Using a process-based model to understand dynamics of Chinese agricultural and water technology development from 8000 bc to 1911 ad

Abstract

Advancements in technology are inextricably bound to our society and the natural environment. However, how the development process of a technology system interacts with both remains unclear. We propose a process model to understand the complex dynamics among technology, society, and the environment via seven interactive elements: technologies, actors, receiving bodies, natural contexts, social contexts, temporal–spatial contexts, and outcomes. The model was applied to agricultural and water technology development in China from 8000 bc to 1911 ad. Our findings show that these elements did not play equally important roles in different periods of the development in ancient China, with social contexts most dominating during the earlier periods and both social and environmental concerns arising towards the later periods. The proposed model, by identifying the elements in the technology development that should be strengthened, can act as an analysis device to assist in reconfiguring a more sustainable socio-technological system.

Electronic supplementary material

The online version of this article (10.1007/s13280-020-01424-7) contains supplementary material, which is available to authorized users.

Introduction

The rapid advancement of technology has generated a “techno-sphere” over the biophysical environment, signaling the entrance of an “Anthropocene” (Zacher, 2017). Without an understanding of the social and natural contexts of a technology, the trajectory of its development cannot be fully understood; nor can technological impacts on society and the environment be properly assessed (Pinch and Bijker 1984; De Dreu and van Dijk 2018; Wei et al. 2018).

The social embedding of technology is usually complicated to understand. Social Construction of Technology (SCOT) theory implores to open the “black box” of historical and contemporary technology development (Bijker et al. 1987). This theory moves away from the technology deterministic view that trajectories of human society are inevitably determined by technology development, and argues that technology development should be conceived not only as physical artifacts but also as constructions of the social, economic, political, and environmental contexts (Pfaffenberger 1992). Resonant with the SCOT theory, the more recently developed socio-technical system (STS) framework emphasizes the co-evolution of technology and society (Winner 1993; De Dreu and van Dijk 2018). In STS, the complex interactions between the physical artifacts and non-physical actors in technology development are considered a complex system (Baškarada and Koronios 2018). The development of such systems is non-linear and continuously driven by policy interventions, institutional reforms, and cultural changes (Fuenfschilling and Truffer 2014). The Actor Network Theory (ANT) and Multi-Level Perspective (MLP) are two major theoretical approaches that can be used to analyze the co-evolution between technology and society. ANT generalizes different actors during technology development and emphasizes associations among them (Latour 1987). The major criticism of ANT is its symmetry assumption, whereby all actors in the network are unanimous (Munir and Jones 2004). MLP analyses technology transformation processes by investigating the interactive relationships between socio-political contexts and technology at the niche, regime, and landscape levels (Geels 2011). This approach also received criticism for limited consideration on actors within the framework and monolithic descriptions on technological trajectories (Fuenfschilling and Truffer 2014). What has been missed in the existing literature is a process-based understanding of what are developed, who/what engage, what are the social and environmental contextual situations, and what are the outcomes in a whole-of-system way (Clark 2002). Without such an understanding, we would fail to reveal the mechanism of the co-evolution between technology and its social–natural environment. Our capacity for directing technology development for a sustainable future will thus be severely compromised.

The aim of this study is to understand the co-evolving dynamics between technology development, the involved key actors, and the social and natural contexts. The agricultural and water technology in ancient China was taken as a case study. Agricultural and water technology, as a key driver of agricultural development, is one of the most ancient technological systems in human history and continues to be the primary means of food production in contemporary society (Weisdorf 2005) and China was one of the most ancient agricultural civilizations in the world (Zong et al. 2007; Shen 2010). In addition, there are more than 8000 years of agricultural practices and abundant ancient documents. Thus, investigating the agricultural technological development in China offers an insight into the evolving process of how agricultural and water technology interact with its social and natural contexts in a long, traceable time frame.

Materials and Methods

Key elements for understanding technology development and its socio-natural contexts

This study proposes a social-technological process concept model to understand the social and natural contexts in which a technology system is developed. The concept model was developed based on the theoretical foundations of the social policy process model by Lasswell (2003) and Clark (2002), who analyzed the interactions among key elements (i.e., participants, perspectives, situations, base values, strategies, outcomes, and interactions) during complex policy developments from a problem-solving perspective. This model has been applied to provide a structured approach to elaborate the effective process in policy development (Knapp et al. 2009) and to facilitate decision-making in environmental governance (Haas 2004).

Our model considers technology development as part of an interactive socio-natural process, which includes seven variables that describe the actors and contextual settings of technology development (Ninan 2009): the technology system itself, participants (i.e., actors) involved, receiving bodies, temporal–spatial contexts, social contexts, natural contexts, and the outcomes of technology development (Table 1). A technology system is defined as a system that contains all physical artifacts, practical measures, and theoretical routines. The “Actor” identifies the human participants involved in the social-technological process, and the “Receiving body” contains the plants and animals that are being mobilized by humans through technology development. These two groups were selected because they are either directly involved in technology development or are influenced by technologies. The temporal–spatial contexts describe the temporal period and locations where the technology is developed and/or used, and are crucial attributes to sociological studies of technology (Wajcman 2008).

