LING150/University of Oregon

Go to Unit 2 Study Plan

Go to Home Page

Web Lecture 2.1
Unpredictable Allomorphy

2.1.1. Allomorphs - Predictable and Unpredictable

2.1.2. One morpheme or two?


2.1.1. Allomorphs - Predictable and Unpredictable

In WebLecture 1.2., we observed that a single morpheme may have two, or even more, alternate forms. The different forms are called allomorphs. There are two kinds of allomorphs:

Unpredictable allomorphs are variant forms which don't seem very systematic to modern English speakers. They are almost always a product of history and did make sense at an earlier point in time. For example, the allomorphs gen , gon and gn meaning 'birth,type,origin' are the leftover remnants of an ancient grammatical process that occurred predictably in Indo-European, the prehistoric language from which most modern European languages developed. This grammatical process has all but disappeared in modern languages. Since the allomorphs are no longer predictable, we have to memorize them. You will always be given all unpredictable allomorphs in the morpheme sets. We'll look at the different types of unpredictable allomorphs in Chapter 4 and WebLecture 2.2.

Predictable allomorphs are systematic. In fact, they occur so regularly that linguists can write rules which describe the alternations. For example, take a look at the words below.

abbreviate

annual

account

appeal

affect

arrive

aggressive

assent

alleviate

attend

All the highlighted prefixes are allomorphs of ad- 'to,toward'. The d in ad- always changes to the sound of any following consonant except m, j, and v (for example, admit, adjourn, and advise). Because this occurs so predictably, we don't need to memorize the various forms. We'll look at these kinds of allomorphs in detail in Unit 3.

 

2.1.2. One morpheme or two?

Suppose you have two different forms with very similar meanings. How do linguists decide whether these are two different morphemes or allomorphs of the same morpheme? The first step involves looking for evidence of a shared history. (We'll see how this is done in Unit 4.) If there is no shared history, the resemblances are accidental and what we have are two morphemes.

Let's suppose that linguists do find that the two forms have originated from the same historical source. However, each form has followed a separate historical path. Perhaps English borrowed the different forms from different languages. Perhaps historical sound changes have affected the two forms in different ways. How different can the forms be to still be considered allomorphs? The answer is : not very different. For example, the form gn, which we mentioned above, came into Latin from the prehistoric Indo-European language. In Latin, it was often used with the suffix -at (which you studied in its Anglicized form, -ate). Over time, the initial g of the root was lost, leaving only n + at, or the morpheme which you've studied as nat 'source,birth,tribe.'

Although linguists can trace the history of this morpheme, the relationship of the two forms isn't very obvious to people who speak the language today. At some point, the changes in the two forms becomes so great that it's just easier to call them different morphemes, rather than allomorphs of the same morpheme. There isn't any clear dividing line, though, which leaves plenty of room for argument among the experts!

Your morpheme sets are constructed based on the consensus of many linguists. Separate morphemes are always presented on separate lines. Allomorphs of a single morpheme are listed together on the same line, with slashes between the different forms.

The picture below shows some of the history of the gen / gon / gn morpheme. The different allomorphs followed different historical pathways in Greek, Latin, and Germanic languages such as Old English. The Modern English words that have arisen from each pathway are shown in red. (Notice that the Old English words have a "c" (pronounced as [k]) instead of a "g"; we'll discuss why in Unit 4.)


Continue with the Unit 2 Reading Assignment 1.

Go to Home Page

Copyright 1998 by the Department of Linguistics, University of Oregon.