Na is a member of the Naish group of languages, a lower-level grouping within Sino-Tibetan that also includes Naxi and Laze. Keeping in mind that family trees only reflect one aspect of the thoroughly complex history of languages, here is the family tree proposed in a historical study of Na, Naxi and Laze.

In Chinese, the Na language is commonly referred to as ‘Mos(u)o’ (摩梭). Book-length linguistic analyses include Lidz (2010) and Michaud (2017), both available online.

In 2010, Na was granted an entry of its own, under the romanized name ‘Narua’, in the inventory of languages maintained by the Summer Institute of Linguistics (code: NRU), whereas it used to be classified as a dialect of Naxi. The former language code NBF is now split into Naxi proper (new code: NXQ), on the one hand, and Narua (code: NRU), on the other.

Most of the documents presented here are from the plain of Yongning 永宁 (Na name: /ɬi˧di˩-di˩mi˩/). Some recordings from the Lake area, and from the village of Labai 拉伯, are also presented. 

Dialect of the Yongning plain

As of 2020, twenty-seven narratives with a phonemic transcription are available. Eight of them have word-level, sentence-level and text-level translations into Chinese and French. English translations are gradually being added.

If you have suggestions for improvements, or if you are willing to contribute a translation of one (or more) of the texts into another language (either English, or another language), you are most welcome to get in touch. A number of narratives have not yet been transcribed; they are nonetheless made available for people who have a command of the language, from native speakers to motivated students.

Three texts have an orthographic transcription added by Roselle Dobbs (on the orthography, see an online User's guide).

In the narratives, the surface-phonological tones are indicated in the transcription at the sentence level, and lexical (underlying) tones in the word-by-word glosses. This makes it possible to study the numerous tonal processes at play in Na.

How the data collection process unfolded over time, with a note about "Unrecorded Tales"

Throughout the course of collaboration with language consultant F4, since 2006, the investigator was keen to collect new narratives, especially folk stories and tales about Na society in the old times: times which the speaker did not experience herself, but knew about from living with witnesses of the times that preceded the demise of the Yongning feudal lord and the direct integration of Yongning under Chinese administration in the 1950s. She was initially reluctant to record narratives. She felt that she did not have the authority or talent to do so, and she was wary of possible misinterpretations and misuses of the recorded data. She nonetheless accepted to record two narratives during the first field trip (2006): "The Sister's Wedding" and "How the Lake was created".

After these two narratives had been transcribed, and as trust and confidence built up over the years, more materials were recorded. The way the collection of texts grew was the following: when one of the topics that came up in our conversations, or in one of the narratives already recorded, seemed to me to be interesting subject-matter for a recording, I would ask the teacher/consultant, F4, whether she could speak about it in a recording session. For instance, I mentioned to her a story recorded in Muli (by a speaker of the Laze language), "Buried Alive". F4 remembered hearing a similar story from her grandmother, and agreed to record it (in August 2011). A second version of this story was elicited (in February 2012) to improve the record: having different versions of the same story sometimes brings out variants that are useful in linguistic studies, e.g. full forms vs. reduced forms of certain morphemes. Midway through this second version, the narrator realized that the names given to the characters in this story happened to correspond with the names of some of her own family members. This she felt to be highly inappropriate, as it associates these real people with some characters involved in tangled family problems. The narrator changed the names at once (midway through the narrative) and completed the tale as best she could, with some hesitations and lapses back into the unappropriate set of names. After the recording was over, she decided that the story should be recorded anew, and a third version was produced in May 2012.

The "Buried Alive" story brought up the topic of ghosts. Later, I elicited some vocabulary in this semantic fields, and suggested making a recording of explanations about different types of ghosts. This resulted in a set of eight recordings about ghosts, spirits and vampires recorded in July 2012.

The recorded narratives as communicative acts

The first of the narratives ("The Sister's Wedding") was told in the presence of another speaker of Na: the narrator's son, Mr. Latami Dashi. For the second story, I was the only audience, and this communicative situation, in which Mrs. Latami Dashilame only faced the microphone and the investigator, was repeated in all the following recording sessions. (An exception is version 3 of "Buried Alive", which was recorded in 2012 in the presence of a team of journalists and of a fellow linguist, Ms. He Jiezhen.) As communicative acts, these narratives can therefore be seen as the consultant's response to a pressing demand on the part of an outsider wishing to learn about Na language and culture. The speaker goes by a simple principle: she will only tell about things which she knows from having heard them in her youth, from elder family members of her grandmother's and her mother's generations. When I narrated to her some tales which were reported in ethnographic reports to be part of the oral tradition in Yongning, she would indicate very clearly whether she was familiar with those or not, and it was clear that she did not want to act as spokesperson for an abstract "Na culture" that did not correspond to her own experience. The stories that she does not have a command of include:

