Pedigree Characteristics and Formation Mechanism of Traditional Dwellings in the Liaoning Coastal Area, China

Abstract

As a key convergence zone between the Circum-Bohai Sea cultural circle and the land–sea interface of Northeast Asia, the Liaoning coastal area has been shaped by multicultural integration, endowing its dwellings with distinctive cultural hybridity and geographic adaptability. This study takes 160 traditional dwellings as samples and integrates field surveys, historical documents, and multi-source geographic data to construct a multi-dimensional feature identification system. Quantitative classification is conducted using principal component analysis and systematic clustering, and external validity is verified through historical document comparison and spatial overlay analysis. The results indicate that five dwelling pedigrees are identified: the Coastal Quadrangle Courtyard Type, the Coastal Flat-Roofed Middle Courtyard Type, the Coastal Gabled-Roof Small Courtyard Type, the Mountainous Gabled-Roof Small Courtyard Type, and the Plain Flat-Roofed Long Courtyard Type. Regarding the formation mechanism, geographic detectors reveal that the coupling effect of migration culture and topographical conditions is the dominant mechanism shaping pedigree differentiation. This study verifies the applicability of integrating quantitative and qualitative methods in dwelling research within multicultural convergence zones, constructs a pedigree framework for traditional dwellings in coastal Liaoning, and provides a theoretical basis for the systematic understanding and sustainable conservation of vernacular architectural heritage in the Circum-Bohai Sea region.

1. Introduction

1.1. Research Progress on Traditional Dwellings

Dwelling research has long been a core topic in the international interdisciplinary dialogue among architecture, anthropology, and geography, as well as an important pathway for analyzing regional cultural landscapes and vernacular built heritage. The trajectory of dwelling research originated in the mid-19th century with anthropological interpretations [1] and empirical studies on social structures [2]. In the first half of the 20th century, represented by the typological regionalization of the French School of Geography [3] and Carl Sauer’s theory of “cultural landscape” [4], an analytical framework was established that positioned dwellings as carriers of regional culture. Concurrently, global ethnographic surveys accumulated extensive knowledge of localized construction practices. The 1960s marked a critical turning point in this field. Bernard Rudofsky, in Architecture Without Architects, established the independent aesthetic and cultural value of vernacular dwellings through visual comparison [5]. Subsequently, Amos Rapoport, in House Form and Culture, systematically proposed the theory of “cultural determinism” [6], establishing a theoretical analytical approach for architectural anthropology. By the end of the 20th century, Paul Oliver’s Encyclopedia of Vernacular Architecture of the World achieved systematic integration and standardized classification of global vernacular dwelling knowledge [7].

This international academic discourse has profoundly shaped the trajectory of Chinese indigenous research [8]. A substantial foundation was established through the pioneering surveys of the Society for Research in Chinese Architecture and subsequent nationwide censuses of traditional dwellings [9]. During the urbanization process, the number of traditional villages decreased by approximately 920,000 between 2002 and 2017 [10]. Facing the plight of rapidly disappearing traditional villages, China has promulgated conservation policies since 2012, integrating the protection of traditional villages into the rural revitalization strategy, with a total of 8171 villages included across six national listings to date. Against this backdrop, domestic research on traditional dwellings has undergone a paradigm shift from singular architectural surveying and documentation toward interdisciplinary integration and the dual emphasis on holistic conservation and cultural inheritance.

In terms of research perspectives, Early research in China focused on the interaction between dwelling culture and regional characteristics [11,12]. Subsequent scholars deepened the study of spatial differentiation and cultural regionalization [13,14,15,16,17]. Some scholars have chosen traditional construction techniques as their research reference [18].

In terms of research methodologies, the study of dwellings has undergone a paradigm shift from qualitative description to quantitative analysis. Early research primarily relied on field surveys and in-depth case studies [19], concentrating on the morphology, craftsmanship, and cultural interpretation of individual buildings or single villages, representing a typical qualitative research approach. With the paradigm transformation, quantitative analysis and spatial technologies have gradually become mainstream. Scholars have widely employed tools such as geographic information systems [20], mathematical statistical models [21], and space syntax [22] to conduct integrated analysis of multi-source data, aiming to reveal core issues including the spatial distribution patterns of traditional villages and dwellings [23,24,25], drivers of morphological evolution [26], cultural landscape differentiation [27,28], and ecological adaptation mechanisms [29]. For example, a multi-method integrated approach has been adopted for studying traditional villages in western Henan Province and the entire province [30,31], while investigations of Qingshan Fishing Village in Qingdao have focused on verifying the correlation between courtyard types and wind environment comfort [32].

In terms of research areas, existing studies exhibit multi-level coverage ranging from geographical units to cultural units, and further to geographic-cultural convergence zones. Research on specific geographical units, such as the Inner Mongolia Plateau [33] and the Lingnan region [34], primarily analyzes the adaptive relationship between natural conditions and settlement morphology. Studies targeting cultural units, such as the comparison between Tujia and Miao villages in western Hunan [35], focus on the intrinsic connections between ethnic culture, social organization, and dwelling forms. Investigations of multi-geographic-cultural convergence zones, such as the border area between Jiangxi and Anhui provinces [36], emphasize the evolution and differentiation of landscape patterns against the backdrop of cultural interaction, diffusion, and integration. Although these studies vary in regional scale and research objects, they universally address core issues including spatial distribution characteristics, formation driving forces, cultural landscape differentiation mechanisms, and cross-cultural comparisons of traditional villages and dwellings, collectively constituting a multi-scale spatial cognition system for the study of vernacular architecture in China [37,38,39].

1.2. Theoretical Framework: An Integrated Perspective for Pedigree Research of Traditional Dwellings

1.2.1. Theoretical Definition of the “Pedigree” Concept

Drawing upon cultural geography [40,41], this study proposes the core concept of “dwelling pedigree” and defines it as: a typological system of traditional dwelling forms, formed through historical layering under the dual influences of multicultural integration and diverse geographical environments, exhibiting spatial differentiation patterns and cultural kinship relationships. This concept encompasses three dimensions: First, the morphological dimension refers to the categorization of identifiable features such as courtyard layout, roof form, and material construction. Based on quantifiable morphological indicators, clusters with significant commonalities are identified through methods such as systematic clustering. Second, the spatial dimension refers to the distribution patterns and spatial differentiation patterns of different types at the regional scale. ArcGIS (version 10.7.0.10450) spatial analysis is employed to reveal the overlay relationships between type distribution, geographical environment, and cultural regionalization. Third, the cultural dimension refers to the “gene” trajectories of migration dissemination, ethnic integration, and regional cultural exchange embedded within dwelling forms. Drawing upon the theory of “cultural diffusion” in cultural geography [42] and the perspective of architectural communication studies [43], this dimension traces the “cultural kinship” of morphological characteristics through proxy variables such as dialect divisions, ethnic distributions, and migration routes, thereby establishing connections between architectural form and historical population movements.

“Pedigree” is distinct from typology, morphology, and cultural regionalization: typology asks “what the form is,” whereas pedigree further investigates the source–flow relationships among different types; morphology traces the evolutionary process of individual elements, whereas pedigree focuses on the “family resemblance” structure that emerges during evolution, viewing different types as kinship branches sharing a common cultural origin. Thus, pedigree takes the morphological classification of typology as its foundation, the evolutionary patterns of morphology as its thread, and the cultural diffusion framework of cultural geography as its explanatory structure, ultimately constructing an analytical system of “morphological type—spatial pattern—cultural origin.” This framework moves dwelling research from static description toward mechanistic explanation, providing an integrated theoretical tool for the study of traditional dwellings in multicultural convergence zones.

1.2.2. Theorizing the Relationship Between Cultural Diffusion, Environmental Adaptation, and Spatial Morphology

To better understand the complex formation of traditional dwelling forms, this study draws on and expands the analytical framework of cultural geography [44,45]. It establishes a theoretical connection between cultural diffusion, environmental adaptation, and spatial morphology. Cultural diffusion and environmental adaptation together shape the logic behind spatial form. Together, they form a dual mechanism of “prototype–variation”.

On one hand, cultural diffusion provides the “prototype” for dwelling forms: diffusion routes determine the direction of spread, while the mode of diffusion affects the completeness of the prototype. The cultural origin serves as the starting point, where core features gradually weaken as they spread outward in space, forming a gradient pattern.

On the other hand, the geographical environment imposes constraints on the incoming “prototype.” Terrain influences how forms are organized, climate drives adaptive changes in construction techniques, and available resources determine material use and structural logic. As a result, the prototype undergoes “variation” during its spread, giving rise to “variants” that suit local conditions.

These two mechanisms do not operate separately in sequence. Instead, they work together in an intertwined process. Diffusion routes determine the initial distribution of the prototype, while the geographical environment sets boundaries and provides direction for variation. At the same time, environmental conditions can also influence the paths and intensity of cultural diffusion.

Thus, spatial form is the result of both the “prototype” from cultural diffusion and the “variation” driven by environmental adaptation. Their interaction over time and space creates kinship relationships among different dwelling types—sharing common origins while evolving into localized variants. Ultimately, this forms a coherent pedigree structure.

1.2.3. The Relationship Between This Study and Traditional Cultural Geography

This study is grounded in cultural geography, but it advances the traditional approach in terms of methodology. Traditional cultural geography often focuses on qualitative description and static divisions of cultural regions in its study of cultural landscapes. Its discussions on cultural diffusion and environmental adaptation also tend to stay at the level of qualitative inference.

Building on this foundation, this study introduces the concept of “pedigree.” It expands the identification of dwelling morphological features from qualitative description to quantitative classification based on principal component analysis and systematic clustering. At the same time, it applies the theory of “cultural diffusion” through quantitative methods such as geographic detectors. The influence of “cultural diffusion routes” and “geographical environmental constraints” on morphological distribution is transformed into a measurable analysis of “culture–geography coupling intensity.” In this way, the study moves from a qualitative divide between “environmental determinism” and “cultural determinism” to a quantitative test of coupling effects.

Furthermore, through a three-dimensional pedigree framework of “morphological type–spatial pattern–cultural origin,” this study expands the static description of cultural landscapes in cultural geography into a tracing of cultural processes. As a result, the groupings obtained through statistical methods gain explanatory depth from the perspective of cultural geography.

1.3. Typicality and Academic Value of the Study Area

The Liaoning coastal area (38°72′–41°90′ N, 118°85′–123°91′ E), as a key convergence zone within the Circum-Bohai Sea cultural circle and the land–sea interface of Northeast Asia [46], holds unique academic value for studying the formation and evolution of dwelling morphology. It serves as a typical case for analyzing the formation mechanisms of vernacular architecture in multicultural convergence zones. The distinctiveness of this region is primarily reflected in the following three dimensions:

1.3.1. The Pivotal and Transitional Nature of Geographical Space

The Liaoning coastal area lies at the convergence of the Central Plains agrarian civilization, the northeastern forest-steppe nomadic and fishing-hunting civilization, and the maritime civilization. It is the junction of three major cultural plates in East Asia. From a transportation perspective, this area is both the eastern terminus of the Liaoxi Corridor—the land throat connecting North China and Northeast China—and the northern endpoint of the sea route between the Shandong Peninsula and the Liaodong Peninsula via the Miaodao Islands—Lüshun—Dalian. This “dual land–sea hub” location has historically made it a place that must be passed through for north–south cultural transmission and east–west ethnic migration. Archaeologist Su Bingqi placed the Circum-Bohai Sea region within the broader archaeological perspective of “from the Circum-Bohai Sea to the Circum-Japan Sea,” noting it as a convergence zone for ancient cultures connecting China, Korea, Japan, and Russia. The study area is precisely the core component of this Northeast Asian land–sea convergence zone.

1.3.2. The Hybridity and Layering of Cultural Composition

Compared to other regions in North China, such as the Hebei Plain and the Shandong Peninsula, which have relatively singular cultural origins, the cultural composition of the Liaoning coastal area exhibits complex characteristics of multi-source layering. The differences are primarily reflected in the following three aspects:

Duality of Transmission Routes: Dwellings on the Shandong Peninsula primarily received cultural radiation from the Central Plains via land routes, representing a relatively singular transmission path. Dwellings on the Hebei Plain relied on the close-range dissemination of metropolitan culture, presenting a concentric diffusion pattern centered on the capital. In contrast, the Liaoning coastal area was simultaneously subjected to the interaction of two major cultural transmission systems. Jiaodong immigrants introduced Qilu architectural techniques and courtyard forms to Liaodong via sea routes, while official metropolitan architectural culture spread northeastward along the post road system via land routes.

Pluralism of Ethnic Composition: Unlike the core Han cultural regions of North China (e.g., Hebei and Shandong, where the Han population exceeds 98%), the Liaoning coastal area has historically been a place where multiple ethnic groups, including Manchu, Mongolian, Han, Xibe, and Korean, lived in interwoven distribution. According to the Gazetteer of Liaoning Province: Ethnic Minorities [47] Manchu populations were concentrated in areas like Xiuyan and Fengcheng, Mongolians were clustered around Chaoyang and Fuxin, and Koreans were mostly distributed in riverside and coastal areas such as Dandong. The layering and integration of ethnic cultures injected multiple genes into the dwelling forms—the kang and wanzi kang layouts of the Manchu, the stone material preference of the Mongols, and the warm kang and lightweight timber structures of the Koreans all left their marks on local dwellings.

Diachronicity of Cultural Layering: The cultural integration in this area was not a one-time “mixing” but a “layering” process spanning millennia. Cross-sea cultural exchanges existed between Liaodong and Jiaodong in the Neolithic period. During the Spring and Autumn and Warring States periods, the Lai people fled across the sea to Liaodong, marking the beginning of the “Chuang Guandong” migration history. In the Qin and Han dynasties, Han people from the Central Plains migrated in large numbers via sea routes. During the Wei and Jin periods, scholars like Guan Ning and Bing Yuan crossed the sea to seek refuge in Liaodong. In the Liao, Jin, and Yuan dynasties, the Khitan, Jurchen, and Mongol peoples moved south through this area. Since the Ming and Qing dynasties, military colony immigrants and the wave of “Chuang Guandong” migration continuously poured in. This millennia-long process of cultural layering caused cultural elements from different periods and origins to sediment, superimpose, and integrate within the same space.

1.3.3. Complexity of Environmental Adaptation: Diversified Geographical Constraints

The Liaoning coastal area faces multiple geographical constraints. The coastal zone must contend with strong winds and tidal influences, experiencing 30–50 windy days annually, with relative humidity often exceeding 70%. The mountainous and hilly areas must cope with steep slopes, soil erosion, and severe winter cold, with extreme low temperatures in the Chaoyang mountain area reaching −30 °C. The Liaohe Plain must address low-lying, flood-prone conditions and cold winter winds. Dwelling forms within this single region must respond to diverse environmental challenges, thus giving rise to a rich system of adaptive techniques, providing an ideal research sample for examining the relationship between architectural form and environmental factors.

1.4. Research Gaps and Positioning of This Study

Although research on traditional Chinese dwellings is quite extensive, systematic studies specifically targeting the Liaoning coastal area—a key convergence zone within the Circum-Bohai Sea cultural circle and the land–sea interface of Northeast Asia [46,48]—remain relatively weak. As a typical area of multicultural integration in northern China, its settlement site selection, spatial organization, and construction methods in coastal and land–sea transitional zones differ significantly from those on inland plains [49]. Historical immigration, coastal defense systems, the fishing and salt economy, and cross-sea interactions have jointly shaped the distinctive cultural hybridity and geographic adaptability of dwellings in this area.

Existing research mainly falls into two categories. The first consists of case analyses at the level of architectural typology, such as descriptions of the form, structure, and decoration of courtyard dwellings in Liaoning [50] and stone masonry dwellings in western Liaoning [51]. The second involves macroscopic investigations from the perspective of cultural geography, such as studies on the correlation between the “Jiaoliao ethnic group” concept and dialect divisions [52,53,54,55]. While these findings have foundational value, they are mostly limited to qualitative descriptions and localized case studies. A systematic pedigree study from a regional holistic perspective is still lacking. This deficiency is primarily reflected in two aspects: insufficient regional integration, characterized by a lack of systematic comparison treating the northern Bohai area as an independent cultural-geographical unit; and weak quantitative support, reflected in the absence of refined analysis and visualization based on GIS and mathematical statistics regarding the spatial differentiation patterns of dwelling morphological features.

Addressing the above research gaps, the core theoretical question of this study can be distilled as follows: In the multicultural convergence zone of the Liaoning coastal area, what pedigree structure do traditional dwelling forms exhibit? How can the patterns of their spatial differentiation and formation mechanisms be revealed through quantitative methods?

Compared to existing regional dwelling studies, this research is characterized by the following aspects: First, in terms of object, it focuses on the Liaoning coastal area within the Circum-Bohai Sea cultural circle, which possesses both mountainous and coastal geographical features, filling a gap in systematic pedigree research for this region. Second, in terms of methodology, unlike purely qualitative cultural regionalization studies in cultural geography, this study achieves “data-driven” regionalization through quantitative clustering, providing a scientifically reproducible basis for delineating cultural area boundaries and types. Third, theoretically, through a three-dimensional definition of the “pedigree” concept, it integrates morphological types, spatial patterns, and cultural origins into a unified analytical framework, offering a new explanatory path for exploring the formation mechanisms of dwellings in multicultural convergence zones. Fourth, at the level of international academic dialogue, this study engages with cutting-edge issues in international vernacular architecture research. Paul Oliver’s Encyclopedia of Vernacular Architecture of the World pioneered the organization of content by cultural background rather than political boundaries, laying a foundation for cross-regional comparative theoretical frameworks [7]. Scholars like Lindsay Asquith, in Vernacular Architecture in the Twenty-First Century, emphasize that vernacular architecture research must balance specific cultural contexts with universal theoretical frameworks [56]. By taking the Liaoning coastal area as a case study to explore identification methods and formation mechanisms of dwelling pedigrees in multicultural convergence zones, this study reflects regional specificity while also pursuing general significance through quantitative methods and theoretical construction, thus helping to situate Chinese vernacular architecture research within the broader perspective of international academic discourse.

2. Materials and Methods

2.1. Study Area

Liaoning Province is situated in the southern part of Northeast China, bordering the Songliao Plain to the north and the Bohai Sea and Yellow Sea to the south [47,54]. Historically, it has long been positioned at the forefront of convergence and interaction among the Central Plains agrarian civilizations, the northeastern forest-steppe nomadic and fishing-hunting civilization, and the maritime civilization. The coastal area of southern Liaoning (38°72′–41°90′ N, 118°85′–123°91′ E) covers a total area of approximately 71,464 square kilometers, including the coastal zones of Chaoyang, Huludao, Jinzhou, Panjin, Yingkou, Anshan, and Dalian. Based on the influence range of sea winds, the study area was delineated using the Liaoning Bohai coastline as a benchmark, extending inland approximately 150 km. Integrating current administrative divisions, a fan-shaped study area encompassing the Liaoning coastal zone was ultimately determined. The topography generally slopes from north to south and from the eastern and western sides toward the central region, which can be summarized as “60% mountains, 10% water, and 30% farmland”. The western Liaoning mountainous and hilly area belongs to the extensions of the Songling and Nuluerhu mountains, while the eastern Liaoning mountainous and hilly area is part of the Changbai Mountain range. The Liaodong Peninsula is predominantly hilly with a indented coastline. The Liaohe Plain, formed by alluvial deposits from the Liao River and its tributaries, features low-flat and open terrain. The coastal area of southern Liaoning has a temperate monsoon climate, primarily located in the cold region IIA zone of China’s building climate classification, while the Chaoyang area in western Liaoning falls within the cold region IIB zone [55]. From the early Hongshan Culture to the present, this region has preserved many traditional dwelling structures. This study focuses on 160 traditional dwelling samples within this area. According to historical records, local chronicles, and extant architectural structures in villages, the identifiable dwellings were constructed during the Ming Dynasty (1368–1644), the Qing Dynasty (1644–1912), and the modern period (1912–1949).

2.2. Data Sources

The selection of traditional dwellings in the Liaoning coastal area was based on three primary sources: (1) one to three representative dwellings were selected from each village officially recognized and listed in the six national batches of “Traditional Chinese Villages” within the Liaoning coastal area [57]; (2) one to three representative dwellings were selected from each provincially recognized traditional village; (3) to ensure comprehensive coverage of the Liaoning coastal area, a stratified sampling method was employed to select one to three dwellings with typical geographical environments and representative cultural characteristics from each township. Through screening and organization, removing duplicates and excluding highly urbanized dwellings, a total of 160 traditional dwellings were ultimately identified as samples.

To facilitate identification and classification, this study employed a geographic location coding method to systematically name the 160 dwelling samples. The coding rules are as follows: the first letter represents the prefecture-level administrative division code (Dalian-A, Yingkou-B, Panjin-C, Jinzhou-D, Huludao-E, Chaoyang-F, Anshan-G); the second letter represents the pinyin initial of the district or county; the suffix number indicates the sequential order of dwellings within that district or county. For example, “AZ8” represents the eighth surveyed sample in Zhuanghe City (Z) of Dalian City (A).

To unify the geographic coordinates of traditional dwellings in the Liaoning coastal area, the MapLocation tool was used to batch convert longitude and latitude data, ensuring all coordinates adopted the WGS-84 coordinate system. These standardized coordinates were subsequently converted into vector data in shapefile (shp) format through ArcGIS, generating a spatial distribution map of traditional dwellings in the Liaoning coastal area. Administrative boundary data were obtained from the National Geomatics Center of China. Basic geographic information data were primarily sourced from the Geospatial Data Cloud (https://www.gscloud.cn, accessed 1 December 2024); these data were subsequently imported into ArcGIS to obtain topographic maps of the Liaoning coastal area. Dialect data were derived from the Chinese dialect maps of Northeastern Mandarin, Jiaoliao Mandarin, and Beijing Mandarin in the Chinese Language Atlas (1987) [58]. Provincial and municipal administrative boundary data were obtained from the DataV.GeoAtlas data visualization platform. The downloaded JSON format files were converted to shp format and imported into ArcGIS (Figure 1).

The samples in this study are not completely evenly distributed across the study area. The numbers of samples from Dalian and Chaoyang are relatively large, while those from Panjin and Anshan are relatively small. This pattern is mainly constrained by objective factors: the spatial distribution of traditional village resources is inherently uneven, with mountainous and hilly areas preserving more complete dwellings while plain areas have fewer; the impact of urbanization on the survival of traditional dwellings varies by region, with some emerging industrial cities having a limited number of surviving traditional dwellings; additionally, during stratified sampling, some townships had no qualified traditional dwellings, resulting in fewer samples from those areas. Therefore, the sample distribution reflects the actual spatial pattern of traditional dwelling resources rather than subjective selection.

Furthermore, this study only selects “representative dwellings” for analysis, which may introduce two types of bias: first, survivorship bias—well-preserved dwellings may not fully reflect the characteristics of ordinary dwellings; second, typological bias—the researcher’s judgment may tend to select samples that conform to expected types. To mitigate the above biases, this study adopts the following measures: (1) record the overall village landscape during field surveys to establish contextual understanding; (2) employ stratified sampling to ensure coverage of all administrative and geographical units; (3) use data-driven methods (principal component analysis and systematic clustering) to make the classification based as much as possible on the intrinsic structure of the indicators rather than the researcher’s presuppositions; (4) validate the clustering results through external data sources such as local gazetteers, dialect divisions, and migration routes. These measures collectively ensure the reliability of the research findings.

