preprocessing/dicts/encoding_syllabi.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Encoding by pronunciation"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sn\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import wn\n",
    "from wn import morphy \n",
    "ewn = wn.Wordnet('ewn:2020')\n",
    "m = morphy.Morphy(ewn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 381,
   "metadata": {},
   "outputs": [],
   "source": [
    "pronouncing_dict_path = './cmudict.dict.txt'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 382,
   "metadata": {},
   "outputs": [],
   "source": [
    "dico = {}\n",
    "with open(pronouncing_dict_path, 'r', encoding='utf-8') as file:\n",
    "\tfor line in file.readlines():\n",
    "\t\tline = re.sub('\\d', '', line) # remove tone number\n",
    "\t\tl = line.strip().split(' ')\n",
    "\t\tword = l[0]\n",
    "\t\tpronounciation = ' '.join(l[1:])\n",
    "\t\t# filter some non standard words\n",
    "\t\tif word.islower() and word.isalpha():\n",
    "\t\t\tdico[word] = pronounciation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 383,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "117490"
      ]
     },
     "execution_count": 383,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dico)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "first check if the word exists in wordnet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 384,
   "metadata": {},
   "outputs": [],
   "source": [
    "def validate(word):\n",
    "\t# check if exists\n",
    "\tres = wn.synsets(word)\n",
    "\tif res != []:\n",
    "\t\treturn True\n",
    "\t# otherwise check if as a derivation, the lemma exists\n",
    "\tfor lemmas_per_pos in m(word).values():\n",
    "\t\tfor lemma in lemmas_per_pos:\n",
    "\t\t\tres = wn.synsets(lemma)\n",
    "\t\t\t# if there exists one, return\n",
    "\t\t\tif res != []:\n",
    "\t\t\t\treturn True\n",
    "\treturn False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "non_words = []\n",
    "for word in tqdm(dico.keys()):\n",
    "\tif not validate(word):\n",
    "\t\tnon_words.append(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 386,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 386,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(non_words)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('./non_wordnet_words.txt', 'w', encoding='utf-8') as file:\n",
    "\tfor non_word in non_words:\n",
    "\t\tfile.write(non_word+'\\n')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, wordnet excludes some function words, they should be taken into account in dictionary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 387,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('./non_wordnet_words.txt', 'r', encoding='utf-8') as file:\n",
    "\tnon_words = file.read().splitlines()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 388,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('./wordnet_stopwords.txt', 'r', encoding='utf-8') as file:\n",
    "\tstop_words = file.read().splitlines()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 389,
   "metadata": {},
   "outputs": [],
   "source": [
    "for non_word in non_words:\n",
    "\tif not non_word in stop_words:\n",
    "\t\tdico.pop(non_word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 390,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "52123"
      ]
     },
     "execution_count": 390,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dico)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that many of them are acronyms, which is pronounced as individual letters, we could remove them directly from the cmu dict, the acronyms will be dealed in another independent dico."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Remove acronyms"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "first collect single letter sound"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 391,
   "metadata": {},
   "outputs": [],
   "source": [
    "letter_sound = {}\n",
    "syllable_count = {}\n",
    "for word, sound in dico.items():\n",
    "\tif len(word) == 1:\n",
    "\t\tletter_sound[word] = list(sound.split(' '))\n",
    "\t\tsyllable_count[word] = len(sound.split(' '))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 392,
   "metadata": {},
   "outputs": [],
   "source": [
    "letter_sound['a'] = ['EY'] # A sounds EY1 not AH0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 393,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({'a': ['EY'],\n",
       "  'b': ['B', 'IY'],\n",
       "  'c': ['S', 'IY'],\n",
       "  'd': ['D', 'IY'],\n",
       "  'e': ['IY'],\n",
       "  'f': ['EH', 'F'],\n",
       "  'g': ['JH', 'IY'],\n",
       "  'h': ['EY', 'CH'],\n",
       "  'i': ['AY'],\n",
       "  'j': ['JH', 'EY'],\n",
       "  'k': ['K', 'EY'],\n",
       "  'l': ['EH', 'L'],\n",
       "  'm': ['EH', 'M'],\n",
       "  'n': ['EH', 'N'],\n",
       "  'o': ['OW'],\n",
       "  'p': ['P', 'IY'],\n",
       "  'q': ['K', 'Y', 'UW'],\n",
       "  'r': ['AA', 'R'],\n",
       "  's': ['EH', 'S'],\n",
       "  't': ['T', 'IY'],\n",
       "  'u': ['Y', 'UW'],\n",
       "  'v': ['V', 'IY'],\n",
       "  'w': ['D', 'AH', 'B', 'AH', 'L', 'Y', 'UW'],\n",
       "  'x': ['EH', 'K', 'S'],\n",
       "  'y': ['W', 'AY'],\n",
       "  'z': ['Z', 'IY']},\n",
       " {'a': 1,\n",
       "  'b': 2,\n",
       "  'c': 2,\n",
       "  'd': 2,\n",
       "  'e': 1,\n",
       "  'f': 2,\n",
       "  'g': 2,\n",
       "  'h': 2,\n",
       "  'i': 1,\n",
       "  'j': 2,\n",
       "  'k': 2,\n",
       "  'l': 2,\n",
       "  'm': 2,\n",
       "  'n': 2,\n",
       "  'o': 1,\n",
       "  'p': 2,\n",
       "  'q': 3,\n",
       "  'r': 2,\n",
       "  's': 2,\n",
       "  't': 2,\n",
       "  'u': 2,\n",
       "  'v': 2,\n",
       "  'w': 7,\n",
       "  'x': 3,\n",
       "  'y': 2,\n",
