates the distribution of categories over interviewees.

Pro-environmental action
In Interview 1, the historical skill of drystone walling with a long tradition in the Peak District was the main topic around tangible and intangible heritage, recognising the potential of creating a sense of identity and belonging through an embodied experience of heritage as part of the cultural landscape (Figure 6).The extraordinary qualities and benefits of this form of heritage were mentioned in the first interview with the longest-serving drystone waller, Trevor.He mentioned that the walls provide shelter for many forms of wildlife from insects and spiders to small mammals.However, other interviews (particularly Interview 10) highlighted that drystone walls are also beneficial in confining and limiting the damage from wildfires.Additionally, drystone walls are artificial barriers for surface water flood prevention.Similarly, the reinstatement of the blanket bog in the north of the Peak District contributes to the water quality and so to the ecosystem services of the region (Interview 9 and 10) (Figure 7).Similar pro-environmental benefits of historical features emerged during Interview 2, mentioning the re-introduction of such walls and hedges for the subdivision of fields into smaller units.Such boundaries were removed over the past century but are still visible on historical maps.Reinstating boundaries benefits wildlife, as mentioned above.Also, the re-introduction of biodiverse plant species, for instance, traditional apple species, orchards, and wildflower meadows, emphasised the need to promote biodiversity for a sustainable future.Initiatives led by engaged farmers support the exchange of wildflower seeds to extend the gene pool of plant species (Interview 2), and the reason for diminishing landscapes lies in the traditional way of farming: I think the other thing is there were lots of wildflower meadows which were farmed environment, a managed environment.And I think we have lost over 90% of them.And most of that again was in pursuit of cash because farmers were encouraged to improve fields and improving means, getting rid of all the wildflowers.(Interview 9) A closely related form of pro-environmental action -renaturalisation of former pastureland (Interview 7) -showed a reorientation of PDNP farming community members.The topic of renaturalisation, rewilding and reforestation, as sub-codes of this category, highlighted the issues around using the term 'rewilding', which was unanimously found to be opaque and misleading.All interview participants saw the need to introduce more biodiverse and natural landscapes, but the term 'rewilding' triggered an immediate negative attitude towards the concept: Not too keen on some of the ideas.We managed without wolves and things … I don't like the idea of like wild boars running through forest particularly.Is it really necessary?(Interview 6) The difficulty is that rewilding to a farmer or a landowner sounds very much like abandonment.And agricultural abandonment has been seen as really, really bad in the farming community.(Interview 8) A question raised by an interviewee asked which point of time to choose for the reinstatement of the 'natural' state: I am involved in rewilding activities and there is a big issue about: so, what is the natural state of the landscape?And I don't think there is an answer to that … Do we want it like it was 200 years ago, 2,000 years ago or 20,000 years ago?So, it is not static.And whatever we do, we are going to interfere with nature.(Interview 8)

Change and Continuation
These Pro-environmental Action topics overlap with the theme of Change and Continuation and the sub-code of traditional farming methods.On the one hand, these methods can create an issue with their opposition to innovative thinking, with practices negatively impacting the landscape.On the other hand, the re-introduction of hedges provides an example of an environmentally positive traditional farming feature, contributing to biodiversity.Small acreage (Interview 6) and pressures from market prices for farming produce led to a continuation of improving the pasture by spraying (Interview 6 and 7) and pressures of overgrazing (Interview 7): No, we are obviously interested in doing things like that [alternative farming methods] but at the moment we haven't done anything like that because we only have a small acreage most of the land is in use constantly.It is like pasture land so the sheep graze it all the time.If you know what I mean, we don't have any that we could set aside for wild meadows or anything like that.(Interview 6) However, changing farming methods can also contribute to a changing landscape.For instance, increasingly, farms are owned by people who are not farming the land.Such land tends to be rented out to farmers, extending businesses on an industrial scale with consequences for work processes.To be more efficient, such farms introduce, for example, larger machinery, demanding a widening of gates and access ways (Interview 3).However, changing conditions can also impact positively, as in the example of the Hope Cement Plant.Vast areas of seemingly devastating quarrying activities to provide minerals for cement production are used as an opportunity to reinstate the exploited areas as renaturalised areas with rich biodiversity (Figure 8  and 9).
Outside the farming and industrial context, Change and Continuation were dominated by the strong influence of intangible and tangible heritage as reasons why people connect to landscapes.For example, keeping traditions alive, such as well dressing, the Castleton Garland, and sheepdog trials proves important for the coherence of communities and the attraction of visitors.Also, traditional skills were important to the interviewees.Interviewees actively work against the loss of skills and traditions, for example, with the creation of a culture centre (Interviews 2 and 7) and the teaching of drystone walling (Interview 1): We have to preserve some of those … you know … those old traditions, we must.Or we have lost them, and it is nice to see some of the older folk, like myself, teaching the young ones that are interested.(Interview 1)