2.3. Research Tools and Methods

The research approach of this paper encompasses the construction of an identification system, systematic clustering, type synthesis and feature extraction, and exploration of formation mechanisms (Figure 2). It employs multidisciplinary theories and methods including literature analysis and field investigation, geographic information system analysis, principal component analysis, and systematic clustering analysis.

2.3.1. Literature Analysis and Field Survey

This study integrates literature analysis and field survey to provide empirical support for the construction of the dwelling pedigree and the analysis of formation mechanisms of traditional dwellings in the Liaoning coastal area. In terms of literature analysis, it systematically reviews historical gazetteers, local chronicles, architectural history and geography literature, as well as genealogies and historical maps. In terms of field survey, field investigations were conducted on 160 traditional dwellings between 2023 and 2025, obtaining data on courtyard layout, building dimensions, and roof forms through architectural surveying and morphological documentation, while collecting information on construction periods, artisan traditions, and migration memories through oral history interviews and the collection of folk documents.

2.3.2. ArcGIS Spatial Analysis Methods

Using the ArcGIS platform for vectorization [59], sample information including geographic and cultural data as well as dwelling pedigree characteristics was input to construct a data-driven geographic information database of traditional dwellings. Distribution maps of different factors were generated to visualize their spatial distribution characteristics.

2.3.3. Principal Component Analysis

Principal Component Analysis (PCA) is a data structure simplification method based on the concept of dimensionality reduction, playing an important role in multi-indicator evaluation systems [60]. This method transforms multiple factor variables related to dwellings into a few mutually independent principal components. While retaining the main information of the original data, it eliminates dimensional differences and multicollinearity issues among indicators, thereby enhancing the validity and comparability of information carried by cultural factors. The extraction of principal components relies entirely on mathematical statistical methods to determine weight relationships among variables, thereby avoiding the subjective bias inherent in traditional “expert scoring” approaches and ensuring a more objective and scientific analysis. The extraction steps are as follows:

$$\mathrm{R}={\left[{\mathrm{r}}_{\mathrm{i}\mathrm{j}}\right]}_{\mathrm{p}\times \mathrm{p}}={\displaystyle \frac{{\mathrm{Z}}^{\mathrm{T}}\mathrm{Z}}{\mathrm{n}-1}}$$

(1)

(1) Calculate the eigenvalues of the correlation coefficient matrix R:

$$|\mathrm{R}-\mathsf{\lambda }{\mathrm{I}}_{\mathrm{p}}|=0$$

(2)

Solve for _P eigenvalues ${\mathsf{\lambda }}_1\ge {\mathsf{\lambda }}_2\ge \cdots \ge {\mathsf{\lambda }}_{\mathrm{P}}\ge 0.$

(2) Determine the number of principal components, m, so that the extraction rate of principal component factor information reaches more than 50%:

$${\displaystyle \frac{{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{m}}}{\mathsf{\lambda }}_{\mathrm{j}}}{{\displaystyle {\sum }_{\mathrm{j}=1}^{\mathrm{P}}}{\mathsf{\lambda }}_{\mathrm{j}}}}\ge 0.8$$

(3)

2.3.4. Systematic Clustering Analysis

Systematic clustering was applied to classify the 160 traditional dwelling samples. The five principal components were used as sample attributes, with the between-groups linkage method and squared Euclidean distance as the distance measure. Based on the dendrogram and the interpretability of spatial distribution, the optimal number of clusters was determined as 5.

3. Distribution Characteristics and Clustering Results

3.1. Establishment of a Feature Identification System for Traditional Dwellings

To systematically identify, classify, and analyze the architectural features of traditional dwellings in the Liaoning coastal area, this study constructed a multi-dimensional feature identification indicator system. This system aims to transcend the limitations of fragmented or qualitative descriptions in previous research by establishing a comprehensive, structured, and comparable analytical framework through integrated quantitative and categorical variables, thereby laying a foundation for subsequent clustering analysis, pedigree feature synthesis, and formation mechanism research.

3.1.1. Selection of Feature Identification Indicators for Traditional Dwellings

The system is organized around three core dimensions: dwelling spatial morphology, geographical environment, and cultural attributes. Among these, dwelling spatial morphology serves as the material carrier of environmental adaptation and cultural expression. The geographical environment dimension analyzes the interactive mechanisms between dwellings and topography. The cultural attribute dimension explains regional similarities and differences in architectural morphology from the perspectives of cultural transmission and ethnic integration. In terms of specific indicator selection, courtyard prototype and relative position of the main hall reflect family structure and the logic of production and daily life, serving as the primary basis for type differentiation. Courtyard area, aspect ratio, and main hall area are quantitative indicators that directly reflect scale differences and are closely related to topographical conditions and land resource endowments. Window-to-wall ratio is a sensitive indicator of climate adaptability, proven in building thermal research to be directly related to winter insulation and summer heat protection. Main wall materials and timber frame form reflect the principle of utilizing locally available materials and the transmission of construction techniques, serving as material evidence for identifying cultural transmission routes. Although decoration type is not a structural element, as a concentrated expression of cultural aesthetics, it has indicative significance for tracing cultural origins. Dialect divisions and ethnic types are used as proxy variables for cultural transmission and are widely adopted in cultural geography research.

In addition to the 13 selected indicators, chimney form, roof slope, and courtyard orientation were excluded due to low discriminative power or high collinearity with retained variables.

3.1.2. Indicator Classification and Standardization Processing

To enhance the scientific rigor and operability of the indicator system, this study systematically classified each identification indicator based on first-hand data from large-scale field surveys, and assigned clear label values to each category according to actual morphological characteristics. For example, courtyard prototype was classified into eight types based on the arrangement of the main hall, wing rooms, and inverted house, with corresponding numerical labels assigned (e.g., main hall-only type = 1), transforming qualitative features into discrete variables applicable for quantitative analysis. Based on data nature, all indicators were systematically divided into descriptive and numerical types: descriptive indicators (e.g., courtyard prototype, main wall materials) were converted into discrete factor variables through preset classification systems and label values; numerical indicators (e.g., courtyard area, window-to-wall ratio) are continuous numerical variables directly reflecting quantitative characteristics such as scale and proportion.

Numerical factors, due to differences in information type and units of measurement, cannot be directly subjected to statistical analysis and require standardization to eliminate the influence of units on the analysis.

(1) Construct the original data sample matrix X of dwelling feature identification factors.

$$X=\left[\begin{array}{c}{x}_1^{\mathrm{T}}\\ x_2^{\mathrm{T}}\\ ⋮\\ x_n^{\mathrm{T}}\end{array}\right]=\left[\begin{array}{cccc}{x}_{11}& x_{12}& \dots & x_{1p}\\ x_{21}& x_{22}& \dots & x_{2p}\\ ⋮& ⋮& \ddots & ⋮\\ x_{n1}& x_{n2}& \dots & x_{np}\end{array}\right]$$

where x_ij represents the value of the j-th indicator in the i-th sample, n is the number of samples, p is the number of indicators, and n > p.

(2) Perform standardization to obtain matrix Z, with data transformation conducted using SPSS (version 27) to eliminate inconsistencies in indicator types and units of measurement within matrix X.

$$Z=\left[\begin{array}{c}{Z}_1^{\mathrm{T}}\\ Z_2^{\mathrm{T}}\\ ⋮\\ Z_n^{\mathrm{T}}\end{array}\right]=\left[\begin{array}{cccc}{z}_{11}& z_{12}& \dots & z_{1p}\\ z_{21}& z_{22}& \dots & z_{2p}\\ ⋮& ⋮& \ddots & ⋮\\ z_{n1}& z_{n2}& \dots & z_{np}\end{array}\right]$$

Standardization formula:

$$Z_{ij}={\displaystyle \frac{x_{ij}-{\overline{x}}_j}{s_j}}$$

where the sample mean $\overline{\mathrm{x}}$_j and sample standard deviation s_j are, respectively:

$${\overline{\mathrm{x}}}_{\mathrm{j}}={\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}},{\mathrm{s}}_{\mathrm{j}}=\sqrt{{\displaystyle \frac{1}{\mathrm{n}-1}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}({\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\overline{\mathrm{x}}}_{\mathrm{j}}{)}^2}$$

It should be noted that the numerical encoding of categorical variables such as wall material, decoration type, and ethnic type (as shown in

Table 1. Feature Identification Indicator System for Traditional Dwellings in the Liaoning Coastal Area.

Identification Factor	Identification Indicator	Corresponding Indicator Label Values	Data Type
Dwelling Spatial Morphology	Courtyard Prototype	Main house only = 1; Main house + east wing room = 2;	Descriptive
		Main house + west wing room = 3;
		Main house + inverted house = 4;
		Main house + east and west wing rooms = 5;
		Main house + east wing room + inverted house = 6;
		Main house + west wing room + inverted house = 7;
		Quadrangle courtyard = 8
	Relative Position of Main House	No inverted house, with backyard, no side yard = 1;	Descriptive
		No inverted house, with backyard, single side yard = 2;
		No inverted house, with backyard, double side yards = 3;
		No inverted house, no backyard, no side yard = 4;
		No inverted house, no backyard, single side yard = 5;
		No inverted house, no backyard, double side yards = 6;
		With inverted house, with backyard, no side yard = 7;
		With inverted house, with backyard, single side yard = 8;
		With inverted house, with backyard, double side yards = 9;
		With inverted house, no backyard, no side yard = 10;
		With inverted house, no backyard, single side yard = 11;
		With inverted house, no backyard, double side yards = 12
	Length-to-Width Ratio of the Courtyard	Numerical value	Numerical
	Courtyard Area	Numerical value	Numerical
	Main House Area	Numerical value	Numerical
	Window-to-Wall Ratio of the Main House	Numerical value	Numerical
	Wall Material	Adobe = 1; Stone = 2; Brick = 3;	Descriptive
		Stone base with brick upper = 4;
		Stone base with adobe upper = 5;
	Main House Structure	Triangular truss = 1; Flat-roofed = 2; Modern structure = 3	Descriptive
	Decoration Type	None = 1; Brick carving = 2; Stone carving = 3;	Descriptive
		Colored patterns = 4; Wood carving = 5; Mixed = 6
Geographical Environment	Site Adaptability	Terrace (coastal) = 1; Terrace (non-coastal) = 2;	Descriptive
		Terrace (riverside) = 3; Flat land (coastal) = 4;
		Flat land (non-coastal) = 5; Flat land (riverside) = 6
	Settlement Morphology	Linear = 1; Clustered = 2; Finger-shaped = 3	Descriptive
Cultural Attributes	Dialect Division	Beijing Mandarin = 1; Northeastern Mandarin = 2;	Descriptive
		Jiaoliao Mandarin = 3
	Ethnic Type	Han = 1; Manchu = 2; Mongolian = 3; Xibe = 4; Korean = 5	Descriptive

) serves only as operational labels. These encodings do not imply a real metric distance between categories. Therefore, the results of the principal component analysis and clustering should be interpreted in typological terms, rather than as strict quantitative relationships between categories.

In summary, by integrating the three dimensions of dwelling spatial morphology, geographical environment, and cultural attributes, and by establishing a classification and quantitative processing method based on field investigations, this system constructs a methodological framework that is clear, well-founded, and highly operable. It provides a systematic analytical foundation for feature identification, pedigree characteristic synthesis, and formation mechanism research of traditional dwellings in the Liaoning coastal area.