Table 1.

Key elements in the socio-technological process model

Technology	Physical artifacts (“hardware” technologies: irrigation canals, etc.)
	Methods and theories (“software” technologies: ridge-furrowing method, water–soil harmony farming theory, etc.)
Actor	Specifically named people (e.g., agricultural expert Xu Guangqi in this study)
	Position-based individuals and organizations (e.g., government officials, farmers)
Receiving body	Food crops (e.g., wheat, rice)
	Horticultural crop (e.g., fruits, vegetables)
	Livestock (e.g., oxen, horses)
Temporal–spatial context	Temporal periods (e.g., dynasties and longer times)
	Spatial locations (e.g., countries, cities, cultural regions, river basins)
Natural context	Weather (e.g., flood, drought, wet, dry, rain) and seasonality (e.g., solar terms, months, harvesting seasons)
	Natural resources (e.g., water, soil)
Societal context	Historical events (e.g., war, migrations, rebellions)
	Socio-economic and political background (e.g., policy, tax)
Outcome	Innovation, diffusion, implementation, and evaluation of technology (e.g., enhanced production)

More importantly, because technology exists in its social–natural settings, any change in societal and natural contexts would influence its development (Preiser 2019). Therefore, social and natural contexts are included in the model, which detail the socio-economic backgrounds and environmental features that may trigger or be re-structured as a result of technology development (Bijker et al. 1987). The “Outcome” was designed to describe how technologies are innovated, diffused, used, and evaluated by the actors regarding the receiving bodies in a certain socio-natural context (Latour 2005). By understanding these key elements of socio-technical systems, we are able to understand the co-evolutionary processes between technologies and their socio-natural contexts. As the seven elements included in the concept model and their interactions are not quantitative variables, a qualitative approach was adopted to apply this model.

Study period

The study period spanned from the dawn of agriculture in ancient China in approximately 8000 bc to the end of the final imperial dynasty in 1911 ad. This period was selected as it covers the full length of pre-industrial agricultural development in China. Eight historical periods based on dynastic changes and technological conditions in history were divided (Table 2).

Table 2.

Division of study period. Revised from Wu et al. (2019)

Dynasty	Historical period	Start	End
Neolithic	Neolithic	8000 bc	2000 bc
Xia	XSZ	1999 bc
Shang			771 bc
Zhou
Chunqiu & Zhanguo	CQZG	770 bc	221 bc
Qin	QH	220 bc	220 ad
Han
Weijin, South and North Dynasties	WJ	221 ad	581 ad
Sui	ST	582 ad	960 ad
Tang
Song	SY	961 ad	1368 ad
Yuan
Ming	MQ	1369 ad	1911 ad
Qing

Data sources and data processing

Three historical encyclopedias were selected as the data sources, all of which covered the full study periods in this study. The History of Chinese Agricultural Technologies by Liang (1989) focused on the physical perspectives of agricultural and water technologies (e.g., farming tools and methods) and the corresponding societal development in different historical periods. The Development History of Chinese Agriculture by Yan and Yin (1993) scrutinized both the social and economic history of agriculture and technologies; and the History of Science and Technology in China—The Agriculture Chapter by Dong and Fan (2000), which is the most up-to-date agricultural encyclopedia for agricultural and water technologies in China. That publication is an improvement from the agricultural volume of the Science and Civilization of China edited by Needham and Bray (1984), and comprehensively discusses the natural and societal history of physical artifacts (“hardware” technologies), methods and theories (“software” technologies) of agricultural development in ancient China. These historical encyclopedias systematically combine first-hand archeological discoveries and ancient literature on agricultural development, and they were compiled by distinguished historians, sociologists, and anthropologists in China. They have been constantly referenced by government reports such as the Chinese Agricultural Yearbooks (Board 1989, 2000) and by other historical studies (Lu 2015; Zhang 2015). Therefore, these data sources provide a comprehensive coverage and cross-validation of the agricultural and water technologies and their socio-environmental contexts.

Text mining was used to extract the relevant information from the chosen data sources. Based on the frequencies of words appearing in the text and how they co-appear with each other, this method was able to identify the key information related to agricultural and water technology development and other elements during the development process (Schwartz et al. 2013). Figure 1 summarizes the procedure of data processing for data reproducibility and verifiability.

Fig. 1.

Flow chart of data processing

To facilitate an understanding of the different focuses in different historical periods, the complete corpus (text body) from all three encyclopedias was converted into digital formats and divided based on their respective historical periods. This approach resulted in 16 documents in total, excluding the forward and concluding remarks of each encyclopedia.

