Ling 450/550: Word List Guidelines The goal of the word list is to

Ling 450/550: Word List Guidelines
The goal of the word list is to demonstrate the phonetic inventory of the selected language. This can be
considered in two aspects: 1) phonemic contrast (including tone and stress) and 2) allophonic variation.
Your word list should be constructed to provide for both, though focusing more on the phonemic
contrasts. Remember that it is meant to be a list, if you feel tempted to write more, then it probably
means you need to rewrite your phonological description. Let’s consider these two pieces independently.
Phonemic Contrast
Phonemes are separate contrastive sounds. That is, they are crucial for lexical contrasts between words;
alternation in phonemes causes changes in meaning. This is most clearly demonstrated with minimal
pairs. In a minimal pair, all of the segments in two words are identical except for the one contrastive
element (typically a segment). Because names of people, places, institutions, recently borrowed words,
archaisms, etc. don’t always follow the language’s current phonology, proper nouns, acronyms, archaic
words, and recently borrowed words are avoided. Here are some examples of minimal pairs in English:
/l~ɹ/: ‘laughter’~‘rafter’
/t~d/ ‘eight’~‘aid’
/n~ŋ/ ‘lawn’~‘long’
/tʃ~dʒ/ ‘batch’~‘badge’
/i~ɪ/: ‘feat’~‘fit’
/eɪ~æ/: ‘fate’~‘fat’
As you can see, what matters is the sound, not the spelling. Also noteː phonemes can be combinations of
IPA symbols, most commonly affricates, diphthongs, and aspirated sounds.
In the actual list you wouldn't use a table like this, but rather have them in a numbered list with at least
three columns (four if you can manage the orthography) as in the following example illustrating nasal
place contrasts in Swahili:
number
gloss
transcription orthography
1
meaning
maːna
maana
2
eight
nane
nane
3
tomato
ɲaɲa
nyanya
4
etc.
gleam
ŋaɾa
ng'ara
It is not necessary to create minimal pairs for each possible pairing of all sounds in the language but
there should be at least one word per phoneme on the list. Instead, the sounds can be grouped and
presented in similar environments. Notice that in the Swahili example, the words aren't in a minimal set,
but the word initial nasals are found in similar environments: word initial position of two syllable words
before the vowel /a/. The most thorough way to cover all of the contrasts for a particular set of sounds is
with a minimal set where all of the words differ by only the contrast of interest. Again, it isn’t necessary
to cover all contrasts in the language in a single set. For example: ‘feat, fit, fate, fete, fat’ nicely illustrate
the front vowel contrasts of American English but the ‘f_t’ bracket does a poor job at illustrating the
back vowel contrasts. The set bracked by ‘c/k_d’ does a better job: ‘cooed, could, code, cawed, cod,
keyed, kid, cad’ illustrate all of the back and many of the front vowel contrasts, and many of the
front~back contrasts. A few more words will be needed to illustrate the missing contrasts like ‘bale,
bowl, boil, bowel’ to illustrate the set of diphthongs. If your language doesn’t permit minimal pairs
(many agglutinative languages won’t), then try to keep the words as similar as possible and avoid using
the same root more than once in the list (ex for English use ‘past’ rather than ‘passed’ if the word ‘pass’
already appears). Remember, you are trying to establish that a lexical contrast depends on the pair of
sounds you are illustrating. When illustrating allophones, you should use pairs of words (or sets if there
are multiple allophones of the same phoneme you are illustrating). For example, in the Northern Cities
dialect of North American English (spoken in cities along the lower shores of the Great Lakes), the
diphthong /aɪ/ is realized as the allophone [ʌɪ] before voiceless obstruents and as the allophone [aɪ] in
other environments. If you were illustrating this in your paper you would want to have one word that
illustrated the [ʌɪ] allophone and one that illustrated the [aɪ] in a sequential pair on the list as in the
example below:
number
gloss
51
right
52
etc.