3.2. Classification and Spatial Mapping of Identification Indicators

Based on the construction of the information database, this study conducted ArcGIS spatial mapping and indicator classification for the 13 dwelling identification factors (Table 2). The results indicate the following:

(a)

Courtyard prototypes can be classified into eight types, among which the main hall-only type is the most numerous and widely distributed. Two-wing courtyard types are relatively dispersed, while three-wing and quadrangle courtyard types are concentrated in ancient towns such as Yixian Ancient City and Qingdui Ancient Town.

(b)

The relative position of the main hall is divided into 12 types, generally exhibiting characteristics of front-back multiple courtyards and single or double side yards adjacent to gable walls. The layout correlates with geographical elevation—for instance, plain areas tend to have side yards, whereas mountainous areas predominantly lack side yards.

(c)

Courtyard aspect ratios are concentrated between 0.5 and 2.5, significantly influenced by geographical environment—courtyards in the Chaoyang mountainous area and coastal Dalian are nearly square, while courtyards in the Liaodong Bay plains are predominantly rectangular. Courtyard areas are mostly concentrated between 250–650 m², with large courtyards primarily found in radiation zones and along the Yellow Sea coast of the Liaodong Peninsula, while small courtyards are distributed in hilly and mountainous areas. The aspect ratio and area show a positive correlation.

(d)

Main hall areas are concentrated between 50–150 m², with small areas predominantly found in the Liaodong Peninsula and large areas concentrated along the coastal plains of Liaodong Bay, correlating with distance from the coast and elevation. Window-to-wall ratios are concentrated between 0.19 and 0.45, with relatively low values mainly distributed in Dalian and Chaoyang, related to elevation, and showing a positive correlation with main hall area.

(e)

Main wall materials primarily include five types: adobe, stone, brick, stone base with adobe upper, and stone base with brick upper. Their distribution reflects local adaptation, utilization of locally available materials, and economic development differences, correlating with window size.

(f)

Main hall timber frame forms mainly include triangular truss, flat-roofed type, and modern structures. The flat-roofed type is widely distributed, while triangular truss is concentrated in the Liaodong Peninsula and southwestern Chaoyang.

(g)

Decoration types are divided into six categories: stone carving, brick carving, wood carving, colored patterns, granitic plaster with colored patterns, and mixed types. Overall, decorations in Liaoning coastal dwellings are relatively simple, while ancient towns exhibit more abundant decorations.

Through analogical overlay, the spatial distribution of dwelling morphology shows strong correlations with cultural and geographical factors, namely site adaptability, settlement morphology, dialect divisions, and ethnic types.

Table 2. Classification and Spatial Mapping of Identification Indicators. Figure (h) illustrates three primary roof structure types [61]: triangular truss, flat-roofed (tunding), and modern structures. The map presents the general distribution characteristics of each type without further subdividing the internal subtypes within each category (e.g., different forms of triangular truss, variations in the flat roof). This is partly constrained by the classification granularity of the current study, while also leaving room for future research—more detailed field investigations and typological analyses can be conducted to further explore the morphological variations and regional adaptation logics within each structural type. Figure (l) is adapted from the dialect division map in the Language Atlas of China (2nd edition) [58]. Based on the original classification system and boundary delineations, the map has been appropriately simplified and locally adjusted for the purposes of the study area and visualization.

Classification of Identification Indicators	Spatial Mapping of Identification Indicators	Legend
(a) Courtyard Prototype
(b) Relative Position of Main House
(c) Window-to-Wall Ratio of the Main House
(d) Courtyard Area
(e) Main House Area
(f) Length-to-Width Ratio of the Courtyard
(g) Wall Material
(h) Main House Structure
(i) Decoration Type
(j) Site Adaptability
(k) Settlement Morphology
(l) Dialect Division
(m) Ethnic Type

Table 2. Classification and Spatial Mapping of Identification Indicators. Figure (h) illustrates three primary roof structure types [61]: triangular truss, flat-roofed (tunding), and modern structures. The map presents the general distribution characteristics of each type without further subdividing the internal subtypes within each category (e.g., different forms of triangular truss, variations in the flat roof). This is partly constrained by the classification granularity of the current study, while also leaving room for future research—more detailed field investigations and typological analyses can be conducted to further explore the morphological variations and regional adaptation logics within each structural type. Figure (l) is adapted from the dialect division map in the Language Atlas of China (2nd edition) [58]. Based on the original classification system and boundary delineations, the map has been appropriately simplified and locally adjusted for the purposes of the study area and visualization.

Classification of Identification Indicators	Spatial Mapping of Identification Indicators	Legend
(a) Courtyard Prototype
(b) Relative Position of Main House
(c) Window-to-Wall Ratio of the Main House
(d) Courtyard Area
(e) Main House Area
(f) Length-to-Width Ratio of the Courtyard
(g) Wall Material
(h) Main House Structure
(i) Decoration Type
(j) Site Adaptability
(k) Settlement Morphology
(l) Dialect Division
(m) Ethnic Type

3.3. Principal Factor Analysis and Systematic Clustering Method

3.3.1. Model Testing

Based on the feature identification system for traditional dwellings in the Liaoning coastal area constructed in this study, an adaptability test prior to principal component analysis was conducted on the data for 13 indicators from 160 sample dwellings [61].The sample data were first standardized to eliminate dimensional influences, followed by assessment of data suitability through the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity (

Table 3. KMO and Bartlett’s Test.

KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy		0.529
Bartlett’s Test of Sphericity	Approx. Chi-Square	242.417
	df (degrees of freedom)	78
	Sig. (significance)	<0.001

). The results show that the KMO value is 0.529, exceeding the acceptable threshold of 0.5; the significance level of Bartlett’s test of sphericity is less than 0.001, indicating significant correlations among the indicators. These results demonstrate that the data meet the basic conditions for conducting principal component analysis.

3.3.2. Total Variance Explained Principal Component Extraction Results

Based on the total variance explained results of the principal component analysis, this study performed dimensionality reduction on the 13 characteristic indicators of 160 traditional dwellings in the Liaoning coastal area (

Table 4. Total Variance Explained.

Total Variance Explained
Initial Eigenvalues				Extraction Sums of Squared Loadings
Component	Total	% of Variance	Cumulative%	Total	% of Variance	Cumulative%
1	2.250	17.309	17.309	2.250	17.309	17.309
2	1.574	12.110	29.419	1.574	12.110	29.419
3	1.351	10.394	39.813	1.351	10.394	39.813
4	1.192	9.166	48.979	1.192	9.166	48.979
5	1.153	8.868	57.847	1.153	8.868	57.847
6	0.973	7.482	65.330
7	0.890	6.846	72.176
8	0.831	6.394	78.570
9	0.782	6.016	84.586
10	0.614	4.725	89.311
11	0.557	4.286	93.597
12	0.516	3.967	97.564
13	0.317	2.436	100.000

Extraction Method: Principal Component Analysis.

). The analysis process extracted principal components based on the criterion of eigenvalues greater than 1, ultimately retaining five components with a cumulative variance contribution rate of 57.847%. In exploratory research concerning regional architectural morphology, as the variable system encompasses multiple complex dimensions such as spatial morphology and cultural functions, certain human geographical heterogeneity exists among variables, leading to a relatively dispersed explanatory power of variance. In such exploratory factor analysis, a cumulative contribution rate of 50–60% is generally regarded as an acceptable standard capable of effectively revealing the internal structure of data and reflecting most of the effective information in the original indicator system.

Specifically, Component 1 has an eigenvalue of 2.250, explaining 17.309% of the variance; Components 2 through 5 have eigenvalues of 1.574, 1.351, 1.192, and 1.153, respectively, with variance contribution rates of 12.110%, 10.394%, 9.166%, and 8.868%. This indicates that the first five principal components jointly constitute the core latent dimensions influencing dwelling characteristics, with relatively balanced contributions from each component and no single component dominating, reflecting the multifaceted composite characteristics of factors influencing traditional dwelling morphology in the Liaoning coastal area. These results demonstrate that through principal component analysis, the original 13 indicators can be effectively condensed into five comprehensive dimensions, systematically capturing the main differences in dwelling characteristics despite a loss of approximately 42% of information, providing a structurally simplified variable foundation for subsequent dwelling pedigree classification, regional comparison, and formation mechanism analysis.

3.3.3. Interpretation of Component Matrix

Based on the component matrix obtained from principal component analysis, this study named and interpreted the five extracted principal components according to the following three criteria: loading threshold (selecting variables with absolute loading values greater than 0.5 as the main explanatory variables for the component), variable clustering (high-loading variables within the same component should have theoretical intrinsic correlations), and explanatory priority (prioritizing variable combinations corresponding to components with higher variance contribution rates). Component 1 (eigenvalue 2.250, explaining 17.309% of the variance) has high-loading variables of Courtyard Prototype (0.725) and Relative Position of Main Hall (0.623), and was named the Courtyard Spatial Organization Factor, reflecting the overall planning of courtyard layout based on family structure and production and living logic. Component 2 (eigenvalue 1.574, explaining 12.110% of the variance) has high-loading variables of Window-to-Wall Ratio (0.583) and Timber Frame Form (0.531), and was named the Technical Construction and Climate Adaptation Factor, reflecting the construction technology responses of dwellings to wind and thermal environments. Component 3 (eigenvalue 1.351, explaining 10.394% of the variance) has high-loading variables of Main Wall Material (0.608) and Dialect Division (−0.600), and was named the Material–Culture Correlation Factor, reflecting the correspondence between material selection and regional cultural background. Component 4 (eigenvalue 1.192, explaining 9.166% of the variance) has a high-loading variable of Decoration Type (0.709), and was named the Decoration Expression Factor, representing the regional characteristics of architectural decoration and cultural aesthetics. Component 5 (eigenvalue 1.153, explaining 8.868% of the variance) has high-loading variables of Courtyard Aspect Ratio (0.555) and Site Adaptability (0.545), and was named the Topographic Adaptation Factor, reflecting the response of courtyard morphology to topographical conditions. These five factors systematically summarize the morphological characteristics of traditional dwellings in the Liaoning coastal area from the dimensions of spatial organization, technical construction, material culture, decorative expression, and topographic adaptation.

The component matrix shows that some variables have moderate loadings on multiple components (

Table 5. Component Matrix.

Component Matrix a
	Component
	1	2	3	4	5
Courtyard Prototype	0.725	−0.252	0.335	−0.085	0.246
Relative Position of Main House	0.431	−0.186	0.472	0.027	0.024
Length-to-Width Ratio of the Courtyard	0.103	0.447	0.085	−0.380	0.555
Courtyard Area	0.513	0.240	0.283	0.139	−0.436
Main House Area	0.274	0.401	−0.090	0.329	0.545
Window-to-Wall Ratio of the Main House	−0.390	−0.471	0.058	0.346	0.198
Wall Material	0.623	−0.309	−0.014	−0.107	−0.228
Main House Structure	0.210	0.583	0.115	−0.151	−0.043
Decoration Type	−0.249	−0.278	0.608	−0.195	0.138
Site Adaptability	−0.317	0.531	0.289	−0.012	−0.419
Settlement Morphology	0.267	0.129	0.199	0.709	0.035
Dialect Division	0.503	−0.022	−0.600	0.143	−0.069
Ethnic Type	−0.344	0.157	0.226	0.444	0.047

Extraction Method: Principal Component Analysis. ^a 5 components extracted.

), indicating certain correlations among the variables. For example, Main Hall Area contributes to both Component 1 (0.513) and Component 5 (−0.436), reflecting that courtyard scale is influenced not only by spatial organization logic but also by topographical conditions. This phenomenon of cross-loading is an inherent characteristic of human geography data—dwelling morphology is itself the product of multiple interacting factors. Principal component analysis decomposes overlapping information into different dimensions through orthogonal transformation precisely to address this complexity.

