Morphology
The branch of linguistics which deals with word structure is called morphology. Since the origin of natural language its users have been conscious of words and their structure. But morphology, the study of word structure, had not been established as a separate branch of linguistics until the nineteenth century. In the twentieth century, morphology has played a dominant role in linguistic analysis. During 1930 -1950, the period of American Structuralism, linguistics or the study of language comprised four levels:
Semantic level (meaning)
Syntactic level (sentence structure)
Morphological level (word structure)
Phonological level (sound system)
These linguistic levels are hierarchical in order, in that the semantic level is at the top and the phonological level is at the bottom of linguistic analysis.
Consider the sentence in (1).
1. Some boys liked bread.
This sentence consists of several units. These linguistic units are sentence, phrases, words, morphemes and phonemes. They are hierarchically organized as in (2).
(2) sentence > phrase > word > morpheme > phoneme
That is to say, a phoneme is the smallest unit while a sentence is the largest unit. Phonemes combine to make a morpheme. For example, look at the following transcription (3) how phonemes combine to form morphemes:
3. sәm bɔɪ laɪkt bred
In (3), the phonemes such as /b/ and /ɔɪ/ make the morpheme (boy) while /z/ stands for the plural morpheme (-s). Similarly, /l/, /ɔɪ/ and /k/ combine to form the word like and the single phoneme /t/ stands for the past morpheme (-ed). Then morphemes together are grouped together to form a word, e. g. (boy) + (s) = boys and (like) + (-ed) = liked. Next, words are arranged to form a phrase. For example, the two words some and boys make the noun phrase some boys while liked and bread constitute the verb phrase liked bread. Finally, phrases combine to make a sentence. For example, some boys and liked bread combine to form the sentence Some boys liked bread.
All languages have phonemes, morphemes, words, phrases and sentences. Of course, units, it needs a good deal of analysis to identify sounds and phrases. Even sentences are not always easy to identify. We do not always speak complete sentences but just a word or a sequence of words.
As the title of this chapter indicates, we will deal with structural morphology in this section. There are three theories to analyze word structure in structural morphology. They are as follows:
i. Item and Arrangement theory: words may be split into linear sequences (arrangement) of morphs (items). For example, consider the following sentence:
The players kicked the ball.
This sentence consists of the sequence the+player+player+s+kick+ed+the+ball. The techniques developed for decomposing words into such a sequence of morphs or items constitute the Item-and-Arrangement (IA) theory. The IA theory, introduced by Hockett (1958), was practiced in linguistics during 1940s and 1950s for morphological analysis.
ii. Item and Process theory: There are some words whose structure can be analyzed as sequence of morphs as done by the IA theory. Let us consider the word took. The word took is derived from take by a process involving vowel change. Hockett (1958) terms this approach as the Item-and-Process (IP) theory. In morphological analysis, the IP theory is used as an alternative to the IA theory to account for analyzing words such as took, mice, etc.
The IP approach assumes that a morpheme would have two forms – an underlying form and a derived form. Thus, in deriving baked and took we treat bake and take as their underlying forms. Then we apply two different processes of to derive their past tense forms.
iii. Word and Paradigm theory: there remain some cases which cannot be accounted for by the IA and IP theories. An alternative theory, called the Word-and-Paradigm (WP) theory., has been developed to explain these residual cases. The WP theory was first elaborated by Robins (1959) and later supported by Mathews (1972). Unlike the traditional approach, it does not arbitrarily select one form of a word as its underlying form and derive the remaining forms in the paradigm from this. Instead, it chooses a common factor (root) within the paradigm and derive the variant forms from this by applying rules.
Morph, morpheme, and allomorph
A morpheme is the smallest unit that has meaning in a given language.
Like ‘build’ and ‘builder’, all other words can consist of one or more than one unit. The words (such as ‘build’) consisting of a single unit are called simple words, while the words (such as ‘builder’) consisting of more than one unit are referred to as complex words.
Now consider the units ‘build’ and -er that make the word ‘builder’. The unit build is a word. In addition to being a word, it is also a unit, called morpheme. The unit -er, on the other hand, is a morpheme but it is, unlike build, not a word. Then what unites these two units? Both of them satisfy the following two criteria:
(i) They seem to assign some sort of meaning, or at least function, to the word they occur with.
(ii) They cannot themselves be broken or decomposed into smaller morphemes.