       "  'z': 2})"
      ]
     },
     "execution_count": 393,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "letter_sound, syllable_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 394,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_acronym(word, sound, threshold=2, show = False):\n",
    "\tmatched = 0\n",
    "\toffset = 0\n",
    "\tsound_list = sound.split(' ')\n",
    "\tfor letter in word:\n",
    "\t\tstep = syllable_count[letter]\n",
    "\t\tif show:\n",
    "\t\t\tprint(letter_sound[letter])\n",
    "\t\t\tprint(sound_list[offset:offset+step])\n",
    "\n",
    "\t\tif letter_sound[letter] == sound_list[offset:offset+step]:\n",
    "\t\t\tmatched += 1\n",
    "\t\tif matched >= threshold:\n",
    "\t\t\treturn True\n",
    "\t\toffset += step\n",
    "\treturn False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 395,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['EY']\n",
      "['AA']\n",
      "['EY']\n",
      "['R']\n",
      "['AA', 'R']\n",
      "['D', 'V']\n",
      "['D', 'IY']\n",
      "['AA', 'R']\n",
      "['V', 'IY']\n",
      "['K']\n",
      "['EY']\n",
      "[]\n",
      "['AA', 'R']\n",
      "[]\n",
      "['K', 'EY']\n",
      "[]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(True, False, False)"
      ]
     },
     "execution_count": 395,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "is_acronym('abc',dico['abc']), is_acronym('good', dico['good']), is_acronym('aardvark', dico['aardvark'], show=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 396,
   "metadata": {},
   "outputs": [],
   "source": [
    "acronyms = []\n",
    "for (word, sound) in dico.items():\n",
    "\tif len(word) > 4:\n",
    "\t\tcontinue\n",
    "\tif is_acronym(word, sound, 2):\n",
    "\t\tacronyms.append(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 397,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "114"
      ]
     },
     "execution_count": 397,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(acronyms)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "metadata": {},
   "outputs": [],
   "source": [
    "for word in acronyms:\n",
    "\tdico.pop(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 399,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'T R IH P AH L EY'"
      ]
     },
     "execution_count": 399,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dico.pop('aaa')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 400,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "52008"
      ]
     },
     "execution_count": 400,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dico)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "now we get a well pronounced, wordnet-validated dictionary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 401,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame.from_dict(dico, orient='index', columns = ['sound'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 402,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>COUNT</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>WORD</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>the</th>\n",
       "      <td>4888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>of</th>\n",
       "      <td>4883</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>and</th>\n",
       "      <td>4878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>to</th>\n",
       "      <td>4873</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>in</th>\n",
       "      <td>4868</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bepride</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>batterable</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>batsome</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>adpugne</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>adaration</th>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>186632 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            COUNT\n",
       "WORD             \n",
       "the          4888\n",
       "of           4883\n",
       "and          4878\n",
       "to           4873\n",
       "in           4868\n",
       "...           ...\n",
       "bepride         0\n",
       "batterable      0\n",
       "batsome         0\n",
       "adpugne         0\n",
       "adaration       0\n",
       "\n",
       "[186632 rows x 1 columns]"
      ]
     },
     "execution_count": 402,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_freq = pd.read_csv('./dict_freq_en.yaml', sep = '\\t', index_col=0, usecols=[1,2])\n",
    "df_freq"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 403,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.join(df_freq.astype('Int64'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 404,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.index.name = 'word'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 405,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.rename(columns={'COUNT': 'freq'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 406,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a</th>\n",
       "      <td>AH</td>\n",
       "      <td>4863</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aachen</th>\n",
       "      <td>AA K AH N</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalborg</th>\n",
       "      <td>AO L B AO R G</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalto</th>\n",
       "      <td>AA L T OW</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aardvark</th>\n",
       "      <td>AA R D V AA R K</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zuni</th>\n",
       "      <td>Z UW N IY</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zurich</th>\n",
       "      <td>Z UH R IH K</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zweig</th>\n",