Robustness checks show that the results of factor analysis and K-means clustering are largely consistent with those of principal component analysis, indicating that the analytical results are robust [62].

In summary, principal component analysis effectively extracted the core comprehensive variables influencing dwelling morphology, providing a robust orthogonalized data foundation for subsequent clustering analysis.

3.3.4. Systematic Clustering Method

Each of the 160 traditional dwelling samples was treated as an independent unit, and the five principal component factors were used as their attributes for systematic clustering analysis using SPSS. The clustering method was Between-groups linkage, and the interval distance measure was Squared Euclidean Distance. The number of clusters was not preset initially, and a clustering dendrogram was generated. From the dendrogram, it can be observed that when the distance coefficient d = 25, all samples merge into one large cluster; when d = 20, the samples are divided into four main categories; when d = 18, one of these categories further splits into two subcategories, ultimately forming five clusters; when d = 15, the number of clusters increases to more than seven. It is evident that different distance coefficients correspond to different numbers of clusters.

To determine a reasonable number of clusters, the analysis was conducted again with the number of clusters set to range from 2 to 10. The cluster assignment codes for the samples were obtained, and these codes were imported into ArcGIS and added to the attribute table of the corresponding samples. Spatial distribution maps for 2 to 10 clusters were generated and comparatively analyzed. The results showed that when divided into 2 to 3 clusters, the spatial distribution of samples within each cluster was highly interwoven, and the morphological similarity within clusters was insufficient to form typological zones with clear geographical meaning. When divided into 4 clusters, although the clustering results initially showed spatial differentiation, one cluster encompassed two geographically distant areas: the southern mountainous area of the Liaodong Peninsula and the Western Liaoning mountainous area. These two areas exhibited different dialect affiliations and cultural origins in subsequent external validity tests (the former belonging to the Jiaoliao Mandarin area and the latter to the Beijing Mandarin area), indicating that the 4-cluster division was too generalized. When divided into five clusters, each category exhibits a clear spatial clustering pattern—as shown in the dendrogram (Figure 3)—and presents significant differences in external data such as dialect divisions and ethnic distributions. When divided into 6 or more clusters, some categories contained only 3–5 samples, mostly located in ethnic integration zones or typical migration areas. Although these subdivisions have some significance for regional differentiation, the small sample sizes cannot support robust statistical inference.

Considering the natural geographical background and historical migration patterns of the Liaoning coastal area, this study considers the 5-cluster division formed at a distance coefficient of d = 18 to be the most reasonable. From a natural geographical perspective, the region exhibits a “coastal–mountainous–plain” geographical gradient, which aligns with the spatial differentiation pattern of the geographical gradient. The distribution of the five dwelling types is highly consistent with this gradient. From a humanistic historical perspective, two major cultural transmission routes—the maritime route of Jiaodong migration and the overland route of official metropolitan architectural dissemination—intersect in this region. Therefore, the 5-cluster structure formed at a distance coefficient of d = 18 can reflect both regional commonalities and identify local subtypes, providing a reasonable clustering foundation for subsequent multi-dimensional interpretation of dwelling characteristic formation mechanisms from perspectives such as physical geography and cultural diffusion. Importing the coordinate points of dwellings under these five clusters into ArcGIS yields the clustering distribution map of traditional dwellings in the Liaoning coastal area (Figure 4).

4. Discussion

4.1. Transformation of Clusters into Architectural “Pedigree”

Its theoretical foundation lies in incorporating the groupings obtained from statistical analysis into the analytical framework of cultural geography and, with the help of quantitative tools such as the geographic detector, achieving a theoretical leap from “data grouping” to a “cultural–geographical coupling structure.” The groupings identified by cluster analysis essentially summarize the similarities of samples across multi-dimensional morphological indicators, answering the statistical question of “which samples are closer in feature space.” However, the concept of “pedigree” in the context of cultural geography requires not only morphological similarity but also that this similarity possesses cultural stability and geographical interpretability—that is, the grouping results should embody a trinity of “morphological type–spatial pattern–cultural origin.” The geographic detector can provide a key verification tool for this purpose. Through factor detection, it can quantify the explanatory power of cultural factors (such as dialect divisions, ethnic distributions) and geographical factors (such as elevation, wind speed) on the grouping results, verifying whether the groupings are genuinely driven by cultural–geographical factors rather than random disturbances. Through interaction detection, it can examine whether there is a synergistic enhancement effect between cultural and geographical factors—if the interaction q-value is significantly higher than the individual factor q-values, it indicates that the morphological differences reflected by the groupings are essentially the spatial projection of the interaction between cultural dissemination and environmental adaptation. When the grouping results simultaneously satisfy spatial clustering and alignment with cultural–geographical factors, the statistically defined “cluster” gains legitimacy as a “pedigree” in cultural geography—its internal samples constitute a morphological type, its spatial distribution constitutes a spatial pattern, and its cultural–geographical coupling mechanism constitutes an explanation of its origin. Therefore, the transformation from “cluster” to “pedigree” essentially involves subjecting the groupings identified by cluster analysis to verification of their cultural–geographical coupling effect via the geographic detector, and then reinterpreting them within the theoretical framework of “morphological type–spatial pattern–cultural origin,” enabling statistical groupings to serve not only classification but also a cultural-geographical explanation of the formation mechanisms of dwelling morphology.

4.2. Discussion on Pedigree Characteristics of Traditional Dwellings in the Liaoning Coastal Area Based on Clustering

Based on the completion of systematic clustering analysis and the identification of five categories, this study further identifies and synthesizes the characteristics of each cluster to form a systematic description system for dwelling pedigree characteristics. First, combining the distribution characteristics of each cluster in the principal component space, architectural features, and courtyard types, the five clusters were named. Drawing on the “three-part” naming method, the clusters were designated as “geographical location—prominent architectural feature type—courtyard type,” resulting in the following five denominations for traditional dwellings in the Liaoning coastal area: Coastal Quadrangle Courtyard Type, Coastal Flat-Roofed Middle Courtyard Type, Coastal Gabled-Roof Small Courtyard Type, Mountainous Gabled-Roof Small Courtyard Type, and Plain Flat-Roofed Long Courtyard Type. During the feature extraction process, for numerical indicators (such as courtyard area, window-to-wall ratio, etc.), median deviation analysis was employed to identify significant tendencies within each cluster relative to the overall sample. For descriptive indicators (such as courtyard prototype, wall materials, decoration type, etc.), attribute categories with frequencies significantly higher than the overall distribution were extracted from each cluster using the frequency proportion threshold method, thereby clarifying their pedigree characteristics. On this basis, integrating the distribution of each cluster in the feature space, combined with graphical representations of typical dwelling courtyard morphology, plan and elevation features, and timber frame forms corresponding to each pedigree, the morphological combination patterns, environmental adaptation strategies, and cultural expression modes of the five dwelling types were systematically summarized.

4.2.1. Overview of Cluster 1

The “Coastal Quadrangle Courtyard Type” represented by Cluster 1 is predominantly distributed in Zhuanghe City, Liaoning Province (Figure 5). From the data indicators, the average courtyard area of this type reaches 800.28 m², the average main hall area is 139.05㎡, and the courtyard aspect ratio is 1.88. All these values are significantly higher than the overall mean, exhibiting hyper-scale characteristics and planar elongation in morphological features. The main hall primarily comprises 3–5 bays, further confirming its relatively large spatial attributes. In terms of morphology and composition, this type predominantly features highly enclosed quadrangle courtyard forms, partially combined with main hall and inverted house configurations, with the inverted house widely present as a typical constituent element. Wall materials predominantly utilize brick and stone, with decorations commonly found on cushioning heads (chí tóu) and entrance pillow stones, forming a simple and elegant facade through lime pointing. The structural system universally adopts triangular truss timber frames. These dwellings are mainly distributed in the gentle transitional zones between the coast and the Liaodong Peninsula mountains, with settlement morphology primarily taking a linear form. Additionally, the ethnic composition of residents shows a relatively high proportion of Mongolian and Manchu populations, and the dialect division is predominantly Jiaoliao Mandarin.

4.2.2. Overview of Cluster 2

The “Coastal Flat-Roofed Middle Courtyard Type” represented by Cluster 2 is scattered in areas such as Pulandian and Wafangdian, located in foothill areas adjacent to water bodies and coastal zones of the Liaodong Peninsula (Figure 6). The settlement morphology is predominantly finger-shaped or linear. From quantitative indicators, its courtyard area is 535.26 m², main hall area is 97.71 m², and courtyard aspect ratio is 1.29, all approaching the overall mean values. The main hall primarily comprises 3–4 bays, exhibiting characteristics of moderate scale. In terms of architectural form, the main hall-type courtyard prototype predominates, with the main hall often maintaining a certain distance from the side walls rather than being arranged flush against them. Wall materials are mainly granite rubble, with some sections using brick. The roofs are predominantly flat-roofed (tunding), among which the distinctive Dalian form, the “guanmao flat roof,” still survives in Pulandian, Wafangdian, and Zhuanghe. Architectural decoration styles are relatively simple, mostly concentrated on woodwork carvings such as sparrow braces, doors, and windows. Additionally, residents are predominantly Han and Manchu, with dialects mainly consisting of Jiaoliao Mandarin and Beijing Mandarin.

4.2.3. Overview of Cluster 3

The “Coastal Gabled-Roof Small Courtyard Type” represented by Cluster 3 is primarily concentrated in the mountainous areas of the Liaodong Peninsula, including Yingkou and Dalian (Figure 7). The settlement morphology is predominantly finger-shaped or linear, following the contour of the terrain. From quantitative indicators, its courtyard area is 321.64 m², main hall area is 79.62 m², courtyard aspect ratio is 1.36, and window-to-wall ratio is 0.24. All these values are below the overall mean. The main hall predominantly comprises three bays, exhibiting characteristics of a small footprint, square-proximate morphology, and narrow window openings. In terms of architectural form, the main hall-type courtyard predominates, with the main hall often arranged detached from one or both side walls, while the north wall is typically positioned flush against the courtyard wall. Wall materials predominantly utilize locally sourced dressed stone alongside later-introduced red or gray bricks. Roofs universally adopt the gabled form, with timber frames primarily employing the “large crossed rafter” (dà chā shǒu) construction method, exhibiting minimal decoration overall. Additionally, residents are predominantly Han Chinese. The above quantitative characteristics are directly derived from the five principal component dimensions upon which the cluster analysis was based, constituting the core morphological definition criteria for this type.

4.2.4. Overview of Cluster 4

The “Mountainous Gabled-Roof Small Courtyard Type” represented by Cluster 4 is mainly distributed in the Chaoyang mountainous area and the Anshan mountainous area (Figure 8). The settlement morphology is primarily linear or finger-shaped, spreading out in response to the terrain. From quantitative indicators, its courtyard area is 753.35 m², main hall area is 112.46 m², and courtyard aspect ratio is 1.42. All these values are above the overall mean, with the main hall predominantly comprising 3–4 bays. In terms of architectural form, main hall-type courtyards predominate, partially equipped with east wing rooms. Courtyard gates are mostly wall-gate type without inverted houses, with the main hall maintaining a distance from one side wall. Wall materials are mainly stone, with some sections adopting a mixed “stone base with brick upper” construction. Roof forms feature the coexistence of gabled and flat roof types. The decorative style is simple, with stone carving applied only to structural nodes such as cushioning heads (chí tóu) and screen walls. Additionally, residents are predominantly Han Chinese, along with Manchu and Mongolian populations. The dialects feature a coexistence and intermingling of Jiaoliao Mandarin and Beijing Mandarin.

