Sound change notation

Sound change is what happens when a people's pronunciation shifts over time. Languages are living systems, and their sounds slowly drift, sometimes so gradually that speakers barely notice it happening. (This is not to say that sound change is the only element of language evolution.)

To describe these sound changes, linguists use sound change notation. The most common form of notation being "SPE" notation, popularised in The Sound Pattern of English (Chomsky & Halle, 1968). "SPE" notation is so common that it is simply called sound change, and is what this article will be about. There are some drawbacks to "SPE" notation that the "non-linear" frameworks can be better at solving, namely Feature Geometry, Autosegmental Phonology and Metrical phonology.

In its most basic form, one sound transforms into another in a "rule". For example:

This means that sound A has become sound B.

Multiple changes in one rule

Two or more sounds can transform, or rather merge, into another. For example both A and B were to transform into C may be written like this:

A series of changes can be written like this:

This means A becomes C; B becomes D; and C becomes E, with an implication that these occur at the same time.

If there are multiple arrows in a rule, it can be interpreted in two ways depending on the author's intentions. For example:

This can either be A becomes B then C. Essentially A has a medial form B before becoming C, a "series of changes".

Or if the author describes it as a "chain shift": B transforms to C while A transforms to B.

The environment

Up to this point, the rules have applied in all contexts. If the rule only applies when it meets certain conditions, the following notation is used:

This means Sound A becomes sound B after C and before D. The underscore is used to represent the position of the sound being changed.

If the rule always applies except when it meets an environment, use a ! instead of a / like so:

Unfortunately the exclamation mark is also used as the IPA phoneme [!].

Special boundaries

To represent not a sound but a word boundary, a hash symbol # is used in the environment:

This means A becomes B only when A is at the beginning of the word.

And this means A becomes B only when A is at the end of the word.

The Greek letter lowercase sigma σ is used to indicate a syllable. To represent a syllable boundary, a closing square bracket followed by a subscript sigma can be used.

Other conventions may have a dollar sign $ as marking a syllable boundary.

This example means that the sound change takes place only at the end of a syllable.

In long-distance sound changes, it is often necessary to represent an unspecified amount of unspecified sounds between definite sounds. A common notation for this uses an ellipses …

This means A becomes B when preceded by C, regardless of any sounds that are between C and A

Brackets

In certain cases, there might be optional properties to a rule. in these cases they are enclosed in round brackets (parenthesis):

This means that A becomes B before D, whether or not C is in the way.

If more than one sound is affected or more than one environment intervenes, then curly brackets (braces) are used.

This means that the change takes place before C or word-finally.

Deletion and insertion

The empty set symbol ∅ is used to represent deletion or epenthesis depending on its location. Unfortunately it looks a lot like the IPA vowel [ø].

This is how deletion (elision) is written:

This means A is deleted before C.

This is how insertion (epenthesis) is written:

This means A is inserted before a C.

Features

Phonemes have distinctive binary features, and each feature forms a natural class of phonemes. For example, the phoneme /t/ has distinctive features such as -nasal, +plosive and -voiced, while the feature +nasal refers to phonemes such as /m/ and /n/

A feature matrix lists these distinctive features inside square brackets. The feature matrix represents all the sounds the features have in common. For example:

This means words such as gud and rub become gut and rup. But a word such as gus remains unchanged.

The result of the change, AKA the "right-hand side of a rule", may be under-specified, and therefore inherit the features from the left-hand side of the rule. For example the last example could be written with redundant features removed:

If there is only a single sound to be changed in a rule, we can write rules using under-specification such as this:

This means /o/ changes to its +high equivalent, /u/

Capital letters

Capital letters can be used as abbreviation of a natural class. These three are widely used:

C = Consonants

V = Vowels

N = Nasals

Capital letters can be modified with IPA diacritics to capture different natural classes:

Ƈ = implosives

Cʼ = ejectives

Ṽ = nasal vowels

N̥ = voiceless nasals

There are other capital letters that authors use, but they vary from author to author. If you are using these or your own you should specify what they mean in your documentation, otherwise your readers won't know what they mean. Here are some of them:

B = Any back vowel

D̪ = Dentals

E = Front vowels

F = Fricatives

G = Glides and semivowels

K = Velars

L = Liquids

O = Obstruents

P = Plosives

Q = Uvular consonants

Ʞ = Click consonants

Subscript numbers

Subscript numbers (₀, ₁, ₂, ₃…) are an indexing device. They tell you that two (or more) classes in a rule refer to the same sound, not just the same type:

This means a consonant deletes before an identical consonant.

If you have more than one indexed sound, start at ₁, then ₂, and so on. If there is only one indexed sound, you could use ₀ only.

Alpha notation

Instead of writing two rules that capture the same assimilation pattern, there is alpha notation. For example, this rule uses alpha notation to say that a plosive takes on the same value of voice as the following plosive, where both are plus or minus:

This is equivalent to a series of rules:

And by using additional Greek letters, we can reproduce more features:

This means a vowel will match the roundedness of a previous vowel when the two vowels share the same height.

Nasals often assimilate with their following stops in all the place of articulation features: labial, coronal, dorsal and all the minor place distinctions such as bilabial/labiodental and the various kinds of coronal. One could write a rule that matches labial, coronal, dorsal, and so on, until you run out of Greek letters.

A more useful approach is to say that the features relating to place of articulation are all part of a special class, a "feature-bundle" called place, and write the rule as if it were a feature:

The other special class, "feature-bundle", is called manner, and contains features relating to manner of articulation.

References

Chomsky, N., & Halle, M. (1968). The sound pattern of English.

Prickett, B. (n.d.). Variables must be limited to a single feature. University of Massachusetts Amherst.

McRobbie, M. (n.d.). Phonological rules (Lecture handout #6). Simon Fraser University.

Zukoff, S. (2019). Class 6: Types of sound changes (and phonological processes).

Gouskova, M. (2016). Features in phonology