Wednesday 31 March 2010

you and you

A query reaches Dear Professor Wells.
Hello! My name is xxx. Im from argentina. I wanted to know when i should use the ju: and ju. It has to do with the vowel or consonat followed?

It’s difficult to know how best to answer this kind of query. The grammatical and spelling errors (reproduced here without correction) make it clear that the person making the query has not achieved a very high standard of English, nor indeed of literacy.
What does she mean by asking when to “use” juː and ju? If she wants to know which words I would transcribe with one and which with the other, the simple answer is to look them up in the dictionary. Possibly, though, the query is a more abstract and sophisticated one: how do I determine when to write one, and when the other. The brief answer to this depends on the distinction I draw in English between strong vowels and weak vowels.
Weak vowels are the vowels that can result from the weakening of strong vowels. They are found only in unstressed syllables. The English weak vowels are i, u and ə), plus sometimes ɪ and ʊ.
So if the vowel we are concerned with is in a lexically stressed syllable, we must write it juː — for example beauty ˈbjuːti, acute əˈkjuːt, amusement əˈmjuːzmənt. The same is true if it is in a strong (though unstressed) syllable: for example avenue ˈævənjuː, attitude ˈætɪtjuːd.
You will find ju only in a weak syllable. Examples include valuation ˌvæljuˈeɪʃən and stimulation ˌstɪmjuˈleɪʃən (though in the latter you can also get as the weak vowel, thus ˌstɪmjəˈleɪʃən). If the preceding consonant is an alveolar plosive, the j element usually coalesces with it to give a palato-alveolar affricate, as in situation ˌsɪtʃuˈeɪʃən: that is, tju→tʃu. In words where weak (j)u is followed by ə, for example annual ˈænjuəl, Samuel ˈsæmjuəl, the two vowels may combine into a diphthong, and there may be further vowel reductions. Study the entries for words such as actual(ly), gradual(ly).
Or perhaps the query is still more sophisticated, and is triggered by the fact that I transcribe in stimulus ˈstɪm jʊl əs but ju in stimulate ˈstɪm ju leɪt. This is a by-product of my analysis of syllabification, and I find that in my own pronunciation and that of many others that the back weak vowel is not as tense/high in closed syllables as it can be in open syllables.
But that isn’t something I want to try to explain to someone who not only confuses ‘following’ with ‘followed’ but doesn’t even know how to formulate a yes/no question in English.
At an elementary or intermediate level you can achieve a perfectly acceptable standard of English pronunciation by totally ignoring the distinctions I draw between strong , strong ʊ and weak u, ʊ. After all, millions of native speakers (those who come from Scotland and Ulster) ignore them.

Tuesday 30 March 2010

FIFA officials

Following on from yesterday’s post about syllabic m...
Beginners tend to look at English phonetics as something static. Each word has its pronunciation. To pronounce a sentence you just string the individual words together, each with its own pronunciation.
But as you move on from the beginner’s stage you realize that this is not enough. Phonetics is dynamic. Connected speech is more than isolated words strung together. To start with, we obviously have to generate an appropriate intonation pattern for the sentence. To do this we have to make decisions not only about tone — arguably, this is one of the least important things we have to decide — but also about which of the lexically stressed syllables to accent. Within each intonation phrase we must choose a syllable to bear the nuclear accent, and possibly other syllables to bear other accents.
But segmental structure, too, is dynamic. We have to pay attention to the phonetic processes of connected speech: form-word weakening, linking, assimilation, elision, syllabic consonant formation, compression. Each of these processes is constrained by the phonetic context, as well as by considerations such as speech rate and degree of formality.
I heard a BBC newsreader recently referring to ˈfiːfrəˈfɪʃl̩z, i.e. FIFA officials. This short phrase is a good example of some of these processes in action.
The citation form or dictionary pronunciation, which mild generativists might call the underlying representation, is
ˈfiːfə əˈfɪʃəlz.
(I disregard the weakening within the lexicon that gives us schwas at the end of FIFA and the beginning and end of official. Deep generativists would no doubt derive official from office. Notice that some varieties of English do not categorically weaken the o- of official as RP does.)
In non-rhotic varieties such as RP this phrase is a prime candidate for r insertion. We have a word ending in schwa closely followed by a word beginning with a vowel sound, which is a phonetic environment in which r insertion is likely. So we get
ˈfiːfər əˈfɪʃəlz (by r insertion)
Any sequence of schwa plus a liquid is a candidate for syllabic consonant formation, whereby the schwa and the liquid coalesce into a syllabic liquid. So we get
ˈfiːfr̩ əˈfɪʃl̩z
(by syllabic consonant formation: the output is syllabic r and syllabic l)
Any instance of a syllabic consonant that is immediately followed by a weak vowel — as for example the syllabic r that we have here — is a candidate for compression, whereby it loses its syllabicity. So we get
ˈfiːfr əˈfɪʃl̩z (by compression)
…which is what I heard. QED.
FIFA ends up as a monosyllable, with a final cluster not allowed in a word in isolation.

