Words don’t just fall out of the mouth. They have to be built – either in the speaker’s or listener’s mind as they are spoken or heard and analysed, or in the history of the language. Which forms are built in the process of speech and which are just historical is the subject of intense ongoing research. The whole process is highly language specific. Learning how it works is difficult. The learning process is prolonged, with many obvious mistakes. Of necessity this is within the learnability space:

There are two sides to the building,

On the one side there are the irregular and unpredictable processes which change meanings or turn one ‘part of speech’ – now commonly known as a ‘syntactic category’ – into another,

• From verbs into nouns , from slip to slipper, from gun to gunner, from chop to chopper, and, by a different morpheme, from eat to eatery
• From nouns into verbs, from beauty to beautify
• From vverbs into adjectives, from play to playful
• From nouns into adjectives. from origin to original
• From prepositions into others with the opposite meaning, from like to unlike;
• From adjectives into nouns, from French to Frenchness , English to Englishness, like to likeness, unlike to unlikeness;
• From adjectives into adverbs,from clever to cleverly, sad to sadly.
• From do to redo, place to replace, changing or reversing the meaning.

Rightly or wrongly, this aspect of morphology is not generally raised as a developmental issue.

And on the other side there are the processses by which what is felt to be a given word is changed to fit in with the rest of the sentence. These processes are mostly regular, but not entirely. This is variously characterised as the grammar, the morpho-syntax, and the syntax.

• Regularly, in the roots of verbs like love, like, die, car, by which past tenses, are formed, as by the -(e)d ‘morpheme’ in loved, liked and died;
• And, without a single exception, the marking of the third person singular present by the s morheme in loves, likes and dies;
• In verbs which allow this what is commonly known as the ‘past participle form’ in passives like loved and liked as in “She was loved by many” or “She was liked by many”
• In verbs which allow this continuous form , as in dying;
• In adjectives with less than three syllables, comparatives and superlatives, by (-er) and (-est) as by “She is luckier than him” or “She is the luckiest woman in the wor;d”;
  
• In nouns like car, bed, house, in the -(e)s plural morpheme which are added on to them to form cars, beds and houses.
• In the 300 or so irregular verbs, including catch and stand, with irregular forms as caught and stood, and sing and take with two irregular forms as sang and sung, and as took and taken, and in the limit case – of the verb be, with am, are, is, was, were, been, and regular being;
• In a small number of nouns, like man, woman, child and mouse, with irregular plurals as men, women, children, and mice. Some words, like scissors and trousers, have no singular form – in these cases for historical reasons. A small number of words, like sheep and deer, have no plural form. Some words, like people, mean different things in the singular and plural;
  
• In a similarly small number of short, root adjectives with a change in the stem of the comparative and superlatives forms, in good as better and best, and bad as worse, and worse;
• In the pronouns as I, me, and my, you and your, he, him and his, she and her, we, us, and our, they, them and their;

Across the world’s languages, English has an uncommonly simple morphology. Amongst the English irregular verbs, the -en form in taken and been is like a regular irregularity. The am form of be displays a commonality across related languages, strongly conserved from the Proto-Indo-European, of marking the first person with a lip gesture, also in the English pronominal forms, me, mine, and my. Apart from be, one other verb, go, uses another root for its past tense, went, historically from wend. Some languages, like the Celtic languages, have many more alternations between entirely different roots. This is known as suppletion.

My late son, Joe, once said “I think I might have misunderstoodended that”, possibly trying to express more than one degree of uncertainty with four of the available past tenses. This is known as ‘multiple marking’. It is a normal aspect of child language.

Unlike about half the languages in the world, English does not use tone or intonation for building words. But in a way that may hard for some first language learners, English uses intonation in every sentence in the language, but not in the building of words.

In the English of older speakers, questions asking for particular sorts of information, beginning with one of the words, who, what, which, where, when, why, and how, all end with a rising intonation. And questions asking for just a yes or no end with a falling intonation. But the intonational difference is currently changing.

A hypothesis

In the morphology of English, there are three phonetically distinct realisations of the singular verb, the plural noun and the possessive form or ‘genitive’ – on the right edge in pats, pads, patches and Pat’s. Brad’s and Rich’s  – as –S, -Z, and –IZ. And there three forms of the past tense – in learnt, weighed, loaded and waited – as –T, –D and –ID. This aspect of English is language-specific and would thus seem to be forcibly within the ‘learnability space‘. The learning here is simplified by the proposal in Nunes (2002), with just two language-specific implementations of a language universal – appealing to a re-developed version of Diana Archangeli’s 1984 Radical Underspecification Theory, RUT, more radical than Archangeli’s, but like Archangeli’s in being available in all languages. By this model, developed directly from RUT, lexical entries consist only in the most minimal information about their sound structure, making as much use as possible of ‘default rules’ to make them pronounceable. On this account, getting unspecified features either by default or being ‘spread’ from the preceding element, the English implementations are as follows:

A) Mark plurals and genitives in nouns or the third person in present tense verbs by the feature known as ‘Continuance’ in the Sound Pattern of English, or otherwise as the property of being a fricative;

B) Mark past tense by a stop, where the articulator is the tongue tip by default

C) Add a vowel between two adjacent instances of what were once known as the ‘spirants’ – in S, Z, SH, or the rightmost spirant element of CH and J, or between T and D.

Significantly, this aspect of English does not appear to raise many developmental issues. An account like the one above avoids the need to build the A, B, and C implementations from scratch. What needs to be learnt is just the three simple steps by A), B) and C). Archangeli’s original proposal identified three vowels which can play this role, including the vowels in him and hem. Interestingly Southern varieties of British English have the him vowel in lashes, bridges and so on, while Northwestern varieties have the hem vowel. An underspecification account of these things is thus economical and both descriptive and explanatory.