Word tokenization was used to determine how the text body was segmented into meaningful words, which is a crucial step in text mining to ensure the quality of further analysis. The textual data in this research were written in Chinese, which is a character-based language and cannot be separated based on space as in English. Some character-based Chinese word tokenization approaches have been developed to identify meaningful words. One commonly used approach is the ICTCLAS algorithm developed by the Chinese Academy of Computer Science (https://ictclas.nlpir.org/), this algorithm uses the Hidden Markov Model to identify and tokenize Chinese words, which has been tested to show more than 90% accuracy (Zhang et al. 2003). The JiebaR package (https://cran.r-project.org/web/packages/jiebaR/index.html) is an R package which integrates multiple algorithms including the Hidden Markov Model, the Maximum Probability Model, the Query Model, and the Mixed Model, which were used in the data processing for this study. To further increase the accuracy in word tokenization, the word bank on agriculture and agricultural technology available online from Sogou (a constantly updated thesaurus database that is supported by one of the largest Chinese keyboard platforms: https://pinyin.sogou.com/dict/) was incorporated into the JiebaR package. These thesauri contain the most commonly used terms in Chinese agriculture. However, as the text data were about agricultural and water technologies in ancient China, a customized list of words was also developed to include the names of technologies in ancient China from the same data source by Wu et al. (2019).

Data cleansing was then conducted to reduce the special characters, and stop words and other meaningless terms in the segmented data. In this study, a pre-defined list of stop words and all non-Chinese characters (arithmetic numbers, English and other language characters) were removed from the analysis. The final number of tokenized words after cleansing was 70,881. Then, this list of words and their appearance frequencies in each document were converted into a term-document matrix, which was the basic unit for further analysis.

The words with high appearance frequencies in text were considered of higher relevance to the central content of the texts (Choi et al. 2018). In this study, these keywords were used to identify the dominant technologies, the central participants (actors and receiving bodies), and the main social and natural contexts of technology development in the different historical periods. However, the meaning represented by words can be limited and difficult to interpret. Topic modeling is another commonly used technique for analyzing unstructured textual data (Zhao et al. 2015; Blei and Smyth 2017). It assumes that some concepts in text are implicitly linked to the others. It interprets data based on the probability of word co-occurrence rather than only counting the appearance frequencies of individual words. Thus, topic modeling was used in this study to identify the interactions between key agricultural and water technologies and their socio-natural contexts in different historical periods (Genkin et al. 2007). The Latent Dirichlet Allocation (LDA) by Blei and Lafferty (2007) was used as the topic modeling algorithm. This algorithm estimates probability distributions of topics appearing in the text body, and the distributions of words appearing in each topic. Although prior knowledge of the text body is not required, pre-specifications of the number of topics are needed (Cambria and White 2014).

To ensure the coverage of sufficient details and maintain the uniformity of the number of keywords to be analyzed, a topic number of 20 and a fixed portion (words with top 1% number of appearances) for each historical period were obtained for further analysis (see Supplementary Methods for how the topic number and the appearance proportion were determined). Table 3 summarizes the initial number of words extracted from the text (after data cleansing), the numbers of the top 1% words, and the respective minimum numbers of appearances for these words in each historical period.

Table 3.

Summary of keywords to be analyzed by historical period

	Total	Neo	XSZ	CQZG	QH	WJ	ST	SY	MQ
Total no. of words	70 881	13 862	13 218	14 622	25 932	25 762	34 376	34 756	34 245
No. of top 1% words	709	139	132	146	259	258	344	348	342
Min. frequency of top 1% words	85	43	42	47	49	48	48	50	50

Grouping the identified keywords for understanding technology development and its socio-natural contexts

The keywords mined from the word frequency analysis and topic modeling were then grouped into the key elements in the socio-technological process model (Table 1) by manual coding to understand the technology development and its socio-natural contexts. Manual coding was conducted not only because human interpretations were required to obtain the latent topics by text mining (Rohrer et al. 2017), but also because humans have better analytical capability in interpreting the key elements inherent in the socio-technological process model (Wei et al. 2017). The manual coding was systematically conducted by two independent coders to ensure problem specific, reliable, and reproducible results (Grimmer and Stewart 2013). The level of ambiguity of the results, represented by Krippendorff’s alpha, was kept above the 80% limit as recommended by Poindexter and McCombs (2000).