• the story of the first ancestors of mankind: three brothers, only one of whom survives the Flood, and seeks a wife in the heavenly world. This story is still very much alive among the Naxi and the Laze. One Na version was collected in the Warm Springs of Yongning (永宁温泉乡) at the end of the 1950s, and is published (in Chinese) in an ethnographic monograph (《永宁纳西族社会及母系制调查》（昆明：民族出版社，国家民委民族社会历史调查云南省编辑组编，1986年，第三本第113-114页）.
• stories about frogs: a frog becoming the son of an elderly couple without children (see the version told by a Naxi speaker from Zhongdian: "Toad Child")

A striking omission is that of any account of the custom for which traditional Na society is famous among ethnologists: its family structure, whereby one stayed in one's mother's house all life long, and the father only had a very minor social role. As ethnologists who conduct research on this topic are well aware, investigations into the private lives of people can easily be intrusive. Moreover, this topic links up with painful memories. In the 1970s the government cracked down heavily on the traditional Na family structure by cutting the grain supplies of people who did not adopt the Chinese-style family structure. Yesterday's forbidden practises are advertised as part of the touristic attractiveness of the area of Yongning and lake Lugu, where mass tourism developed at a staggering pace since the beginning of the 21st century. The amount of publicization of data about 'love among the Na/Mosuo', and common misinterpretations of the facts, have created no small amount of resentment. This is mentioned, for instance, in Shih Chuan-Kang's 2010 book: Quest for harmony: the Moso traditions of sexual union and family life (Stanford, California: Stanford University Press; see especially pp. 132-134). Many anecdotes are told locally about people's strategies to deal with indiscreet questions, resorting to humorous inventions to protect their private lives while catering for the visitors' expectations. In this context, I could easily understand my consultants' choice not to broach this topic, and I never tried to go against what I interpreted as a deliberate avoidance of this topic.

Other topics that were avoided because the consultant felt that they were not appropriate for publication include past conflicts between ethnic groups, and rites that are perceived to be related to these conflicts. For instance, the Na believe that the red dots on the cakes eaten by the Naxi on the 15th day of the eighth lunar month (the Chinese Mid-Autumn Festival) stand for the blood of the Na chieftain, whom the Naxi symbolically devour to celebrate their victory over him at some grim juncture in the distant past. Some Naxi consultants do remember that, in their youth, mooncakes used to be dotted with red spots, and that there must be at least five such spots on each cake. But the only explanation they can offer is that it was "for decoration". Chinese-style decorations on mooncakes are now chasing this old custom out of memory.

Informed consent for data communication / ethical issues

The investigator's home institution (CNRS) did not have formal ethical guidelines at the time of the investigation. The responsibility for relating with language consultants in the best possible way is the researcher's own: adopting a culturally appropriate behaviour, respecting the consultants, valuing the knowledge that they share, explaining the process of language documentation, conservation and research, giving them a fair compensation for their time and effort, and allowing them access to the documents and to the researcher's productions.

The research centre where I work, LACITO, consistently emphasizes this important responsibility. Researchers need to be highly respectful of the individuals and communities whose language and culture they study. They should have a strong commitment to helping in projects for the communities to the greatest extent possible. This includes contributing to language maintenance through the creation of literacy materials and teaching. Colleagues' attitudes in the field reflect a passionate personal implication, and a strong professional sense, that go way beyond the call of duty, and beyond the application of a set of laws and regulations. Ethical guidelines stated in legal terms -- those which are enforced by universities and other institutions --, however fine-grained, cannot but be tilted towards the (culturally relative) principles that hold in the countries where they were elaborated, and they can prove inappropriate or inapplicable under certain cultural circumstances. (A blistering attack on ethics protocols was published by George van Driem in 2016, pointing out that "some granting agencies have (...) compelled aspiring field linguists to engage in absurd and even unethical exercises in laying the groundwork for their linguistic fieldwork. Fundamental cultural differences and disparate histories of diverse language communities are essentially ignored" (in: "Endangered language research and the moral depravity of ethics protocols", Language Documentation and Conservation 10: 243-252).

However, going into the opposite extreme and leaving it up to the researchers alone to declare that they are up to the highest standards is not too good either, as it encourages patronizing attitudes and self-righteousness. There is no one-size-fits-all solution. It is necessary to work out solutions that are culturally appropriate both in the place of data collection and at the researchers' home institutions. It appears advisable for researchers to gather some pieces of evidence that an individual and/or a community agree to having materials about their language and culture published – and in many cases, that they not only allow the conservation and publication of materials, but wish that these materials get preserved and are made available, in a context where languages are under threat. 

In the case of Yongning Na, the speaker (F4) herself never uses writing, and never signs her name, entrusting all administrative matters to other family members. As a university graduate and a member of the Chinese administration, her son is not only familiar with these matters, but also knowledgeable about computers and digital technologies generally. He agreed to sign such a document on her behalf, on each field trip. I am very grateful to him for this, since it is good to have a written document to show when the topic is brought up within 'Western' institutions. The list of online documents in the Pangloss Collection is likely to be examined at various points in time in light of different critera, and a person in charge (at some level within the administrative hierarchy, within a research institution or an archive) may have to take personal responsibility, testifying that the documents comply with all the relevant laws, regulations, and recommendations. 

