Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

Behind the Counter: Naming Shopkeepers in Shanghai’s Newspapers

This study of shopkeepers takes the press as its main source. This is not because we are unaware of the value of other types of sources, as discussed in our literature review. Our focus on the press is a deliberate choice for several reasons:

  • Newspapers are one of the most important historical sources in which shopkeepers are likely to appear.
  • Newspapers are also the most immediately accessible source. In the case of China, other types of materials—such as official records or guild archives—would require consultation at the Shanghai Municipal Archives.
  • We want to map how the press represented shopkeepers: as economic actors, civic leaders, troublemakers, or victims.
  • We also want to examine the semantic field of “shopkeeper” in the Shenbao: how were shopkeepers designated, which variants appeared, in what contexts, and with what connotations?
  • Methodologically, we aim to blend computational methods with conventional historical inquiry, demonstrating how large-scale press analysis can bring overlooked categories of urban actors into view.

The main objective of studying shopkeepers through the press is to recover their social, economic, and cultural role in Shanghai between the late Qing and Republican periods. Our documentary base includes the Chinese-language Shenbao and the English-language newspapers published in Shanghai.[1]

To achieve this objective, our first step was to examine terminology itself: how did the press designate shopkeepers, and which keywords would allow us to retrieve them systematically? Constructing a press corpus thus required identifying the full range of terms by which shopkeepers were described, both in Chinese and in English.

In English, the range of terms we identified is quite limited. It consists of three core expressions: shopkeeper, shop owner, and grocer. Variants appeared only in different spellings—shop owner, shop-keeper, shop keeper, etc.—which did not pose a problem, as our search algorithm automatically normalized such variations. Using the single term shop generated too many false positives, since our target was the individuals who ran the shops rather than the establishments themselves. We might also have used the specific designations of trades—such as tailor, dressmaker, or butcher—but these terms were often employed in the press to refer interchangeably to the proprietor or to the shop. We therefore chose not to pursue this path, as it would have required constructing an exhaustive list of specialties, and instead relied only on explicit designations of shopkeepers.

The dominant and most consistent term was shopkeeper, which served as the general label across the period (2,777 articles). Alongside it, grocer gained strong visibility from the 1890s onward, peaking in the 1920s with more mentions than any other term (2,068 articles). This prominence reflects the particular salience of food supply and market disputes for expatriate communities. A more formal expression, shop owner, appeared later, in the 1910s–20s (325 articles). This simplified vocabulary in the foreign press stands in sharp contrast to the far greater diversity of terms in the Shenbao, where identifying shopkeepers posed greater challenges. The contrast highlights how Chinese journalism reflected a complex indigenous commercial world, while the foreign press offered simplified, recognizable figures that aligned with expatriate perspectives and priorities.

Identifying shopkeepers in the Shenbao proved more challenging. The most common contemporary term for shopkeeper in Chinese is dianzhu (店主, shop owner). Yet we were aware that other terms, now largely obsolete, were also in use. With the advice of a fellow Chinese historian, we added three further expressions: dongjia, dongzhu, and dianjia (東家, household head or employer, 東主, master, 店家, shop owner/proprietor). We also consulted ChatGPT-5, which suggested four additional terms, including one false positive: zhanggui, shanghu, and yezhu (掌櫃, shop manager, 商戶, merchant 業主, business owner). Although the last two appeared somewhat generic, we chose to incorporate them into our search set. Our principal method for expanding the list of relevant keywords, however, was to rely on word embeddings, a function included in HistText. This approach yielded a richer list of ten terms, including one that explicitly designated female shopkeepers.

Table 1. Shopkeeper Query Terms in Chinese


Shopkeeper
坊主 店主 棧主 莊主
號主 行主 鋪主 館主
店主婦 店東 東家 東主
店家 掌櫃 商戶 業主

Table 2 records the frequencies of the various terms used to designate shopkeepers or proprietors. The language of description reveals both the persistence of older mercantile categories and the gradual consolidation of a modern commercial vocabulary. A handful of terms dominate usage. Foremost among them is dianzhu (店主, shop owner), which accounts for more than a quarter of all references. Its overwhelming prevalence suggests that it became the default expression for proprietors in newspaper reporting, a general and versatile term applicable across a wide range of businesses. In this sense, dianzhu represents a process of lexical standardization, whereby a single word came to embody the figure of the shopkeeper as it is understood in today’s public discourse.