Results

Key agricultural and water technologies and their socio-natural contexts in different historical periods

We identified the key features relating to ancient Chinese agricultural and water technology development (Fig. 2). It can be seen from Fig. 2 that the appearance frequency of technology-related words ranged between 193 and 2000, with “furrowing,” “timing,” “farming field,” “rake,” “fertilization,” and “cultivation” of highest frequency. There were limited appearances of keywords that indicated the actors (on average, less than 500 times), and included only “farmers,” “landlords,” and specifically named agricultural experts (e.g., “Wang Zhen”) who were directly involved with implementations and documentations of agricultural technologies in ancient China. Likewise, the receiving bodies (with a slightly higher appearance range between 193 and 1000) mainly included the crops planted (rice and wheat), and oxen and horse which were crucial power sources in traditional agriculture. Keywords from the social context also had high appearance frequencies (between 236 and 2712), and were “development,” “economy,” “tradition,” and “farming experiences.” Keywords related to the natural context were limited (190–730) and contained basic natural elements (e.g., “sun,” “moon,” and “water”) and seasonality (e.g., “spring” and “phenology”). The frequency for temporal–spatial context ranged between 194 and 1388 and highlighted the QH, SY, ST, and MQ periods in both southern and northern regions of China (especially the Yellow River Basin). The outcome was mainly related to direct actions and improvements of agriculture (e.g., “enhance” and “utilize”) (frequency between 190 and 1377).

Fig. 2.

Frequency distribution of words by socio-technological process model groups in all periods

As shown in Fig. 3, different periods had different core technologies. The new keywords in each period highlighted technologies and their contexts specific to that period (Fig. 4) (refer to Supplementary spreadsheet for detailed lists of keywords and corresponding appearance frequencies). We explain the technologies and their contexts in each period as follows:

Fig. 3.

Frequency distribution of words by socio-technological process model grouped in each historical period

Fig. 4.

Frequency distribution of new words appeared in each historical periods grouped by the socio-technological process model

Neolithic period (8000–2000 bc)

The major focus in the Neolithic Period was primitive farming. The technologies identified include furrowing and soil cultivation, which indicated the early understanding of the relation between soil and agriculture in primitive farming practices. In this period, the key actors included farmers and tribe leaders, and the major receiving bodies were the crops (millets, beans, and coarse rice) and the animals (e.g., dog) domesticated. The major social context was human migration and production, mainly in the Yellow River Basin, whereas the natural context focused on weather features (e.g., “moon,” “spring,” “weather,” and “climate”). The major outcome of the agricultural system at the time was primitive and based on verbal (e.g., talk) inheritance.

XSZ period (1999–771 bc)

During the XSZ Period, improved technologies emerged that were related to mobilization of soil. The “gully drainage and irrigation system” was the most significant technology development, which greatly transformed agriculture at the time. Bronze agricultural tools were developed, leading to better cultivation methods and increased varieties of crops. Agricultural production continued to be the main focus of the social context, with farmers being the major actors and millets and crops as the major receiving bodies. The Yellow River Basin emerged as the center of agricultural development as the spatial context, whereas no significant outcome was observed. The previous focus on weather observations in the natural context had evolved to better determine the timing of agricultural practices and the development of phenology, implying the increasing recognition by farmers to farm according to seasonality.

CQZG period (770–221 bc)

The technologies identified were more explicit during the CQZG Period, the iron tools (e.g., rake, hoe), cultivation methods (e.g., sowing, weeding, and fertilization), and theories (e.g., using simple tools for intensive soil tillage and cultivation). Additional agricultural literature developed in this period (e.g., Lv’s Spring and Autumn) was considered as summaries of technological theories in agriculture. “Relationship,” “tradition,” and “foundation” became keywords as part of the social context that stressed social collaborations in agricultural practice. The natural context, temporal–spatial context, actors, and receiving bodies remained similar to previous periods. Keywords such as “utilization,” “change,” and “improvement” were also mentioned as outcomes to indicate increasing use of technology to intervene with agriculture and the natural environment.

QH period (220 bc–220 ad)

The technologies developed in this period highlighted “furrowing,” with more details on agricultural practices: oxen furrowing, ridge furrowing, moisture conservation; irrigation and agricultural tools: plow, hoe, and rake. These keywords were indicative of more systematic and comprehensive technological development in agriculture. The most frequent keywords that were newly appearing in this period was the renowned agricultural literature, The Book of Qi People, which provided comprehensive coverage on agricultural tools and methods from previous periods. Landlords and specifically named agricultural experts (e.g., Si Shengzhi) had appeared as the key actors. As the social elite class, they controlled more advanced agricultural technologies and larger farmlands, marking the dawn of a bureaucratic feudal society. Crops included varieties of beans and wheat, which diversified food consumption in this period. The social context focused on establishing law and order for regulating agricultural production, whereas the promotion of technology and the organization of large-scale farming activities appeared as part of the outcome. The natural context indicated recognition of the importance of water (“water,” “rain,” and “swamp”). The Yellow River Basin and parts of northern China were still central in the spatial context, whereas southern China was also mentioned.

WJ period (221–581 ad)

Following the QH Period, keywords related to technologies were similar, demonstrating systematic farming procedures that included furrowing, sowing, cultivations, weeding, irrigation, and harvest. The Book of Qi People continued to guide agricultural development in this period, with keywords that described the paddy field farming and rice growing in China that were beginning to emerge. As a result, while the key actors and the outcomes remained unchanged, rice, vegetables, wheat, beans, and ramie emerged as the key receiving bodies. Although both the social context and the spatial context were similar to the QH period, attention was increasingly paid to the natural context, especially to the months and seasonal phenology, including March, June, August, and the Summer Solstice, which formed a systematic “twenty-four solar terms, seventy-two phenology” system. These keywords indicated the ideal timing for farming practices (e.g., sowing, furrowing).

