C. Armand, Rotarians in transcolonial networks in early twentieth-century Shanghai
Ce billet propose l’introduction du texte de Cécile Armand. La version pdf complète est téléchargeable ici-même (2 Mo).
Les Rotariens dans les réseaux transcoloniaux à Shanghai dans la première moitié du XXe siècle
Abstract. Taking the Shanghai Rotary Club as a case study and tapping historical newspapers as a valuable source for drawing relational data, this paper proposes a multimodal approach to persons in networks, in which persons are conceptualized as club members and treated alternatively as nodes and edges. Relying on three sample periods (1919-20, 1930-1, 1938-9), we first analyze two-mode networks of meeting attendance (persons-events joined by roles) in order to identify core persons and events. We then supplement this bimodal analysis by a dual projection method, which consists in projecting the initial two-mode network into two one-mode networks, and jointly analyzing the resulting projections. Our analysis of co-attendance (persons-to-persons joined by events) allows us to identify three main personal profiles (leaders, brokers, outsiders) and to locate them within clustered networks (2). In the networks of multiple attendance (events-events joined by persons), we turn to persons as links between social events (3). We pinpoint meetings that have participants in common and try to clarify the underlying principles of participant sharing. In the last projection (roles-roles by persons) (4), we eventually reveal structural combinations of roles, but also alterations that reflect early experiments or wartime disruption. Further research is needed to develop a fully longitudinal approach and to improve the extraction and detection of positional roles within networks.
Keywords: club membership; dual projection; meeting attendance; person; role; social event; two-mode network; weighted network.
Outline of the paper
1. Meeting attendance (two-mode networks of persons-events, or affiliation networks)
2. Co-attendance (one-mode networks of persons-persons joined by events)
Degree distribution (personal influence)
Degree and betweenness (influence and intermediary)
Dyads and edge weight (co-attendance frequency)
3. Multiple attendance (events-events joined by persons)
4. Multiple role holding (roles-roles joined by persons)
List of figures
List of tables
This paper is part of a larger investigation on the transformation of elite sociability in early twentieth-century China.1 It is situated in a specifically transcolonial setting – from 1842 to 1943, Shanghai was a treaty-port with a large foreign community and two foreign settlements under French and British-American administration – during a transitional period characterized by political and social instability (abolition of the imperial regime replaced by a Republic in 1911, warlords, Sino-Japanese war in 1937-45, civil war in 1945-9). The Rotary club is an American-born elite club aimed at cultivating friendship and fellowship among businessmen and professionals in modern cities. The first club was founded in Chicago in 1905. The movement spread rapidly and eventually turned to a worldwide organization in the first half of the 20th century. The first club in China was established in Shanghai in July 1919. The Shanghai Rotary Club offers an ideal case to study the sociability of global elite in modern China. Despite its American origins, it was open to Chinese and other foreign nationals. It is fairly well documented in local newspapers (meeting reports), Rotary International Archives and personal papers (rosters of members) [Fig.1a, 1b, 2a, 2b].2 The Shanghai Rotary Club hold regular meetings every week on Thursday at lunch time. These “tiffins” usually consisted in luncheons and lectures covering a wide range of topics (from business to science, education or international relations). Occasionally, the club also organized special events (farewell dinner, Christmas party, recreational outing). These meetings were widely advertised in the local press. Newspaper reports usually list the participants and the roles their performed (a minima). At best, some articles provide very detailed information on attendants’ behaviors. Since regular attendance to weekly meetings was a duty for every member, we focus on meeting attendance as a starting point for this research. We argue that club meetings offer the best site to identify club members and to observe concretely how they participated in club activities. Many Rotarians, however, did not appear in newspapers.3 Missing members either did not attend or were just omitted in press reports. In both cases, their absence suggests weak membership. On the opposite, newspapers provide information on participants who were not members of the club, but occasional visitors, special guests or invited speakers. These persons offered the Rotary an entry point to the outside world.
