Dear Petr --

You unfortunately know all too well that keeping silent can be a life-preserving skill. Silence can however also bring trouble - what is it with those intellectuals pursuing abstract impractical ideas about "silent words", "silent morphemes" or "silent features"; shouldn't they pick some potatoes instead? What's a man to do - how much silence is too much silence?

Skepticism about "silent words" in fact permeates even adult educated linguists. I have perhaps even inadvertently contributed to this trend: if a word (or morpheme) can be the realisation of multiple syntactic heads, we don't need silent words (or morphemes) for all these heads - so maybe nanosyntax allows us to get rid of silence?

But we should know better. What is a word (or morpheme)? Before "intro to linguistics", it is a bunch of sounds. After, it is an (abstract!) association between sound and meaning. A bit later still, it becomes an association between various types of information: melodic information, prosodic information, syntactic information, conceptual information and probably more.

Are all these various types of information always there? Of course not! Every linguist understands the concept of 'function word'. Every linguist understands them to be 'grammatical markers' in opposition to 'content words' who dabble with concepts rather than pure grammar. Function words are therefore words which "lack" some type of information. If there was a need for it, we could summarise the situation this way (omitting prosody et al for brevity):

(1)	a. content word: melody <-> syntax <-> concept b. function word: melody <-> syntax

Some would probably argue that the grunters, sighers and wailers also have:

(2)

melody <-> ∅ <-> concept

In other words, even the most down-to-earth potato picker accepts that words (or morphemes) are abstract associations which sometimes lack some piece. Since we all abhor stipulations, who would want to stipulate that one particular piece is not allowed to be missing? And hence, we should all agree that

(3)

silent word: ∅ <-> syntax <-> concept

are as innocent as function words.

But how much is too much? I worry about that. There is a natural upper boundary: language needs to be learnable. Silent entities are going to be severely limited by how much they can be deduced from their environment.

The silent external argument of your dear Italian and Czech sentences have great chances of surviving - I have not heard convincing competitors for pro, and no amount of nanosyntax will help here (similarly for PRO, for that matter). But what of more subtle cases? Your and my once-upon-a-time-student and now colleague of many years, has famously argued that each noun comes with a whole band of little functional friends, so that their syntax looks something like:

(4)

noun k1 k2 k3 k4 k5 k6 ... k70

This band of friends has its own melody in Czech or Sanskrit, which we call case morphemes. In other languages, they are vocalized by prepositions. But the first two in that band have a funny restriction: they happily come wrapped in case morphemes (nominative and accusative), but refuse to appear as prepositions (in some rare cases they come under the name of differential subject/object markers, but even those may not be true cases of what we're looking for).

So what happens to k1 k2 in languages without case morphemes, such as in the language I am writing in right now, or my own native language? And similarly, what happens to verbal inflection features in languages with barely any verbal inflection pronounced? Well, you see, this has been a bit of a taboo question in my circles, which people find cautious to stay silent about.

It is a silence I intend to break, but maybe this is not the place to break taboos, and we have much more important things to not keep silent about right now: happy 70th birthday, dear Petr!

Michal