Table 2. Distribution of Term Usage in the Shenbao (1872-1949)

Term

Number

Percentage

店主

23682

27,3%

行主

9894

11,4%

館主

9877

11,4%

業主

8359

9,6%

號主

6376

7,4%

作主

6061

7,0%

棧主

4573

5,3%

店東

4036

4,7%

莊主

2583

3,0%

店家

2437

2,8%

東家

2430

2,8%

鋪主

1363

1,6%

店主婦

1210

1,4%

坊主

1027

1,2%

商户

991

1,1%

東主

986

1,1%

掌櫃

799

0,9%

Total

86684

100,0%

 

Yet other terms reflect the diversity of Shanghai’s economic life. Xingzhu (行主, firm owner) is strongly associated with wholesale houses and guild-linked enterprises, pointing to the visibility of large-scale trading firms. Guanzhu (館主, proprietor of a guan) more likely refers to innkeepers, tavern owners, and other catering-service figures, reflecting the importance of establishments that also served as places of socialization. Haozhu (號主, owner of a firm name) evokes the traditional mercantile world in which the reputation of the “house name” remained a central asset. Finally, yezhu (業主, business or property owner) marks the entry of a more legalistic and formal category into journalistic language. Its significant share of usage indicates how notions of ownership were increasingly framed in terms of law, property, and enterprise, reflecting the modernization of economic practice. Together, these five terms account for roughly 70 percent of all references.

The remaining terms occur less frequently, but they still reveal semantic nuances. Less common but still notable are fangzhu (坊主, owner of a workshop), tied to neighborhood production; zhanzhu (棧主, depot or inn owner), associated with storage depots and transport inns; and puzhu (鋪主, shop owner), a variant that never rivaled dianzhu in frequency. Older hierarchical designations such as dongzhu (東主, master) and dongjia (東家, household head or employer) appear only sporadically, suggesting their gradual eclipse in the Republican era. Likewise, zhanggui (掌櫃, shop manager), though familiar in late imperial commerce, is relatively rare, indicating the press’s preference for highlighting ownership rather than management.

While these figures establish the overall hierarchy of usage, closer examination of the terms themselves reveals how vocabulary mapped onto the structures of Shanghai’s economy. On the one hand, Shanghai’s economic modernity fostered the rise of standardized, versatile expressions such as dianzhu (店主, shop owner) and yezhu (業主, business or property owner), which aligned with new legal and commercial practices. On the other hand, the continued presence of terms like guanzhu and haozhu underscores the persistence of older commercial forms, rooted in guild culture, service establishments, and long-standing mercantile traditions. The lexical choices in the Shenbao thus map onto broader transformations in the city’s economy and society, where old and new commercial worlds coexisted, overlapped, and gradually gave shape to the language of modern urban commerce.

The distribution of terms as measured in Table 2, however, does not capture changes over time. These aggregate patterns, however, tell only part of the story. To fully grasp how commercial vocabulary evolved, we must also examine its distribution across time. Beneath the surface of aggregate figures lay a quiet transformation in commercial vocabulary that reflected broader shifts in Chinese society. Through the pages of the Shenbao, we can trace how the language of commerce evolved alongside the city’s modernization (Fig. 1). Beyond the crude numbers—the 86,684 mentions of seventeen different terms for shopkeeper across seventy-eight years—we can discern patterns that speak to wider social, economic, and cultural change. The data tells a story of linguistic adaptation, gender inclusion, and the gradual displacement of traditional commercial categories by more modern business concepts.

Figure 1. Distribution of Term Usage over Time

The overwhelming dominance of dianzhu suggests a fundamental continuity in how Chinese society conceptualized retail ownership. Dianzhu peaked in 1927 with 682 mentions, coinciding with the height of the Republican era’s commercial expansion. What makes this dominance particularly significant is its consistency across the entire seventy-eight-year period. Unlike other terms that experienced dramatic rises and falls, dianzhu maintained steady usage, averaging over 303 mentions per year. This linguistic stability likely reflects the term’s practical utility: it was specific enough to denote ownership while flexible enough to adapt to new commercial contexts.