For example, the morpheme build means 'to construct' and cannot be further broken into smaller morphemes. We may, however, split up morphemes like build into phonological units such as phonemes /b/, / ɪ/, /1/ and /d/ and syllables. But then these phonological units are not only not meaningful but also belong to a different level of analysis, viz. phonology. Similarly, another morpheme in builder is -er, which means 'someone who does something, i.e., 'doer' and cannot be further broken into smaller morphemes.
Consider the word development, however. This word is comprised of two morphemes: develop and -ment. Develop has a meaning ('to grow or change') and cannot itself be divided into smaller morphemes. It is rather difficult to tease out the meaning of the morpheme -ment. But it is clear that it assigns a function, in that it is added to some verbs that refer to actions, processes, and states and turns it into a noun, viz. development. Similarly, consider books, where book is a meaningful unit referring to an item while -s expresses the grammatical function of plurality.
We have thus noticed that a morpheme is a meaningful or functional unit of grammar. We have also observed that it cannot itself be split up into smaller morphemes and is thus minimal. To sum up, a morpheme refers to the 'minimal units of meaning or grammatical function' that make up words in language. We thus arrive at the following definition of morpheme:
"The morpheme is the smallest difference in the shape of a word that correlates with the smallest difference in word or sentence meaning or in grammatical structure." (Katamba, 1993:24)
(Note: Thus, the definition of morpheme as just 'the minimal meaningful unit' of grammar is inadequate.)
We thus know what is meant by a morpheme. When we analyze a word into morphemes, we break it into morphs. A morph has a physical shape consisting of a phoneme or a sequence of phonemes and realizes a morpheme. For example, /-s/ in /bʊks/ ‘books’ is a phoneme standing for the English plural morpheme while /id/ (-ed) in /wantıd/ wanted is a sequence of two phonemes representing the English past morpheme. Forms such as /-s/ and /id/ are thus examples of morphs. Thus, When the word can be segmented into parts, these segments are referred to as morphs.
Now, consider the following examples of morphs:
a. /id/ if a verb ends in /d/ or /t/
b. /d/ if a verb ends in any voiced phoneme other than /d/
c. /t/ if a verb ends in any voiceless phoneme other than /t/
The /id/, /d/, and /t/ are the three morphs in English. All of them stand for a single morpheme, viz. the past morpheme with regular English verbs. If various morphs represent a single morpheme they are grouped together and are referred to as the allomorphs of that morpheme. Thus, the three morphs - /id/, /d/, and /t/, which belong to the same past morpheme in English, are the allomorphs of that morpheme.
Like the past morpheme in English, there are some other morphemes which are not constant, as they change their forms in different contexts. For example, an article is used in English to refer to an indefinite noun. Such an indefinite article is a morpheme which has two forms: a /ǝ/ and an /ǝn/, e.g.
a boy an apple
a girl an umbrella
a shop an orange
We select the form ‘a’ before words beginning with a consonant and the form ‘an’ before words beginning with a vowel. In the above example, ‘a’ occurs before the consonants /b/ (in boy /bɔɪ /), /g/ (in girl /gɜ:l/) and /s/ (in shop /sǝp/). ‘An’, on the contrary, is used before the vowels /æ/ (in apple /æpl/), /A/ (in umbrella /ʌmbrelǝ/) and /ɒ/ (in orange /ɒrindz/). Thus, and an are the two variants’ forms of the same morpheme, namely, the indefinite article. These forms occur in different phonological contexts: ‘a’ before consonants and ‘an’ before vowels. Such variants of a single morpheme occurring in different phonological contexts are called allomorphs.
There are, on the other hand, some morphemes which remain invariant in their shapes. For example, the morph -ly (pronounced as /lɪ/) in nicely is a morpheme. It does not change but retains the same form in other words also, e.g., sharply (/saplı/), slowly (/slǝʊli/), etc. Thus, the morpheme -ly has an invariant form. Such other invariant morphemes are -hood /-hʊd/ (as in childhood), -ness /-nǝs/ (as in kindness), dis-/dɪs-/ (as in dislike), etc.
(Note: each morph represents (or is the exponent of a particular morpheme but each morpheme doesn't have a morph.)