ride
transcription orthography
rʌɪt
raɪd
right
ride
Phonation-Airstream Mechanisms
You want as many minimal pairs and minimal sets as possible, but you don't need to have all place by
manner by voicing by phonation by airstream mechanism by length by... contrasts. As a rule of thumb,
you should aim for at least one pair illustrating phonation contrasts etc. but not all combinatorial
possibilities. At least one example per phoneme is a minimum to aim for as a rule of thumb. Remember
that your verbal description of the language will help the reader fill in the blanks (ex: “…there is a two
way contrast consisting of plain-voiced vs plain-voiceless for all stops, affricates, and fricatives with the
exception of bilabial stops where there is only a voiced segment /b/.”). In our ‘sample language’ Eastern
Freedonian, there are no contrastively voiced consonants at any place of articulation. In English,
phonemically we have only a plain voiced~voiceless contrast for stops, fricatives, and affricates;
therefore, for English we would include a sequential pair on the list for each: ex: ‘pet’~‘bet’,
‘sewn’~‘zone’, ‘cheap’~ ‘jeep’. If there are no obvious minimal pairs, near minimal pairs are acceptable.
For example in English, there is no obvious minimal pair for [ʃ]~[ʒ], so one could use:
‘pleasure’~‘pressure’. In English, there is no phonation (voicing) contrast for nasal stops or
approximants. In Korean, as you saw in the VOT lab, there is a three-way contrast for stops; we would
need to provide three minimal pairs: fortis-lenis, fortis-aspirated, lenis-aspirated. Alternately, we could
use the minimal set used in that lab, which provides the same minimal pairs (if you use the lab pairs, be
sure to cite the lab). For a four-way contrast, you would need six minimal pairs; a five-way contract
would need ten minimal pairs, etc.
Place
Provide appropriate minimal pairs for each place of articulation. If there is contrastive phonation, it is
preferable to use one phonation type within each manner set. Let’s take English as an example. For
stops, there is a three-way place of articulation contrast. So, we could use either [b d g] or [p t k] for our
minimal pairs/set; for exampleː ‘bet’, ‘debt’, ‘get’. This will cover contrasts between labial, alveolar, and
velar stops. The fricatives display a four-way contrastː [f θ s ʃ] or [v ð z ʒ]; for exampleː ‘fin’, ‘thin’,
‘sin’, ‘shin’. This adds labiodental, dental, and palato-alveolar to our list of places of articulation
covered. The nasals have a three-way contrast, [m] / [n] / [ŋ]; however, the velar is found only at the end
of syllables, so the velar can only be shown to be contrastive in a set if all of the words have the nasals in
the same position as in the set ‘sum’~‘sun’~‘sung’. The remaining places of articulation are palatal and
labio-velar. In English, there is a single palatal consonant, the approximant [j], and a single labio-velar
consonant, the approximant [w]. For approximants, there is a four-way contrast, [j] / [ɹ] / [l] / [w]; for
example: ‘yore’, ‘roar’, ‘lore’, and ‘war’ would be sufficient (for most speakers).
Manner
Similarly, for manner of articulation, pick the place(s) of articulation that will demonstrate the phonemic
contrasts. For example, in English, the alveolar place of articulation has covered all of the different
manners of articulation with the exception of palato-alveolar for which place is not contrastive: oral stop,
fricative, nasal stop, central approximant, and lateral approximant. This would require ten minimal pairs,
or a minimal set, like: ‘dip’, ‘zip’, ‘nip’, ‘rip’, and ‘lip’.
Vowels
For vowels, try to create minimal sets for reasonable groups of vowels; for example high/mid/low
vowels, or front/central/back vowels, or rounded/unrounded vowels. In the best of all possible worlds,
we would have a minimal set for the full phonemic inventory of vowels. However, as discussed earlier,
this can be very difficult when you have a large vowel inventory like English.
Other Contrasts
Also, if there are other contrastive phonological elements in the language, these should be demonstrated
by minimal pair / minimal set. Here are some examples.
