top of page

License Grammaticism

Back in the days of my early education (and probably most people of my generation who went through the U.S. education system), English teachers used to love to have students create sentence diagrams. This pedagogical exercise was designed to help students identify the major parts of speech (subject, verb, object) in an English-language sentence, as well as the other ancillary words used in composing that sentence. The point of the exercise was to identify where certain words were incorrectly or confusingly used in a sentence. After learning how to do a sentence diagram, most students (except maybe those who went on to become English teachers) forgot the whole thing.

For the legal draftsperson, understanding parts of speech can actually be a useful exercise, as it can elucidate issues in the way a particular legal document may be incomplete. An excellent example of how that issue works out in actual, legal, practice, it the debate around the BSD license (or at least, the -0, -1, -2, -3 & -4 clause variants) and the scope of the grants in that license. Several years ago, there was a vigorous debate, which actually made it into the legal academic literature, about whether the grammar of the BSD 1-4 clause licenses conferred a patent license to the recipient. One set of authors, primarily relying upon statements from the license authors/stewards, argued they do not; another author, relying at least in part upon the actual grammar in the license – and in particular the verbs used – argued they do.

During the time that this debate was most vigorous, I did a short presentation including a grammatical analysis of a group of OSI-approved non-copyleft (also known as “permissive”) licenses: BSD, MIT, Apache 2.0, UPL and BSD+Patent (full disclosure: I authored BSD+Patent and shepherded it through the OSI approval process). The point of the analysis was to show which licenses used which verbs from U.S. copyright and patent law (which outline the exclusive rights under copyright and patent, and thus those rights that a copyright or patent holder may confer via a license). Some licenses use but a few of these verbs; the absence of a verb might allow a licensor to later argue the rights that that verb encompasses were not granted. And there's at least one license, apparently a BSD-variant, which has chosen and omitted different verbs, for reasons which are not entirely clear. Of additional interest to the legal grammatician are the objects of the verbs in these licenses, which are also outlined in the presentation.

In an international context, it can be even more interesting. Different jurisdictions use different verbs, which may (or may not) correlate with one another, and some of those verbs may not be used in licenses primarily drafted with U.S. law in mind. I have prepared a separate set of charts outlining how those verbs are used internationally (one of these charts will be included in a book chapter I wrote for a book coming out later in 2022).

Another factor which potentially comes into play – at least in the U.S. – is whether the use of certain verbs without the recitation of all statutory or treaty verbs nevertheless should be interpreted as implying a license to the exercise the rights in the unrecited verbs. Implied patent licensing has been a subject of analysis for both the MIT and GPLv2 licenses.

Whether these questions will ever be the subject of a live legal dispute, and if so, how it will be resolved, remains to be seen. Nevertheless, it has certainly become a practice in more recently-developed FOSS licenses to include a full list of operative verbs to reduce the risks that the licensor will argue later that certain rights were reserved or not conferred by omission. Perhaps the best way to curtail such arguments is to make sure the license involved is as grammatically robust as possible.

225 views0 comments

Recent Posts

See All


bottom of page