Types of morphs and morpheme
Types of morphs
As noted above, morphemes are realized as morphs. Thus, morphemes and morphs hold certain relations. There may be a one-to- one relation between a morpheme and its morph. For example, the morpheme {-nəs} (spelt as -ness as in goodness, whiteness, etc.) is always realized by the morph /-nəs/ and /-nǝs/ always represents the morpheme {-nǝs}.
Besides this simple relation, there may be a many-to-one relation between morphemes and morphs. For example, the single phonological form /raɪt/ stand for four separate morphemes, e.g., right, write, wright, and rite. Such forms, which are pronounced alike but differ in their meaning or grammatical function, are called homophones.
Conversely, there may be one-to-many relation between morphemes and morphs. For example, the single plural morpheme in English is realized by three morphs-/-s/ (as in books), /-z/ (as in girls), and /ɪz/ (as in buses). Similarly, the negative morpheme in English, namely, /in-/ is realized in three ways: /in-/ (inactive), /Im-/ (impossible) and /ɪ-/ (illogical) in different contexts.
Next, a single morph may stand for several morphemes. For example, the suffix -s in an English verb such as walks (as in He walks) realizes three morphemes at the same time, viz. simple present, third person, and singular. Such a morph which represents a bundle of morphemes simultaneously is called a portmanteau morph.
It has been assumed earlier that morphemes and morphs may hold one-to-one relation. Such assumption runs into difficulty if they do not match with one another. There are two situations in which this difficulty arises. First, the number of morphemes may be greater than that of their morphs. For example, the past tense morpheme in English is -ed for regular verbs such as walked, played, etc. But it is a zero morpheme for verbs like cut, put, etc. Thus, a single morph represents more than one morpheme.
Conversely, the number of morphs may be greater than that of morphemes. It happens when a surplus morph occurs without representing any morpheme. For example, an adjective in English is derived by suffixing -al to a noun, as person +-al- personal. But factual is made up of fact + u + al, where the morph -u- stands for no morpheme. Such a morph is called empty morph.
Types of morphemes
Free morphemes
Look at the following morphs:
a. play
b. happy
c. boy
d. nice
The morphs in (a-d) do not have to attach to some other morphs to make words. They consist of single morphemes and are words by themselves. Thus, one and the same form can be both a morpheme and a word. For example, play is a morpheme as well as a word; so are happy, boy, and nice. These 'monomorphemic' forms independent words as they can be used on their own in sentences:
(1) a. We play football in the afternoon.
b. Mina looks happy.
c. This boy is bright.
d. The house is nice.
In the above sentences (a-d) all the words consisting of single morphemes are used without the help of any other elements attached to them. Such a morpheme making an independent word and occurring in its own right is said to be a free morpheme. That is to say, a free morpheme can stand as a word.
All free morphemes are the roots that stand on their own. These roots are lexical morphemes such as nouns, verb, adjectives, postpositions (i.e., prepositions as in English or postpositions as in Nepali), and adverbs. They have 'semantic content' and express most of the meaning of a sentence. Nouns, for example, refer to entities (e.g., boy, Ram, book, etc.: verbs indicate activities (e.g., swim, walk, etc.), events (e.g., fall, happen, etc.), states (e.g., know, believe, etc.), etc.; adjectives assign properties (e.g., good, honest, etc.); and adverbs indicate manner (e.g., nicely, rudely, etc.).
Apart from lexical morphemes, there are other roots which can be free morphemes. These roots are function words. The function words, unlike the lexical morphemes, do not often have semantic content. Instead, they mainly express grammatical information. They include articles (e.g., a, the), pronouns (he, it, they, etc.), conjunctions (e.g., and, but, or etc.), and so on. The number of function words is limited in a language.
We have so far observed that only roots - lexical or functional - can be free morphemes. It is, however, to be noted that not all roots are free morphemes. There are several roots which cannot be free morphemes since they cannot occur on their own. They have to be attached to other morphs to make free morphemes or words. For example, -mit is a root which is attached to another morpheme per- to form the word permit.