4.2.5. Overview of Cluster 5

The “Plain Flat-Roofed Long Courtyard Type” represented by Cluster 5 is concentrated in the inland plain areas of western Liaoning, including Panjin, Jinzhou, and Huludao (Figure 9). The settlement morphology is predominantly regular linear or clustered. From quantitative indicators, the average courtyard area is 681.42 m², the average main hall area is 121.38 m², the courtyard aspect ratio is 1.95, and the window-to-wall ratio is 0.37. All these values are above the overall mean, with the main hall primarily comprising 3–5 bays, exhibiting characteristics of a long and narrow courtyard form and a high window-to-wall ratio. In terms of architectural form, main hall-type courtyards predominate, with the main hall often detached from both side walls and equipped with a backyard, forming a courtyard system with front-back connectivity. Wall materials are mainly brick, stone, or a combination of both. Roofs are predominantly flat-roofed (tunding), with some adopting modern structures. The overall form is simple with almost no decoration. From the perspective of human geography data, residents are predominantly Han Chinese, with dialects mainly being Northeastern Mandarin.

4.3. External Validity Testing of Clustering Results

To ensure that the five dwelling pedigrees obtained through systematic clustering are not a fortuitous product of the algorithm but rather reflect the actual pattern of morphological differentiation, this study conducted external validity tests on the clustering results from the following three aspects: comparison with the distribution of residential buildings recorded in historical documents, spatial overlay analysis with dialect divisions and ethnic distributions, and in-depth case studies of representative samples.

4.3.1. Verification Through Comparison with Historical Documents

The spatial distribution of the five dwelling pedigrees was overlaid with migration routes and architectural features recorded in local chronicles to verify the consistency between the clustering results and historical documents.

Coastal Quadrangle Courtyard Type: Cluster 1 is concentrated in Qingdui Ancient Town and the surrounding gentle coastal areas of Zhuanghe City. The 1921 Gazetteer of Zhuanghe County records that Qingduizi Port was opened in 1737 and became a major transit point for Shandong refugees [63], which aligns with the formation of hyper-scale quadrangle courtyards.

Coastal Flat-Roofed Middle Courtyard Type: Cluster 2 is scattered in foothill areas adjacent to water bodies such as Pulandian and Wafangdian. The 1907 Gazetteer of Fuxian County and the 1908 Gazetteer of Gaiping County record that most residents were descendants of Shandong immigrants who preserved their native customs, which corroborates the timber frame practices (e.g., “large crossed rafter”) in Cluster 2 [64,65].

Coastal Gabled-Roof Small Courtyard Type: Cluster 3 is concentrated in the mountainous areas of the southern Liaodong Peninsula, such as Lüshun and Jinzhou. The Gazetteer of Lüshunkou District (Qianlong–Jiaqing periods) and the 1921 Gazetteer of Zhuanghe County record that Shandong immigrants built simple stone huts with thatched roofs in the mountains, aligning with the compact courtyards and gabled roofs of Cluster 3 [64,66].

Mountainous Gabled-Roof Small Courtyard Type: Cluster 4 is mainly distributed in the Chaoyang mountainous area and the Anshan mountainous area. The 1930 Gazetteer of Chaoyang County records that dwellings imitated the metropolitan style but preserved only the essence due to financial constraints, which corroborates the simplified translation of Beijing courtyard forms in Cluster 4 [67].

Plain Flat-Roofed Long Courtyard Type: Cluster 5 is concentrated in the inland plain areas of western Liaoning, including Panjin, Jinzhou, and Huludao. The 1908 Gazetteer of Jinzhou Prefecture and the 1931 Gazetteer of Yi County record that immigrants built wide and long dwellings on the vast Western Liaoning Plain, aligning with the long and narrow courtyards and high window-to-wall ratio of Cluster 5 [68,69].

4.3.2. Spatial Overlay Analysis with Dialect Divisions and Ethnic Distributions

The clustering results of the five dwelling types were subjected to spatial overlay analysis with independently obtained data on dialect divisions and ethnic distributions to further verify the external validity of the classification.

Dialect Overlay Test: Statistics on the dialect affiliations of the villages where the 160 samples are located show that the overlay rate of the Coastal Quadrangle Courtyard Type with the Jiaoliao Mandarin area is 89.4%. For the Coastal Flat-Roofed Middle Courtyard Type, Jiaoliao Mandarin and Northeastern Mandarin account for 58.3% and 31.6%, respectively, exhibiting transitional characteristics. The overlay rate of the Coastal Gabled-Roof Small Courtyard Type with the Jiaoliao Mandarin area is 90.9%. For the Mountainous Gabled-Roof Small Courtyard Type, Beijing Mandarin and Jiaoliao Mandarin coexist (64.7% and 26.5%, respectively). The overlay rate of the Plain Flat-Roofed Long Courtyard Type with the Northeastern Mandarin area is 73.7%. This distribution pattern aligns highly with historical migration routes—Jiaodong immigrants brought Jiaoliao Mandarin via sea routes, while metropolitan immigrants brought Beijing Mandarin via land routes. The convergence of these two dialects in the Western Liaoning Corridor forms a transitional zone, which is entirely consistent with the dialect overlay results for the various clusters.

Ethnic Distribution Test: The clustering results were compared with records of ethnic distributions in the Gazetteer of Liaoning Province: Ethnic Minorities and local chronicles. According to the Gazetteer of Zhuanghe County revised in 1921, the distribution area of the Coastal Quadrangle Courtyard Type is characterized by significant concentrations of Manchu and Mongolian populations. The Gazetteer of Fuxian County revised in 1907 records that the proportion of Manchu population in the distribution area of the Coastal Flat-Roofed Middle Courtyard Type is significantly higher than the regional average. The distribution area of the Mountainous Gabled-Roof Small Courtyard Type is generally consistent with the historical distribution area of the Manchu “Eight Banner Reclamation.” According to the Gazetteer of Xiuyan County: “During the Kangxi reign of the Qing Dynasty, the imperial court recalled over a thousand Eight Banner soldiers from within the pass to station in Xiuyan. Later, some were assigned to the Dalian area. The Manchu Eight Banner troops brought their families, claimed land, and formed settled villages.” The distribution area of the Plain Flat-Roofed Long Courtyard Type is predominantly inhabited by Han Chinese, consistent with the scope of the civilian (Han immigrant) reclamation area during the Qing Dynasty. According to the Gazetteer of Yi County: “Yizhou initially had no permanent Han households. During the Qianlong reign, people from Shandong, fleeing famine, began to settle here, followed by merchants who came and settled, gradually forming villages.” This correspondence indicates that the clustering results effectively capture the influence of ethnic factors on dwelling morphology.

4.3.3. In-Depth Case Validation of Representative Samples

Representative samples were selected from each of the five dwelling types for in-depth validation, combined with records of family migration in local chronicles.

Representative Case of Coastal Quadrangle Courtyard Type–Li Residence, Qingdui Town, Zhuanghe City: Taking a residence in Qingdui Town, Zhuanghe City, as an example, according to the family genealogy, the ancestors crossed the sea from Dengzhou Prefecture, Shandong, during the Qianlong reign of the Qing Dynasty and settled there, engaging in commerce and trade for generations. The residence features a typical Jiaodong-style quadrangle courtyard layout: three main hall rooms in the center, with wing rooms symmetrically arranged on both sides. The courtyard gate is located in the southeast corner, and the stone-paved courtyard is square and open. The spatial order is identical to the “southeast entry” pattern found in the Mu Family Manor in Qixia, Shandong. The cushioning heads (chí tóu) feature layered, rhythmically receding lines, perfectly matching the official Jiaodong style. The walls are constructed with grey bricks laid in running bond, with lime pointing as fine as hair; the mortar joints are smooth and clean, with no spillage. This mutual corroboration among pedigree records, spatial layout, decorative motifs, and masonry techniques confirms the close relationship between this type and commercial migration.

Representative Case of Coastal Flat-Roofed Middle Courtyard Type–Wang Residence, Pulandian District: According to the Gazetteer of Fuxian County revised in 1907: “Most residents in the area are descendants of Shandong immigrants… Some live together in clans, preserving the customs of their place of origin.” The flat roof (tunding) construction in this village is consistent with that found in Rongcheng and Wendeng, Shandong. The rubble masonry wall technique aligns with the “use of local materials” recorded in local chronicles, corroborating the transmission of native techniques by Jiaodong immigrants.

Representative Case of Coastal Gabled-Roof Small Courtyard Type–Zhao Residence, Lüshunkou District: According to the Gazetteer of Lüshunkou District: “During the Qianlong and Jiaqing reigns, people from Deng and Lai prefectures who crossed the sea often built simple huts in the mountains, using stone for walls and thatch for roofs.” The “large crossed rafter” timber frame used in the dwellings of this village is directly descended from the construction methods of dwellings in the mountainous areas of Jiaodong, Shandong. The characteristics of compact courtyards and low window-to-wall ratios are consistent with the “simple dwellings” recorded in the county gazetteer, confirming that this type represents a simplified form created by Jiaodong immigrants in the mountainous interior.

Representative Case of Mountainous Gabled-Roof Small Courtyard Type–Sun Residence, Chaoyang County: According to the Gazetteer of Chaoyang County revised in 1930: “Dwellings in the county often imitated the metropolitan style, but due to financial constraints, only the essence was preserved.” The residence in this village has a wall-gate layout without an inverted house. The courtyard is centrally symmetrical but with a simplified form, which aligns highly with the gazetteer’s record of “imitating the metropolitan style but preserving only the essence,” corroborating the localized translation of Beijing quadrangle courtyard forms in the mountainous environment.

Representative Case of Plain Flat-Roofed Long Courtyard Type–Meng Residence, Panshan County: According to the Gazetteer of Panshan County: “From the late Qing Dynasty to the Republic of China period, dwellings in the area were predominantly rammed earth houses. Wealthy families built brick and stone tiled houses. Courtyards were mostly rectangular, with a threshing ground in front, adapted to agricultural reclamation and fishing-hunting livelihoods.” The characteristics of the rectangular courtyard, backyard, and high window-to-wall ratio in this village’s dwellings align with the “adapted to agricultural reclamation” record in the county gazetteer, confirming that this type is a product of the integration of North China immigrant culture with the needs of the plain reclamation area.

4.3.4. Summary of External Validity Testing

Through the above three tests, the following conclusions can be drawn: (1) The spatial distribution of the five dwelling pedigrees aligns highly with the migration routes and dwelling characteristics recorded in historical documents; (2) The clustering results show significant spatial overlay relationships with independently obtained dialect division and ethnic distribution data, with overlay rates generally exceeding 75%; (3) In-depth validation of representative cases indicates that the morphological characteristics of each type have a real cognitive basis in local historical records. These external pieces of evidence collectively confirm that the five dwelling pedigrees obtained through systematic clustering are not a fortuitous product of the algorithm but an effective revelation of the actual pattern of morphological differentiation of traditional dwellings in the Liaoning coastal area.

5. Conclusions

5.1. Findings on the Pedigree Characteristics of Traditional Dwellings in the Liaoning Coastal Area

Based on systematic clustering analysis, this study classified 160 traditional dwelling samples in the Liaoning coastal area into five dwelling types with significantly differentiating characteristics: Coastal Quadrangle Courtyard Type, Coastal Flat-Roofed Middle Courtyard Type, Coastal Gabled-Roof Small Courtyard Type, Mountainous Gabled-Roof Small Courtyard Type, and Plain Flat-Roofed Long Courtyard Type.