Place History
In view of the historical dimension influencing place attachment coded as Place History, we identified a strong trend in all interviews -independent of whether the interviewee lived or worked in the Peak District or if the interviewee was born or moved into the area -to develop either a strong connection to place through family history or a strong interest in the local history.This was found in the lifelong connection to the landscape or particular characteristic features, such as drystone walls (Interview 1) (Figure 6).As another example, there was a deep connection to the 'plague village' of Eyam for one interviewee whose family history can be traced back to one of the survivors of the plague -creating deep roots to the location (Interview 5): … it is the landscape, we have to keep it as it was and try to preserve a lot … we must respect our elders, what they did … But, I still say, it was better then as it is now.(Interview 1) Contrasting attitudes become apparent when we compare this with residents who moved in, describing the landscape as 'frozen' (Interview 9) or preserved 'in aspic' (Interview 7), referring to the aspect of 'dynamic' landscapes (Interview 8).There is a notable need to accept changing and dynamic landscapes.However, changes and developments in infrastructure and the built environment are identified as having a negative effect on the quality of landscapes in the National Park: You know the motorway that crosses the South Pennines just north of the national park boundary.You wouldn't look at that and say: Well that has been a marvellous addition to the landscape, wasn't it?But you would look at Howden Reservoir and most people visit the Upper Derwent valley because it has reservoirs in it and they find this an attractive feature … if we hopped on a hundred years, 200 years we wouldn't look back at the M62 and say: "What a fine piece of industrial architecture that is.(Interview 10) Comparably, moving into the National Park from outside led to a strong interest in local history, especially the specific history of people's houses of residence and the immediate area around them.Residents developed a keen interest, deep knowledge, and active engagement in the local history -a deep sense of belonging (Interviews 2 and 7).
Correlating the category Change and Continuation with Place History trends showed that participants with a long family connection to the Peak District were more in favour of continuation and preservation of the status quo (Interviews 1, 5, and 6), while interviewees with a shorter period of residence or people working in the National Park suggested that they were more open to change (Interviews 7, 8, 9 and 10).Such change included the diversification of farms, including holiday cottages, culture centres and renaturalisation action (Interviews 2, 7 and 8), a strong awareness of climate change and loss of biodiversity (Interviews 4, 9 and 10), and the associated need for pro-environmental action.

Landscape Quality
The strong connection to place was also visible in the perception of Landscape Quality with regard to the familiar, aesthetic and natural qualities.The landscape character proved to have potential for building a strong sense of 'home': But coming back into Ashbourne, into the south.As soon as I saw these hills … ah … I felt at home.It was the landscape.(Interview 1) Opposed to the former concept is the perception of 'man-made' (sic.)versus 'wilderness'.All interviewees acknowledged the notion that the whole of the National Park landscape has in one or another form been impacted by past human action, despite the seeming 'wilderness' of some parts of the landscape: So, the major contribution that it [the national park] has made is that we have the last bits of wilderness England.From a people perspective that is actually quite good for our mental health.Because you can't get anywhere in this country where people say: look, this is a wilderness!… It certainly looks like a wilderness in January in the snow.(Interview 9) This, again, overlaps with the strong connection to the history of the National Park and the aspect of a living and working landscape: So, our national parks … tend to be looked upon as being up in some conservation of wilderness.And the Peak District and the South Pennines are very much a varied, managed landscape.And I sort of quite like that, and I think a lot of people who visit this area like that.They like that industrial heritage aspect that surrounds it.That story-telling of past human activity which is not quite as strong in other landscapes in England.(Interview 10) I think knowing that we have to feed … there are so many demands on the Peak District.It is not a park in the sense of a wilderness.It is a lived … a living landscape.Living and working landscape.But it could be … it continues to be for me the lungs of Sheffield and we need to protect it … But I would also like to see more bold decisions around land use (Interview 4)

