Further evidence for self-organization in English spelling

Abstract

Further evidence for self-organization in English spelling Kristian Berg and Mark Aronoff Geoffrey Sampson (2018) makes two claims in his response to our article about self-organization in spelling (Berg & Aronoff 2017): our explanation is redundant, and the phenomena we address can be better captured in more traditional terms. In our article, we showed first that the relation between graphemic form and morphological function is isomorphic for some suffixes, for example, the adjectival suffix -ous: in today’s English, all and only adjectives with this morpheme are spelled with final <ous>, even though phonographically, more words could be spelled that way in American varieties (cf. nervous/service). We also demonstrated that this system evolved gradually in an unsupervised process of self-organization. Sampson does not contest our synchronic statement, but brings forth a variety of distinct accounts for our diachronic findings. Overall, he suggests that phonology and etymology are sufficient to explain why the spellings are the way they are: in British English, nervous and service have (and more importantly, had) distinct reduced vowels. The distinct spelling is thus explicable from phonology here. Other spellings, Sampson argues, follow a widespread convention preferring etymological spellings: service has an <i> in Latin, and <ous> stems from Latin -os-; that is why *<servous> is not the accepted form. No need for a distinct principle of self-organization, says Sampson, because all the data are accounted for already. There are three misunderstandings in Sampson’s response. We take responsibility for the first, though not for the second and third. Let us start with the first. Synchronically, we showed that graphemic form and morphological function are isomorphic. We then ask how this situation arose historically, but the fact is that we looked at only half of the historical story. To show this, we need to step back a little and look at the relation between form and function. It is fine to talk about isomorphism as above—but it can be helpful to take the relation apart: we can distinguish between uniformity and uniqueness. Uniformity denotes the degree of consistency with which a function is represented in its form: is there more than one spelling for the suffix? Today, -ous is spelled (almost) uniformly <ous>,1 but in earlier stages of English, there were variants (e.g. <us>, <ouse>), so we can say that the spelling of this adjective suffix has become more uniform over the centuries. Uniqueness flips the perspective; it denotes the degree of consistency with which a form represents a function. Is there more than one interpretation for this string of graphemes? While <ous> is unique in representing this one adjective suffix, other suffix spellings are not; <er> represents (among others) the agentive nominal suffix in <singer> and the comparative suffix in <longer>, but also a nonmorphological word ending in <hammer>. Both uniformity and uniqueness are independent of phonology: they deal with the relation between spelling and morphology. When we take phonology into account, it is as circumstantial evidence: the fact that a number of words in today’s American English could potentially be spelled with <ous> (e.g. service) is a further argument for the uniqueness. [End Page e48] Synchronically, we showed that -ous/<ous> is both unique and uniform—but diachronically, we investigated only the uniformity of the suffix. We asked whether Latinate loans like status were responsible for the decline of the <us>-variant of -ous like <humerus> (short answer: probably not)—that is a question of uniqueness. But we did not investigate the uniqueness of each spelling variant in each time period, simply because the investigation was extensive enough already. That does not mean the question is uninteresting. A superficial search in the Helsinki corpus (a selection of English texts produced before 1710) indeed shows a number of words whose endings are like variant spellings of the suffix -ous (<ous>, <ouse>), but that did not contain the suffix, for example, <almous dedes> ‘alms’, <felouse> ‘fellows’, <alehous> ‘alehouse’, or <sparous> ‘sparrows’. It remains an open question how unique the respective spellings were at any given time. We now have data that show a striking case of the emergence of uniqueness over time in the spelling of the nominal...