From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Linguistics / Phonetics  (Rated Start-class)
WikiProject icon This article is within the scope of WikiProject Linguistics, a collaborative effort to improve the coverage of Linguistics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 ???  This article has not yet received a rating on the project's importance scale.
Taskforce icon
This article is supported by the Phonetics Task Force.
Note icon
This article has been marked as needing immediate attention.


The claim that "kn" does not exist in English is a little unclear: what does it actually mean? Certainly the sequence of consonants kn appears ("knight", "knife") although not pronounced in a reasonable way. But the sounds appear: strychnine (pronounced strik-nine). So perhaps this is a more subtle issue than it appears?

I just corrected the "kn" issue to make it correct. The article speaks of phonemes, not letters. Matt gies 01:59, 6 Mar 2004 (UTC)
Would the words "canoe" or "canuck" not count as a "kn"? I know there is an a there but I personally don't even put a schwa between them (maybe that's just how it is in my parts, if you haven't already figured out where I'm from). - Chris, May 1st 2006

Another good example would be that in English, /vl/ is forbidden word-initial, but /fl/ is allowed. Sanxiyn 05:53, 18 Jan 2005 (UTC)

"slips" and "pusl" are is not very good examples since you get "wrestle", "nestle" etc. /sl/ at the coda seems to be very English. -- 06:41, 28 Jan 2005 (UTC)


The sequence of consonants may occur, but who ever said that english pronunciation follows any rules? Well ok, maybe somebody, but anyway, the issue at heart here is phonetics. Phonetically, english knight and knife do knot begin with /k/ the voiceless velar stop. So, the claim that the english language phonotactically avoids the /kn/ consonant cluster (phonetically) is still true.

By the way, in the word "wrestle", what "le" signifies is a syllabic "l", that is, /l=/ in X-SAMPA. This is not the same sort of "l" that was being spoken of in the case of "pusl", as it's acting effectively as a vowel, rather than as a consonant. -- 13:49, 12 Apr 2005 (UTC)

(I edited this little discussion and put it in a subheading rather than just being on the page here) Chris: I'm from Minnesota, and I've definitely heard canoe pronounced [knu]. Still, this is very localized and not at all generalizable to American English, so the statement that [kn] does not exist in "English" is still, in general, true.

What is not mentioned in the article and is extremely relevant to the discussion here is the issue of the sonority scale. This is a list of manners of articulation arranged in terms of sonority:

oral stop > fricative > nasal > approximant

And depending on the language it may further break up the categories (i.e., "liquid > glide" in lieu of "approximant") of or have slight rearrangements depending on the sonority with which a particular class of sounds is produced. This scale is useful because rather than argue about what clusters are permissible, you can say that within an onset you may have a minimum separation of 3 in English: You can't have a stop+nasal (separation of 2) onset ([kn-]) and this also means you can't have an fricative+nasal (separation of 1) onset ([fn-]), but with a separation of 3 in English, you can have stop+approximant onsets ("play," [.pleɪ.] "dry," [.dɹaɪ.], "address," [.ə.dɹɛs.].

Additionally: phonotactics deals with the syllable, and thus using examples like strychnine, hackney, or acknowledge are irrelevant: the k and n in each do not occur within the same syllable, as predicted by the sonority sequencing in the previous paragraph. These words would be syllabified as [stɹɪk.naɪn], [hæ], and [æk.nɔ.lɪdʒ], respectively. This also explains why the claim that /sl/ occurs in the coda is unfounded: [sl] is not a coda in the word wrestle because this word is not comprised of only one syllable. It would be syllabified [ɹɛ.sl̩], with /s/ rather forming an onset to the syllable whose nucleus is a syllabic [l].

And additionally, if you point out that the [s] at the beginning of "strychnine" would seem to violate the sonority sequencing principle, you would be right. Languages may have an "index," which refers to a sound ([s]) or type of sound ("liquid") which may violate the sonority sequencing and only occurs on a word boundary.

Okay, I think that's all I've got. Someday this should be incorporated into the article, but I don't have time just yet. --Coyne025 05:41, 13 November 2006 (UTC)

"pack a marbles"[edit]

convenience store much? people pronounce marlboro "marble" quite a bit. possibly 40% of the time. w/ "marlboro light" - you'r notably more likely to hear "marble light". someone incorporate that. KzzRzzKnocker 06:11, 16 April 2006 (UTC),,,

If I heard someone make such a mistake I would assume that they are not native English speakers. Or that they're retarded. Nino137.111.47.208 (talk) 01:53, 26 June 2008 (UTC)

spw, stw, and skl[edit]

Hi, I can easily think of examples such as:

  • spring
  • spew
  • string
  • stew
  • scream
  • skewer
  • squeak
  • splash

but I can't think of any words that begin with /spw, stw,/ or /skl/. If nobody can think of any examples, I think the article should be edited to remove spw, stw, and skl. --Kjoonlee 13:47, 29 January 2007 (UTC) 06:55, 3 March 2007 (UTC) : Hey, wait! I know a guy who's name starts "SKL" plus some vowels, and nobody has any trouble pronouncing it. It might not be of English origin, but there is nothing about the rules of English phonology which prevents that onset from occurring. The only evidence necesary is that English speakers have no trouble pronouncing it. My medical advisor says that "Scleroderma" is a valid English word for a medical condition. Compare, for example, the native speaker of Japanese. Regardless of effort, he can't force his mouth to produce these sounds. (Well, perhaps with years of practice, he might.) And the Czech language is loaded with consonant clusters that an English speaker just plain can't wrap his lips around. 06:55, 3 March 2007 (UTC)

This leaves:
/s/ + /p t k/ + /ɹ j/
/s/ + /k/ + /w/
/s/ + /p/ + /l/
... which brings us to another question. Is there a natural class that only includes /ɹ/ and /j/? --Kjoonlee 14:02, 29 January 2007 (UTC)
Non-lateral non-coarticulated approximants? Possibly it makes more sense to have the scheme as /s/ + /p t k/ + /ɹ j w l/, and note that /w/ is forbidden after /sp/ and /st/, and /l/ after /st/ and /sk/. –EdC 16:34, 29 January 2007 (UTC)
Very good idea. I'd put it in myself, but I'm afraid WP:NOR is stopping me. --Kjoonlee 10:26, 30 January 2007 (UTC)
It's in Crystal, p. 243 in the second edition. (He references the English Pronouncing Dictionary and Gimson 1970.) Also, for /skl/ there's sclerotic, sclerosis, and for /smj/ there's smew. I'll add those in. –EdC 12:29, 30 January 2007 (UTC)
Thank you very much for your attention to detail and thoughtful contribution. :) --Kjoonlee 14:12, 30 January 2007 (UTC)

Subject to language?[edit]

Phonotactic rules are often waved off as something that changes from language to language, but shouldn't a moreorless strict adherence to the IPA soon show what consonant clusters and vowel combinations are physically possible to vocalize without inserting unintended vowels or comitting elision? Wouldn't it be something as simple as seeing if the mouth can transition from one shape to another without passing through and vocalizing and unintended configuration? For example, try to pronounce "hn" or "lk' without additional sounds (vowels? semivowels? approximants?) before or between the consonants. Can you say "sbort" without saying "sport"? If this is not something covered by phonotactics, what is it called? 10:03, 8 May 2007 (UTC)

That's simply phonology: in English, "voiced" stops are distinguished from "voiceless" stops in most environments by aspiration and/or length of preceding vowels rather than by actual voicing. If you take a recording of "spin" spoken by a native-English-speaker and start it just barely after the "s", it will sound exactly like "bin" to a native-English-speaker. For that reason "sport" and "sbort" are indistinguishable for English speakers. There are no doubt some universals in phonotactics, but the realization of those universals is intertwined with language-specific phonological processes and constraints to the point that it's just easier to describe phonotactics as mostly language-specific. Chuck Entz (talk) 22:04, 19 October 2012 (UTC)

Phonotactics does deal with this type of thing, and when formalizing rules you can see that languages may exhibit similar limitations, though the limitations as a whole set might be difficult to find in multiple languages. If in the examples /hn/ and /lk/ you're saying that one can't pronounce them without an intermittent vowel, I disagree. Since sounds are just a specific positioning of articulators, most sequences of sounds are possible, though they might be really cumbersome and therefore not very likely permitted by a languages phonotactic constraints. Talking about what sounds "can" be pronounced together is not so fruitful, since the answer is essentially "any," but talking about what different languages do permit and the reasons for this (ease of articulation, historical changes, etc.) are where it gets really fun. As for /sbort/, this voicing distinction is neutralized after voicelessness by English phonological rules. Both would normally surface as [spɔɹ̯t], but if you try to violate these rules to voice the /b/ you can say [sbɔɹ̯t]. Russian would handle such a cluster in an opposite way: while /st/ could surface as a voiceless cluster ([stalʲin]), Russian has a rule of anticipatory voicing, so the cluster /sd/ would surface as /zd/ ([zdʲelatʲ], to do), which may initially be difficult for an English speaker but can be pronounced with practice. Or perhaps I'm not quite getting at what you're talking about... Coyne025 19:45, 29 May 2007 (UTC)

for expansion[edit]

some suggested things to mention:

  • phonotactics are related to wordlikeness (how typical of a word or typical as a word?) and word acceptibility (a possible/probable word?) judgments
  • these judgments are often gradient — for example cat is a more typical English word than sphere and blick would be a better English word than bwick and both of these would be better than bzick
  • these judgments have been studied in numerous psycholinguistic experiments (which should be cited)
  • these judgements correlate with both phonotactic probability (measured in different ways: co-occurrence probabilities, positional probabilities, transitional probabilities) and lexical neighborhood density (this deserves a separate article discussing what this is) and although phonotactic probability & neighborhood density is highly correlated with each other a few studies show that both independently contribute to wordlikeness and/or acceptability
  • this statistical information may be influenced by the frequency of word usage
  • this statistical information could arguably be considered not "grammatical" knowledge
  • discuss psycholinguistic processing models that have been proposed to account for this knowledge
  • phonotactic probability is connected to a number of other things besides word-likeness/acceptability, such as word recognition, phoneme identification, nonword repetition, how fast children learn new words, infant word segmentation, naming, etc. Discuss these experiments and the relation (if any) between these and the word-likeness/acceptibility judgments
  • introduce the term morpheme structure constraints
  • formal grammatical models that account for phonotactics

ishwar  (speak) 02:52, 16 May 2010 (UTC)

Really bad example[edit]

The example given in spanish "pllegue" does neither exist nor does it comply to spanish phonotactis. Who invented this word? The IPA transcription is also utterly wrong as <ll> in spanish is never pronounced [lj]. — Preceding unsigned comment added by (talk) 06:43, 19 December 2011 (UTC)

The text says pliegue, not *pllegue. Perhaps those look the same on your screen, so once again in monospaced all-caps: the text says PLIEGUE, not PLLEGUE. Angr (talk) 08:00, 19 December 2011 (UTC)

No affricates in complex onsets?[edit]

What about the word tree? I have a Cleveland, Ohio, accent and I (we) pronounce it more or less as /t͡ʃɹi/. The fifth rule for English phonotactics says that there should be "No affricates in complex onsets" but isn't /t͡ʃɹ/ just that? I think the same can be said of the words trill, trampoline, trail at least in vulgar pronunciation. Biocrite (talk) 05:24, 12 July 2014 (UTC)

I think the claim is being made at the phonemic level, where these words have /tr/ even though they may surface as [tʃr]. —Aɴɢʀ (talk) 06:03, 12 July 2014 (UTC)

No onset /ŋ/ or /ʒ/?[edit]

What about pleasure and azure? Those are the classic examples of /ʒ/ and both have syllables that begin with /ʒ/. (talk) 23:10, 31 July 2014 (UTC)

Maybe. Many phonologists believe that the /ʒ/ in those words is in the coda of the first syllable rather than the onset of the second. —Aɴɢʀ (talk) 19:01, 1 August 2014 (UTC)

No Japanese /st/ is bad example[edit]

Some sources claim that the "u" in, e.g., ストライキ (sutoraiki/"strike", more usually shortened to "suto") is pronounced, but devoiced (whispered), but to my ear it's more commonly dropped altogether in favor of a lengthened "s" sound, meaning that Japanese could be argued to have an /st/ sound after all. It seems a poor example at minimum: it's also imprecise to claim that consonant clusters "do not occur" in Japanese, as obviously clusters beginning with Japanese syllabic "n" are a counter-example. Perhaps a better consonant cluster could be chosen, such as "gd", or even "kt", since in that case you really do consistently get the unvoiced "u" between the consonants in a Japanese approximation. — Preceding unsigned comment added by Micahcowan (talkcontribs) 19:55, 3 March 2015 (UTC)

@Micahcowan: I think the reason for using st as an example is that it occurs in English but (allegedly) not in Japanese. Gd or kt would be worse examples, because these clusters don't occur initially in English. If you can think of an initial consonant cluster in English that is clearly not allowed in Japanese, such a cluster would be an acceptable substitute for st. I think Japanese phonology seems to indicate that sut- is pronounced with a voiceless vowel, not a lengthened sst-, but if there's a source that says the latter, then it should certainly be mentioned. — Eru·tuon 20:42, 3 March 2015 (UTC)