Communities
A strong reason for place attachment could be found in the close-knit community as shown in the example of Eyam: Or I have just grown tomatoes for instance, … I had about twelve or fourteen left over … and I would put them at the gate in a box with a note on "free to good home" and they all went within two days.And so we share things across community as well.(Interview 5)

Challenges
The interviews gave insight into the issues of residents in the National Park, mostly associated with loss of local housing to holiday and second homes and an associated loss of community (Interviews 3, 5, 6, and 8) and pressures from the increased influx of visitors (Interviews 3, 5, 6, 9 and 10) with damage to local landscapes, disturbance of wildlife and farmstock and daily life of residents.

People and Place Engagement
This category showed the connection between activities that connect people and places.One aspect mentioned in this regard was the stories associated with places that reinforced the connection: … so, I think if we would use the landscape to tell the story I think that then gives us … it's a story of place that is … I think that is important.(Interview 4) In summary, the interview excerpts show the sometimes opposing attitudes amongst people living and working in the Peak District and the different values that people afford to places within its landscape (Figure 5).Manifold demands and the diverse needs of people are putting pressure on the National Park, but they also offer opportunities to see the places through the eyes of the residents, to improve the quality of place and lives.After reflecting on the limitations and biases, we will discuss these results and their meaning.

General discussion of results
The connection between people and places has been subject of research with a practical focus on landscape characterisation (e.g.Dalglish & Leslie, 2016;Koblet & Purves, 2020;Primdahl & Kristensen, 2016), and geographically on national parks (e.g.Maguire, 2017;Petrova, � Ciha� r, & Bouzarovski, 2011;Ramkissoon, Weiler, & Smith, 2012), using interviews as a data source (e.g.Polfliet, 2020;Wartmann et al., 2018;Wit, 2013) and deploying AI tools for qualitative data analysis (e.g.Goerz & Scholz 2010;Purves et al., 2022;Sassolini & Cinini, 2010;Sporleder, 2010).However, approaches in heritage studies have so far not focused on landscape assessments (e.g.Fiorucci et al., 2020;Matrone et al., 2020;Bordoni et al., 2016).Our research builds on the previous work and combines elements of previous approaches.This paper presents one part of a wider research project exploring social value in the views on place attachment using social media (Tenzer, 2022), surveys (Tenzer & Schofield, 2023) and in-depth interviews to get both broad and deep insights into factors for place attachment on a landscape scale that can provide background for socially sustainable, inclusive, and transparent management of cultural landscapes.
For this study on TM, we used ten in-depth interviews with people living and working in the PDNP.Case studies provide a deep insight into occurrences that can only be achieved by analysing the particular, 'to draw otherwise inaccessible conclusions' on the perception of specific phenomena (in Creswell, 2017, p. 99;Flybjerg, 2011).
The Council of Europe advises testing the deployment of AI in all sectors (Traviglia, 2022), and we utilised TM (after Jones, 2021;Jones et al., 2021) as a specific application.This tool enabled us to extract themes latent within the empirical data to set a framework or structure using labels based on the words used in the interviews for the subsequent manual assessment and categorisation of the data.With this approach, we adhere to Grounded Theory elements (Charmaz, 2006).This supports the discovery of themes emerging from the data that might otherwise not be apparent.
From the initial TM clusters and labels, we inferred seven overarching themes with 27 subcategories of reasons for the connection between the interviewees and places.The analysis showed that 'Change and Continuation' (24.83%) and 'Landscape Quality' (22.48%) dominated the perception of people living and working in the PDNP.The strong connection between the people and the places was evident from the close-knit community of villagers in Eyam and the community centre for local history, landscape and knowledge education.The interviewees were strongly attached to their places in the PDNP through their love for the landscape, which is partly created through the aesthetic qualities and the seeming 'wilderness' and, at the same time, the human traces and achievements inscribed in the historic landscape.Main insights from the data showed that lifelong residents are more resistant to change than long-term residents who moved in (e.g.changing farming methods, change in local population, renaturalisation).All interviewees were aware of pressing problems such as climate change and biodiversity loss and demonstrated action for nature conservation (benefits of drystone walls for wildlife, planting old apple species and hedges, preserving blanket bog).Despite these positive attitudes towards the various places in the PDNP, challenges are part of the life and work of the interviewees, as shown by issues such as increasing footfall of visitors and associated damage to the environment, loss of community coherence through second-home ownership, and negative impact of traditional farming practices.While we have demonstrated that a larger number of individually held social values create patterns across landscapes based on the same values existing across areas or through the application of different values to one place (Tenzer & Schofield, 2023), this paper deepens the insight into the manifold values of individual people.
These deep insights support the understanding of reasons for people's connection to the places where they live or work in the PDNP, which correlate with the findings of the broader approach through survey data as detailed in Tenzer and Schofield (2023).These insights from various data sources can provide essential insider knowledge and background for planning and decision-making in the PDNP, but also beyond this landscape, in other environments, for example, in urban areas or conservation areas.The sample size used in this paper can be scaled up based on time-and resource-efficient TM clustering.
AI tools and methods have applications in a wide range of disciplines.Adoption and adaptation into the field of heritage studies is in its early stages of development.Inter-and transdisciplinary collaboration can aid professionals and practitioners in developing tools for the integration of social values and engagement of the public for a more transparent, inclusive, and socially sustainable planning and management of living and working landscapes.

