Morin argues that a standardization-oriented account best explains the nonexistence of “generalist, self-sufficient ideographies” (target article, sect. 1, para. 12). As against the competing learnability accounts his argument is compelling, but the space available means the notion of “standardization” is unavoidably somewhat underspecified. A more granular account will provide insight into the precise mechanisms at work. Morin's article looks directly at a variety of codes, but also argues analogically from histories of standardization in other technologies, and I will proceed similarly.
“Standardization” it is not a simple property. First, at a minimum (and as glanced at in the article), it often is a scalar rather than a binary. Metrological standardization, for instance, takes place within some range of tolerance. Second, the thresholds vary with how extensive the scope of any standard is in time and space. So, metrological standardization in the Classical Mediterranean world as a whole was quite loose (Riggsby, Reference Riggsby2019, pp. 100–114), but for individual construction projects it could be extremely precise (Riggsby, Reference Riggsby2019, pp. 115–116). Third, outside of pure metrology, standardization is multidimensional; ancient Mediterranean transport vessels standardized volume, shape, and perhaps weight (Olmer, Reference Olmer, Corti and Giordani2001). Even weight systems sometimes standardize shape as well as mass, often to the detriment of precision in the latter (ancient Rome: Luciani & Lucchelli, Reference Luciani and Lucchelli2016; ancient Near East: Schon, Reference Schon2015; early modern Burma: Gear & Gear, Reference Gear and Gear1992). Finally, multidimensional standardization often requires the choice of dimensions along which difference will be ignored. For instance, the rise of the modern commodity grain trade in the American Midwest involved the collapsing of numerous qualitative distinctions into a small number of standard categories to permit large-scale, automated handling of the product (Cronon, Reference Cronon1991, sects. 2485–2621).
These considerations then raise the following question in the context of the target article. How much standardization of what sort(s) across what sort of group(s) would an ideography require to succeed? In the following I consider in these terms two cases of codes that have been successful as a way of organizing the inquiry.
Scripts that allow the writing of languages are codes and thus they should also require standardization. (That the scripts are simpler than the languages themselves, vastly so in most cases, presumably reduces but does not eliminate the demand here.) Current research on evolution of scripts points immediately to a limit on the standardization required. For instance, Miton and Morin (Reference Miton and Morin2021) choose to ignore “font” or merely “typographic” variants of character shapes (e.g., sans-serif or italic forms of Latin script). This seems correct, but what are the limits?
Consider the attested variation in Sumerian cuneiform and hieroglyphic Mayan forms shown in Figures 1–3.
The reader may compare the more exhaustive listings for Sumerian (Mittermayer, Reference Mittermayer2006) and Mayan (Kettunen & Helmke, Reference Kettunen and Helmke2020; Prager, Reference Prager2014) to see that this level of diversity is common and that vastly more varied cases also exist (e.g., Sumerian MAḪ or ĜIR2 or Mayan /u/ or TZ'AK “whole”). The diversity cannot be significantly reduced by choosing to ignore features (as in the commodity case above). Moreover, although there is considerable evolution over time in both scripts (cf. Miton & Morin, Reference Miton and Morin2021, for some general principles at work), there is also a clear local dimension. For Mayan, this can be seen directly in the site-specific syllabaries collected by Boot (Reference Boot2010a, Reference Boot2010b, Reference Boot2010c), and for cuneiform the time-limited sign lists collected by the Cuneiform Digital Library Initiative (2016). In this context I would suggest that the standardization of early-complex scripts is perhaps surprisingly low, certainly extending well beyond what is recognizable as variation in “font.” This nonstandardization appears to be one of the reasons it has been particularly to develop computer character recognition for cuneiform scripts (Bogacz, Gertz, & Mara, Reference Bogacz, Gertz, Mara, Wohlhart and Lepetit2015; Bogacz & Mara, Reference Bogacz and Mara2022).