With a view to strengthening the case for open access to the data, I also wished to collect the speaker's own views on the topic of data diffusion. An oral consent was therefore recorded. The idea was to record a short message addressed to listeners of the recordings placed online for public access: the speaker would indicate to them that she agreed to have anyone listen to these recordings and get to know more about her native language and culture. This recording would testify to the speaker's willingness to have outsiders listen to these materials. I had a feeling that I was twisting the speaker's arm by eliciting this message, but proceeded nonetheless, because I felt a threat that for want of such an oral consent I would be accused of ignoring a recommendation often made in language documentation and conservation.

This piece was more than usually difficult for the speaker. When recording narratives, she addressed the investigator, knowing that a much broader audience would have access to the recording. When recording a consent, she was talking about our collaboration, in my presence, but addressing, in her own language, an abstract, potentially very distant audience. The situation was predictably awkward, and I am especially grateful to the consultant for complying with this strange request, explained to her by myself and by her son. What she eventually recorded was along the lines of: "My son introduced ti˧ɖo˥ [the name given to me in Yongning] to me, to study the Na language; I told him all that I know; he is a good student, transcribing what I tell him day after day, and developing a command of the language; by means of the online recordings, you ("outsider friends") are able to listen and see that he is doing a good job".

This is said in a language few people can understand, and I have not transcribed the recording. I am not keen to wade again through this strange piece with the consultant: it is spoken in a vacuum, and it is highly personal at the same time. It was produced at my own instigation, but it is probably too different from the intended message to be much use to the intended audience: someone who would wish to verify the data's compliance with laws and regulations. I could go back to the consultant and record a different piece, closely tailored according to a model, but that would seem strange (putting words in someone's mouth). So the document is left as is, accepting the consultant's interpretation of the task that was required of her.

In 2012, Xinhua Press reporters came to my home: they wanted to report about a foreigner working in Lijiang, and had heard about me from colleagues at the local research institute (Dongba Culture Research Institute). The consultant and I did our best to explain our work together. On that occasion, in front of the camera, as she had done when I recorded her "informed consent", she lay emphasis on the linguist's achievements. She had received instructions to this effect from family members, and would probably have adopted the same stance by herself, under the pressure of answering official media. The report was considered drab and unattractive by the journalists' hierarchy, who had precise expectations on what they wanted to produce, and it was not publicized in the end. An interesting aspect of this brief video, to me, is that her answer to the journalists' questions was in Na. (I hope I will be able to get round to transcribing and translating it at some point, and make it available online as a contribution to the corpus.)

By contrast with these awkward productions, the narratives told to me by the consultant in a quiet and familiar one-to-one setting sound pretty good to me, even though an element of 'stage fright' or sudden sense of awkwardness can always come up when recording with audio or video.

Documenting the morphophonology of Yongning Na

In addition to narratives, data from phonological investigations are being put online, as part of documentation of the language's morphophonology. The tone system is analyzed in full in a book, available online.

The lexical tones of nouns

The complexities of the Yongning Na tone system make it necessary to use two nonstandard symbols in addition to the International Phonetic Alphabet 'tone-letters':

• the pound symbol # stands for the last syllable of the lexical word: thus #H is a High tone that can only be realized after the last syllable of the word to which it is lexically attached. This tone therefore does not surface when a word is spoken in isolation.
• the dollar symbol $ is used in the expression H$, which refers to a lexical tone category distinct from both H# (a final H tone) and #H (a floating H tone).

The following table recapitulates all the tonal categories of monosyllabic and disyllabic nouns. 

tone pattern in isolation	tone pattern when followed by the copula	proposed phonemic analysis: underlying tone (=lexical tone)	transcription in the texts (underlying forms), using a real example
LM	L+M	LM	bo˩˧ 'pig'
LH	L+H	LH	ʐæ˩˥ 'leopard'
M	M+L	M	lɑ˧ 'tiger'
M	L+LM	L	jo˩ 'sheep'
M	M+H	#H	ʐwæ#˥ 'horse'
MH	M+H	MH#	hwɤ˧˥ 'cat'
M.M	M.M+L	M	go˧mi˧ 'younger sister'
M.M	M.M+H	#H	gi˧zɯ#˥ 'younger brother'
M.MH	M.M+H	MH#	hwɤ˧li˧˥ 'cat'
M.H	M.M+H	H$	hwɤ˧mi˥$ 'she-cat'
L.LM	L.L+M	L	kʰv̩˩mi˩ 'dog'
M.L	M.L+L	L#	dɑ˧ʝi˩ 'mule'
L.MH	L.M+H	L+MH#	ʝi˩ʈʂæ˧˥ 'waist'
L.M	L.M+H	L+#H	nɑ˩hĩ˥ 'Naxi'
L.M	L.M+L	LM	bo˩mi˧ 'sow'
L.M	L.M+L	LH	bo˩ɬɑ˥ 'boar'
M.H	M.H+L	H#	ʁæ˧ʈv̩˥ 'neck'

Verbs and affixes

The documents made available here are useful to study the tone categories of verbs, and also to study the tonal behaviour of affixes. Materials include tokens of reduplication, verbs in the frame /le+V+ze/ (accomplished+verb+perfective), verbs in the frame /tʰi+V+dʑo/ (durative+verb+progressive), verbs preceded by the prohibitive, and many more.

