Morphology, Word by Word

Morphology is the study of the internal structure of words and the rules that build them. Its basic unit is the morpheme, the smallest piece of form that carries meaning or grammatical function. Unbreakable has three: un-, break, -able. Speakers create and understand new words all the time (unfriend, doomscroll), so morphology is really about what speakers know, not just a list of words.

There is no single test. A phonological word has one main stress. A grammatical word behaves as one unit in syntax. An orthographic word sits between spaces. A semantic word names one concept. These often disagree: ice cream has a space but one concept, don't is one written word but two grammatical words.

A lexeme is the abstract dictionary item (written in capitals, SING). A word form is one of its realisations: sing, sings, sang, sung, singing. Tokens are every running word; types are the distinct forms.

Try the borderline cases. Tap each to see why it is hard.

Morphology lets a language make new words from old ones without memorising each, mark grammar such as tense, number and case on the word itself, and save effort by reusing small pieces. It also lets words carry more meaning, so sentences can stay short. Languages differ hugely in how much of this work they hand to morphology (see section 7).

Free morphemes stand alone (book). Bound morphemes need a host (-s, un-). A root carries the core meaning; a base is whatever an affix attaches to; a stem is the base for inflection. Some bound roots exist too (-ceive in receive, perceive).

1 Collect related forms. 2 Compare them and find recurring pieces with recurring meaning. 3 Separate roots from affixes. 4 Check for allomorphs (in- becomes im- before p: impossible, il- before l: illegal). 5 Test your segmentation on new data. Try it here:

A bound morpheme attached to a base: prefix (before), suffix (after), infix (inside, marginal in English: fan-bloody-tastic), circumfix (both sides, as in German ge-lieb-t).

Some suffixes are stress-neutral (-ness, -ful, -ment) and some shift stress and trigger sound changes (-ity, -ic, -ion). Compare:

Affixes are picky about their base: -ize likes nouns and adjectives (modernize), -able likes verbs. Affixes stack in a fixed order. Press to grow a word layer by layer:

Most affixes are polysemous. -er makes agents (teacher), instruments (opener) and objects (keeper). un- negates adjectives (unhappy) but reverses verbs (untie). Meaning is often predictable but rarely exact.

Tests: stress (BLACKboard vs black BOARD), inseparability (no insertion: *a black wet board), lexicalised meaning, and only weakly spelling. Parts can be noun, verb, adjective, preposition or even a phrase.

Canonical English compounds are right-headed: the last element sets category and core meaning. An endocentric compound is a kind of its head (a houseboat is a boat). An exocentric one has no head inside (a redhead is not a head). Tap a compound.

The link between the two nouns is not stated. Listeners infer it:

Nominal compounds usually take stress on the first element: GREENhouse (a glasshouse for plants) versus green HOUSE (a house that is green). Exceptions exist (apple PIE), so stress is a strong tendency, not a law.

Adjectival: user-friendly, bittersweet, snow-white. Verbal (mostly by backformation or conversion): babysit, to sleep-walk; synthetic compounds like truck driver and heart-breaking contain a deverbal head with an object. Neoclassical compounds join Greek or Latin combining forms that are not English words:

Words are normally lexically integral: syntax cannot reach inside them. You cannot pluralise the first part (*rats-catcher) or refer to it with a pronoun (*a car-lover who buys them). Yet phrasal compounds (a pain-in-the-neck attitude) blur the boundary, so many linguists treat compounding as a zone shared by both systems.

Inflection shows grammatical information without making a new lexeme: number, case, person, tense, aspect, mood, degree, gender. Regular patterns apply by rule, irregular ones (mouse, mice) are stored.

English has only eight inflectional suffixes. Tap for an example.

Inflection is highly productive and near-obligatory. Children apply -s to invented nouns (the classic wug test), showing the rule is a real mental process. Irregulars survive because frequent forms are memorised and block the rule (went blocks *goed).

Inherent inflection carries meaning chosen by the speaker (plural on nouns, tense on verbs, comparative). Contextual inflection is required by syntax (third person -s in she walks, case, agreement).

Diagnostics: derivation changes category or meaning and is optional, sits closer to the root, and is less regular. Inflection completes a word for syntax, is outer, and is regular. The two are a cline, not a wall.

Early scholars such as Humboldt and Schlegel tried to classify the "genius" of each language. Modern typology asks two questions together: what is possible in every language (universals) and how do languages cluster (particulars)? Your toolkit is the set of morphological resources a language uses.

Drag the slider along the index of synthesis (morphemes per word).

Two other axes: fusion (do affixes stay separable or blend several meanings?) and head or dependent marking.

Related languages often share morphological habits (Slavic case systems, Semitic root-and-pattern). Neighbouring languages also converge through contact: South Asian languages share retroflexes and verb-final order, and the Balkan languages share features across family lines.

Syntax decides which forms appear; morphology supplies them. Case, agreement, tense and mood are the clearest meeting points. Whether they belong to the lexicon or to syntax is a central theoretical debate.

Ask: can you insert material inside? (*black very board). Can one part be replaced by a pronoun or coordinated alone? Is the order fixed? Does it have one stress? Clitics such as 's and n't sit awkwardly between the two categories.

Some words mean the sum of their parts, others only partly. Tap each on the scale.

Affixes add semantic roles and templates (agent, instrument, state, event, result). Words also drift: nice once meant "foolish". Polysemy (-ing as action, result or thing: building) and lexicalisation freeze meanings into the lexicon.

Distributed Morphology spreads word building across syntax. Lexeme-based morphology (Aronoff, Beard) treats rules as operating on whole lexemes. Paradigm Function Morphology (Stump) formalises the Word-and-Paradigm idea. Construction Morphology (Booij) uses form-meaning schemas. Network Morphology and usage-based models stress inheritance and frequency. Optimality Theory handles competition between forms.