We can also look a little more closely at the role of scope of standardization. That is, given some notional target level of standardization, how broadly is it achieved across spaces or populations? In the Sumerian and Mayan cases, writing and reading were done principally by a professional scribal class (Delnero, Reference Delnero2010; Houston, Reference Houston, Boone and Mignolo1994), and most of our surviving texts were meant for consumption within the same circles in which they were produced. This should be a congenial environment for the development of small islands of high standardization. At the same time, the same scripts were also used for communication outside those immediate contexts (Michalowski, Reference Michalowski2011; Tedlock, Reference Tedlock2010, pp. 146–164), so their success was not purely an artifact of isolation.
A similar pattern arises from looking at another set of codes: Languages themselves. Morin is undoubtedly right that face-to-face language (oral or signed) has self-standardization mechanisms that no graphic code does, but the leap from “the level of the pair” to “entire populations” (target article, sect. 6, para. 8) of course elides some complexity. Of particular relevance here is the phenomenon of dialect continua, in which chains of mutually comprehensible language forms cumulate large differences from end to end (Chambers & Trudgill, Reference Chambers and Trudgill1998). In both cases (scripts, spoken dialects) overall standardization is low but the codes in question remain useful. This is apparently sustainable because of the existence of pockets of relatively high standardization.
None of the above seems to me to cast any doubt on the fundamental truth of Morin's historical argument. Nonetheless, these issues are worth raising for two reasons. Historically, how much standardization has been achieved by individual codes in particular contexts? In particular, to what extent have small communities been able to serve as nuclei to produce and diffuse complex standards as an intermediate step to universalization? Prospectively, the answer to that helps us calibrate an answer to Morin's question whether emoji (or anything else) could evolve into the world's first general-purpose ideography.
You have
Access
Morin argues that a standardization-oriented account best explains the nonexistence of “generalist, self-sufficient ideographies” (target article, sect. 1, para. 12). As against the competing learnability accounts his argument is compelling, but the space available means the notion of “standardization” is unavoidably somewhat underspecified. A more granular account will provide insight into the precise mechanisms at work. Morin's article looks directly at a variety of codes, but also argues analogically from histories of standardization in other technologies, and I will proceed similarly.
“Standardization” it is not a simple property. First, at a minimum (and as glanced at in the article), it often is a scalar rather than a binary. Metrological standardization, for instance, takes place within some range of tolerance. Second, the thresholds vary with how extensive the scope of any standard is in time and space. So, metrological standardization in the Classical Mediterranean world as a whole was quite loose (Riggsby, Reference Riggsby2019, pp. 100–114), but for individual construction projects it could be extremely precise (Riggsby, Reference Riggsby2019, pp. 115–116). Third, outside of pure metrology, standardization is multidimensional; ancient Mediterranean transport vessels standardized volume, shape, and perhaps weight (Olmer, Reference Olmer, Corti and Giordani2001). Even weight systems sometimes standardize shape as well as mass, often to the detriment of precision in the latter (ancient Rome: Luciani & Lucchelli, Reference Luciani and Lucchelli2016; ancient Near East: Schon, Reference Schon2015; early modern Burma: Gear & Gear, Reference Gear and Gear1992). Finally, multidimensional standardization often requires the choice of dimensions along which difference will be ignored. For instance, the rise of the modern commodity grain trade in the American Midwest involved the collapsing of numerous qualitative distinctions into a small number of standard categories to permit large-scale, automated handling of the product (Cronon, Reference Cronon1991, sects. 2485–2621).
These considerations then raise the following question in the context of the target article. How much standardization of what sort(s) across what sort of group(s) would an ideography require to succeed? In the following I consider in these terms two cases of codes that have been successful as a way of organizing the inquiry.