Verbs: spatial orientation

The data are shown in table form below. Monosyllabic indications of spatial orientation can be analyzed as prefixes.

prefix	tone	meaning	tone of verb
			Ma	Mb	H	La	Lb	MH
			li˧a	tsi˧b	se˥	kwɤ˩a	ɻ̩˩b	mi˧˥
			to look	to set	to walk	to throw	to turn	to push
gɤ˩-	L	upward	gɤ˩-li˧(-ze˧)	gɤ˩-tsi˧(-ze˧)	gɤ˩-se˥(-ze˩)	gɤ˩-kwɤ˧(-ze˩)	gɤ˩-ɻ̩˧(-ze˩)	gɤ˩-mi˧˥
mv̩˩-	L	downward	mv̩˩-li˧(-ze˧)	mv̩˩-tsi˧(-ze˧)	mv̩˩-se˥(-ze˩)	mv̩˩ kwɤ˧(-ze˩)	mv̩˩-ɻ̩˧(-ze˩)	mv̩˩mi˧˥

Disyllabic indications of spatial orientation are analyzed as adverbials.

adverbial	tone	meaning	tone of verb
			Ma	Mb	H	La	Lb	MH
			li˧a	tsi˧b	se˥	kwɤ˩a	ɻ̩˩b	mi˧˥
			to look	to set	to walk	to throw	to turn	to push
ɬo˧tɑ˧	M	to the side	ɬo˧tɑ˧ li˧(-ze˧)	ɬo˧tɑ˧-tsi˧(-ze˩)	ɬo˧tɑ˧ se˧(-ze˩)	ɬo˧tɑ˧ kwɤ˩	ɬo˧tɑ˧ ɻ̩˩	ɬo˧tɑ˧ pʰæ˧˥
ʁo˧dɑ˧	M	to the front	ʁo˧dɑ˧ li˧(-ze˧)	ʁo˧dɑ˧ tsi˧(-ze˩)	ʁo˧dɑ˧ se˧(-ze˩)	ʁo˧dɑ˧ kwɤ˩	ʁo˧dɑ˧ ɻ̩˩	ʁo˧dɑ˧ mi˧˥
ʁwæ˧-gi#˥	#H	leftward	ʁwæ˧-gi˧ li˩	ʁwæ˧-gi˧ tsi˧(-ze˩)	ʁwæ˧-gi˧ se˧(-ze˩)	ʁwæ˧-gi˧ kwɤ˥(-ze˩)	ʁwæ˧-gi˧ ɻ˥(-ze˩)	ʁwæ˧-gi˧ pʰæ˩
jo˩lo˩	L	rightward	jo˩lo˩ li˥(-ze˩)	jo˩lo˩ tsi˥(-ze˩) / jo˩lo˩ tsi˩	jo˩lo˩ se˩	jo˩lo˩ kwɤ˥(-ze˩)	jo˩lo˩ ɻ̩˥	jo˩lo˩ pʰæ˥(-ze˩)
jo˩gi˩	L	rightward	jo˩gi˩ li˥(-ze˩)	jo˩gi˩ tsi˥(-ze˩) / jo˩gi˩ tsi˩	jo˩gi˩ se˩	jo˩gi˩ kwɤ˥(-ze˩)	jo˩gi˩ ɻ̩˥	jo˩gi˩ mi˥
ʁo˧tʰo˩	L#	backward	ʁo˧tʰo˩ li˩	ʁo˧tʰo˩ tsi˩	ʁo˧tʰo˩ se˩	ʁo˧tʰo˩ kwɤ˩	ʁo˧tʰo˩ ɻ̩˩	ʁo˧tʰo˩ mi˩
gɤ˩tɕo˧	LM	upward	gɤ˩tɕo˧ li˧(-ze˧)	gɤ˩tɕo˧ tsi˧(-ze˧)	gɤ˩tɕo˧ se˧ (-ze˩)	gɤ˩tɕo˧ kwɤ˩	gɤ˩tɕo˧ ɻ˩	gɤ˩tɕo˧ mi˧˥
mv̩˩tɕo˧	LM	downward	mv̩˩tɕo˧ li˧(-ze˧)	mv̩˩tɕo˧ tsi˧(-ze˧)	mv̩˩tɕo˧ se˧(-ze˩)	mv̩˩tɕo˧ kwɤ˩	mv̩˩-tɕo˧ ɻ˩	mv̩˩tɕo˧ mi˧˥
gauche	H#	leftward	ʁwæ˧lo˥ li˩	ʁwæ˧lo˥ tsi˩	ʁwæ˧lo˥ se˩	ʁwæ˧lo˥ kwɤ˩	ʁwæ˧lo˥ ɻ̩˩	ʁwæ˧lo˥ pʰæ˩

Numeral-plus-classifier phrases were also an object of systematic elicitation. Their tone patterns are of some complexity: there exist 9 tonal categories of classifiers (2 subcategories for H, 2 for M, 2 for MH, and 3 for L), and some combinations allow two variants. (Before I understood that there were variants, I was led to believe that there must be as many as 11 categories.) The data set contains recordings of various lengths, some of them only covering the range of numerals from 1 to 10, others from 1 to 30, from 30 to 100, or from 1 to 100. Errors are not infrequent, due to the fact that the task was not very familiar to the language consultant. These are indicated by an asterisk in the transcription, and the correct form is also indicated.