Among these, the Coastal Quadrangle Courtyard Type is concentrated in the coastal areas of Zhuanghe and surrounding Qingdui Ancient Town, exhibiting hyper-scale characteristics with highly enclosed quadrangle courtyards and brick-stone walls as typical forms. Residents are predominantly Mongolian and Manchu, with its formation closely related to its historical role as a commercial hub. The Coastal Flat-Roofed Middle Courtyard Type is scattered in foothill areas adjacent to water bodies on the Liaodong Peninsula, featuring moderate scale with main hall-type courtyards and highly distinctive flat-roofed (tunding) construction as core characteristics, reflecting climatic adaptation to strong winds and abundant rainfall in the Bohai Bay region. The Coastal Gabled-Roof Small Courtyard Type is concentrated in mountainous areas of the Liaodong Peninsula including Yingkou and Dalian, exhibiting significantly compact spatial scale characterized by three-bay main halls, gabled roofs, and simple “large crossed rafter” timber frames. This type serves as direct material evidence of Jiaodong populations migrating northward by sea and bringing their native construction techniques to Liaodong. The Mountainous Gabled-Roof Small Courtyard Type is mainly distributed in the Chaoyang mountainous area and Anshan mountainous area, predominantly featuring main hall-type courtyards, partially equipped with east wing rooms and without inverted houses, representing a localized translation of Beijing quadrangle courtyard forms within mountainous environments. The Plain Flat-Roofed Long Courtyard Type is concentrated in the western Liaoning plains including Panjin, Jinzhou, and Huludao, featuring markedly rectangular courtyards characterized by flat-roofed structures and spatial layouts emphasizing lighting, reflecting the integration of North China courtyard traditions brought by “Chuang Guandong” migrants during the Qing Dynasty with local needs. These findings systematically reveal the construction wisdom of “adapting to local conditions and governing according to customs” embodied in traditional dwellings of the Liaoning coastal area.

5.2. An Integrated Explanatory Framework for Formation Mechanisms

The previous section identified five dwelling pedigrees through systematic clustering. This section further asks: How did these morphological differences come about? How do cultural and geographical factors interact to shape the spatial differentiation pattern of traditional dwellings in the Liaoning coastal area? To address these questions, this study constructs a “cultural–geographical” integrated analytical framework, introduces the GeoDetector for factor detection and interaction detection, and verifies the transmission and variation hypotheses through spatial overlay analysis of migration routes and dwelling type distributions.

5.2.1. Construction of the Cultural–Geographical Integrated Analytical Framework

Based on the results of the principal component analysis, this study extracted the core factors influencing dwelling morphology and grouped them into two major systems: the cultural dimension and the geographical dimension.

The cultural dimension includes: (1) Migration routes (using dialect divisions as proxy variables: the Jiaoliao Mandarin area corresponds to the maritime migration route from Jiaodong; the Beijing Mandarin area corresponds to the overland migration route from Jiaozhou and the metropolitan region; the Northeastern Mandarin area corresponds to the multi-source migration routes via land and sea into the interior regions of northern Liaoning, Jilin, Heilongjiang, etc.); (2) Ethnic composition (Han, Manchu, Mongolian, etc.); (3) Dissemination of official metropolitan architectural forms (using distance from Ming and Qing post roads as a proxy variable, reflecting the spillover of northern official architectural culture. Basic geographic information data are primarily sourced from https://www.ditushu.com/book/1131 (accessed on 27 September 2025).

Ming and Qing post roads refer to the official postal and transportation road systems established during the Ming and Qing dynasties, used for transmitting official documents, transporting supplies, receiving officials, and military mobilization. These post roads were equipped with facilities such as relay stations (for rest and changing horses) and courier depots (for transmitting documents), forming a transportation network that covered the entire country. In Northeast China, the post road system centered on Shengjing (Shenyang), extending west through the Western Liaoning Corridor to Shanhaiguan and Beijing, east to Jilin and Heilongjiang, and south to seaports such as Lüshun.

The geographical dimension includes: (1) Topographic conditions (elevation, slope); (2) Climatic conditions (annual average wind speed); (3) Distance from the coastline.

We constructed a multiple regression model and a GeoDetector analytical framework using the above factors as independent variables and the type assignments of the five dwelling pedigrees as the dependent variable, aiming to quantify the individual contributions and interaction effects of each factor.

5.2.2. Quantitative Analysis of Factor Contribution Rates Using the GeoDetector

The GeoDetector was employed to quantitatively measure the explanatory power of cultural and geographical factors. The core principle of this method is: if a factor significantly influences dwelling type differentiation, its spatial distribution should be significantly consistent with the spatial distribution of dwelling types. The factor detection power (q-value) ranges from 0 to 1, with larger values indicating stronger explanatory power of the factor for dwelling type differentiation.

The factor detection results show that, in the ranking of individual factor explanatory power (

Table 6. Factor Detection Results of GeoDetector.

Factor Dimension	Detection Factor	Proxy Variable	q-Value	Significance p-Value	Ranking of Explanatory Power
Cultural Dimension	Migration Route	Dialect Division	0.327	<0.001	1
	Official Architectural Dissemination	Distance from Ming and Qing Post Roads	0.241	<0.001	3
	Ethnic Composition	Ethnic Type	0.112	<0.05	7
Geographical Dimension	Topographic Conditions	Elevation	0.293	<0.001	2
	Topographic Conditions	Slope	0.187	<0.001	4
	Climatic Conditions	Annual Average Wind Speed	0.203	<0.001	5
	Resource Endowment	Distance from Coastline	0.156	<0.01	6

), dialect division (q = 0.327) and elevation (q = 0.293) rank the highest, indicating that migration culture and geographical environment are the core driving forces shaping the dwelling pedigrees in the Liaoning coastal area. These are followed by distance from Ming and Qing post roads (q = 0.241), slope (q = 0.187), annual average wind speed (q = 0.203), distance from coastline (q = 0.156), and ethnic type (q = 0.112) (Table 6).All factors passed the significance test, indicating that dwelling morphological differentiation is the result of the combined action of multiple factors. Notably, dialect division has the highest q-value, confirming that historical population movements such as Jiaodong immigration and metropolitan immigration have had a significant impact on dwelling morphology.

Interaction detection further reveals the synergistic effects among factors (

Table 7. Key Factor Interaction Detection Results.

Interaction Factor Combination	Interaction q-Value	Individual q-Values	Interaction Type
Dialect Division ∩ Elevation	0.472	0.327, 0.293	Nonlinear Enhancement
Dialect Division ∩ Annual Average Wind Speed	0.438	0.327, 0.203	Nonlinear Enhancement
Dialect Division ∩ Distance from Post Road	0.415	0.327, 0.241	Bivariate Enhancement
Distance from Post Road ∩ Elevation	0.396	0.241, 0.293	Bivariate Enhancement
Dialect Division ∩ Ethnic Type	0.351	0.327, 0.112	Nonlinear Enhancement
Elevation ∩ Slope	0.312	0.293, 0.187	Bivariate Enhancement
Annual Average Wind Speed ∩ Distance from Coastline	0.278	0.203, 0.156	Bivariate Enhancement
Distance from Post Road ∩ Annual Average Wind Speed	0.352	0.241, 0.203	Bivariate Enhancement
Elevation ∩ Annual Average Wind Speed	0.376	0.293, 0.203	Bivariate Enhancement
Dialect Division ∩ Slope	0.403	0.327, 0.187	Nonlinear Enhancement

). The interaction types in the GeoDetector can be classified as nonlinear enhancement, two-factor enhancement, and so on.

The interaction detection results show that the interaction q-value for any two factors is greater than the individual q-values of the factors, indicating a nonlinear enhancement or bivariate enhancement relationship among the factors. Cultural factors and geographical factors do not act independently but jointly shape the spatial differentiation pattern of dwelling morphology through coupling effects. Among them, the interaction q-value of “Dialect Division ∩ Elevation” reaches 0.472, significantly higher than their individual q-values (0.327 and 0.293), exhibiting a nonlinear enhancement relationship, and has the strongest explanatory power among all factor combinations. This indicates that the coupling effect of migration culture and topographical conditions is a key factor influencing the differentiation of dwelling pedigrees in the Liaoning coastal area.

This statistical result is mutually corroborated by spatial analysis evidence. The spatial distribution characteristics presented earlier (Section 3.2) show that the distribution of dwelling types in the Jiaoliao Mandarin area exhibits a significant topographic gradient—the gentle coastal zones are predominantly characterized by the Coastal Quadrangle Courtyard Type, the foothill areas adjacent to water by the Coastal Flat-Roofed Middle Courtyard Type, and the mountainous interior by the Coastal Gabled-Roof Small Courtyard Type. Combining the interaction results of the GeoDetector with this spatial distribution pattern allows for further inference: during the transmission of Jiaodong migration culture to the Liaodong Peninsula, its interaction with different topographical conditions contributed to the type differentiation of the coastal series of dwellings. Similarly, the interaction q-value of “Dialect Division ∩ Annual Average Wind Speed” reaches 0.438, exhibiting nonlinear enhancement. Combined with the concentrated distribution of flat-roofed (tunding) technology in the coastal high wind speed zone, it can be inferred that the synergistic effect of migration culture and climatic conditions reinforced the continuity of flat-roofed technology.

Other factor combinations also exhibit synergistic enhancement effects: the interaction q-value of “Distance from Post Road ∩ Elevation” is 0.396, exhibiting bivariate enhancement. Combined with the distribution characteristics where the Plain Flat-Roofed Long Courtyard Type is predominant in plain areas near post roads and the Mountainous Gabled-Roof Small Courtyard Type is predominant in mountainous areas far from post roads, this reflects that during the transmission of metropolitan official culture along the post roads, topographical conditions played a moderating role in the localized variation in official architectural forms. The interaction q-value of “Dialect Division ∩ Ethnic Type” is 0.351, exhibiting nonlinear enhancement. Combined with the differences in decorative features exhibited by the Coastal Quadrangle Courtyard Type and the Mountainous Gabled-Roof Small Courtyard Type within the Manchu-Mongolian settlement areas, this indicates an interaction between migration culture and ethnic settlement (

Table 8. Comparison of Architectural Decorative Features between the Xu Hexiu Residence in Gaojiazhuangzi Village, Zhaoyuan City, and the Sun Jiadong Residence in Xutun, Wafangdian City, Dalian.

At the level of construction craftsmanship, taking the Sun Jiadong Residence in Xutun, Wafangdian, as an example, its rubble masonry technique, when compared with the Xu Hexiu Residence in Gaojiazhuangzi Village, Zhaoyuan City, Shandong, exhibits consistency in construction methods: the height of the first three courses of the wall foundation follows a specific proportion, measuring 270 mm, 250 mm, and 210 mm from bottom to top. The mortar pointing at the stone joints is extremely fine, dense, and uniform, with virtually no spillage of mortar. The depth of the main hall in this residence is uniformly set at 2 zhang 4 chi (approximately 7.6 m), which aligns closely with the construction dimensions of dwellings in the Jiaodong region. These consistent choices in masonry techniques and dimensional standards suggest a transmission pathway in which traditional Jiaodong masonry techniques were brought northward across the sea by migrant artisans and subsequently preserved and followed in the Liaoning region.

).

5.2.3. Spatial Comparative Verification of Migration Transmission Paths

To verify the specific paths of cultural transmission and their impact on dwelling morphology, this study conducted a spatial overlay analysis of the spatial distribution of the five dwelling pedigrees with historical migration routes and the Ming and Qing post road network.

(1)

Verification of the Jiaodong Migration Transmission Path

Historically, there were two main pathways for population migration and cultural transmission in the Liaoning coastal area: Jiaodong migrants crossed the sea to land on the Liaodong Peninsula, while migrants from the metropolitan region and North China entered the Northeast interior overland along the Western Liaoning Corridor. Some Jiaodong migrants, after landing, may have continued northward overland, while some metropolitan migrants may have also taken sea routes (e.g., via Tianjin Port) and then transferred to coastal shipping to reach the shores of the Liaodong Bay. These two pathways intersected and interacted spatially in the Liaoning coastal area, jointly shaping the region’s population and cultural landscape. To verify the influence of these two transmission paths on dwelling morphology, this study conducted a spatial overlay analysis of the spatial distribution of the five dwelling pedigrees with historical migration routes(Figure 10), and quantitatively tested the gradient changes in morphological characteristics along the transmission paths.