ST period (582–960 ad)

One of the significant differences of technology development in this period was the focus on methods of paddy field farming (e.g., construction of lower ground paddy fields by dykes, methods of rice transplanting), and tools specific for paddy field cultivation, sericulture, and irrigation infrastructure for higher grounds (e.g., water wheels). The key actors had changed, with new agricultural experts (Chen Fu and Wang Zhen) and with the government which actively promoted paddy field farming. The key receiving bodies had also changed to rice, and various crops in the south included cotton, mulberry, lychee, tea, and buckwheat. The keywords in social context included “population” and “labor force,” which were crucial for the agricultural system in ancient China. The spatial context indicated that the center of agriculture had shifted to the south, including the Tai Lake, the Yangtze River Basin, and even the Fujian Province. The natural context had also expanded to consider different seasons (e.g., “autumn”) and terrains that were suitable for farming, whereas multiple land uses and farming promotions were highlighted as key outcomes during this period.

SY period (961–1368 ad)

With rice continuing to be the major food crop and the south of China being the major region of agricultural practice, there also appeared keywords related to diversified furrowing technologies to ensure better weeding and moisture conservation, and more intensive cultivation methods such as transplantation of seedlings, manure fertilization, and soil treatment methods. These more technology-specific keywords suggest that standardized practice procedures for paddy field farming had been developed. The key actors (“agricultural experts” and “the government”) and receiving bodies (“rice” and “wheat”) identified in the SY Period closely aligned with those in the ST Period, whereas “life,” “family,” and “structure” emerged as new keywords in the social context, implying a family-based unit of agricultural practice. The spatial context also indicated that the Yangtze River Basin continued to be the center of agriculture. However, the outcome remained mostly unchanged.

MQ period (1369–1911 ad)

During the MQ Period, in addition to the focus on paddy field farming, keywords on the development of systematic agricultural theory, method (e.g., indicating the depth to which soil should be cultivated), and irrigation related to technologies increased. The key actors included Xu Guangqi (the author of the Book of Agricultural Politics), farmers, and landlords. The mention of Emperors Qianlong and Wanli indicated increasing centralized control of agriculture during the MQ Period. Both rice and wheat were the major receiving bodies, whereas crops from overseas (e.g., potatoes, corn) were also imported. Mention of foreign countries (e.g., Japan, the West) indicated possible exchanges between China and the rest of the world. Keywords from the social context including “government system” and “agricultural policy” indicated the systematic management of agricultural practices from the government, whereas “experience,” “research,” “theory,” and “knowledge” highlighted the early exploration of agriculture as a contemporary scientific discipline and learning from the West. These keywords echoed with those in the outcomes (e.g., adaptation, introduction, and improvement). Conversely, the natural context remained philosophical, which considered nature to be under constant influences of Solar (positive) and Lunar (negative) forces and the five basic natural elements of metal, wood, water, fire, and earth.

Interactions between key agricultural and water technologies and their socio-natural contexts in different historical periods

We identified the interactions among the key process elements regarding agricultural and water technology development in different periods as different topics using the LDA topic modeling (Fig. 5 and refer to Supplementary Fig. S2 for more details). It should be noted that more than one topic can belong to the same historical period. We explain these interactions in each period as follows:

Fig. 5.

The interactions of key process elements for agricultural and water technology system in China from 8000 bc to 1911 ad based on topic modeling results

Neolithic period (8000–2000 bc)

The major interaction during the Neolithic Period was identified as “Primitive agricultural technologies in a tribe-based social structure.” During this period, primitive tools for farming were mainly made of wood, stone, bone, and shell and focused on primitive slash-and-burn and till-and-lay furrowing using a Leisi (primitive plow). Technology development was closely tied with the social context, involving multiple key actors (“tribe member,” “leader,” “slave,” and “family member”) that formed a patriarchal society with a tribal social structure. The tribal culture was closely linked to settlement, exploration, migration, hunting, and gathering as major social activities. The limited food productivity could not support a fully settled lifestyle and required additional food sources and migration for more fertile land. The importance of the natural environment in agricultural practice was not yet recognized, as no keyword in the natural context were linked. Major settlement clusters were located both in the Yellow River Basin in the north and in the Yangtze River Basin in the south, including “Yangshao,” “Dawenkou,” “Hemudu,” “Cishan,” “Peiligang,” and “Longshan.”