Monday 29 March 2010

a loaded question

A loaded question is a question with a false or questionable presupposition, and it is "loaded" with that presumption. The question "Have you stopped beating your wife?" presupposes that you have beaten your wife prior to its asking, as well as that you have a wife. If you are unmarried, or have never beaten your wife, then the question is loaded.
Russ Stygall emailed me to ask
When did syllabic_M disappear in English?

Assuming that by “syllabic_M” he means syllabic m, I replied
It has not disappeared. It remains an alternative to əm in words such as blossom, organism etc., and can arise through progressive assimilation in words such as happen, ribbon.
So this is a loaded question. What made Russ think it had disappeared? Perhaps the fact that I do not list it among my list of phonetic symbols for English. That’s because it’s not a phoneme, but a segment derived (when it occurs) from an underlying sequence əm by an optional or variable rule (the syllabic consonant formation rule). I present this in LPD by writing the schwa raised, implying that it is optional. (The absence of any compression mark ‿  implies that the number of syllables is unaffected, and therefore that on the loss of the schwa the nasal becomes syllabic.)

underlying ˈblɒsəm blossom
by s.c.f. ˈblɒsm̩

underlying ˈhæpən happen
by s.c.f. ˈhæpn̩
by progressive assim. ˈhæpm̩

Note that progressive assimilation can happen only phrase-finally or before a following consonant sound. It is OK in happened but not in happening.
In some kinds of English you can change the sequence pm into ʔm.

underlying ˈhæpən happen
by s.c.f. ˈhæpn̩
by progressive assim. ˈhæpm̩
…by glottalling ˈhæʔm̩

Friday 26 March 2010

Agatha and Helena

In Britain the woman’s name Agatha is pronounced ˈæɡəθə, with stress on the first syllable. As far as I know, this is the stress pattern of this name in most kinds of English. In Montserrat, though, I found to my surprise that people stress it on the second syllable. I have never heard the pattern əˈɡæθə anywhere else.
Etymologically, the name is of Greek origin. It looks as if it ought to be the feminine form of the adjective ἀγαθός agathós meaning ‘good’, though in Attic Greek the feminine nominative singular is actually ἀγαθή agathē. So it must have passed via Latin to acquire the Latin feminine ending -a. In English it is also subject to the Latin stress rule, which means that since the penultimate vowel is short the stress goes on the first syllable — except in Montserrat.
Perhaps the best-known Greek female name is Helen(a). It too has in Greek: Ἑλένη Helénē. Here too, if we don’t thoroughly anglicize it as Helen or Ellen, we have Latinate -a and some uncertainty about the stress. In the name of the island St Helena it is (h)ɪˈliːnə, but as a personal name we usually have ˈhelənə. The latter is what the Latin stress rule produces, given that the vowel in the second syllable is short in Greek. The state capital of Montana, too, is ˈhelənə.

As a scientific unit, a “millihelen” is that degree of facial beauty which suffices to launch a single ship (see discussion here).

Thursday 25 March 2010

(n+1)st? (n+1)th?

What is the ordinal numeral corresponding to the cardinal numeral (n+1)?
The ordinal numeral corresponding to 1 is first, so we write 1st.
The ordinal corresponding to 2 is second, so we write 2nd (or perhaps 2d if American).
The ordinal corresponding to 3 is third, so we write 3rd (or AmE 3d).
The ordinal numerals corresponding to 4 and upwards are formed with the suffix th, so we write 4th, 5th, 6th, 7th etc.
But after 20th come 21st twenty-first, 22nd twenty-second and 23rd twenty-third.
When we use algebraic expressions instead of ordinary numerals, the ordinal corresponding to n is nth, pronounced enθ. That corresponding to x is xth, pronounced eksθ. One of the teachers at my secondary school had the nickname kjuːθ, i.e. qth.
But it is not clear how to form ordinals for expressions such as x2, (x+1), and (x-2).
I was quite taken aback when, in the course of reading Seeing Further (ed. Bill Bryson, London: HarperPress for the Royal Society, 2010), I came across the sentence (p. 379)
[May, Oster and Yorke] identified simple features displayed by wide classes of difference equation relating the (n+1)st to the nth state of a system as it made the transition from order to chaos.

Do we really say en plʌs fɜːst? It feels wrong to me.
Imagine there is a queue of people waiting to enter a club. Your job is to pick which ones can go in. You might decide to take person number one (the first person), person number two (the second person), and person number three (the third person). More generally, you might decide to take person number n (the nth person), person number (n+1), (the (n+1)??? person), and person number (n+2) (the (n+2)??? person).
If pushed, I think I’d go for saying en plʌs wʌnθ, en plʌs tuːθ and writing (n+1)th, (n+2)th.
Any mathematicians have views on this?