If vowel length is contrastive, as in Japanese (two-way contrast), one minimal pair is sufficient:
[kiːtɛ] ‘listening’ 聞いて kiite
[kite] ‘coming’ 来て kite
If geminate consonants are contrastive, as in Italian (two-way contrast), one minimal pair is sufficient:
[fatːɔ] ‘fact’ fatto
[fatɔ] ‘fate’ fato
If tone is contrastive there should be a minimal pair per contrast. For example there are four lexical tones
in Standard Mandarin Chinese (a four-way contrast), this would require six minimal pairs or one
minimal set to illustrate the contrasts (for the “5th tone” no minimal pairs are possible since it is only
found on certain types of syllables):
[tɕʰiŋ˥] ‘agile’ 轻 qing1
[tɕʰiŋ˧˥] ‘passion’ 情 qing2
[tɕʰiŋ˨˩˦] ‘6.67 hectares’ 顷 qing3
[tɕʰiŋ˥˩] ‘celebrate’ 庆 qing4
As with the other contrasts, indicate how widespread or restricted the given contrast in the text. For
example, the geminate contrast in Italian can occur with any of the consonants except
[z] and [ŋ] which are never geminate and [ɲ], [ʃ], and [ʎ] which are always geminate. Also, the geminate
consonants only occur between two vowels or between a vowel and an approximant, never word initially
or word finally.
Allophonic Variation
Allophonic variation refers to alternate phonetic realizations of a single phoneme. These phones can be
in complementary distribution or free variation. In general, in the full project, allophonic variation is
more typically uncovered when the recording of word list is analyzed acoustically. For the mini-project,
you may optionally include examples of major allophones. As with contrastive variation, one example of
a class of variation is sufficient.
In complementary distribution, the occurrence of the given phones is patterned; that is, predictable. For
allophones in complementary distribution, provide examples of each of the allophones. In our ‘sample
language’ Eastern Freedonian, the alternation between mid-high and mid-low vowels is complementary.
In this case, if desired, a single pair showing the mid-high / mid-low alternation for the front or back
vowel is sufficient. Similarly, the alternation between voiced and voiceless stops and fricatives is
complementary. Here, a single set showing the four allophones of /t/ (voiceless unreleased, flap,
aspirated, and unaspirated) for any one of the three places of articulation would be sufficient. Here is an
example from English:
[leɪt̚] ‘late’ /let/
[leɪɾɚ]̩‘later’ /letɹ/̩
[tʰeɪk] ‘take’ /tek/
[steɪk] ‘stake’ /stek/
As with contrastive variation, the accompanying text (in the phonological description section) should
describe how wide-spread the pattern of allophonic variation is.
In free variation, the occurrence of a given phone is not predictable. You cannot provide examples which
will ‘guarantee’ the use of allophones in free variation. In our ‘sample language’ Eastern Freedonian, the
alternation between velar and uvular allophones is in free variation. Examples cannot be given which
will always show [k] and [q].
Format and Structure: Mini-Projet
For the word list for the mini-project, group the words together to show the phonemic contrasts or
allophonic variation. If you find it useful, you can put a heading at the beginning of each group, but
number them continuously so that you can refer to the numbers easily in your previous sections (if you
don't number continuously, you will have multiple number-1 etc. on your list and it will be confusing).
You are allowed to recycle the same word in multiple groups if that word is useful for illustrating
multiple contrasts or allophones (for example 'pit~spit' illustrate the aspiration alternation, but you might
also use 'peat~pit~pate~pet~pat' as a minimal set for the front vowels). For the phonemic entries, give
the word in IPA and the English gloss (translation). For allophonic alternations, give the word in IPA,
the underlying phonemic transcription, and the English gloss. As an optional feature, you can add the
word in native orthography. There are some examples in the preceding sections.