XSZ period (1999–771 bc)

During the XSZ Period, the major interaction was identified as “Improved agricultural technology system with focus on interaction with natural system.” A diverse range of iron tools and irrigation infrastructure (e.g., the spade, hoe, sickle, shovel, and harrow) were improved and facilitated cultivation and tilling for farmers to sow in-line ridges. Additional technologies included construction of extensive irrigation infrastructures: Shao Reservoir, Zhengguo Channel, and Dujiangyan Weir, which were connected to the development in the middle and lower reaches of the Yellow River region. With increasing keywords such as wind, summer, sun, and months, people began to gain awareness of the importance of climatic conditions for planting. During this period, the key actors (e.g., landlords, academics, craftsman, and water engineers) were primarily specialized for millet, proso millet, wheat, and soybeans production. These actors were also involved in the establishment of laws for agricultural production, reflecting the transition from a slavery-based society to an early-stage feudal society.

CQZG period (770–221 bc)

The major interaction during the CQZG Period indicated “A farmland tax-based society that was founded on agriculture.” Keywords on technology focused on soil cultivation, soil selection, as well as on farming theories that indicated the importance to maintain soil fertility. These technologies were mainly linked to the social context involving the elite class (academics, philosophers, government officials, and landlords) as major actors. These actors were interacted under the social context (“tradition,” “relationship,” “foundation,” and “policy”) for tax and public policies development founded on agriculture. This was most evident in development of the farmland gully system, which divided the land into uniform blocks and required farmers to farm blocks of land that were public for tax payments.

QH period (220 bc–220 ad)

Two major interactions were identified during this period. The first one was referred to as “Agricultural technology system with wide range of food and cash crops.” Technologies developed included ironmaking, irrigation, cultivation, soil treatment, weeding, and seed selections. Fertilization and harvest technologies also emerged. These extensive improvement of technologies were closely linked to more diverse receiving bodies (proso millet, wheat, soybeans, rice, foxtail millet, and other horticulture crops including mulberry, hemp, and vegetables) that supported the increasing populations. The second interaction was identified as “Agricultural technology system with weather observations,” which indicated that knowledge derived from intuitive observations of weather (e.g., “cold,” “rainy,” “warm”) guided the timing of farming in the northern China, especially the Yellow River Basin and the Silk Road.

WJ period (221–581 ad)

During the WJ Period, the first interaction indicated “A technology system focused on tools and their effects on plants.” An extensive ranges of food and horticultural crops were mentioned as the receiving bodies, ranging from vegetables (e.g., soybeans and Chinese cabbages), green manure, to sunflowers. In line with increasing crop varieties, the types of agricultural tools had diversified to include plows with improved plow disks and curved beams, multi-shaped harrows, shovels, and new farming methods (e.g., the ridge-furrow method and the divisional cultivation method). The second interaction was identified as “Agricultural technology system focused on different actors’ roles,” which highlighted the farmers, landlords, government officials, emperors, and some specifically named agricultural experts (e.g., Dong Zhongshu). Their knowledge on agricultural practices were evidenced in published literature: the Book of Qi People; the Book of Si Shengzhi; and the Calendar of Simin. These actors were also linked to the policies that encouraged agricultural practices via implementation, collaboration, and adaptation (as outcomes). These interactions mainly took place in the south, with Guangdong, Sichuan, Huai River Basin appeared as the key spatial context and highlighted the increased farming in paddy fields.

ST period (582–960 ad)

Three major interactions were identified in the ST period. The first one indicated “A technology system focused on paddy field farming in southern China,” which linked detailed technologies with the spatial contexts. Irrigation infrastructure was the major focus and included water wheels and large-scale ponds that facilitated effective irrigation in paddy fields. The curved plows, harrow, and field leveler were specifically designed for furrowing in the paddy fields for both food crops and cash crops. Spatial locations, including lakes, swamps and rivers, were mostly located in the southern part of China, and these locations were referred to as emerging locations of paddy field farming. The second interaction highlighted “Government-level and farmer-level socio-natural interactions,” which covered a wide range of actors: emperors and their families, government officials, veterinarians, and professionally trained soldiers for irrigation infrastructure maintenance. The wide range of actors indicated in-depth, society-wide interactions with both ways of top-down (from emperors enacting policies to officials guiding practices) and bottom-up (from farmers to technicians and scholars). Reform of the tax system and encouragement for land reclamation by the government also facilitated agricultural practices, whereas wars and chaos were mentioned as disturbances to agriculture. These factors led to human migrations which, conversely, facilitated exchanges of agricultural technologies, especially paddy field farming in the south. The final interaction (“Emerging awareness of natural hazards”) linked the social context with natural context. Increasing observations of weather and seasonality (e.g., snow, summer) co-appeared with natural hazards key words (e.g., droughts and floods).