Limitations, bias, and the advantage of Topic Modelling
Bias is introduced in various stages of the project, e.g. in the selection of interview partners, the questions, the manual assessment process and the choice of algorithm during the analysis.We fully acknowledge that this selection of interview partners and questions 3 limits the wider representational qualities of this study, but as this is part of a wider project exploring place attachment and social values in the study area, the case studies were intended to explore a few typical cases in-depth as opposed to the study where a larger survey provided a shallower but wider overview of the reasons for place attachment (Tenzer & Schofield, 2023).
TM was chosen as a method to explore the data in a first assumption-free approach.The theme-based clustering of LDA is based on statistical principles.In other words, the researcher does not influence the outcome of the computational phase and the method is, therefore, more objective and replicable than manual labelling using predefined codes.As this step is based on complex mathematics and statistics, it is often treated as a black box.Therefore, we need to document the choice of model algorithm and parameter settings, such as number of topics and iterations.
The manual observation and categorisation of the data in the second stage of the process introduced further bias in this analysis.While this process was based on the assumptionfree approach through TM, the human factor introduces subjectivity and researcher bias at this stage.Documentation of this process can ensure rigorous and transparent procedures during this step.
While the interview dataset is too small for reliable statistical analysis, TM has advantages that can bring new insights and perspectives that would otherwise perhaps not be discovered.It allows empirical data to 'speak for themselves' (Tenzer & Schofield, 2023).Furthermore, it enables researchers to scale their research by applying an algorithm and workflow tested on a small dataset to larger datasets (see also Tenzer & Schofield (2023).Such analysis can subsequently be used on a rolling basis reviewing similar data for future analysis.This process can help understand changing attitudes towards places over time as a longitudinal study.

Conclusion
This paper began with the recognition that case studies provide deep insight, adding 'insider' knowledge as background for decision-making and planning, which is still dominated by 'objective' outsider assessments.Sustainable development depends on buy-in of residents and communities through recognising people's visions, demands and needs.People-centred approaches offer opportunities for conversation and collaboration to increase the qualities of place and strengthen the appreciation of places.
To gain insight into themes latent in the interviews -following principles of Grounded Theory -we applied Topic Modelling to cluster and label topics emerging from the empirical data.This approach provided a framework from which we developed categories of social values to explore and analyse reasons for developing a sense of place, identity and belonging.The discussion around socially sustainable change and development can be aided by active reflection on the historic landscape and help to develop a better and more nuanced understanding of social values related to working and living landscapes in the agenda-setting and planning of the historic environment.Understanding the reasons for people's practical and emotional connection to places can foster care for the natural and cultural environment, promote inclusivity, and enable socially sustainable landscape management that does not alienate people but recognises and understands place attachment.