Perhaps the most revealing pattern emerges when we examine peak usage by decade. The 1920s stand out as the pivotal moment in shopkeeper terminology, with six different terms reaching their historical peaks in that decade. This clustering is hardly coincidental. The 1920s marked a surge in Shanghai’s commercial prosperity, the “golden age” of the Shenbao, and a period of relative political stability that fostered business growth. The terms that peaked in the 1920s—dianzhu, haozhu, dianjia, and fangzhu—evoke a commercial landscape in full bloom. The diversity of these terms reflects the diversity of Shanghai’s economy, encompassing both traditional family shops and emerging manufacturing concerns.

The temporal analysis reveals a fascinating pattern of linguistic modernization. Traditional terms such as zhanzhu (棧主, inn or warehouse owner) and puzhu (鋪主, an older form of shop owner) dominated the early period, with more than 60 percent of their usage occurring before 1912. Zhanzhu peaked as early as 1873, while zhanggui (掌櫃, shop manager) reached its peak in 1891 and accounted for less than 1 percent of total mentions. These terms evoked older forms of commercial organization—inns that served a sojourning population, or traditional shops with their characteristic architectural features.

In contrast, modern terms such as yezhu (業主, business or property owner) and diandong (店東, shop proprietor) emerged as dominant in the later period, with more than 80 percent of their usage occurring after 1911. The rise of yezhu is particularly significant. As a more abstract, corporate term, it suggests a shift from personal, family-based commerce toward more systematic and institutionalized business practices.

Beyond these shifts in vocabulary, the corpus also reveals the emergence of a gendered category that deserves particular attention: dianzhufu (店主婦, female shop owner), which peaked in 1914 with 1,210 mentions across the corpus. While representing only 1.4 percent of all occurrences, this term’s existence and timing are historically significant. The 1910s marked a period of growing women’s rights activism in China, and the Republican era opened new opportunities for women’s economic participation. The fact that dianzhufu warranted its own distinct terminology indicates that female shop ownership was common enough to require recognition, yet differentiated enough from male ownership to merit a separate category. This reflects both the changing gender roles and the new commercial realities of early twentieth-century China.

The vocabulary extracted from the Shenbao, while rooted in Shanghai, likely reflects broader patterns in urban commercial China. Shanghai’s role as a treaty port and commercial hub made it a laboratory for new forms of business organization and terminology. The linguistic patterns observed here probably spread to other major cities, contributing to a shared vocabulary of modern commerce. The persistence of certain traditional terms alongside newer ones suggests that Chinese commercial culture adapted rather than simply replacing older forms. The evolution of shopkeeper terminology in the Shenbao thus offers a unique window into the transformation of Chinese commercial culture between 1872 and 1949. The data reveals not only changing vocabulary, but also shifting concepts of business ownership, gender roles, and commercial organization.

The dominance of dianzhu (店主, shop owner) throughout this period suggests fundamental continuities in Chinese commercial culture, while the rise and fall of specific terms trace the country’s journey from traditional to modern forms of business. The 1920s emerge as a crucial decade in which old and new vocabularies coexisted, creating a rich linguistic landscape that reflected Shanghai’s position as China’s commercial capital. Most significantly, this study demonstrates how language functions as a sensitive indicator of social change. The terms used to describe shopkeepers reveal contemporary assumptions about commerce, gender, and social organization. In the bustling marketplace of words, we can hear the echoes of a society in transition, adapting its long-standing commercial practices to the demands of a modern world. The story of shopkeeper terminology is ultimately one of adaptation—of how a culture maintained its identity while embracing change, how language evolved to meet new needs, and how even the most mundane commercial vocabulary can illuminate the larger sweep of historical transformation.

Placing these findings alongside the evidence from the English-language press brings the contrasts into sharper relief. The Shenbao preserved a remarkably diverse semantic field that captured the coexistence of different business forms—some rooted in older mercantile traditions, others tied to emerging modern practices. By contrast, the foreign press reduced shopkeepers to a handful of familiar categories, framed by expatriate perceptions and priorities. Taken together, these strands underline a broader conclusion: language not only reflected but also actively shaped how commercial actors were imagined in Shanghai’s public sphere. By tracing these words across time and linguistic communities, we uncover how everyday commerce was woven into the story of China’s transition from late imperial traditions to the practices of a modern urban economy.

