https://d2ng68cf36kxsk.cloudfront.net/store/public/post/1265/post_images/Slide407.JPG">

The Unit’s focus for these decades was computational work that had a rigorous philosophical basis. For this, she worked on the development of a thesaurus based on semantic principles, in the form of a computerized lexical database. The challenge, she believed, was not just to develop single-level dictionaries or ontologies that classified lexical sets, but to develop a view of the relations between sets. Lexicographers never got beyond single-level naming.

The thesaurus Masterman proposed would offer a more sophisticated, multilayered account of meaning, as well as connecting its constituent parts with semantic primitives. It would also function as an interlingua for machine translation and other applications of computers to information and meaning.¹²² But importantly, unlike the Chomskyans, her work was grounded in the empirical messiness of language and the meanings to which language referred. The thesaurus would emerge from the analysis of the practice of language.¹²³ Today, this idea would be called “machine learning,”^§AS1.4.6c and her thesaurus would be called an “ontology.”^§MS3.3a

An early collaborator in this endeavor was Richard Richens. In his regular job, he was Director of the Commonwealth Bureau of Plant Breeding and Genetics at Cambridge University, where his main claim to fame was to create an exhaustive taxonomy of the elm tree. As a taxonomist, he was interested not just in the definitions of terms, but their systematic inter-relations.

In 1958, Richens wrote a paper, “Interlingual Machine Translation.” Here, he said, “Linguistic and translation problems are, to one way of thinking, more clearly and usefully formulated in terms of a standard language, devised, as Wittgenstein once suggested, to mirror the logical multiplicity of the state of affairs which is being represented. Thus the twelve English terms stallion, bull, ram, mare, cow, ewe, colt, calf, lamb, horse, ox, sheep can obviously be replaced by three terms for the animal species and terms, respectively, for sex, masculine, youth and contrariety. It is redundant to allocate a term for female, which can be defined in terms of sex, male and contrariety.” Margaret Masterman was a vocal champion of women’s rights, so perhaps it is with her in mind that he added, “If preferred, feminists could define male in terms of sex, female and contrariety, but it is not possible to dispense with both male and female. Natural languages recede from formal simplicity in using homonyms. These have no place in an efficient interlingua.”¹²⁴ An interlingua, Richens concluded, would need to be minimalist and non-redundant.

With these principles in mind, Richens embarked on the NUDE project, (“naked ideas for the bare essentials of language”), a notational interlingua which he first created for plant biology.¹²⁵ In later developments, NUDE adopted some ideas from Charles Fillmore’s^§MS2.1.2a case grammar.¹²⁶

Also closely associated with the Unit at the time was Michael Halliday,^§MS0.3a then University Lecturer in Chinese at Cambridge University. Masterman took from Halliday the idea that grammar was fundamentally semantic and pragmatic, or about meaning in the play of social purposes or interests. Working with Halliday, she also became interested in the Chinese ideograph, seeing in it something with which Wittgenstein had been wrestling in his picture theory.^§MS2.1.1d An arrow could represent movement; objects could be arranged to express facts; an utterance can point to a situation.¹²⁷ The interlanguage she was proposing would be ideographic, or in our terms, multimodal and transpositional.

In the context of his work with Masterman’s Unit, Halliday had also been thinking of a “mechanical thesaurus” taking an “interlingual form” capable of, among other things, supporting machine translation. Because “the present writer happens to be interested in railways,” he said, he took a number of sentences about train travel and looked at what it would take to disambiguate the terminology of “station,” “stop,” “platform,” “track,” “train,” “line,” “rail,” and “coach” in English, French, Italian, and Chinese. The thesaurus would leverage “contextual determination within the language … The principle on which it rests, that of ‘make the language do the work,’ can only be fully applied after the linguists have done the work on the language.”¹²⁸

Halliday also published with Richens on “Word Decomposition for Machine Translation,” articulating principles of “stemming,” or reducing words to semantic elements – “jumps,” “jumping,” and “jumped” can all be stemmed to “jump.” This technique was to become an integral part of the discipline now called Natural Language Processing.¹²⁹

When Masterman died, the Research Unit’s offices were cleared out, and its trove of research papers thrown into a dumpster. Today, she is credited as having been a foundational thinker in the development of artificial intelligence.¹³⁰

---

¹²¹ Ludwig Wittgenstein, 1933–34 [1958], Preliminary Studies for the “Philosophical Investigations,” Generally Known as the Blue and Brown Books, New York: Harper and Row.

¹²² Margaret Masterman, 1967, “Mechanical Pidgin Translation,” pp. 195–227 in Machine Translation, edited by A.D. Booth, Amsterdam: North-Holland Publishing; Margaret Masterman, 2005, Language, Cohesion and Form, Cambridge: Cambridge University Press, pp. 83–145.

¹²³ Yorick Wilks, 2000, “Margaret Masterman,” pp. 279–97 in Early Years in Machine Translation: Memoirs and Biographies of Pioneers, edited by W.J. Hutchins, Amsterdam: John Benjamins; Frank Knowles, 1987, “Margaret Masterman: Her Life and Work,” Computers and Translation 2(4):198–203.

¹²⁴ R.H. Richens, 1958, “Interlingual Machine Translation,” Computer Journal 1(3):144–47, p. 144.

¹²⁵ Karen Sparck Jones, 2000, “R.H. Richens: Translation in the Nude,” pp. 263–78 in Early Years in Machine Translation: Memoirs and Biographies of Pioneers, edited by W.J. Hutchins, Amsterdam: John Benjamins.

¹²⁶ Yorick Wilks, 2000, “Margaret Masterman,” pp. 279–97 in ibid., pp. 287–88.

¹²⁷ Ibid., p. 285.

¹²⁸ M.A.K. Halliday, 1956 [2005], “The Linguistic Basis of a Mechanical Thesaurus,” pp. 6–19 in Computational and Quantitative Studies: The Collected Works of M.A.K. Halliday, Volume 6, edited by J.J. Webster, London: Continuum, pp. 12–14, 19.

¹²⁹ R.H. Richens and M.A.K. Halliday, 1957, “Word Decomposition for Machine Translation,” paper presented at the Eighth Annual Round Table Meeting on Linguistics and Language Studies, Georgetown University, Washington, DC (mt-archive.info/GURT-1957-Richens.pdf).

¹³⁰ Yorick Wilks, 2000, “Margaret Masterman,” pp. 279–97 in Early Years in Machine Translation: Memoirs and Biographies of Pioneers, edited by W.J. Hutchins, Amsterdam: John Benjamins.

---

Kalantzis, Mary, and Bill Cope. 2020. Adding Sense: Context and Interest in a Grammar of Multimodal Meaning. Cambridge UK: Cambridge University Press. pp. 224–226. https://doi.org/10.1017/9781108862059

Margaret Masterman’s “Mechanical Thesaurus” §AS2.2.1c