In this paper, we propose a multimodal approach to persons in networks, in which persons are conceptualized as club members or meeting attendants. In this approach, persons are treated alternatively as nodes and edges. We pursue two main purposes. Substantially, how did participants behave during meetings? What roles did they perform? How did they (re)shape club rules and conventions? How did they interact with each other? How did roles and behaviors translate into positions in networks? Conversely, can we infer participants’ roles and status from their positions in networks? Did personal attributes (status, nationality, gender, profession) affect social behaviors locally (members’ positions) and the structure of the network globally (i.e. Rotary’s organization) over time? Methodologically, what can we learn from a multimodal approach to persons in networks? Can affiliation (two-mode) networks bring a new light on members’ status and personal behaviors in social life? What can we gain from, and what are the challenges of applying Social Network Analysis (SNA) techniques to historical newspapers – a source that has remained neglected in the current SNA-driven historical scholarship to date?
Methodologically, we proceeded in several successive steps. We first constructed three sample datasets extracted at different points in time. Each sample covers a one-year period which corresponds to a presidential term selected for its historical and documentary significance. The first sample under Petit’s presidency covers the first year of existence of the club (1919-20). It was a decisive period for setting rules and members’ roles. The second sample (Fitch, 1930-1) marks a period of intense activity and a turning point in term of membership, with an increasing number of Chinese members joining the club after the establishment of the Nationalist regime in Nanjing (1927-1937). It is also the best documented period in both newspapers and rosters. The last term (Harkson, 1938-9) situates during the Sino-Japanese war (1937-1945). It is a critical juncture in the local history of the club and of Shanghai more generally, as many foreigners began to leave the city and Japanese membership became problematic. This sample will serve to address the issue of whether the war disrupted the life of the club and its members’ roles. While newspapers reports became scant, we can rely on a richly detailed roster of members for this term. For each sample, we extracted the following information. From newspapers, we extracted the dates of meetings, participants’ names and the roles they performed during meetings (used as basic information for identifying nodes and qualifying edges). We also coded the types of events (regular meetings, special events, outings, etc.) and topics of discussion (used as events attributes). From the rosters, we extracted additional information on members’ status, nationality and social-professional occupations (used as persons’ attributes). The following table summarizes the data available for each period [Table – see the pdf file].
Drawing on these data samples, we built four sets of networks. Starting from a bimodal network of meeting attendance (persons-events joined by roles), we first propose a direct method analysis aimed at identifying the most important persons and events (nodes), and at examining the specific roles that participants performed during events (edges). The second step consists in projecting the initial two-mode network into two one-mode networks of persons-persons joined by events (co-attendance) and events-events joined by persons (multiple attendance). In order to supplement the bimodal approach, we offer a dual projection approach, which consists in jointly analyzing the two projections in order to offset the loss of data resulting from the projection (Borgatti 1997, 2013). In the second section, we examine co-attendance networks (2) in order to refine our analysis of personal profiles, to identify the most frequent co-attendants and sub-communities based on attendance. The next section focuses on the networks of multiple attendance in order to investigate how persons (participants) could serve as links between events (3). We eventually built a network of roles joined by persons to examine more closely how participants could perform multiple roles, to detect specific patterns of role combination and to how these patterns changed over time (4).
1 This research is made possible by the ERC advanced grant project “Elites, networks and power in modern China” (https://enepchina.hypotheses.org/). This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No 788476).
2 In this research, we relied on the digital versions of English-language newspapers accessible on the database “ProQuest Historical Newspapers – Chinese newspaper collection”, which includes the North-China Herald, China Weekly Review, China Press (just to mention the most important). As for rosters, we relied in two editions issued in 1930 and 1940, which we found in Calder Papers (1911-1956), Hoover Institution Archives, Stanford, California.
3 Only 36% of members listed in the roster issued in 1930 appeared in newspapers, and 44% in 1940.
4 All the original files and figures mentioned in this paper are available on the research platform MADSpace (https://madspace.org/).