Historical records indicate that during the “Chuang Guandong” migration in the Qing Dynasty, migrants from the Jiaodong region primarily reached the Liaodong Peninsula by sea, with major landing points including coastal ports such as Lüshun, Jinzhou, and Fuzhou. Spatial overlay analysis shows that the distribution area of the Coastal Flat-Roofed Middle Courtyard Type (Pulandian and Wafangdian coastal areas) is generally consistent with the distribution of secondary landing points for Jiaodong migrants, with an average distance of 15.7 km from the landing points. The distribution area of the Coastal Gabled-Roof Small Courtyard Type (the mountainous areas of Lüshun and Jinzhou) is located in the hinterland behind the landing points, with an average distance of 22.4 km from the landing points, exhibiting a “landing–diffusion” gradient characteristic. Along the “landing point–hinterland” path, dwelling morphology shows a regular pattern of evolution: courtyard scale decreases, the proportion of flat roofs decreases, and the proportion of gabled roofs increases. This gradient change verifies the transmission mechanism of “landing–diffusion–localization” of Jiaodong architectural culture.

(2)

Verification of the Overland Transmission Path of Official Metropolitan Architectural Forms

Historical records indicate that during the Ming and Qing dynasties, Beijing’s official architectural culture spread northeastward along the post road system, with the main corridor being the Western Liaoning Corridor (roughly along the present-day Beijing–Harbin Railway). Spatial overlay analysis shows that the distribution area of the Plain Flat-Roofed Long Courtyard Type (Panjin, Jinzhou, and Huludao) is highly consistent with the post road network of the Western Liaoning Corridor(Figure 11), with an average distance from the Ming and Qing post roads that is relatively small. The distribution area of the Mountainous Gabled-Roof Small Courtyard Type (the mountainous areas of Chaoyang and Anshan) lies in the extended zone of post road branches. Constrained by topographical and economic conditions, these dwellings omitted inverted houses and simplified courtyard layouts, resulting in a wall-gate layout without an inverted house. The distance from the post road is negatively correlated with the degree of preservation of official architectural elements such as wall-gate layouts and central axis symmetry. Along the “post road–hinterland” path, official architectural forms exhibit a trend of “simplification–localization,” reflecting the adaptive variation in official culture during its transmission process.

(3)

Formation Mechanism of the Cultural Convergence Zone

In the convergence zone between the Western Liaoning Corridor and the Liaodong Peninsula (the Jinzhou–Panjin–Yingkou line), the maritime transmission path of Jiaodong migration and the overland transmission path of metropolitan migration intersect, forming the distribution area of the Coastal Flat-Roofed Middle Courtyard Type. The dialect in this area is Northeastern Mandarin, and the dwelling morphology combines Jiaodong flat-roofed (tunding) construction with metropolitan courtyard layouts, representing a product of the interaction between the two major cultural transmission systems.

5.2.4. Integrated Explanation of the Formation Mechanism

Synthesizing the above analyses, this study proposes an integrated explanatory model for the formation of the traditional dwelling pedigrees in the Liaoning coastal area.

(1)

Core Driving Mechanism: Cultural–Geographical Coupling

The GeoDetector results indicate that the dominant mechanism for dwelling pedigree differentiation is the coupling of cultural and geographical factors, rather than the independent influence of a single factor. The interaction q-value of Dialect Division ∩ Elevation reached 0.472, significantly higher than the explanatory power of either factor alone, indicating that the matching relationship between migration culture and topographical conditions is the core mechanism shaping the dwelling pedigrees. Specifically, this is manifested as follows: Jiaodong migrants landing on gentle coastal areas formed the Coastal Quadrangle Courtyard Type; entering foothill areas adjacent to water, they shaped the Coastal Flat-Roofed Middle Courtyard Type; diffusing into the mountainous interior, they evolved into the Coastal Gabled-Roof Small Courtyard Type. Metropolitan migrants entering the Western Liaoning Plain along the post roads formed the Plain Flat-Roofed Long Courtyard Type; extending to the mountainous areas along post road branches, they transformed into the Mountainous Gabled-Roof Small Courtyard Type.

(2)

Spatial Gradient Mechanism: Dual Diffusion Paths

Dwelling morphology exhibits two major spatial gradient differentiation patterns: Maritime gradient (from the coastline inland), where courtyard scale decreases, the proportion of flat roofs first increases and then decreases, and the proportion of gabled roofs increases, reflecting the spatial attenuation effect of “landing–diffusion–localization” of migration culture. Overland gradient (from post roads to hinterland), where the degree of preservation of official architectural elements decreases, and courtyard forms evolve from complete quadrangle courtyards to simplified courtyard forms, reflecting the adaptive variation in “transmission–simplification–localization” of official culture.

(3)

Climate Adaptation Mechanism: Rational Logic of Technological Choice

Annual average wind speed is the strongest negative predictor of window-to-wall ratio (β = −0.324) and the strongest positive predictor of flat-roofed (tunding) choice (OR = 2.28), indicating that the combination of flat roofs and low window-to-wall ratios is a rational technological response to strong winds in the Bohai Bay. The positive influence of elevation on gabled roofs (β = 0.294) reflects the shaping effect of mountainous drainage needs on roof form.

(4)

Ethnic Integration Mechanism: Layered Deposition of Multiculturalism

Although the explanatory power of the ethnic factor alone is relatively weak (q = 0.112), its interaction with dialect divisions and topography is significant. The distribution areas of Manchu and Mongolian populations spatially overlap with the Coastal Quadrangle Courtyard Type and the Mountainous Gabled-Roof Small Courtyard Type. Their residential cultures have left imprints in material selection (e.g., preference for stone) and decorative details, constituting a phenomenon of “cultural layering” within the dwelling pedigrees.

It is important to clarify that the “causal relationship” revealed in this study is essentially a statistical causal inference based on observational data, rather than a strict causal relationship under experimental control. The interaction detection method employed in this study can effectively identify the independent contributions and synergistic effects of factors, but it cannot completely exclude endogeneity issues such as omitted variables or reverse causality. Therefore, the mechanisms revealed in this paper should be understood as strong correlations with causal directionality, rather than absolute causal determinism. Future research could further introduce methods such as instrumental variables and natural experiments to deepen the identification of causal mechanisms.

5.3. Methodological Validation and Application: A Reference for Other Multicultural Regions

At the methodological level, this study is not merely a regional application of the typology of traditional dwellings in the Liaoning coastal area, but also a validation and advancement of the research paradigm for vernacular architecture in multicultural geographical convergence zones. The study adopts a technical workflow consisting of “indicator system construction-data collection and processing-principal component analysis-systematic clustering-synthesis of type characteristics-GeoDetector-based mechanism analysis.” Based on 160 typical dwelling samples, this approach achieves a deep integration of qualitative and quantitative methods.

Through principal component analysis and systematic clustering, the study identifies five dwelling pedigrees and their characteristics in a data-driven manner, effectively avoiding the subjective bias that may arise from over-reliance on qualitative descriptions in traditional research. Cultural factors (dialect divisions, ethnic composition, dissemination routes of official architecture) and geographical factors (topography, climate, distance from the coastline) are incorporated into a unified analytical framework. Using the GeoDetector, the contribution rates and interactions of these factors are quantitatively measured, providing statistical validation for the mechanism of “cultural–geographical coupling.” This methodological system moves beyond the qualitative dichotomy of “environmental determinism” versus “cultural determinism” in cultural geography, offering a reproducible analytical path for similar multicultural convergence zones.

In terms of application value, the three-dimensional pedigree framework of “morphological type—spatial pattern—cultural origin” established in this study, along with the operational procedure of “data-driven classification-spatial visualization-coupling mechanism testing,” can be extended to other areas of the Circum-Bohai Sea region, the land–sea interface of Northeast Asia, and even regions worldwide that exhibit similar characteristics of cultural superimposition and geographical gradients. Researchers in different contexts may flexibly adapt the methodological framework of this study by adjusting the indicator system and proxy variables according to local migration history, ethnic composition, and environmental conditions. In doing so, this study helps advance vernacular architecture research from case-specific descriptions toward regional comparison and mechanistic explanation.

5.4. Limitations and Future Prospects

This study has conducted a preliminary exploration of the formation mechanism of traditional dwellings in the Liaoning coastal area from the perspective of cultural geography. Based on the above findings, the following limitations must be acknowledged:

First, the cumulative variance contribution rate of the principal component analysis was 57.847%, meaning that approximately 42% of the original information was not explained by the first five principal components during dimensionality reduction. This is partly attributable to the inherent complexity of human geography data: nonlinear relationships between dwelling morphological characteristics and geographical/cultural factors may exist, which linear dimensionality reduction methods cannot fully capture. Additionally, although the current indicator system encompasses 13 indicators across three dimensions, some indicators have a coarse granularity (e.g., decoration type is classified into only six categories), limiting the ability to distinguish subtle differences. This loss of information may result in insufficient sensitivity of the clustering results to some micro-level features, but the identification of core pedigree types remains highly credible.

Second, the sample size is limited (160 sites), and the data sources are primarily based on traditional village listings, which may affect the representativeness and generalizability of the research findings. The distribution of traditional village resources within the study area is inherently uneven, and areas with more rapid urbanization have fewer surviving traditional dwellings, leading to a sample distribution that does not fully and uniformly cover all geographical units.

Third, the research framework is mainly based on cultural geographical factors, without systematically incorporating the influence of multiple factors such as economy and society, resulting in an incomplete explanation of the formation mechanisms of dwelling morphology. For example, the shaping effects of socio-economic factors such as the fishing and salt economy, trade networks, and land systems on dwelling morphology were not sufficiently quantified in this study.

Fourth, the selected cultural–geographical factors (such as dialect and ethnicity) remain relatively macroscopic and could not be further decomposed into quantifiable and operational sub-variables, limiting the depth and precision of the analysis. For instance, dialect division, used as a proxy variable for migration routes, cannot distinguish the specific origins of different counties or families within the same dialect area.

Fifth, this study does not report internal clustering quality indices (e.g., silhouette coefficient, within-group/between-group variance). The validation of the clustering results is primarily based on the structure of the dendrogram, spatial coherence, and cross-comparison with external data such as historical documents, dialect divisions, and ethnic distributions. This constitutes an interpretive and spatial validation rather than a formal assessment based on internal metrics. This limitation should be addressed in future research.

In response to the above limitations, future research can be advanced from the following aspects: First, expand the sample scope and data sources to enhance the diversity and typicality of research objects, particularly by strengthening the rescue investigation and documentation of endangered dwellings undergoing urbanization. Second, construct a more refined quantitative indicator system by further decomposing some descriptive indicators into more discriminative sub-variables; simultaneously, explore the introduction of nonlinear dimensionality reduction methods (such as t-SNE, kernel principal component analysis) to complement traditional principal component analysis and enhance the ability to capture complex nonlinear relationships. Importantly, the potential future use of nonlinear methods does not negate the validity of the current results, which are robust for the main typological differentiation; such methods would only help to explore highly specific or localized variations. Third, attempt to incorporate multiple factors such as economy and society to build a multi-level, multi-factor comprehensive analytical framework, deepening the systematic understanding of the formation mechanisms of dwelling morphology. Finally, integrating the concept of sustainable development, explore specific pathways for the active conservation and adaptive renewal of traditional dwellings, providing more operational theoretical support and practical guidance for the preservation and inheritance of regional architectural heritage.