In 1924, the Classical Investigation, a study undertaken by the American Classical League, reported a shift in the focus of teaching Latin from the ability to write to the ability to read. As a result, the amount of prose composition in textbooks diminished and was replaced by a variety of reading-oriented exercises. But while the goal of Latin instruction became reading rather than writing, an accompanying change in teaching methods was lacking. There were no specific instructions on how to go about reading Latin or translating from Latin into English. That missing component stimulated Waldo E. Sweet and his colleagues Gerda M. Seligson and Glenn M. Knudsvig, to develop an approach to Latin pedagogy which has come to be known as ‘Michigan Latin’.
Twenty-five years after the Classical Investigation, Waldo Sweet was teaching Latin at the Penn Charter School in Pennsylvania. He became intrigued with the application of structural linguistics to the teaching of foreign languages. Applied structural linguists used organized descriptions of the phonological, morphological, and syntactic systems of languages to highlight differences between the target language and the student’s native tongue. Sweet was particularly influenced by the work of Charles Fries, began to consider the possibility of applying Fries’ theories of second-language acquisition to the teaching of Latin. He received support from the Carnegie Corporation for two Latin workshops held at the University of Michigan in the summers of 1952 and 1953.
In 1950, Gerda Seligson was teaching at the Brearley School in New York when she attended an Independent Schools meeting and heard a paper given by Sweet on the application of structural linguistics to the teaching of Latin. The paper was received by the audience at large with some scepticism, but Seligson stood up (the only woman in the audience to speak) and said “That man is right!”. She was subsequently invited to be a participant at the 1952 and 1953 workshops in Ann Arbor, where participants collected and analyzed data and produced a structural description of the Latin language.
In 1953 Sweet came to the University of Michigan as Associate Professor of Latin and the Teaching of Latin. He continued to explore the application of linguistics to Latin pedagogy and in 1957 he published Latin: A Structural Approach (LASA), explaining:
After using the Experimental Materials for one year, I plunged further ahead in unorthodox lines, resulting in the publication…of Latin: A Structural Approach. My purpose was not so much to write a textbook for the average teacher as to try out new techniques…At one time or another I have discarded almost all of the traditional methods, but if I could find nothing better to replace them with I have returned to them.
In his description of the Latin language in LASA, Sweet replaced 19th century notional categories with a set of categories arrived at through the observation of their formal distribution. The distinctive features of ‘Sweet Latin’ include a horizontal paradigm showing the forms of a single case in all declensions (rather than the traditional introduction of all forms in a single declension), pattern practices typical of modern language instruction applied to, and the use of ‘real’ vs. ‘made-up’ Latin from the beginning.
In 1956, Seligson also came to Michigan to teach at the University School and in the Department of Classical Studies, and continued collaborative work with Sweet. Seligson’s contributions in those early years were refining a method of teaching strategies for translation and reading (called metaphrasing). Her classroom experiences led her to realize that while Sweet’s pattern practices were an effective drilling procedure, there was still a need for morpho-syntactic analysis in order to achieve comprehension. LASA was revised in 1966 with Seligson and Ruth Craig as co-authors. This text continued to be used at Michigan until 1981, when it was replaced by the first version of Latin for Reading.
Sweet produced a variety of pedagogical materials during these years. One of the most well-known is his unique programmed course Artes Latinae (described by Seligson “his magnificent monument to behaviorism at its strictest”) which shows the influence of Skinner and the behaviorists. Artes is still used extensively by students learning Latin in a home-school environment.
Structural linguistics formed the basis for Sweet’s approach. He explains:
A structural linguist assumes that a student of any language will have difficulty wherever the target language differs from his own. He further assumes that wherever the student has difficulty a contrast between the two languages must exist. It has also been observed that partial contrasts are more difficult to grasp than complete contrasts. The student apparently is misled by the partial similarities…The traditional approach has been the exact opposite. …The traditional approach tried to explain the contrasts by pointing out the similarities, which were in many cases superficial and in all cases were the chief sources of difficulty, since the students assumed that because these items were like English in some respects they must be like English in them all.
Two technical terms introduced by Sweet and still in use in Michigan Latin are metaphrasing and kernel. Seligson described metaphrasing as a technique “by which the replacement of Latin structural signals with English signals can be taught,” that is, a technique for teaching, practicing translation, and reading. With metaphrasing, a minimum syntactic context for each Latin unit is provided in English and the final result of metaphrasing a sentence would be a structural translation in English word order. The term kernel (the obligatory elements of subject and predicate) first appears in the 1966 LASA, although the notion of “sentence kernel” is found in earlier articles.
During the 1960s Sweet’s and Seligson’s interest in transformational grammar gave rise to an emphasis on sentence-level phenomena, most notably gapping. Gapping, traditionally called ellipsis, is the absence of common elements in one of two clauses, or, in linguistic terms, ‘unfulfilled syntactical expectations.’ This notion was included in the 1966 LASA, but the term gap came into use only following an 1970 article by John Robert Ross on gapping. Like the kernel and metaphrasing, the concept of the gap proved highly significant, as it allowed a systematic treatment of an area of Latin which follows rules distinctly different from those of English.