The full set of recordings is described in the tables below. 

H1 category of classifiers
classifier for...	recordings WITH electroglottographic component		recordings WITHOUT electroglottographic component
	30 to 100	1 to 100	1 to 10	1 to 30
chunks	AUDIO + EGG	AUDIO +  EGG		AUDIO
handspan		AUDIO + EGG

 

H2 category of classifiers
classifier for...	recordings WITH electroglottographic component		recordings WITHOUT electroglottographic component
	30 to 100	1 to 100	1 to 30
days	AUDIO + EGG	AUDIO + EGG	AUDIO

 

M1 category of classifiers
classifier for...	recordings WITH electroglottographic component		recordings WITHOUT electroglottographic component
	30 to 100	1 to 100	1 to 30
knives	AUDIO + EGG	AUDIO + EGG	AUDIO
handfuls		AUDIO + EGG
heaps		AUDIO + EGG

 

M2 category of classifiers
classifier for...	recordings WITH electroglottographic component	recordings WITHOUT electroglottographic component
	1 to 100	1 to 30
months		AUDIO
pairs	AUDIO + EGG
round objects	AUDIO + EGG

 

MH1 category of classifiers
classifier for...	recordings WITH electroglottographic component		recordings WITHOUT electroglottographic component
	30 to 100	1 to 100	1 to 30
members of a pair			AUDIO
years	AUDIO + EGG	AUDIO + EGG	AUDIO
pounds		AUDIO + EGG

 

MH2 category of classifiers
classifier for...	recordings WITH electroglottographic component	recordings WITHOUT electroglottographic component
	1 to 100	1 to 30
people	AUDIO + EGG	AUDIO

 

L1 category of classifiers
classifier for...	recordings WITH electroglottographic component		recordings WITHOUT electroglottographic component
	30 to 100	1 to 100	1 to 10	1 to 30
money (currency)	AUDIO + EGG	AUDIO + EGG		AUDIO
sections (of road, etc)			AUDIO
sets of two		AUDIO + EGG

 

L2 category of classifiers
classifier for...	recordings WITH electroglottographic component			recordings WITHOUT electroglottographic component
	1 to 10	30 to 100	1 to 100	1 to 10	1 to 30
birds		AUDIO + EGG			AUDIO
ropes/long objects		AUDIO + EGG	AUDIO + EGG		AUDIO
armspans			AUDIO + EGG (1st recording); AUDIO + EGG (2nd recording)
meals			AUDIO + EGG
times			AUDIO + EGG
ounces	AUDIO + EGG
trees				AUDIO

 

L3 category of classifiers
classifier for...	recordings WITH electroglottographic component
	1 to 100
bundles of hay	AUDIO + EGG (1st recording) AUDIO + EGG (2nd recording)

 

One last file contains audio recordings of some verifications of various combinations.

Tone patterns in subject+verb and object+verb combinations

In view of the range of tone categories found on nouns and verbs in Yongning Na, it is no wonder that their combinations yield a wealth of diverse patterns. Combinations between verbs and subject or object nouns were systematically elicited. Combinations between a noun and a verb do not yield the same tone patterns depending on whether the noun is a subject or an object.

Determinative compound nouns

In Yongning Na, no tonal change takes place in possessive constructions, whereas tonal changes take place in compounds. In possessive constructions, the possessive /bv̩/ is added after the determiner, before the head, e.g. /hw˧li˧˥/ ‘cat’, /ɬv̩˧˥/ ‘brains’, /hwɤ˧li˧-bv̩˥ ǀ ɬv̩˧˥/ ‘brains of the cat’. The tone pattern of the head (the second noun) remains the same as in isolation. In determinative compounds, the order of constituents is, again, determiner plus head, but the tonal strings that result from compounding are not simply the concatenation of those found in isolation. The data set presented here is the result of a systematic investigation into the tones of compound nouns in Yongning Na.

The data set presented here is the result of a systematic investigation into the tones of compound nouns in Yongning Na. Combining animal names with names of body parts appeared as a useful means of eliciting most combinations. A second set was elicited using names of peoples (ethnic groups) and objects. Numerous peoples live in the area: Na (Mosuo), Prinmi (Pumi), Tibetan, Bai, Han (Chinese), Yi (Lolo), Lisu... Not all combinations refer to existing objects: for instance, not every group builds 'sanctuaries', /tse˧kʰo˩/, on the mountains. As the main language consultant understood that the purpose was to explore linguistic combinations, not record random expressions, she agreed to record compounds which to her were unusual, such as "Lisu knife". The Lisu are not known for a specific type of knife, and so the phrase does not make sense in the same way as "Lisu clothing", which refers to something specific and familiar. At times she even agreed to record some combinations that are slightly shocking, such as "Lisu blood". The artificial nature of the data goes some way towards explaining the tonal slips of the tongue that occasionally crept in: these are indicated with an asterisk * in the annotation.

Numerous topics can be investigated on the basis of these data. Among phonetic topics, I can think of the following.

(i) Phonation type at voicing offset. In the absence of any phonologically specified phonation type at voicing offset, how often does the utterance end in glottal constriction, laryngealization, and whispery voice/devoicing?

(ii) Fundamental frequency 'resetting' at the juncture/boundary between phonological groups. In a group such as /hĩ˧-bv̩˧ ǀ ʁo˧qʰwɤ˩/ 'human head', to what extent does one observe a greater discrepancy between successive tones than within a phonological group? Which linguistic uses are made of this latitude of variation?

Morpho-phonological data: coordinative compound nouns

The tone patterns of coordinative compounds are not identical with those of determinative compounds. Two productive sources were found: pairs of animal names of the two sexes, and their offspring, such as ‘ewe and ram’ and ‘ewe and lamb’; and pairs of kinship terms, such as ‘uncle and nephew’ and ‘mother and daughter’. The recording "Coordinative Compounds (2012)" presents about 100 tokens, covering a number of tonal combinations, though not all of them. The language consultant (F4) clearly preferred to remain within the bounds of common sense, and semantically inappropriate pairs such as ‘mother and nephew’ or ‘grandmother and brother’ were avoided. Another recording, "Coordinative Compounds 2", presents compounds based on 'day', 'month' and 'year': 'one or two years', 'two-three years', from 1-2 to 9-10.

The study of these items suggests that pronunciation habits develop for certain items, which come to have a habitual tone pattern to the exclusion of others. The existence of highly diverse tone patterns would thus reflect the semantic diversity of coordinative compounds. It is not always possible to arrive at the meaning of coordinative compounds simply on the basis of their two constituents. For instance, /hwɤ˧zo˧-hwɤ˧mi˥/, made up of ‘kitten’ and ‘she-cat’, does not mean ‘kitten and she-cat’ (the child and the mother), but refers to cats in general, as a species. More spectacularly, the terms for male and female puppies, /kʰv˧zo#˥/ and /kʰv˧mv#˥/ respectively, are used as names for human newborns: an unlovely name is purposedly chosen to repel demons who may be lurking around to take their lives. The real name is only given after a couple of months; sometimes as late as one full year after birth. The two terms /kʰv˧zo#˥/ and /kʰv˧mv#˥/, and their compound /kʰv˧zo˥-kʰv˩mv˩/, have become culturally specialized and cannot be used to refer to real puppies. There is thus a broad range of situations, from elicited combinations which the consultant may never have conceptualized before (such as ‘male and female jackal’) to highly lexicalized expressions.

Pronoun+noun possessive constructions

Pronouns do not behave like nouns in possessive constructions. The recording "Possession - Pronouns: possessive constructions with pronouns, without an intervening particle" (2012) presents three categories of pronouns with the various tonal categories of nouns.

Nouns: spatial orientation

Postpositions indicating spatial orientation combine tonally with nouns. The recording "Nouns_Spatial orientation" (2012) presents the various categories of nouns with 'beside', 'behind', 'to the left', and 'to the right'.

Nouns followed by 'even'

A recording presents the various categories of nouns followed by 'even' ('even a tiger', 'even a dog'...).

Nouns followed by 'and/only'

A recording presents the various categories of nouns followed by 'and/only' ('...and a dog', etc).

Vowels and consonants: Illustrations of some of the phonemes

With a view to avoiding large discrepancies between the transcription in International Phonetic Alphabet and the actual pronunciation, the transcription adopted is not strictly phonemic. In particular:

• the palatal nasal [ɲ] can be analyzed as an allophone of /ŋ/


• onset-less syllables receive an empty-onset-filler: /i/ is realized as [ʝi], /o/ as [wo], /ɯ/ as [ɣɯ], etc. (This is similar to Naxi.) These syllables are here transcribed with their phonetic onset, as [ʝi], [ɣɯ], etc.

On the other hand, the apical vowels [ʅ] and [ɿ] are phonemicized as /ɯ/. For details on the phonemic analysis, the reader is referred to Appendix A of the 2017 book about Yongning Na.

The documents presented here contain examples illustrating the phonemic contrasts that proved most challenging. The first is one between two different rhymes after alveolo-palatal affricates: /dʑi/ vs. /dʑɯ/; /tɕi/ vs. /tɕɯ/; /tɕʰi/ vs. /tɕʰɯ/. The vowel is more strongly apicalized for the latter than for the former. See "Palatalized Apicalized": words containing /i/ or /ɯ/ after an alveolopalatal affricate.

Dialect of Labai 拉伯

1. Vocabulary exploration: a transcribed elicitation session

See the recording of an exploratory elicitation session. The items were transcribed on the fly; the speaker and the investigator discussed in Chinese over the meaning of words, cases of homophony, issues of transcription, and so on.

2. A set of five untranscribed narratives

title	summary
"Dog"	Once upon a time, dogs and men exchanged their lifespan. Another version of this narrative was recorded in the Yongning plain.
"Trickster"	This is an anecdote about a colourful character of the village of Labai with whom the narrator was familiar
"Dape"	This is an excerpt from a ritual of the local religion (Dape, /dɑ˧pɤ˧/; Chinese rendering: 达巴). The text is first said, then chanted (out of memory).
"Sister"	Once upon a time, there were two orphans; the brother went hunting for a living, and his sister did not hear of him again; she eventually married; on the day of the marriage, her elder brother happened to come back to the village, but she did not recognize him and he was treated like a beggar. Other versions of this narrative were recorded in the Yongning plain, and in the village of Fengke.
"Flood"	This is a story of the flood and the origin of the Na people.

The data from the Labai dialect presented here were collected in two work sessions with the distinguished writer Lamu Gatusa 拉 木•嘎 吐萨 (Chinese pen-name: 石高峰), on August 11th and August 12th, 2009. The second session was recorded. For reasons of time, the narratives could not be transcribed. The only document with complete annotations is the word list. The purpose was to devise a transcription system in the International Phonetic Alphabet, which Mr. Lamu would then use (and develop further) for transcribing Na stories, proverbs and rituals and preparing annotated bilingual editions.

In the end, Mr. Gatusa was not able to prioritize the phonetic transcription of texts over his various other projects and commitments. It nonetheless appears useful to make this small data set available to all interested. The sound file of the vocabulary session was edited, deleting asides, long explanations, and some taboo items of vocabulary. It must be emphasized that this document is not based on an in-depth phonemic analysis, unlike the Yongning Na data: the transcription of tones needs to be verified and deepened.

Technically, the quality of the recordings is not very good: the recording took place at Mr. Lamu's home in the city of Kunming; the distance to the microphone varies greatly in the course of the recording.

Explanations about the format of the annotation: notes, grammatical glosses,  borrowings, full-text translations...

The annotation comprises numerous notes, inserted in the XML code with the following markup:

<NOTE message="This is a comment."/>

These include comments about verifications that were made about form and meaning, and comments about phonetic implementation. The earliest notations, which were later corrected systematically, are closer to phonetic realizations; surface-phonological transcriptions abstract away from the actual acoustics. Information about earlier transcriptions may therefore be useful to understand the analysis process. Moreover, notes contain indications about possible variants. 

For instance, in the 6th sentence of Sister1, "and the boy went out hunting", the High tone that constitutes the endpoint of the /MH/ sequence on the verb /hɯ˧˥/ is projected onto the following particle, /tsɯ˧˥/. The sequence [kʰv̩˧ʂæ˧-hɯ˧-tsɯ˥-mv̩˩] is the product of a phonological reassociation of the H tone to the right of its original position. A note was added to indicate that it would also be possible to end the tone group before the final particles, yielding /...kʰv̩˧ʂæ˧-hɯ˧˥ ǀ tsɯ˧˥.../.

The grammatical glosses provided here are essentially based on Liberty Lidz's dissertation (Lidz 2010); needless to say, the differences with L. Lidz's glosses are my own responsibility. The standard abbreviations are used for the concepts included in the Leipzig Glossing Rules (Comrie, Haspelmath, and Bickel), whereas the full word is provided for all the others. Technical glosses are preceded by a special symbol: °.

The following conventions are used for passages to be added or removed:

[ ]	square brackets indicate an addition to be made to the text (as indicated by the speaker when the transcription was done)
< >	angled brackets indicate a 'false start' or mistaken use of terms, and hence a passage to be suppressed (again, as indicated by the speaker when the transcription was done)

The letter F is added after a word that is intonationally focalized. Focalization is realized by phonetic correlates that include lengthening, a tilt in fundamental frequency, and a slight centralizing diphthongization of the vowel.

Borrowings: Due to increasing exposure to Mandarin Chinese, borrowings appear here and there in the narratives. These borrowings are transcribed in IPA like the others; their glosses are formatted as: 

ChineseCharacters::gloss

e.g. 事情::affairs/matters

(XML code: <TRANSL xml:lang="fr">事情::affaire/tâche</TRANSL>
                    <TRANSL xml:lang="en">事情::affairs/matters</TRANSL>)

I considered using English words in the French translation, since French speakers' command of English is comparable to Na speakers' command of Chinese. This would yield translations such as «le grand frère, il a été tout sad». But this convention was not used in the end, because it was found that this made the texts harder to read. Also, the same strategy could not be extended to Chinese and English translations in any simple way. 

Punctuation: Some punctuation is indicated in the transcription of sentences, following punctuation conventions for French or English. As pointed out by Liberty Lidz (p.c.), there is a discrepancy with the conventions used for Chinese. These indications are nonetheless kept in the present version of the transcriptions, for want of a satisfactory system for transcribing phrasing and intonation. This part of the transcription gets extremely complex as one attempts to encode more and more refined information. At the present stage, in addition to informally used punctuation symbols, I propose a division into phonological groups (within which phonological regularities apply, e.g.: (i) there can be at most one H tone, (ii) there cannot be only L tones, (iii) a H tone can only be followed by L tones). The IPA symbols for boundaries are used: mostly | but also another symbol for looser junctures: ||.

Full-text translations are available for all documents that have sentence-level translations. The text-level translations are intended to be legible, offering a convenient way of getting the gist of a story before beginning to decipher the original. News (July 2012): it is now possible to select the language of the full-text translation in the online interface.  

References

Comrie, Bernard, Martin Haspelmath, and Balthasar Bickel. Leipzig Glossing Rules. http://www.eva.mpg.de/lingua/resources/glossing-rules.php

Jacques, Guillaume, and Alexis Michaud. 2011. "Approaching the historical phonology of three highly eroded Sino-Tibetan languages: Naxi, Na and Laze." Diachronica 28 (4): 468-498.

Lidz, Liberty. 2010. A descriptive grammar of Yongning Na (Mosuo). Austin: University of Texas, Department of linguistics. https://repositories.lib.utexas.edu/bitstream/handle/2152/ETD-UT-2010-12-2643/LIDZ-DISSERTATION.pdf

Lidz, Liberty. 2016. "Yongning Na (Mosuo)". In Graham Thurgood & Randy J. LaPolla (eds.), The Sino-Tibetan languages, 840–855. 2nd edition. (Routledge Language Family Series). London: Routledge.

Michaud, Alexis. 2017. Tone in Yongning Na: lexical tones and morphotonology. (Studies in Diversity Linguistics 13). Berlin: Language Science Press. http://langsci-press.org/catalog/book/109.

Michaud, Alexis, and Latami Dashi. 2012. A description of endangered phonemic oppositions in Mosuo (Yongning Na). In Issues of Language Endangerment, ed. by Xu Shixuan, Tjeerd de Graaf and Cecilia Brassett. Book series: 16th World Congress of IUAES. Beijing: 知识产权出版社 (Intellectual property publishing house), pp. 55-71.

中文介绍 / presentation in Chinese

永宁摩梭语，也曾经按照和即仁先生和姜竹仪女士两位关于此地区语言的划分，称作“纳西语东部方言永宁土语”。但因为对纳语的方言调查资料目前还不算很完全，“东部方言”和“西部方言”的区分至今只能算是一个初步假设，没有经过充份的科学证明。

《泛语资料库》里的摩梭语资料，至今（至2020年）主要来自米可与杜玫瑰的调查。米可2006年10-12月开始调查云南丽江市宁蒗县永宁乡的摩梭语（纳语），一直到2018年几乎每年或每两年调查一次。一个重要的研究方向是分析此语言的声调系统。

丽江市宁蒗县民族文化研究所副所长拉他咪达石招待米可在他家住，发音人是他的家人，拉他咪达石本人也参加了调查工作。拉他咪达石是人类学家，纳族（摩梭人）文化专家。

在历史上，简单来说，金沙江（长江）两岸的纳族自元朝以来有了不同的发展方向，靠近丽江的纳族地区受到内地汉族影响越来越深，越来越直接，而永宁等地区受中央政府影响不太直接，而受西藏文化的影响多一些。(见：郭大烈、和志武的《纳西族史》，四川民族出版社，重庆，1992年。1999年第二版。)

“纳西”方言自称中的“纳”这个成份，来指此民族，而“西”意思为‘人’，在丽江坝说的是/hi˧/，在永宁，奉科等等说的是/hi͂˥/。杨福泉教授提出，“‘纳西学’的内容，实际上已经不仅仅包括丽江自称‘纳西’的纳西族(…)按照标准的说法，实际上应该是‘纳学’研究，‘纳学’是这门学科更为标准的定义”(《摩梭社会文化研究论文集(1960-2005)》序二，云南大学出版社，主编：拉他咪达石，2006年，4-5页)。

“摩梭”是它称，是1960年代中华人民共和国定义少数民族之前，外界对于整个纳族的称呼。1990年代以来，云南省接受了“摩梭”这个称呼，现在永宁纳族人算是属于“纳西族的摩梭人”，等于说“摩梭”是纳西族里面有特点的人口。每个称呼的选择，自然有其在政治等方面的考虑，这方面的情况比较复杂，在此就不多加赘言。(关于族称问题，以上提到的《摩梭社会文化研究论文集》的序言二有详细的描述。)从语言学的观点来看，通过研究我们可以证明，永宁摩梭语与丽江坝纳西语有清楚的同源关系。

目前，宁蒗纳族地区的变化很快，泸沽湖是旅游热点，来永宁的游客也越来越多。这当然影响着当地的传统社会，在语言方面，老一代和年轻一代的差异非常明显，有的孩子不讲当地的语言了，只讲普通话和西南官话。总的说来，跟过去相比，今天，永宁的生活条件有了较大改善。但是对语言学家来说，调查摩梭语方言的任务也越来越迫切。