Scripts that allow the writing of languages are codes and thus they should also require standardization. (That the scripts are simpler than the languages themselves, vastly so in most cases, presumably reduces but does not eliminate the demand here.) Current research on evolution of scripts points immediately to a limit on the standardization required. For instance, Miton and Morin (Reference Miton and Morin2021) choose to ignore “font” or merely “typographic” variants of character shapes (e.g., sans-serif or italic forms of Latin script). This seems correct, but what are the limits?
Consider the attested variation in Sumerian cuneiform and hieroglyphic Mayan forms shown in Figures 1–3.
Figure 1. Variant forms of AN “sky, heaven” (Mittermayer, Reference Mittermayer2006, p. 5).
Figure 2. Variant forms of phonetic /li/ (Prager, Reference Prager2014).
Figure 3. Variant forms of logographic HUUN “writing material” (Prager, Reference Prager2014).
The reader may compare the more exhaustive listings for Sumerian (Mittermayer, Reference Mittermayer2006) and Mayan (Kettunen & Helmke, Reference Kettunen and Helmke2020; Prager, Reference Prager2014) to see that this level of diversity is common and that vastly more varied cases also exist (e.g., Sumerian MAḪ or ĜIR2 or Mayan /u/ or TZ'AK “whole”). The diversity cannot be significantly reduced by choosing to ignore features (as in the commodity case above). Moreover, although there is considerable evolution over time in both scripts (cf. Miton & Morin, Reference Miton and Morin2021, for some general principles at work), there is also a clear local dimension. For Mayan, this can be seen directly in the site-specific syllabaries collected by Boot (Reference Boot2010a, Reference Boot2010b, Reference Boot2010c), and for cuneiform the time-limited sign lists collected by the Cuneiform Digital Library Initiative (2016). In this context I would suggest that the standardization of early-complex scripts is perhaps surprisingly low, certainly extending well beyond what is recognizable as variation in “font.” This nonstandardization appears to be one of the reasons it has been particularly to develop computer character recognition for cuneiform scripts (Bogacz, Gertz, & Mara, Reference Bogacz, Gertz, Mara, Wohlhart and Lepetit2015; Bogacz & Mara, Reference Bogacz and Mara2022).
We can also look a little more closely at the role of scope of standardization. That is, given some notional target level of standardization, how broadly is it achieved across spaces or populations? In the Sumerian and Mayan cases, writing and reading were done principally by a professional scribal class (Delnero, Reference Delnero2010; Houston, Reference Houston, Boone and Mignolo1994), and most of our surviving texts were meant for consumption within the same circles in which they were produced. This should be a congenial environment for the development of small islands of high standardization. At the same time, the same scripts were also used for communication outside those immediate contexts (Michalowski, Reference Michalowski2011; Tedlock, Reference Tedlock2010, pp. 146–164), so their success was not purely an artifact of isolation.
A similar pattern arises from looking at another set of codes: Languages themselves. Morin is undoubtedly right that face-to-face language (oral or signed) has self-standardization mechanisms that no graphic code does, but the leap from “the level of the pair” to “entire populations” (target article, sect. 6, para. 8) of course elides some complexity. Of particular relevance here is the phenomenon of dialect continua, in which chains of mutually comprehensible language forms cumulate large differences from end to end (Chambers & Trudgill, Reference Chambers and Trudgill1998). In both cases (scripts, spoken dialects) overall standardization is low but the codes in question remain useful. This is apparently sustainable because of the existence of pockets of relatively high standardization.
None of the above seems to me to cast any doubt on the fundamental truth of Morin's historical argument. Nonetheless, these issues are worth raising for two reasons. Historically, how much standardization has been achieved by individual codes in particular contexts? In particular, to what extent have small communities been able to serve as nuclei to produce and diffuse complex standards as an intermediate step to universalization? Prospectively, the answer to that helps us calibrate an answer to Morin's question whether emoji (or anything else) could evolve into the world's first general-purpose ideography.
Acknowledgment
I would like to thank Mallory E. Matsumoto for invaluable guidance on the Mayan material. Any errors of interpretation are my own.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interest
None.