Methodological Note

The initial corpus was built around the term 店主 (dianzhu, “shop owner”), which served as the primary entry point into the lexical field of shopkeepers. To expand beyond this single term and capture the full range of vocabulary used to describe shopkeepers in Republican-era Chinese, I employed the HistText Word Embeddings function, a computational tool that identifies semantically related terms based on their contextual usage patterns in historical texts. This process yielded thirteen related terms.

Using the search_documents() function from the histtext R package, I queried the Shenbao for all sixteen terms simultaneously. To understand the individual contribution of each term, I also conducted separate searches for each term. The combined search yielded 69,851 articles, which formed the foundation of the working corpus. The dataset was deduplicated to ensure that articles appearing in multiple search results (due to containing more than one shopkeeper term) were counted only once.

To understand the chronological distribution of shopkeeper discourse, I performed temporal analysis at multiple levels. First, I aggregated the corpus by year, counting the number of articles mentioning shopkeepers annually. This produced a time series that revealed broad patterns in the newspaper’s attention to shopkeepers across different historical periods, including the late Qing dynasty, the early Republic, the Nationalist era, and the wartime period. Beyond the aggregate analysis, I examined the temporal distribution of each individual term to identify potential shifts in vocabulary over time. For example, I created detailed visualizations showing how frequently 店主 appeared in articles across the decades, which helped identify whether certain terms gained or lost prominence during particular historical moments. To facilitate comparative analysis, I processed each of the sixteen term-specific datasets by converting their date fields into standard year values, aggregating the counts annually, and labeling each with its corresponding term. These datasets were then combined into a unified structure that enabled multi-line visualization, with each line representing the trajectory of a different term across time. A custom color palette was applied to ensure visual distinction among the fifteen terms in the final plot.

To move beyond simple document counts and examine how shopkeepers were actually discussed in the press, I employed concordance analysis. Using the search_concordance() function, I extracted each occurrence of the sixteen terms along with 120 characters of surrounding context (approximately 60 characters on each side). This context window was designed to capture complete sentences or substantial phrasal units that would reveal the discursive contexts in which shopkeepers appeared, whether in crime reports, commercial disputes, advertisements, or other narrative frames. The initial concordance extraction yielded 107,287 occurrences, substantially more than the 69,851 articles in the corpus. This discrepancy indicated that many articles contained multiple mentions of shopkeeper terms, suggesting that some pieces were centrally focused on shopkeeper activities or issues.

Concordance data required careful cleaning to remove false positives. Chinese characters often function as components in different words, and certain character combinations that matched our search terms were actually parts of unrelated words or phrases. Through examination of the concordance results, I identified problematic patterns where the matched characters were followed by specific punctuation marks or were embedded in larger compounds that changed their meaning. Using regular expression filtering, I removed instances where the matched text was followed by markers indicating it was part of a different grammatical or semantic unit. For example, matches followed by certain punctuation patterns (。, ·, ), 、) were eliminated as they often indicated the character was functioning in a different capacity. After filtering false positives, I merged the three concordance columns (the text before the match, the matched term itself, and the text after) into a single continuous text field. This facilitated easier reading and analysis of the contextual passages.

While concordances provided valuable snapshots of shopkeeper discourse, full article texts were necessary for more comprehensive content analysis. Using the get_documents() function, I retrieved the complete text of all 69,851 articles identified in the initial search. Each article’s length was calculated in characters to enable length-based analysis and filtering. Several cleaning operations were performed on the full text corpus. Articles with missing text fields were removed, as were extremely short entries (fewer than seven characters) that likely represented metadata artifacts or incomplete records. At the upper end, I removed outliers exceeding 23,336 characters, which often represented erroneously concatenated entries or other data anomalies. These filtering steps reduced the corpus to 69,591 articles, representing a minimal loss of approximately 0.4% of the original material while substantially improving data quality. To facilitate movement between micro-level concordance analysis and macro-level full-text reading, I created a merged dataset joining the concordance extracts with their corresponding full article texts. This integrated resource enables researchers to quickly locate specific passages within longer documents, moving fluidly between close reading of key passages and broader contextual understanding of entire articles.