Notes
1. https://www.peakdistrict.gov.uk/learning-about/news/70-years-of-the-peak-district-national-park/peak-district-facts 2. From here, the following terminology will be used: Topics created by the algorithm will be referred to as topic_number and for the label: keyword_keyword.Manually created topics and labels for the code book will be presented in the format: Topic þ number and "Code Label".The format: Topic number1/number2 shows the correlation of automatic and manual labels.For example, Topic 16/1 refers to the Topic Model label topic_16 correlated with the manual category 1. 3. Semi-structured interviews used the same "general questions" (part 1) and "other questions" (part 3).Part 2 of the interviews focused on the specific cases and associated themes.This free conversational part introduced researcher bias and themes linking to other interviews, e.g., Did they use traditional farming methods?What did they think about quarries, drystone walls, wildflower meadows, rewilding/renaturalisation? impact, Artificial Intelligence in heritage management, mapping and visualisation of complex, abstract concepts, QGIS.
John Schofield teaches cultural heritage management and contemporary archaeology in the Archaeology Department, University of York (UK).Prior to this he worked for Historic England's Characterisation Team, where he developed a particular interest in understanding people's social and communal values for everyday places.John holds adjunct status at the universities of Turku (Finland) and Flinders and Griffith (Australia).He is a Corresponding Fellow of the Australian Academy of the Humanities and a Fellow of the Society of Antiquaries in London.His latest book, Wicked Problems for Archaeologists, with Oxford University Press, is due for publication in 2024.

Figure 1 .
Figure 1.A flowchart for the Topic Modelling methodology developed in this research.The aim is to create the topics based on interviews and develop a general observation of landscape perception (Key: white: manual process, grey: automated process).

Figure 2 .
Figure 2. Map of the study area: Peak District National Park.The coloured parts of the park show the location of Protected Areas (PA: SSSI, National Nature Reserve, Ancient Woodland).Location of the interview participants are marked in red.Source: PA map data: https://designatedsites.naturalengland.org.uk/GreenInfrastructure/Map.aspx.Basemap # Crown copyright and database right 2022, Map tiles by Stamen Design, under CCBY 3.0.Data by OpenStreetMap, under ODbL.

Figure 3 .
Figure 3. Model creation with the Topic Modelling algorithm.Sixty Models were created (k).The best topic coherence in a model can be assessed by the coherence score (association of relevant keywords in the documents and their best fit to each other).High coherence is achieved at 37 topics with a flattening of the curve for more topics.
building material, object' (Topic 32/14) (Supplementary Material 3).The manual evaluation resulted in 27 sub-codes ranging from personal life history, connection to the past through landscape history, pro-environmental action, and climate change awareness to perception of working conditions in a national park and the advantages and disadvantages of using traditional skills and methods and summarised in seven overarching topics: Pro-environmental Action (PEA) (n ¼ 52), Challenges (CHA) (n ¼ 25), Change and Continuation (CAC) (n ¼ 74), Communities (COM) (n ¼ 13), People/Place Engagement (PPE) (n ¼ 26), Place History (PLH) (n ¼ 41), Landscape Quality (LSQ) (n ¼ 67) (Figure4).Supplementary Material 3 shows how the Topic Modelling and the manual topics correlate and are categorised.

Figure 5 .
Figure 5. Heatmap demonstrating the distribution of categories over the interviews.Each interview contains between three and six categories in varying degrees of intensity.The count shows the number of mentions of each category within each interview.

Figure 6 .
Figure 6.Typical character of the Peak District National Park landscape with the small village of Castleton in the background, Peveril Castle to the right and the typical drystone walls and field barns.Source: photo by M. Tenzer.

Figure 7 .
Figure 7. Black Hill triangulation point in the blanket bog area, Dark Peak moorland, part of the Pennine Way long distance trail and part of the 'Moors for the Future' partnership for the restoration of blanket bog.Source: photo by M. Tenzer.

Figure 8 .
Figure 8. Cement plant quarry adjacent to the north of the plant, current state of work area.Source: photo by M. Tenzer.

Figure 9 .
Figure 9. Cement Plant building and the former area of mineral extraction now reinstated with a vegetation of mature small trees and shrubs.Example for the renaturalisation of the quarry area as envisaged for the remainder of the extraction areas.Source: photo by M. Tenzer.