SY period (961–1368 ad)

Five major interactions were identified. The first interaction: “Development of systematic technology and interactions with government and academics” focused on a technology system that was guided and encouraged by officials while academics focused on improvement of influential technologies, with various agricultural literatures (e.g., Keynotes in Farming and Sericulture; the Book of Wang Zhen) published. The second interaction: “A wide range of plants supported by diversified technologies” highlighted connections between technology and great varieties of receiving bodies (e.g., “double-season rice,” “ramie,” “cole,” “flowers,” and “cotton”). Bees and ants were mentioned as a sign of insect breeding as agricultural by-products. These receiving bodies were linked to farming methods and theories that aimed to keep harmony among weather, soil fertility, and plant productivity, along with water-, wind-, and animal-powered tools. The third interaction focused on the spatial context, which highlighted that the southern part of China (specifically the Yangtze River Basin and Tai Lake) as the center of agriculture. The remaining two interactions focused on the social contexts. One indicated additional autumn taxes and growing populations. The other further highlighted rebellion wars by farmers, which severely harmed agricultural development.

MQ period (1369–1911 ad)

There were five major interactions identified in the MQ Period that focused on different perspectives of socio-natural contexts in technology development. The first interaction (“Development of a technology system focusing on people-soil-weather harmony”) was about extensive improvement of technology with outcomes of increased productivity. Agricultural tools and methods (e.g., new tools for inter-tillage, deworming, soil leveling, and methods that utilized manure, inter-cropping, crop rotation, and mixed cropping), and multiple land transformation methods (e.g., stone field, mulberry fish ponds, and floating fields) were adopted, with conceptualization of soil fertility as “Qi”: an invisible energy flow that needs to be carefully maintained. The second interaction: “Farmers and scholars mediated knowledge exchange” mainly linked the actors in agriculture, covering not only the farmers in general, scholars (e.g., Xu Guangqi), or landlords, but also farmers who farmed rented land and those in the mountains. With new actors including missionaries and Christians from foreign countries in the West (e.g., England, Italy) and Southeast Asia, frequent exchanges were occurring as one of the major interactions. The third interaction indicated “Increased concerns on farming in extreme adverse environment,” which linked natural hazards and harsh environment (e.g., strong wind, droughts, and floods) with the social context, indicated as concerns of famine in the ruling class. The fourth interaction (“Diversification of agricultural products due to foreign trade”) highlighted another major social context linked to the receiving bodies, of which there was a trend to adapt more productive farming crops. While rice continued to be the major food crop, other grains and imported crops (e.g., peanuts, sweet potato, and tobacco) were also mentioned. These crops greatly assisted in alleviating food pressure of the population boom. The final interaction was “Development spanning across different regions,” with farming regions expanded all over China, especially in the south (e.g., Yangtze River Basin, Guangdong, Zhejiang, Fujian Province) and water abundant regions (riverbanks and lakes).

Discussion

Understanding the interactions between the different elements in the process-based concept model (Table 1) allowed to identify what were more likely to be the key elements for technology development during different historical periods in pre-industrial China. The major reconfiguration process of the agricultural and water technology system and the corresponding societal and natural contexts in China from 8000 bc to 1911 ad are summarized below:

In the Neolithic Period, the beginning of agriculture was marked by primitive soil furrowing technologies. The key social context was hunting and gathering activities, which complemented the low-efficient agricultural production. The technology system expanded to include soil cultivation tools and drainage infrastructure in the XSZ Period, and to a set of technological theories that highlighted the harmony among labor inputs, soil conditions, and farming seasonality in the CQZG Period. The major actors were lower class farmers who directly engaged in both the natural contexts (soil conditions and the seasonality of farming), and the social contexts (law and order establishment that prioritized agricultural practices). Social contexts were thus paramount in shaping (e.g., defining farmland structures) and being shaped (e.g., taxing based on agricultural activities) by technology developments. The advancements of iron tools, use of fertilizer technologies, and multiple power sources sparked agricultural efficiencies in the QH Period. Greater institutional controls became the major social context under which these more effective technologies were developed, with emperors and government officials serving as the major actors. The WJ and SY Periods were considered as transition periods that did not demonstrate significant innovations. Rather, diversification of the existing technologies continued to be closely associated with the social contexts. In the WJ Period, such diversification was triggered by prolonged political disturbances and wars, whereas in the SY Period it was due to increasing control over a unified nation and family-based farming structures. Finally, in the ST and MQ Periods, reconfiguration to paddy field farming technologies continued to be actively accompanied by the social contexts of more competent land taxes and centralized governance. Knowledge exchange from the western world further supported a highly productive, intensive technology system. Yet, it was not until these periods that increasing concerns regarding the natural contexts, including the deterioration of soil fertility and more frequent occurrences of natural disasters were triggered. The major actors during this process were social elite groups including academics and government officials with a centralized yet singular focus to improve and promote technologies. These findings from our socio-technological process model are evident from a wide range of disciplinary studies including archeological (Zhao 2010; Hu et al. 2013; Zhuang et al. 2016; Kidder and Liu 2017), anthropological (Fei et al. 2012; Zhang 2013; Fang et al. 2015), and paleoclimatic (Dearing et al. 2008; Li et al. 2018) studies.