In 1963 Glenn Knudsvig brought his interests in Latin, psycholinguistics, and reading theory to Michigan. In 1974 he began writing with Seligson and in the early 1980s they began work on a new textbook in the linguistic tradition – Latin for Reading. The revised edition of LfR (currently in use at Michigan) was published in 1986. In a l983 article, Seligson discusses the motivation for writing LfR:
The first consideration was our gradual realization that in spite of lack of results as measured by performance the standard traditional approach flourishes again. Disappointed with some inefficient features of experimental courses teachers have sought refuge in prestructuralism, throwing out the baby with the bathwater. A talented studented acquires mastery of phonology, morphology, syntax and vocabulary at a high price in nervous energy while learning to read almost accidentally. But the less talented students, handicapped by today’s lack of basic verbal education, does not learn to read. To teach this student to read Latin is our goal….The second consideration was the fact that the theoretical work of linguists inside and outside of the United States seems to confirm and validate many of our conclusions and practices.
A comparison of LfR with Sweet’s work shows a number of developments: the pattern practices are dropped; metaphrasing and translation strategies are more prevalent throughout; gapping is formally treated; the systems of morphology, syntax, and semantics are distinguished; kernel and modification categories are expanded; the notion of syntactic equivalents is included for the first time. Semantic categories with syntactic consequences, such as animacy, were introduced. A major advance in LfR was the formalization of expectations for the reader. Seligson again:
We have tried to combine the facts of structure on all levels and of gapping and deletion with the notion of expectation. The formation of expectations is the condicio sine qua non for learning to read your own or any other language.
Knudsvig’s contribution to Michigan Latin was two-fold. First, he clarified the categories of morphology and syntax, made further refinements in the area of semantics, and added categories in pragmatics, discourse phenomena, and word order. With Knudsvig’s work, significant advances were made in moving from teaching students to read at the sentence level to teaching them to read at the level of connected text.
With the publication of LfR, the treatment of Latin syntax was refined to the point that, 19 years later, we generally satisfied with its descriptive accuracy. The vast area of semantics – issues of meaning - was a natural next step. Starting in 1985, Seligson and Knudsvig began working more extensively in this area, attempting to describe the semantic system of Latin in an organized and pedagogically useful way.
When linguistic research showed that Latin word order was largely governed by the organization of information at the level of connected text, Knudsvig’s attention turned toward the study of discourse analysis. With the development of theories of discourse analysis, Knudsvig began looking at the discourse features of Latin and analyzing them with the intention of training students to recognize and be aware of the characteristics of connected text, and thus improve their comprehension.
Knudsvig’s other major contribution to Michigan Latin was the development of specific strategies for reading Latin, which offer students a systematic way of approaching a text. Metaphrasing, which had been hitherto primarily a static written exercise, matured into a dynamic process using a simple series of three questions a student could reiteratively ask and answer: “what do I see?” (eliciting part of speech and morphological information), “therefore, what do I have?” (moving to the level of syntactic function), and “therefore, what do I expect?” (the verbal equivalent of a metaphrase). This habit of thought gave learning readers a series of things to consider, with consequential decisions.
Knudsvig’s untimely death in 1998 cut short his personal contributions, but those of us still here at Michigan continue to develop the linguistic approach. We still use the 1986 version of Latin for Reading, but a revision is underway that will incorporate information about semantics, word order and ambiguity formalized since its publication.
Michigan Latin has a presence not only here in Ann Arbor, but across the country and beyond. Starting in about 1995, an increasing number of presentations on aspects and applications of Michigan Latin began to be given at national conferences, most by people who did their pedagogical training at Michigan, but some by paper and workshop attendees who became intrigued. Work is currently being done on critical thinking in the Latin classroom, Latin vocabulary, pedagogical applications of word order research, techniques for making the initial transition into authentic connected texts, and developing Michigan Latin templates which can be used as a linguistic overlay with traditional or reading approach textbooks. Michigan Latin concepts such as kernel, metaphrasing, gapping and expectations have found their way recently into articles and materials in the outside world: a middle school teacher in Texas uses metaphrasing with the Cambridge Latin course; a handbook for intermediate readers incorporates readingquestions; an introductory Aeneid text uses the term ‘gapping’.
We continue, both here at Michigan and elsewhere, to follow Sweet’s dual vision of employing the principles of linguistic science to inform our teaching of the language, and to teach students to read Latin as Latin. There is now, almost 55 years later, a large and growing number of teachers across the country at all levels – from elementary school through university – who use the Michigan method either in a pure or modified form. We at Michigan feel that this approach carries with it the message that all students can learn something using this method, and that students and teachers together can learn to think about language in new and exciting ways.
Deborah Pennell Ross
Elementary Latin Program
Department of Classical Studies
The University of Michigan