Bound morphemes
Now look at the following words:
(2) a. played
b. unhappy
c. boys
d. nicely
These words can be analyzed into smallest meaningful.
elements (i.e., morphemes) as follows:
(3) a. play+ed = played
b. un+happy = unhappy
c. boy+s = boys
d. nice+ly = nicely
We have noticed earlier that play, happy, boy and nice are free morphemes as they can stand on their own as words. See the sentences in (1a-d) above. By contrast, the other morphemes attached to these free morphemes, namely, -ed in (1a), un- in (2b), -s in (3c) and -ly in (3d), behave differently. That is to say, they, unlike free morphemes, cannot stand as words on their own. They only appear as a part of words. Thus, we can have sentences (1a-d), but we cannot have the following sentences:
(4) a. *We-ed football in the afternoon.
b. *Mina looks un-.
c. *This -s is bright.
d. *The house is -ly.
These morphemes are attached to free morphemes or words. Thus, we can have the following sentences:
(5) a. We played football in the afternoon.
b. Mina looks unhappy.
c. The boys are bright.
d. This book is written nicely.
Such morphemes which cannot occur alone as words but are always attached to other forms to form words are said to be bound morphemes.
Bound morphemes can be affixes, roots, or clitics. All the bound morphemes, viz. -ed, -s and -ly, which are attached to forms such as play, boy and nice, are affixes. Apart from the affixes, there are a number of roots which cannot stand on their own are thus bound morphemes. They are always attached to some other elements in word building. Consider the following examples:
(5)
a. -ceive e.g., conceive, deceive, perceive, receive.
b. -mit e.g., admit, commit, permit, remit, submit.
c. pred- e.g., predator, predatory, predation
d. sed- e.g., sedan, sedent, sediment, sedentary
In (5) -ceive, -mit, pred-, and sed- are the bound roots and are attached with other forms.
The third type of bound morphemes can be clitics. Clitics look like words, but they cannot stand on their own in an utterance. Instead, they are attached to a phrase or clause and are thus bound morphemes. In English, the contracted forms of be such as I'm, he's, they’re are the examples of clitics or bound words.
As mentioned above, roots have semantic content. But it is sometimes very difficult to tease out their meaning. The roots listed in (5) have been borrowed into English from Latin. It is, therefore, necessary to have the knowledge of Latin to know that -ceive means 'take' and -mit means 'send, do'. In contemporary English these meanings have undergone change; hence, they cannot be clearly determined.
If so, the requirement that the morpheme must have a clear, constant meaning (or grammatical function) seems to be strong. The bound roots like -ceive, -mit, etc. do not convey any clear, constant meaning. It is the word (e.g., receive, permit, etc.) which carries a full meaning of its own. Thus, the semantic criterion of a morpheme is shaky.
In case the semantic criterion is not decisive in identifying a morpheme, its distribution can come to rescue. As Harris (1951) suggests, it is not the meaning but the distribution (i.e., position of occurrence) that determines the identity of a morpheme. For example, consider -mit. As mentioned by Aronoff (1976:12-13), the morpheme -mit is always realized as the allomorph [miʃ] or [mis] if it precedes the suffixes -ion, -ory, -or, -ive, and -able.
(6) Latinate root [mit] [miʃ] before -ion [mis] before -ive, -ory
admit admission admissive
permit permission permissive
remit remission remissory
In other phonetic environments the morpheme -mit remains unchanged phonetically, i.e., /t/ does not change to /s/, as in dormitory, vomitory, etc. The reason is that unlike -mit in (6), -mit in dormatory and vomitory is not Latinate, i.e., it does not have Latin origin.
To determine whether a morpheme is bound or free, linguists criterion involves grammatical dependence, i.e., a morpheme is bound use syntactic, semantic or phonological criteria. The syntactic if it must always occur with a particular grammatical category such as noun, verb, adjective, etc. Thus, the plural morpheme in English always occurs with a noun, e.g., boys, books, buses, etc. and is bound Similarly, the past morpheme, which is always attached to a verb, e.g., walked, is bound.
Grammatical dependence is not sufficient to determine whether a morpheme is bound or free. For example, the articles in English -a, an, and the must occur before a head noun, e.g., a book, the boy on the tree. But they are not considered bound morphemes for certain reasons. First, other words can intervene between the article and the noun, e.g., a colourful book. Secondly, the article can receive its own stress as in This is the book for me. Thirdly, the article can be separated by a pause in hyperarticulated speech, e.g. He asked_me_to_read_the book. Here, the pause between the and book is natural. These phonological reasons tell us that it is phonology, not just syntax, that helps us to determine whether a morpheme is bound or free.