It is found from the development process of agricultural and water technology systems in ancient China that not every element in the process model, including technology itself, was engaged in every period. In other words, these elements did not play equally important roles in different periods. This finding is a reflection of the historically conscious and unconscious interactions between technology, human society, and the natural environment. Our proposed model can therefore help identify the key elements that have influenced and/or have been shaped by technology developments and the elements that should be strengthened and provide insights into future design of a more sustainable technology system based on the historical patterns (Turnheim et al. 2015). Our process model is different from the existing models including the Actor Network Theory (ANT), Multi-level Perspective (MLP) framework, and technology diffusion models but it does not conflict with them; rather, complement to them. Specifically, our model could add the dynamic processes that exist between people and technology and bring differentiation of actors and receiving bodies of technology into the MLP framework and ANT. Furthermore, our model engages the natural environment as part of the context within which a technology system is developed, which has received less consideration in previous studies.

Our findings show that computer-based text mining and topic modeling could enhance the capacity of the conventional content analysis to capture the interactions among the key elements in our socio-technological process model. This enables the detailed analysis of vast documents on technology development, thus providing more process-based understanding (Ravi and Ravi 2015; Rohrer et al. 2017). However, the limitations in our study should be noted. Only historical encyclopedias were studied in our study and there may be a tendency for historians to focus on “technological winners” (Geroski 2000). When our model is applied in the modern technology systems and other regions, the process elements (e.g., actors, social contexts) are expected to be different. It would be an interesting research direction to compare the social and natural factors influencing the development of the same technology in different temporal–spatial contexts. In addition, quantitative data (e.g., productivity, GDP) can be adopted to validate our model outcomes in modern technology systems.

Conclusions

This study aimed to understand the socio-technological dynamics of a technology system by using a process-based concept model comprising seven elements: technologies, actors, receiving bodies, natural contexts, social contexts, temporal–spatial contexts, and outcomes. The model was applied to agricultural and water technology development in China from 8000 bc to 1911 ad with a qualitative approach. It was found that the technology system was developed from primitive furrowing, specific cultivation tools, irrigation infrastructures, farming theories that promote harmony between the environment and humans to reconfigure these technologies from dryland farming to paddy field farming. The corresponding social contexts ranged from primitive hunting and gathering social activities to a more centralized, higher level political structure, then to exchanges from foreign countries by the social elite group. Not every element in the process model played equally important roles in the different periods of technology development. Social development was dominating during the earlier periods, with farmers being the actors, and both social and environmental concerns arose towards the later periods. The proposed model can act as an analysis device for the reconfiguration of a socio-technological system to achieve a more sustainable human society and natural environment.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Biographies

Shuanglei Wu

is a doctoral candidate at the School of Earth and Environmental Sciences, University of Queensland, Australia. Her research interests include coupled human–natural systems, socio-technical studies, global environmental change, and sustainable water resources management.

Yongping Wei

is an Associate Professor at the School of Earth and Environmental Sciences, the University of Queensland. She is leading the catchment governance system science laboratory covering three innovative disciplines of catchment governance system science, which includes catchment governance mechanics, catchment system connectivity, and catchment social hydrology. Her research interests are global environmental change, sustainable water and land management, and interaction between human system and environmental systems.

Brian Head

is a Professor in the Center for Policy Futures, Faculty of Humanities and Social Sciences, The University of Queensland. His major interests are evidence-based policy, complex or ‘wicked’ problems, program evaluation, early intervention and prevention, collaboration and consultation, public accountability, and leadership. He also has strong interests in applied research across many areas of public policy and governance, and is committed to building closer links between the research and policy sectors.

Yan Zhao

is a postal doctoral research fellow at the Centre for Crop Science, Queensland Alliance for Agriculture and Food Innovation, the University of Queensland. His research interests include remote sensing applications of catchment management and forest ecosystem assessment.

Scott Hanna

is an Adjunct Professor at the School of Earth and Environmental Sciences, the University of Queensland. His primary research focus is in the areas of environmental and social safeguard assessments, due diligence assessments, environmental and social impact assessments, natural resource and environmental management, regulatory management, strategic planning, and environmental auditing and training for large capital projects in the energy, transportation, and mining sectors.

Author contributions

S. Wu and Y. Wei conceptualized and designed the study; S. Wu collected and analyzed the data, and wrote the initial version of the manuscript; S. Wu, Y. Wei, B. Head, Y. Zhao, and S. Hanna interpreted the data and edited previous versions of the manuscript. All authors read and approved the final manuscript.

Data availability

All data produced in this study have been mentioned in the main text or provided in the Supplementary Materials. Codes used to process the data have been deposited in GitHub: https://github.com/SLWU423/Code-for-text-mining-historical-encyclopedias.git.

Conflict of interest

Authors declare that there is no conflict of interest.