Dangerous questions about morphology

A recent short paper by Laurie Bauer ask 6 questions about the morphology of English. Bauer shows that each of them carries unhelpful presuppositions that are ultimately not likely lead to a sustainable theoretical position.

1 What is the plural of mouse?

Grammars of English generally say that the plural form of mouse is mice. (It is one of a small number of English nouns whose plural is formed by a historical process known as umlaut.)

Bauer agrees that mice is the form most likely to be used in standard English for serious discussion of rodents. But he points out that it has a range of other plurals, such as mouses, meeces and mices. For example, people sometimes use the plural mouses in talking about an input device attached to a computer or about characters (eg Mickey Mouses), or meeces or mices when playing with language. They are not all equally likely, but even in context we may not be able to predict absolutely which one will occur.

Bauer gives some other examples of words where different speakers produce different plurals, or where some speakers hesitate over the plural:

  • words derived from Latin or Greek for which the original ending is not transparent as a plural suffix for many modern English speakers: criteria / criterions / criterias; phenomena / phenomenons / phenomenas; octopuses / octopodes / octopi
  • Hebrew plurals whose plural form is not apparent to most modern English speakers: goyim; kibbutzim

He concludes that the forms filling paradigm slots are not necessarily unique. So, any theory predicting only a single form is likely to be insufficient.

2 When do irregularly inflected words become regular?

Some people assume that rules can specify the cases when a regular inflection must (or may) be used for a word that normally bears a regular inflection. Rules that have been proposed include ones referring to the following environments:

  • within a name (eg Mickey Mouses, not Mickey Mice).
  • within a ‘bahuvrihi’ or exocentric compound, such as lowlife (not a type of life) or sabretooth (not a type of tooth).
  • on a noun with a figurative meaning (eg computer mouse);
  • when an irregular verb is converted into a noun and then again converted into a verb (eg to grandstand, past tense grandstanded)

Bauer concludes that some constructions which regular inflection more likely. But variability is endemic, and neither constructions nor lexemes provide sufficient information to allow prediction of regular versus irregular inflection. There may be preferences and tendencies, but there are no rules.

3 How do you make an adjective negative?

The 3rd question is which form is used to convert an adjective into the corresponding negative adjective. English as at least 3 prefixes used for these conversions: a-, i- and -un-.

In many cases, 2 or even all 3 of these prefixes can be used. Examples include:

  • atypical / untypical
  • improper / unproper
  • alogical / illogical / unlogical

It is not always possible to predict:

  • which of these prefixes can attach to any given adjective.
  • whether and how much the available prefixed forms differ in meaning, or overlap in meaning.

4 Which allomorph do we use when?

Some affixes have different forms applying in different phonetic environments. For instance, the prefix en- has an allomorph em- appearing before bilabials (-m, -b, -p). For example, we have entrain but embark and emplane.

But this rule has exceptions, including enmesh and enmire. Similarly, the prefix -in has allomorphs ir- (before -r-), il- (before -l-), im- (before bilabials) and in- (elsewhere).

As expected, those allomorphs produce the forms irreparable, illegitimate, improbable, inevitable.

But dictionaries and corpora also supply unexpected forms such as inbalance, inbearable, inbelievable, inperfections, inpraticable and inpronounciable.
Bauer points out that <n> and <m> are next to each other on an English keyboard, so some of these forms could just be typos.

5 Is -ness more productive than -ity?

The 5th question is whether the suffix -ness is more productive than the suffix -ity. Bauer provides data which I have summarise in tables 1 (where -ity is more common) and 2 (where -ness is more common).

-ality / -alness>1,00091
-arity / -ariness26446
-bility / -bleness>1,00081
-icity / icness41035
-inity / -ineness1189
-ority / -oriness22615
-anity /-anness8657
Table 1. Where -ity is more common than -ness

-antness /-anity61
-esqueness /esquity30
-fulness /-fility1540
-ishness /-ishity1380
-ousness / -osity435133
-someness / – somity320
Table 2. Where -ness is more common than -ity

Bauer points out that it is not enough to say that one of these prefixes is more productive than the other. In some domains, one of the prefixes is more productive. In the contexts captured in table 1, -ity is more productive. In other contexts, the other prefix is more productive.

Bauer also points out that the tables do not show how productive the prefixes now. Instead, they show how productive they were when the words were formed.

6 Is -TH a productive suffix?

A few English nouns end in the suffix -th that creates a noun from an adjective, as in warmth. That suffix does not widely form new words today. According to Bauer, the most recent formations recorded in the Oxford English Dictionary are gloomth, greenth, blueth, builth (from build) and illth, all first attested between 1753 and 1862. (I haven’t come across any of those 5 words.)

Bauer discusses the word coolth. He says many people view it as a recent recreational formation, but points out that the OED first records it in 1547. He suggests that this form, though rare, ‘has been in the linguistic environment’ and is not being reinvented in every generation.  He suggests that one factor that may maintain its survival is the fact that users can see a transparent pattern (as with warmth). (Personally, I don’t remember ever coming across coolth, so I don’t see it as part of our living vocabulary today.)

From this 6th question, Bauer concludes that ‘the borderline between productive and unproductive is often very difficult to spot, marginal productivity is still important and still productivity and it is dangerous to assume that any recognizable pattern is completely defunct’.


Bauer concludes that it is worth challenging some of the notions that morphologists hold most dear. In particular, rules demanding fixed outcomes may not be the best way of formulating insights into linguistic behaviour. Real language use allows for more synonymy and homonymy than even practised observers may believe. Good linguistic description has to allow for tendencies as well as for definitive outcomes.

Bauer also has a more general goal. He warns that our presuppositions about the way the world is, and about the way a language is, can lead us to ask the wrong questions when setting up our theories. Very often, the most fundamental parts of our theoretical base are the most deeply embedded in our assumptions, and yet may be the hardest to explain and justify.  

Close examination of the fundamental data of linguistic descriptions can show us where we have gone wrong, and where we may need to stop and rethink our theoretical machinery. We must always be prepared to jettison any parts of our theory which are misleading, or which impede scientific progress.


‘What is the plural of mouse?’ and other unhelpful questions for morphologists, Laurie Bauer (2021), Cadernos de Linguística, 2(1), p. e303. doi: 10.25189/2675-4916.2021.v2.n1.id303.

Leave a comment

Your email address will not be published. Required fields are marked *