Semantic criterion has also been used to determine whether a morpheme is bound or free. For example, we may say that if a morpheme refers to an entity it is a free morpheme (e.g., man, book, river, etc.) but it is a bound morpheme if it refers to time (e.g. -ed in walked). But free and bound morphemes cannot always be distinguished on the basis of semantic criterion. For example, consider the semantic notion 'definiteness'. It is marked by a free morpheme in English such as the but by a bound morpheme in other languages such as Danish, e.g., dag 'day' -dagen 'the day'.
Languages vary in their predominance of free or bound morphemes. There are languages such as Mandarin Chinese which have just free and very few or no bound morphemes. Languages like English have a limited set of bound morphemes. On the other extreme, there are languages such as Greek and Sanskrit which mainly contain bound roots.
Types of allomorphic alternations
As discussed earlier, a morpheme can have one or more than one pronunciation. For example, the morpheme play has just a single pronunciation [pleɪ] while the past tense morpheme (written as ‘-ed’) are pronounced in three ways: /id/, /d/, and /t/. All the three morphs stand for a single morpheme, viz. the past morpheme with regular English verbs. Such morphs representing a single morpheme are grouped together and referred to as the allomorphs of that morpheme. Thus, the three morphs - /1d/, /d/, and /t/, which belong to the same past morpheme in English, are the allomorphs of that morpheme. When a morpheme such as English past tense is realized by various allomorphs this phenomenon is said to be its allomorphy.
Allomorphy was discovered by ancient Indian grammarians and called sandhi. The term sandhi is a Sanskrit word meaning 'joining, putting together'. It was more developed and systematized by Panini, the ancient Sanskrit grammarian and the first scientific linguist. Sandhi has been the most widely adopted and used of all the Sanskrit terms borrowed by modern western linguists. Since its first instance of adoption in western linguistics in 1891 by Georg von der Gabelentz it has been used in Structural Grammar by Bloomfield (1933) and Hockett (1970) and other linguists such as Allen (1962), Kiparsky (1979), Anderson (1986), and others.
In structural grammar (Hockett, 1970:277) the allomorphic alternations of words and morphemes have been handled in terms of sandhi. These alternations have been classified into two broad categories of sandhi: internal and external sandhi. When two forms are combined with each other to make words there may take place allomorphic alternations within these words. Such alternations are called internal sandhi. The internal sandhi mainly includes loss of phonemes (loopasandhi), addition of phonemes (aagamasandhi), assimilation (aadeeshesandhi), dissimilation, and stress shift. Besides, there are allomorphomic alternations in language which operate across word boundaries. Such alternations are called external sandhi. For example, the intrusive (linking) r chiefly in connected speech in RP accent of English as a form of liaison which has no basis in the pronunciation of the words on their own, e.g., law(r) and order, draw(r)ing, a/an in a man and an ostrich.
The allomorphy of a morpheme thus exhibits its variant phonological shapes or allomorphs in different contexts. These shapes, called allomorphic alternations or variations, are not random or arbitrary. They are conditioned or governed by a number of linguistic factors associated with them. Accordingly, there are four
types of allomorphomic alternations or allomorphy in language. They are listed below.
(i) Phonologically conditioned allomorphy
(ii) Grammatically conditioned allomorphy (iii) Lexically conditioned allomorphy
(iv) Suppletion
Let us try to understand each of them in some detail.
(i) Phonologically conditioned allomorph
Allomorphs of different morphemes combine to form words, as in ‘in- + active’ = inactive. When such allomorphs combine the phonological properties of an allomorph may become similar to those of sounds used in another allomorph. For example, consider the three allomorphs of the morpheme in- /In-/ in English.
a. im- /in-/+ possible /posibl/→impossible /Imposibl/
(The alveolar /n/ becomes the bilabial /m/.)
b. in- /iŋ-/+ complete /kǝmpli:t/→incomplete/iŋkǝmpli:t/
(The alveolar /n/ becomes the velar/ŋ/.)
C. in- /in-/+tolerable/tɒlǝrǝbl/→ intolerable/intɒlǝrǝbl/
(The alveolar /n/ remain the /n/.)
These examples illustrate the three allomorphs of the morpheme in-, viz. [Im], [in], and [iŋ]. They occur in three different phonological contexts. [Im] with the labial [m] occurs before a labial consonant, e.g. [p] in possible; [Iŋ] with the velar [ŋ] occurs before a velar consonant, e.g. [k] in complete whereas [in] occurs elsewhere. We thus select an allomorph in terms of the phonological properties of the following sound in another morpheme. Such a choice of allomorphs is said to be phonologically conditioned. In other words. the final consonant in the negative prefix becomes similar to the first consonant in the morpheme to which it is attached. Such similarity is referred to as assimilation.
