The speech sounds or ‘phonemes’, which we recognise as S and Z, occur in most languages, and are yet more problematic than any others,  not just for children, but for adults. How is it that in most languages something so nearly universal is so problematic? How is it that sounds which occur in most of the world’s languages are systematically mispronounced in a similar ways, so much so that the mispronunciations have a name in English, a lisp? How is human speech so maladapted?

In all frameworks, phonemes can be broken down into components, by the framework here, by what are known as ‘distinctive features’. One distinctive feature of S and Z is that they are what are known as ‘continuants’ or ‘fricatives’, meaning that the airstream is forced through a narrow gap, creating what is known as ‘aperiodic’ acoustic noise. ‘Periodic’ sounds have a well-defined lowermost frequency. Aperiodic sounds don’t. The whistle of a kettle is (more or less) periodic. The hiss of steam is aperiodic, what sound engineers call ‘white noise’. Most fricatives are ‘strident’ with the gap tightly narrowed and with the aperiodicity in a relatively high register. Two English fricatives, TH in think and this, are not strident. All the other fricatives, those in fin, sin, and shin, veal, zeal, and these, are strident.  The feature of stridency was introduced by Noam Chomsky and Morris Halle in their 1968 magnum opus, The Sound Pattern of English, SPE  (although they oddly don’t mention acoustic frequency in their characterisation of stridency.)

The main distinctive features of English S are as follows:

• The sound is a consonant, at, or close to, the outermost edge of its syllable;
• In the case of S, unlike Z, the vocal chords are relaxed and separate, so they are not vibrating against one another;
• In what is known as a ‘fricative’ or what SPE calls a ‘continuant’, there is no actual closure; a small space is left between the tongue and the roof of the mouth;
• The opening to the nose is closed off;
• The articulation is (for most speakers) by the tongue tip against the ‘alveolar ridge’, the fleshy area behind the upper front teeth, and positioned to within a millimetre;
• Unlike the first sound in shine, shine, Shah and shore. the tongue tip is kept flat, not grooved, and very far forward, almost touching the teeth;
• The sound is strident, distinguishing S and Z from TH in thing and then.

For most people the easiest way to achieve stridency is with the highly mobile and densely innervated tongue tip.

Although features are not, in the framework here (and in most competitive theories), matters of degree, S and Z are in fact the most strident of fricatives.

Almost all issues with S and Z, commonly known as ‘lisps’, involve the loss of stridency. At least among children learning English in the South East of Britain, there are two common ways of misconfiguring the tongue position. The commoner of the two is by positioning the tongue too far forward, actually touching the upper front teeth, or between them. With an incomplete closure, the result is a sound similar or identical to the sound in TH. By a less common misconfiguration, the tongue is brought into contact with the roof of the mouth in the midline and the air stream is allowed to pass on both sides. The effect is very similar to the sound of what us written as LL in the Welsh place name Llangollen. This sound is often described as a ‘lateral’. This sort of lisp is much more difficult to treat than one with tongue tip too far forward. But it should be noted that allowing the air stream to go round the two sides of the tongue is quite different from shifting the tongue to the side of the mouth, something that is very rare across the world’s languages. (It may be by a misunderstanding of the point here that proponents of non-speech exercises encourage tongue waggling exercises.) Significantly, both of these two common misconfigurations for S and Z have the same effect, of reducing stridency. So by my proposal here, most, indeed almost all, errors with S and Z are by ‘not hitting the stridency button’. And the task of getting a normal S or Z is best addressed by looking at the target phonemes in this kind of ‘featural’ way.

There is a view by which problems with S and S are by mispositioning, the tongue, putting it either too far forward or (less often) too far back. But there are four problems with this essentially articulatory account:

1. They don’t explain why both of the commonest misconfigurations have the same effect, namely the loss of stridency.
2. There are many ways in which children might misconfigure S and Z to avoid the supposedly difficult tongue tip configuration, and still get a variation of the acoustic noise spectrum sufficient to contrast with F, TH and SH. But they don’t tend to do any of these things.
3. From information from the Welsh speech and language therapist, Olwyn Rhys, children learning Welsh and English sometimes have difficulty with S and Z. But hearing the lateral sound as they do in Welsh, they do not use this particular misconfiguration for S and Z.
4. As ventriloquists know, a phonetically normal S can be achieved in various ways, other than using the tongue tip to form a narrow gap for the airstream to be forced through. Some people, of whom I am one, say S and Z by making the channel for the air stream, not with the tongue tip, but with the blade of the tongue, positioning the tongue tip behind the lower front teeth. I can produce something like a standard S or Z with my tongue tip just behind the upper front teeth, but only with great effort and concentration.

There is a plausible explanation of these problems by a version of Underspecification Theory: Phonemes are BUILT one feature at a time. Underlyingly, S may be represented by just one of the features listed above, namely continuance or the property of being a fricative. All of the other features may be filled in by what Archangeli (1984) calls ‘default rules’, allowing the lexicon to store words very economically. By the proposals of Archangeli, Nunes (2019) and here, these rules are necessarily ordered, with the strident feature, keeping the first sounds in sing and thing apart, ordered last. But it may be defined INCOMPLETELY, either too late or not at all. Or the positioning gesture may be mistimed with the effect of turning the sound into a Welsh-style lateral. This explains the variety of apparent realisations by a single model. and the non-occurrence of lateral lisps by Welsh speaking children, if Welsh LL is already ‘booked’. If this proposal is correct, or at least on the right lines, a lisp is primarily, not by a mispositioning of the tongue, but by a failure in the implementation sequence. The mispositioning of the tongue may be a secondary effect.

This linguistic approach may be very relevant for those lisping children, particularly those with lateral lisps, who seem not respond to treatment on the basis that they are mispositioning their tongues.

But

In treating this sort of issue, as the child reaches the point when he or she can mostly say the target sound correctly, I often find it useful to show the child a 10 x 10 grid and tick the boxes as the target sound is said correctly. The child can see his or her achievement. There is an articulatory and habitual aspect to lisping. Derivation is not the B all and end all.