Wednesday 24 March 2010


Serbian is unusual among languages in that it can be written in either of two different alphabets, Cyrillic (like Russian) or Latin (like English). As Wikipedia puts it, “Serbian is a rare and excellent example of synchronic digraphia, a situation where all literate members of a society have two interchangeable writing systems available to them.” Compare the related languages Bulgarian, which is written only in Cyrillic, and Slovenian, which is written only in Latin letters.
On the streets of Belgrade some advertisements or names of businesses are in one alphabet, some in the other. The same shop window may display written messages in both. Many road signs show names of destinations in both, first in one and then in the other, thus for example Београд Beograd.
The letter-to-letter correspondence between the two Serbian alphabets is mostly one-to-one: to each Cyrillic letter there corresponds a Latin letter (perhaps bearing a diacritic), and vice versa. But not entirely so, because there are certain Cyrillic single letters that correspond to Latin digraphs (two-letter sequences). For example, Cyrillic њ corresponds to Latin nj and Cyrillic џ to Latin . (Phonetically, these stand for [ɲ] and [dʒ] respectively.)

Another notable characteristic of the Serbian writing system is that the spelling is very close to being perfectly phonemic (at least at the segmental level — though stress and possible vowel length are not indicated). No child has to learn how to spell individual words: you just write them the way you say them, and say them the way you write them. (This is of course notoriously not the case in English.)

My picture shows the cathedral of St Sava (Храм Светог Саве / Hram Svetog Save) in Belgrade. (Photo: Ana B)

Tuesday 23 March 2010


Quite a few people in Montserrat have the surname Boatswain. Presumably that means they had some ancestor whose job was just that: boatswain on a sailing ship. He was the officer in charge of equipment and crew.
A ship’s boatswain is usually pronounced ˈbəʊs(ə)n, as correctly though awkwardly shown in the OED. The word is often spelt correspondingly as bo’sun, bosun, or bo’s’n. These spellings are not shown in the OED under boatswain, although it has a separate entry for bosun, bo’sun.
However the Montserratian surname Boatswain is pronounced as spelt, which locally means ˈboːtsweːn.
I don’t know anyone without Montserratian connections who bears this name.

I shall be in Belgrade for the next few days, at the Second Belgrade International Meeting of English Phoneticians. Providing my hotel has decent connectivity, I will continue posting from there.

Monday 22 March 2010

Joe and Joel

While visiting the old people’s (senior citizens’) home in Montserrat, I started chatting to a young Montserratian man (pictured) who was visiting his grandfather there. I asked him his name.
dʒo, he said, with a very short vowel. If my ears had been better attuned to Montserratian pronunciation, I would have recognized this as Joe. (Montserrat phonetics includes a rule shortening long vowels in final position.) But, still half attuned to London (where a mid back monophthong can only be ɔː), I wasn’t sure.
—George, did you say?
—No, dʒoːʊ, he replied, accommodating to an English visitor (me) by making his vowel longer and diphthongal.
I registered that as Joe, as intended. But my companion, of Montserratian origin but having lived in London for fifty years and even more in London mode than me, thought he had said Joel. After all, in London Joe is dʒʌʊ (RP dʒəʊ), while the fully back and diphthongal dʒɒ(ʊ)o is indeed Joel (RP ˈdʒəʊəl).
But in Montserrat Joel would be ˈdʒoːel.
Confusing, isn’t it?

Friday 19 March 2010

simplifying double affricates

The island of Montserrat, where I have spent the last three weeks (pictures), has the usual anglophone Caribbean spectrum of varieties ranging from a creole (which can be pretty impenetrable to those not familiar with it) up to the local variety of Standard English.

The syntax of Caribbean English Creole includes a vivid method of topicalization by fronting and repeating. You can do this not only with noun phrases but also with verb phrases:
a kɪl i mi a kɪl i (is kill him I PROG kill him) ‘I’m going to KILL him!’

Less dramatically, here’s a neighbour telling us about ongoing building works further down our road:
a ˈkaŋkriːt dem a ˈkjaːs doŋ ˈdeə
which means ‘they’re casting concrete down there’ and like the first example demonstrates two distinct uses of a, first as the topicalizer particle at the beginning (‘it’s concrete that they're casting’) and then as the progressive particle with the verb (a kja:s = (are) casting). Notice the combination of Standard English technical language (casting concrete) with Caribbean grammar and phonetics.

In the local pronunciation of Standard English one striking feature is the reduction of affricate sequences. In most varieties of English an affricate is preserved as such when followed by another affricate: we say each chair with -tʃtʃ- and orange juice with -dʒdʒ-. But in Montserrat and are reduced to unexploded t and d respectively when followed by another affricate. Not only do we get -ttʃ- in each chair and -ddʒ- in orange juice, we also get -ttr- in each trip and -ddr- in a large drop.
For months I was trying to puzzle out a radio jingle played on the local radio station, Radio ZJB, to advertise the Bank of Montserrat,
But for iːt transaction
You will get swift action…

— until suddenly I realized that what they were singing was for each transaction.