Romanization System

I know that the 1960’s course from the Defense Language Institute for 台山話 uses a really old romanization system.  And I know it gave me a lot of trouble at first.  So I made this write-up to:

  • help explain some of the differences between the romanization scheme in this textbook and the Yale and jyutping romanization schemes used in Cantonese
  • help explain some of the patterns I observed in the similarities between Cantonese and Hoisanwa

The romanization system is really old (like 1960’s old, to be exact).  I had to do the first 4 lessons of drills several times before I really got the hang of the romanization system.  So I thought writing a short guide on this might help.

I have learned Cantonese through romanization (Yale, and a little 粵拼 jyutping), but I have not learned Mandarin nor hanyu pinyin.  So my advice is coming from that background.


Tone Numbering System

For those familiar with Yale and 粵拼 jyutping, the 6 tones in Cantonese are:

Tone Name Tone Numbering System
High Level (and High Falling) 1
Medium Rising 2
Medium Level 3
Low Falling 4
Low Rising 5
Low Level 6

This is shown at

An Important Point: There is actually a 7th tone of High Falling, but these two main romanization schemse of Yale and 粵拼 jyutping do not make a distinction between the High Level and High Falling tones.  This is an important point, because in 台山話 Hoisanwa, these are treated as two distinct tones, and it’s important to hear and speak them as distinct tones.  I know this personally still gives me a lot of trouble.

Since I really like this number system for tones, I denote the High Falling tone as 1`.

So in my own system, I use the following as tone numbers for 台山話 Hoisanwa:

Tone Name My Tone System Textbook Tone Diacritics
High Level 1 no mark
High Falling 1` ^ above letter
Medium Level 3 ` above letter
Low Falling 4 – above letter and . under
Low Level 6 – above letter

Comparison to Cantonese

Of the 6 Cantonese tones, Hoisanwa has all of the same tones, except for the 2 rising tones.  Then Hoisanwa additionally makes a distinction between the High Level tone # 1 and the High Falling tone # 1`.

Tone Sandhi

In addition to the 5 tones mentioned in the previous section, Hoisanwa has some tone changes.  I am still trying to figure out when these tone changes occur.  In the textbook, they are marked by adding  an asterisk * after the syllable, in addition to the regular tone diacritics.

I’m even having trouble accurately pronouncing these tones.  They seem to be that you should perform the tone as normal, but then half way through the syllable, you start to make it a rising tone instead.

This is explained further here in the section on “changed tones”.  Although, what they call a “Medium Falling” (31) tone has sounded to me more like a “High Falling” (51) tone.

Similarities to Cantonese

There are no hard and fast formulas for converting between the tone of a word in Cantonese and it’s corresponding tone in Hoisanwa.  But if I know the tone in Cantonese, I can occasionally guess the tone in Hoisanwa.  Here are the “rules” I have noticed.  But I just want to say again, there are almost as many exceptions to the “rule” as there are examples of it.

Cantonese Tone Hoisanwa Tone Example
2 – Medium Rising 1 – High Level “to speak” : gong2 -> kong1 , “to give” : bei2 -> pi1
6 – Low Level 1` – High Falling “to be” : hai6 -> haai1` , “side” bin6 -> ping1`
1`* – High Dipping
3 – Medium Level
3* – Medium Rising
4 – Low Falling
4* – Low Dipping
4 – Low Falling 6 – Low Level “not” : m4 -> m6 , “person” : yan4 -> ngin6
6* – Low Rising

Syllable Initials

Table of Hoisanwa initials:

unaspirated aspirated nasals fricatives semi-vowels
labials p p’ m f
dentals t t’ n hl l
palatals ch ch’ s y
velar k k’ ng h
labialized-velar kw k’w w

Table of Cantonese initials:

unaspirated aspirated nasals fricatives semi-vowels
labials b p m f
dentals d t n l
palatals j ch s y
velar g k ng h
labialized-velar gw kw w

So there are 19 Cantonese initials and 20 Hoisanwa initials.  Hoisanwa has all of the Cantonese initials plus an extra lh initial, explained pretty well here:

I had trouble with this lh sound for a long time.  At first I thought it was a cross between a hard c sound (or “k” sound) and a “l” sound.  Then I went through a phase where I thought it was like the German “ch” sound and a “l” sound.  Now I’m more sure that it’s somewhere between a “th” sound and a “l” sound.  I kept asking my wife which of these it was, but she had a hard time describing it too.  So I guess my best advice to those trying to learn it is to just try out all 3 of these sounds and see which sounds to you more like the recordings.

Comparison to Cantonese – Romanization

Except for the lh initial, Hoisanwa has the same initials as Cantonese.  The only difficulty is that the Romanization system is very old.  So for “b” sounds they use the initial p, and for “p” sounds they use the initial p’.  Similarly for other unaspirated and aspirated initials in the table above.

Comparison to Cantonese – Pronunciation

As in the previous section, there are some really common patterns I have noticed, but these are by no means formulas that can be used to convert from Cantonese to Hoisanwa.

Perhaps the most noticeable are the following:

pattern example character Cantonese Hoisanwa
ch -> t’ cho3 t’o3
chaan1 t’aan3
t -> h tau4 haau6
toi4 hoi6
tai2 haai1
j -> t jeui3 tooi1`
jau2 taau1
jau6 tiu1`
d -> <null> deui ooi3
daai6 aai1`
din6 ing1`
s -> lh sin1 lhing3
sai2 lhaai1

NOTE: Remember in the above that a t initial in Hoisanwa is actually a “d” sound, and the t’ initial is actually the “t” sound.