Now, consider the plural morpheme -s as another example of phonologically conditioned allomorphs.
The plural morpheme-s
Singular plural
cat /kæt/ cats/kæt/
dog/dɒg/ dogs/dɒgz/
rose /rǝʊz/ roses /rǝʊziz/
The suffix -s has the same meaning in all these instances: 'more than one'. However, it is variously realized as /-s/, /-z/ or //-Iz/. The rules for these realizations are given below:
The plural suffix -s is realized as
1. /- Iz / if it is preceded by an alveolar or alveo-palatal sibilant (/s, z, ʃ, ʒ,
tʃ or /dʒ/), e.g., roses /rǝʊziz/;
2. /-z/ if it is preceded by a voiced nonstrident sound (i.e., vowels and voiced consonants) other than /z/, /ʒ/ and /dʒ/), e.g., rags /rægz/, fans /fænz/, etc.; an
3. /-s/ elsewhere (i.e., if it is preceded by a voiceless nonstrident consonant), e.g., map /mæps/, pats /pæts/, etc.
Like the allomorphs of the morpheme in-, the allomorphs of the plural morpheme /-Iz/, /-z/ and /-s/ are in complementary distribution. They occur in different phonological environments and are thus phonologically conditioned.
Thirdly, the three allomorphs of the past morpheme -ed in English regular verbs, viz. /id/, /d/, /t/, are also phonologically conditioned. This morpheme is realized as:
1. /id/ if it is preceded by /d/ or /t/, e.g., wanted /wantId/
2. /d/ if it is preceded by any voiced sound except /d/, e.g., called /kɔːld/
3. /t/ if it is preceded by any voiceless sound except /t/, e.g., pushed /pʊʃt/
Here also, the choice of an allomorph depends on a particular phonological context and is thus phonologically conditioned. It is, however, to be noted that some linguists do not accept the phonologically conditioned allomorphy as a 'genuine' or 'real' allomorphy.
(ii) Lexically conditioned allomorphy
We have just noticed that allomorphy is phonologically conditioned. However, this is not true of all allomorphy. Some words take unusual inflections. For example, consider the following examples:
singular plural
ox oxen
child children
Thus, the plural inflection is -en in oxen but -ren in children. Such inflections occur only with specific words or lexemes. Hence,
Apart from
conditioned
this type of allomorphy is said to be lexically conditioned. unusual inflections, lexically allomorphy also involves difference in stems or roots to which inflections are attached. Consider the following examples in this regard.
(7) singular plural
a. index [Indeks] indices [Indesɪ-z]
b. wife [waɪf] wives [waɪv-z]
Here the plural has the same form [z]; so, it has no allomorphy. But the root for singular is phonologically different from the root for plural. Compare [Indeks] vs. [IndesI] and [waIf] vs. [waIv]. This means that the root exhibits allomorphy. Such allomorphy is also called lexically conditioned in that it is specific to a lexical item.
(iii) Grammatically conditioned allomorphy
The choice of an allomorph sometimes depends on the presence of a specific grammatical element. Such an allomorphy is called grammatically conditioned. Look at the following the following examples:
(8) Present tense Past tense
a. kiss /kis/ kissed /ksɪt/
b. weep /wi:p/ wept /wept/
c. take /teɪk/ took /tʊk/
In (8a) the presence of the past tense morpheme has no effect on the choice of the allomorph of the verb since it remains /kis/ as a root and also with the presence of the past tense morpheme. In (8b) and (8c), however, the presence of the past tense needs the choice of a particular allomorph of the verb, namely, /wep/ (instead of /wi:p) and /tʊk/ (instead of /teɪk/), respectively.
(iv) Suppletion
When we add a morpheme to a stem the resulting form sometimes becomes entirely different in phonetic shape. For example, when we add past to the verb stem be it becomes was or were. Such allomorphs are phonetically unrelated and cannot be related by a rule. They are called suppletion. Other typical instances of suppletion are good-better (not gooder), bad-worse (not *badder), go went (not *goed). The number of suppletive forms is rather limited.