Based on our previous experience with text extraction from the Shenbao where articles segmentation is faulty, I divided the corpus into two subcorpora based on a 400-character threshold. Articles of 400 characters or fewer (the shorter subcorpus) were predominantly standalone news items, brief notices, or advertisements, relatively discrete textual units ideal for developing a complete workflow pending the furtehr segmentation of the articles exceeding 400 characters. These shorter pieces tended to be more focused and less likely to mention shopkeepers incidentally within broader narratives. Articles exceeding 400 characters (the longer subcorpus) will be processed at a later stage of this research.

To enable word-level analysis, I applied Chinese word segmentation to the shorter subcorpus. Unlike alphabetic languages where words are separated by spaces, Chinese text requires computational segmentation to identify word boundaries. I employed the cws_on_df() function using a transformer-based model specifically trained on Shenbao texts (trftc_shunpao_23:zh:cws). This domain-specific model is crucial for accurate segmentation, as general-purpose Chinese word segmentation tools often perform poorly on historical texts due to differences in vocabulary, orthography, and grammatical patterns. The tokenization process converted each  continuous text into a space-separated sequence of words, creating a dataset suitable for downstream analyses such as term frequency analysis, collocation studies, or topic modeling. By focusing initially on the shorter subcorpus, I can test and refine analytical methods on more tractable material before scaling up to the full corpus.

Throughout the corpus construction process, I maintained multiple versions of the dataset at different processing stages, saving each as CSV files for interoperability with various analytical tools. Key outputs included the complete article metadata (shopsAll.csv), all concordance extracts (shopsConc.csv), the integrated concordance-fulltext dataset (shops_concfull.csv), the full text corpus (shops_ftext.csv), both subcorpora (shops_ftext400.csv and shops_ftext401.csv), the temporal comparison data (combined_data.csv), and the tokenized texts (shops_400ft_tok.csv). This preservation strategy ensures both reproducibility and flexibility, allowing the corpus to be analyzed using R’s native tools while remaining accessible to other platforms and methods. The layered approach to data processing, from initial search through cleaning, enrichment, and tokenization, created a robust foundation for addressing diverse research questions about shopkeepers in Republican-era Chinese print culture.

Statement on AI-Assisted Development

In the course of developing this essay, I made use of a large language model (ChatGPT5) as a support tool for both analysis and writing. I turned to it at several moments when structuring, interpreting, or presenting my material. For example, I first asked the model to transform a list of Chinese terms for shopkeepers into tabular form. This operation, repeated in different formats, allowed me to better visualize lexical variety before moving on to interpretation. Most substantively, I drew on the model for interpreting data about shopkeepers in the press. This included frequency distributions of Chinese terms in Shenbao, absolute and relative timelines of mentions in both Shenbao and the English-language press, and breakdowns by publisher. At each stage, I asked for a structured analysis of the visualizations and then reformulated these results into narrative prose. In this way, I was able to move from technical data to readable historical argument more efficiently. Finally, I relied on the model’s capacity for stylistic support when crafting titles and framing my blog posts, ensuring consistency across the two parts of the series. In all these cases, the LLM functioned not as a source of information but as a tool to structure, interpret, and narrate my own research results. It helped me experiment with presentation formats, clarify analytical trade-offs, and prepare prose that moved smoothly from data to interpretation.


[1] For both the Shenbao and the English-language newspaper, we relied on the pre-processed collections on the Modern China Text Base (SolR server). The source for the Shenbao is the commercial version of the Zhongguo Jindai Baokan collection. The source for the English-language newspapers is the ProQuest Chinese Newspapers Collection (1832-1853).


OpenEdition suggests that you cite this post as follows:
Christian Henriot (October 1, 2025). Behind the Counter: Naming Shopkeepers in Shanghai’s Newspapers. Blog Virtual Shanghai. Retrieved November 13, 2025 from https://doi.org/10.58079/14txn


You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.