Format and Structure: Full-Projet
Ultimately, there should be at least one word to illustrate each phoneme (including tones, pitch accents,
etc.) on the list. This will probably be covered by all of the minimal pairs etc. But there may be some
gaps. Just look for those gaps and fill them in. If there’s space (there might not be if you have picked a
language with a large number of phonemic contrasts) you might want to consider adding some words
with environments that cause allophonic variation in many languages. Here are a few example
environments you might want to try one or more of the following strategies:
•
coronal (alveolar, dental, retroflex) and dorsal (velar, uvular) stops and fricatives before both /i/
and before /a/ to see if you get palatalization and/or spirantization (turns into a fricative or
affricate) before /i/
•
•
•
•
•
•
vowels before tautosyllabic (same syllable, usually as a syllable coda) nasals to look for vowel
nasalization
nasals preceding obstruents of various places (usually across a syllable boundary) to look for
nasal place assimilation. For this you might want to try prefixes or compounds.
voiceless stops in the middle of words in intervocalic position (between vowels) to look for
voicing
voiceless stops at the beginning of words to look for aspiration
voiced obstruents (stops, fricatives, affricates) at the ends of words to look for word final
devoicing
words with the same phonemic tone in a sequence to look for tone sandhi (changes in the tone
caused by adjacent words with identical tone).
If you don’t add the words that are fishing for allophonic variation, you should have words that illustrate
allophonic variation (at least 5 pairs) as described in your sources.
There are two versions of the wordlist that you will need: 1) the version for the recordings, 2) the version
that you use for the transcriptions. You don’t have to hand in the wordlist that you use for the recordings.
Recording wordlist
The recording list consists of three columns: a number column, a column with only the native
orthography in large type (large enough to read from 1 meter away ex: 24 point Times), and a third
column with the English translations (no IPA or other information; it will confuse your consultant). It is
fine for the native orthography to be hand written by your consultant for the recording list. You should
start the recording by saying your name, the date, the purpose of the recording (550 final project), the
name of the language and dialect, the region and potentially town/city where your consultant is from
(and if your consultant desires it his/her name). During the recording, you’ll read the number and the
translation, then your consultant will read the word two or three times with a brief pause between each
word. Once you have finished the entire list that way, you consultant will read through the entire list
reading each word once without the numbers or translations. This will give you plenty of opportunity to
hear the difference between reading errors, slight random variations, and real allophonic variation. You
should start the list with two or three "dummy" words at the beginning and end of the list (that you're not
interested in) to minimize the list effect in the reading. Number them but don't worry about transcribing
them.
If you have time, you are encouraged to have your speaker read a short passage, tell a brief story, or say
a couple of sayings so that you can hear the intonation of the language in connected speech (although
you don't have to transcribe the saying, try to have your speaker translate it and write it out in their
native orthography so you can see where the word boundaries are. If you decide to do this hand it in as a
separate section of the paper following the wordlist. If you do this you'll get some bonus points.
Here’s an example with a only a few words (the first two would be dummy words):
1
2
3
4
庆
情
轻
顷
celebrate
passion
agile
unit of measure: 6.67 hectares
Transcription wordlist
The transcription list is the one you’ll hand in and it has the following format: 4 or five columns:
1. numbers the words are in the same order with numbers as on your recording list (although the
dummy words will occur on both lists, you won’t have to transcribe them, just mark them
“dummy”)
2. English translation,
3. broad IPA transcription,
4. narrow IPA transcription
5. 5- optionally, if you can manage to enter it electronically, the word in the native orthography.
The difference between a broad and a narrow transcription should be obvious to you when we get to this
point in the class, but just to summarize, the broad notes the major allophonic effects (such as aspiration
in English), the narrow notes idiosyncratic pronunciation and sub allophonic details. It is common for
students to note more than one pronunciation in the narrow column and indicate where in the recording
(by number and repetition) where the two variants come from.
Here’s an example of the transcription list that will be handed in with the paper and the recording
No. translation
broad
narrow
orth
1
celebrate
dummy
庆
2
passion
dummy
情
3
agile
轻
[tɕʰiŋ˨˩˦]
[tɕʰ ḭ̃ŋ ˨˩]
4
unit: 6.67 hectares
顷
[tɕʰiŋ˥˩]
[tɕʰ ĩŋ˥˩]