Syllable Finals

Table of Hoisanwa finals:

aa ia e i ie o oo u
<null> aa e i o oo
i aai ei oi ooi
u aau iau iu
m aam im iem
n aan in ien on oon
ng aang iang ing ong ung
p aap ip iep
t aat it iet ot oot
k aak iak ik ok uk

Table of Cantonese finals:

aa a e eu i o u yu
<null> aa e eu i o u yu
i aai ai ei eui oi ui
u aau au iu ou
m aam am im
n aan an eun in on un yun
ng aang ang eng eung ing ong ung
p aap ap ip
t aat at eut it ot ut yut
k aak ak ek euk ik ok uk

So there are 51 Cantonese finals, but only 38 Hoisanwa finals.  The differences are:

  • no eu, eui, eun, eung, eut, and euk finals
  • no short “a” finals, only long “aa” finals
  • no yu, yun, and yut finals
  • a few new finals such as iau, iang, iak, iem, ien, iep, and iet

Similarities and (Subtle) Differences

At first, I thought that the “iak” final in Hoisanwa was pronounced identically to the “ek” final in Cantonese.  Similarly for other pairs, such as “iang” / “eng” and “iem” / “im”.  However, there are slight differences in the pronunciation, so be careful to listen carefully to the drills for these.

On the other hand, as far as I can tell the “oo” finals such as oo, ooi, oon, and oot appear to be exactly the same pronunciation as the finals u, ui, un, and ut in Cantonese.

Comparison to Cantonese

Just as above, there are no definite formulas for predicting the Hoisanwa final based on the Cantonese final.  However, here are some patterns I have noticed.

eui -> ooi

I have seen several examples of this, such as:

Character Yale Romanization Hoisanwa Romanization
heui3 hooi3
geui3 gooi3
deui3 ooi3
jeui3 tooi3

short “a” finals -> “oo” finals

Character Yale Romanization Hoisanwa Romanization
fan3 foon3
bat1 poot1
hang4 haang6
bak1 baak1


8 Responses to Romanization System

  1. Brian says:

    Just what the doctor ordered 😀
    This is really great, it combines real life experience and stated definitions.
    You should write a textbook!
    Hrm… and perhaps creating a youTube video linking to this site, should boost your reader count.

    It’s a pity that I can only understand spoken Toisanese, and not speak it. Stupid brain.

  2. Brian says:

    Oh, and since it’s a wordpress blog. Try the SexyBookMarks plugin .. so you can get this blog a bit more recognized.
    Or maybe…
    …You get the idea!

  3. Brian says:

    Oops stupid Firefox copying system/my ctrl key not working properly.

  4. Aaron says:

    I’m glad to see you exploring this course book in depth. I haven’t gotten around to doing so, even though I’ve owned the book for some time. Here are some tips that I can provide on two issues you note.

    First, you rightly observed that the voiceless lateral fricative is a complex consonant—acquisition studies have noted that this sound is one of the last consonants children master. Young learners often pronounce this sound similar to “kl” before they get the hang of it. Pronouncing the lateral fricative like “thl” is a very decent approximation. The best tip that I can give is that you want to fuse “th” and “l” into as much of the same sound as possible. Just as we think of the tch in “patch” as a single sound—even while it’s produced similar to a t followed by a sh—so you should think of the lh in “lham” 三 as a single sound. I’ll post about this sound in more depth on my blog next week.

    As for the mid falling tone. In most cases, it should be fine to call it a high falling tone. After all, there are usually only two falling tones, and one is higher than the other. If you do an acoustic study, you’ll find that this high/mid falling tone starts out lower than the high level tone, but higher than the mid level tone. In some local variants, however, there is also a high falling tone in addition to the mid and low falling tones. This high falling tone is derived from a high tone verb followed by the word 到, which carries a mid tone. So for these speakers, 睇到 can be pronounced “hai-oh” (a high tone followed by a mid tone), or “hai” with a high-to-mid falling tone—contrast with 係, which carries a mid-to-low falling tone.

  5. Ben says:

    Thanks for giving the linguist’s perspective on this, which is much more precise than my attempt.

    And thanks for the comment on the falling tones. I have been having trouble getting the “high falling” tone to sound just right, I’ll try out your advice and think of it as a mid falling tone instead.

  6. Jonathan says:

    In my opinion, you’re lucky not to have learned Mandarin/pinyin. I have no clue why. And you’re lucky to have learned Cantonese because, as you have noticed, it will help you with 台山話.
    台山話 was my first language, but I’ve forgotten most of it and am (very slowly) trying to relearn it. Thank you and anyone else on the internet for the resources and info! My family has the initial “v” (which I think they talk about) and some others that sound like they fall into the gw and kw category, most notably “mw” at the beginning of 乜 (mwot).
    Good luck with “thl”.

    Thank you for appreciating our language and taking time to learn it.

  7. Leigh says:

    Thanks for confirming some of the patterns I’ve noticed in nineteenth century English language records and inscriptions here in Australia. Also, I’ve noticed that the lh sound was often rendered as ‘sl’ eg ‘sling’ for 生.

  8. idastravels says:

    I recently started attending the free hoisan/cantonese class again at the 1st Chinese Baptist Church in NYC Chinatown. Kim does a great job. It inspired me to try working with the Dept. Of Defense materials again. They are horribly organized – makes me wonder how many people actually learned it.

    Anyway, since I learned Hoisanese at home and Cantonese from my friends, any talk of tones and linguistic nomenclature makes me go blank. However, since I did formally study standard Chinese, I can handle 4 formal tones.

    If I get to the point of posting concersational videos, I hope you guys will jump in with corrections and contributions

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: