Initial commit

index.html

@@ -0,0 +1,376 @@
+<div class="tabu">
+<h1 class="unnumbered" id="language-skills">Maxime Fily, Linguist and Engineer</h1>
+<img alt="Mugshot" src="/home/mfily/Documents/ancien_espace/personal_files/perso/maxime_neige_Datsha_crop.jpg"  width="360" />
+<h1 class="unnumbered" id="language-skills">Language skills</h1>
+<table>
+<tbody>
+<tr class="odd">
+<td style="text-align: left;">French:</td>
+<td style="text-align: left;">Native speaker</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;">English:</td>
+<td style="text-align: left;">C1</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;"></td>
+<td style="text-align: left;"><span>↪</span> Bright test: 4.5/5
+(2007)</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;"><span>↪</span> TOEFL test: 627/677
+(2005)</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;">Mandarin:</td>
+<td style="text-align: left;">B2/C1</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;"><span>↪</span> DCL, B2 level, (2017)</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;"></td>
+<td style="text-align: left;"><span>↪</span> HSK level 5 : 214/300
+(2017)</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;"><span>↪</span> HSK level 3 : 299/300
+(2015)</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;">Spanish:</td>
+<td style="text-align: left;">Intermediate</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;">Na (Narua):</td>
+<td style="text-align: left;">Working knowledge</td>
+</tr>
+</tbody>
+</table>
+<p style="width: 500px;"><br />
+</p>
+<h1 class="unnumbered" id="computer-skills">Computer skills</h1>
+<table>
+<tbody>
+<tr class="odd">
+<td style="text-align: left;">General IT:</td>
+<td style="text-align: left;">GNU/Linux systems administration (Ubuntu,
+Mint)</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;">LaTeX, Libre Office, MS Office</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;">Programming:</td>
+<td style="text-align: left;">Linux/Unix : Bash, csh, Ksh</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;">General: Python, C, xml, JuPyteR</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;"></td>
+<td style="text-align: left;">Speech processing: Praat</td>
+</tr>
+<tr class="even">
+<td style="text-align: left;"></td>
+<td style="text-align: left;">NLP: basic knowledge of Transformers-based
+neural networks</td>
+</tr>
+<tr class="odd">
+<td style="text-align: left;"></td>
+<td style="text-align: left;">Statistics: R</td>
+</tr>
+</tbody>
+</table>
+<p style="width: 500px;"><br />
+</p>
+<h1 class="unnumbered" id="soft-skills">Soft skills</h1>
+<ul>
+<li><p style="width: 500px;">Autonomy, adaptability</p></li>
+<li><p style="width: 500px;">Reliability, solicitude</p></li>
+<li><p style="width: 500px;">Teamwork, collaborative work</p></li>
+<li><p style="width: 500px;">Oral and written communication</p></li>
+</ul>
+<h1 class="unnumbered" id="education">Education</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">PhD in Phonetics, Phonology and Speech Sciences, Université Paris III
+Sorbonne Nouvelle (defended on Dec. 2, 2022).<br />
+</p>
+</div>
+<p style="width: 500px;"><span>↪</span> <u>PhD Dissertation</u>:</p>
+<div class="refsection">
+
+</div>
+<p style="width: 500px;"><u>Abstract</u>: This thesis constitutes a description and analysis
+of the dialect of the Na (Mosuo) language (Sino-Tibetan/Trans-Himalayan
+family) spoken in the village of Shekua. After a brief presentation of
+the segmental phonology (consonants and vowels), the main part of the
+analysis focuses on the tonal system. Based on previous work and
+first-hand immersion fieldwork, it is confirmed that the system is based
+on two tonal levels (High and Low). The system is explored in a
+systematic manner, starting out from the tones of nouns and progressing
+to those of compound nouns, verbs, and various morphosyntactic
+constructions. The analysis of the tonal system brings out categories
+that can only be revealed by combining information from several
+contexts. Seven categories are found for monosyllabic nouns, and no less
+than twelve for (monosyllabic) verbs. A contrastive use of the HL
+contour, not found among previously documented Na dialects, makes
+functional sense inside a two-level system, as it puts the available
+phonological units to maximum use. Three-level systems (contrasting H,
+M, and L) such as that of the village of Alawa (Yongning) allow for a
+wider combinatorial range, and hence are under less pressure to exploit
+every nook and cranny of this combinatorial range. This monograph on a
+Na tonal system, combined with a fully open access corpus, adds to the
+literature on a language whose traditional chain of transmission is
+undoubtedly threatened.<br />
+</p>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Master’s degree in Linguistics, with a specialization in the field of
+experimental phonetics and phonology, Université Grenoble Alpes (<span
+class="smallcaps">uga</span>), “mention Très Bien”.<br />
+</p>
+</div>
+<p style="width: 500px;"><span>↪</span> <u>Master’s thesis</u>:</p>
+<div class="refsection">
+
+</div>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Level 3 University degree (“DU niveau 3”) in Chinese language and
+culture, Université Lyon 3, “mention Très Bien”.<br />
+</p>
+<p style="width: 500px;">Master of Engineering, <span class="smallcaps">enspg</span> (“École
+Nationale Supérieure de Physique de Grenoble”, renamed PHELMA in 2008),
+member of Grenoble INP group (Grenoble Institute of Engineering, a
+public sector technology university), “mention Assez Bien”.<br />
+</p>
+</div>
+<h1 class="unnumbered" id="professional-experiences">Professional
+experiences</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Post-doc at <span class="smallcaps">cnrs</span>, <span
+class="smallcaps">llf</span>, Laboratoire de Linguistique Formelle and
+<span class="smallcaps">Lacito</span>, Langues et Civilisations à
+Tradition Orale.<br />
+</p>
+<p style="width: 500px;">Supervisor for the internship of Université Sorbonne Nouvelle BA
+student Berthilde Biard, on the subject of data extraction and analysis
+for the buildup of verbal and nominal paradigms in Na languages
+(3months).<br />
+</p>
+<p style="width: 500px;">Doctoral student at <span class="smallcaps">lacito</span> (Villejuif)
+and <span class="smallcaps">Gipsa-lab</span> (Grenoble).<br />
+</p>
+<p style="width: 500px;">Mandarin teacher, private classes, 2h/week.<br />
+</p>
+<p style="width: 500px;">Temporary lecturer for the class “Introduction to linguistics and
+language families”, <span class="smallcaps">uga</span> (12 hours).<br />
+</p>
+<p style="width: 500px;">Research Internship at <span class="smallcaps">Gipsa-lab</span> in
+the frame of my Master 2 in Linguistics. Design of an oral-nasal signals
+separate recording plate for the acoustic study of the oral and nasal
+cues of nasalised sounds in French and Taiwan Mandarin.<br />
+</p>
+<p style="width: 500px;">Confirmed Engineer at <span class="smallcaps">Areva NP</span>.
+Neutronic design of nuclear fuel assemblies, Lyon, France.<br />
+</p>
+<p style="width: 500px;">Confirmed Engineer at <span class="smallcaps">Wecan</span>, an <span
+class="smallcaps">Areva-CGN</span> (<em>China General Nuclear</em>)
+joint-venture specializing in nuclear design and safety, Shenzhen,
+China.<br />
+</p>
+<p style="width: 500px;">Engineer (“Ingénieur d’étude”) at <span class="smallcaps">Areva
+NP</span>, in nuclear safety analyses, Paris, France.<br />
+</p>
+</div>
+<h1 class="unnumbered" id="linguistic-fieldwork">Linguistic
+fieldwork</h1>
+<p style="width: 500px;">Since the beginning of my work as a field linguist, I have focused on
+the Shekua variety of the Na language, also known as Lataddi Narua.
+Shekua is a small village situated in proximity to the Grass Sea, and
+the speakers in this village use a variety that is closely related to
+Yongning Narua, whose tone system has been described in Michaud (2017).
+The detail of my fieldwork experience is outlined below:</p>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Interviews of Shekua Na speakers : narratives, phonological
+confirmation paradigms, dialogues (Yunnan, 2 months).</p>
+<p style="width: 500px;">Interviews of Shekua Na speakers : phonetics and phonology,
+narratives (Sichuan and Yunnan, 3 months).</p>
+</div>
+<h1 class="unnumbered" id="foreign-collaborations">Foreign
+collaborations</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018:</span></p>
+<p style="width: 500px;">I undertook a research trip to Kunming “Yunnan Minzu University”,
+organized at the invitation of M. He Likun and M. Liu Jinrong (School of
+Ethnic Cultures). During this visit I attended seminars that covered a
+range of descriptive works on Yunnan languages. Additionnally, I
+presented on the phonological system of Shekua Na.</p>
+</div>
+<h1 class="unnumbered" id="technical-achievements">Technical
+achievements</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018:</span></p>
+<p style="width: 500px;">Creation of a fully customizable keyboard for Linux users interested
+in writing with the International Phonetic Alphabet. See <a
+href="https://lacito.hypotheses.org/3086"
+class="uri">https://lacito.hypotheses.org/3086</a></p>
+<p style="width: 500px;">Realization of a solution to allow the conversion of praat textgrids
+to <span class="smallcaps">xml</span> format to accelerate Pangloss
+deposits. See <a href="https://github.com/maxime-fily/TXTGRD2XML"
+class="uri">https://github.com/maxime-fily/TXTGRD2XML</a></p>
+<p style="width: 500px;">Design of the acquisition module for a separate recording of oral and
+nasal tract output, by modifying the Glottal Enterprise nasalance
+plate.</p>
+</div>
+<h1 class="unnumbered" id="grants-and-projects">Grants and projects</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>Since 2018:</span></p>
+<p style="width: 500px;">I obtained a grant to perform part of my research abroad, via the
+<span class="smallcaps">uga idex</span> International Mobility Program.
+The funding allowed a field trip and a research visit to Yunnan and
+Sichuan.</p>
+<p style="width: 500px;">Participant in the Labex-EFL program, within the “Typology and
+dynamics of linguistic systems” strand. Since 2023, I also participate
+in the “Computational Linguistics” strand. The goal is to address
+computational linguistics through a cross-disciplinary approach,
+enabling the deployment of NLP methods across the various linguistic
+strands of the Labex.</p>
+</div>
+<h1 class="unnumbered"
+id="publications-in-conferences-with-proceedings">Publications</h1>
+
+<div class="refsection">
+<p style="width: 500px;"> See my Google scholar profile : </p>
+<a href="https://scholar.google.fr/citations?user=XYvTKbIAAAAJ&hl=fr&oi=ao">Maxime Fily</a>
+</div>
+<h1 class="unnumbered" id="ongoing-research-activity">Ongoing research
+activity</h1>
+<h2 class="unnumbered" id="comparative-work-on-na-dialects">Comparative
+work on Na dialects</h2>
+<p style="width: 500px;">I am currently in the field, collecting data to finalize a study on
+tonal correspondences between Yongning Na and Shekua/Lataddi Na. The
+study focuses on noun and verb categories in the two dialects. The main
+difference between these two dialects lies in their tone systems:
+Yongning has a three-level tone system, while Shekua/Lataddi has a
+two-level tone system. Lexically, the two dialects are very close, and
+moreover, they are mutually intelligible. This pushes for a tonal
+comparison: our approach aims to identify precise tonal correspondences
+between a two-level system and a three-level system, to potentially shed
+light on rearrangement mechanisms. A preliminary study covered 52 verbs
+and 51 nouns, and the objective is to expand this dataset to 100 verbs
+and nouns for a comprehensive study. The challenge in this task arises
+from the <em>morpho-tonological complexity</em> of tonal categories in
+Na, which can only be accessed in combination with other morphemes. For
+instance, verbs are examined in combination with TAME particles, nouns
+within Object-Verb structures and compound nouns.</p>
+<p style="width: 500px;">For this study, we utilize carefully curated data and rely on
+statistical tools such as Sankey Diagrams and Agglomerative Hierarchical
+Clustering. These tools offer evidence-based and interpretable graphs,
+contributing to opening discussions on categories. Additionally, the
+study draws on first-hand field data, all of which is accessible online
+via the Pangloss website.</p>
+<p style="width: 500px;">The results of the study mark a first step towards reconstructing the
+diachronic processes that led to contemporary Na tone systems.
+Simultaneously, a search for cognates is underway between Naish
+languages and the conservative languages of the Trans-Himalayan
+(Sino-Tibetan) family, particularly Gyalrong languages. This research is
+conducted with guidance from Alexis Michaud (<span
+class="smallcaps">Lacito</span>) and advice from Guillaume Jacques
+(CRLAO). Furthermore, we closely follow the works of Li Zihe (Peking
+University), who has undertaken the task of assembling that cognate list
+in Naish, Gyalrong, Old Burmese, Old Tibetan and Written Tibetan
+languages.</p>
+<h2 class="unnumbered" id="nlp-for-underdocumented-languages">NLP for
+un(der)documented languages</h2>
+<p style="width: 500px;">My post-doctoral activities over the past year (and until August
+2024) focus on evaluating the deployment cost of NLP methods for
+un(der)documented languages. Our work is grounded in the understanding
+that the rapid disappearance of rare and undocumented languages poses a
+monumental challenge for linguists involved in documenting these
+languages. Language models can assist linguists, but since they are
+pre-trained on a closed list of resource-rich languages, they need to be
+<em>fine-tuned</em> to be able to perform NLP tasks on a language not
+seen in pre-training. Before I started this position, a successful
+fine-tuning experiment was conducted on Yongning Na (glottocode:
+yong1234) and Japhug (glottocode: japh1234), with Word-Error Rates well
+within acceptable values for an Automatic Speech Recognition (ASR) task
+(Guillaume, Wisniewski, Macaire, et al. 2022a). However, the study
+revealed that the quality and diversity of the fine-tuning data impacted
+the system’s performance. For instance, a system fine-tuned on
+elicitations exhibited poor performance on texts, and vice versa.</p>
+<p style="width: 500px;">To better design minimal sets of audio data for input into a language
+model, specifically for ASR tasks – such as phonetic and phonological
+transcriptions in our case –, my assignment consisted in developing
+distance measurement methods in the vectorial representation space. This
+allowed us to assess similarities between recordings. With the
+assistance of the developed toolkit, our aim is to predict the
+performance of a fine-tuning task based on the characteristics of the
+input and tested materials.</p>
+<p style="width: 500px;">This research is highly interdisciplinary, engaging both NLP
+scientists seeking a better understanding of language models based on
+neural networks and linguists who stand to benefit from language models
+better tailored to their needs. Neural networks are producing
+increasingly powerful models for academia, but the process needs to be
+better understood, especially by the linguists, who have the expertise
+to assess the accuracy of the language models produced recently. The
+community can benefit from these models by accelerating time-consuming
+stages of work, such as transcription. Thus, it seemed essential to
+enter the NLP arena as a linguist. Guillaume Wisniewski (LLF) is the
+principal investigator for the project, with a dedicated focus on the
+rare languages of Yunnan and Sichuan, leaving ample room for future
+collaborations within my area of linguistic interest.</p>
+<p style="width: 500px;"><br />
+</p>
+<h1 class="unnumbered" id="academic-service">Academic service</h1>
+<h2 class="unnumbered" id="article-reviews">Article reviews</h2>
+<p style="width: 500px;">Peer-reviews in the field of phonetics and phonology, for the
+following conferences: Interspeech, “Journées d’Étude de la Parole”.</p>
+<h2 class="unnumbered" id="collective-duties">Collective duties</h2>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Representative for the doctoral students at <span
+class="smallcaps">lacito</span>.</p>
+<p style="width: 500px;">Representative for the doctoral students at <span
+class="smallcaps">Gipsa-lab</span> and <span
+class="smallcaps">uga</span>.</p>
+<p style="width: 500px;">Representative for the doctoral students at <span
+class="smallcaps">Grenoble-INP</span> <span
+class="smallcaps">clhsct</span>.</p>
+<p style="width: 500px;">Workplace First-Aid Rescuer (retraining planned in 2024).</p>
+</div>
+<p style="width: 500px;"><br />
+</p>
+<h1 class="unnumbered" id="learned-societies">Learned societies</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>2018 - 2022:</span></p>
+<p style="width: 500px;">Member of the “Association Française de Communication Parlée”.</p>
+<p style="width: 500px;">Member of the “Société de Linguistique de Paris”.</p>
+<p style="width: 500px;">Member of the International Phonetic Association.</p>
+</div>
+<h1 class="unnumbered" id="references">References</h1>
+<div class="compactlabel">
+<p style="width: 500px;"><span>Linguistics </span></p>
+<p style="width: 500px;">M. Alexis Michaud<br />
+Research Director, <span class="smallcaps">cnrs</span><br />
+<span class="smallcaps">Lacito</span>, 7 rue G. Môquet, Villejuif<br />
+Phone: (+33) (0)1 49 58 37 78<br />
+E-mail: [email protected]<br />
+</p>
+<p style="width: 500px;">M. Guillaume Wisniewski<br />
+Assistant Professor, Université Paris-Cité<br />
+LLF, 8 rue A. Einstein, Paris<br />
+Phone: (+33) (0)1 57 27 57 64<br />
+E-mail: [email protected]<br />
+</p>
+</div>
+</div>