       "      <td>Z W AY G</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zydeco</th>\n",
       "      <td>Z AY D AH K OW</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zygote</th>\n",
       "      <td>Z AY G OW T</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>52008 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    sound  freq\n",
       "word                           \n",
       "a                      AH  4863\n",
       "aachen          AA K AH N  <NA>\n",
       "aalborg     AO L B AO R G  <NA>\n",
       "aalto           AA L T OW  <NA>\n",
       "aardvark  AA R D V AA R K  <NA>\n",
       "...                   ...   ...\n",
       "zuni            Z UW N IY  <NA>\n",
       "zurich        Z UH R IH K  <NA>\n",
       "zweig            Z W AY G  <NA>\n",
       "zydeco     Z AY D AH K OW     0\n",
       "zygote        Z AY G OW T     7\n",
       "\n",
       "[52008 rows x 2 columns]"
      ]
     },
     "execution_count": 406,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transcription"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "def encode(phonemes):\n",
    "\tvowels = re.compile('(AO|UW|EH|AH|AA|IY|IH|UH|AE|AW|AY|ER|EY|OW|OY)\\d*', re.VERBOSE)\n",
    "\tphonemes = re.sub(vowels, 'Q', phonemes)\n",
    "\tcomplex_consonant = re.compile('CH|DH|HH|JH|NG|SH|TH|ZH|')\n",
    "\t# complex consonants\n",
    "\tphonemes = re.sub('HH', 'H', phonemes)\n",
    "\tphonemes = re.sub('CH', 'C', phonemes)\n",
    "\tphonemes = re.sub('JH', 'J', phonemes)\n",
    "\tphonemes = re.sub('TH', 'X', phonemes)\n",
    "\tphonemes = re.sub('DH', 'X', phonemes)\n",
    "\tphonemes = re.sub('SH', 'U', phonemes)\n",
    "\tphonemes = re.sub('ZH', 'O', phonemes)\n",
    "\tphonemes = re.sub('NG', 'I', phonemes)\n",
    "\t\n",
    "\t# remove whitespaces\n",
    "\tphonemes = re.sub(' ', '', phonemes)\n",
    "\n",
    "\treturn phonemes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "def transcribe(phonemes_list):\n",
    "\tencoding_list = []\n",
    "\tfor phonemes in phonemes_list:\n",
    "\t\tencoding_list.append(encode(phonemes))\n",
    "\treturn encoding_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "encoding_list = transcribe(df.sound.astype('str'))\n",
    "df['encoding'] = encoding_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>freq</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a</th>\n",
       "      <td>AH</td>\n",
       "      <td>4863</td>\n",
       "      <td>Q</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aachen</th>\n",
       "      <td>AA K AH N</td>\n",
       "      <td></td>\n",
       "      <td>QKQN</td>\n",
       "      <td>QKN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalborg</th>\n",
       "      <td>AO L B AO R G</td>\n",
       "      <td></td>\n",
       "      <td>QLBQRG</td>\n",
       "      <td>QLBRG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalto</th>\n",
       "      <td>AA L T OW</td>\n",
       "      <td></td>\n",
       "      <td>QLTQ</td>\n",
       "      <td>QLTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aardvark</th>\n",
       "      <td>AA R D V AA R K</td>\n",
       "      <td></td>\n",
       "      <td>QRDVQRK</td>\n",
       "      <td>QRDVRK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zuni</th>\n",
       "      <td>Z UW N IY</td>\n",
       "      <td></td>\n",
       "      <td>ZQNQ</td>\n",
       "      <td>ZNQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zurich</th>\n",
       "      <td>Z UH R IH K</td>\n",
       "      <td></td>\n",
       "      <td>ZQRQK</td>\n",
       "      <td>ZRK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zweig</th>\n",
       "      <td>Z W AY G</td>\n",
       "      <td></td>\n",
       "      <td>ZWQG</td>\n",
       "      <td>ZWG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zydeco</th>\n",
       "      <td>Z AY D AH K OW</td>\n",
       "      <td>0</td>\n",
       "      <td>ZQDQKQ</td>\n",
       "      <td>ZDKQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zygote</th>\n",
       "      <td>Z AY G OW T</td>\n",
       "      <td>7</td>\n",
       "      <td>ZQGQT</td>\n",
       "      <td>ZGT</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>51950 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    sound  freq encoding vowel_free\n",
       "word                                               \n",
       "a                      AH  4863        Q          A\n",
       "aachen          AA K AH N           QKQN        QKN\n",
       "aalborg     AO L B AO R G         QLBQRG      QLBRG\n",
       "aalto           AA L T OW           QLTQ       QLTQ\n",
       "aardvark  AA R D V AA R K        QRDVQRK     QRDVRK\n",
       "...                   ...   ...      ...        ...\n",
       "zuni            Z UW N IY           ZQNQ        ZNQ\n",
       "zurich        Z UH R IH K          ZQRQK        ZRK\n",
       "zweig            Z W AY G           ZWQG        ZWG\n",
       "zydeco     Z AY D AH K OW     0   ZQDQKQ       ZDKQ\n",
       "zygote        Z AY G OW T     7    ZQGQT        ZGT\n",
       "\n",
       "[51950 rows x 4 columns]"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## delete non-initial vowels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_vowels(encoding_list):\n",
    "\tres = []\n",
    "\tfor encoding in encoding_list:\n",
    "\t\tif len(encoding) > 1:\n",
    "\t\t\tres.append(encoding[0] + re.sub('Q','', encoding[1:-1]) + encoding[-1])\n",
    "\t\telse:\n",
    "\t\t\tres.append(encoding)\n",
    "\treturn res"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Detect inefficient encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 420,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['vowel_free'] = remove_vowels(df.encoding) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>aug</th>\n",
       "      <td>AO G AH S T</td>\n",
       "      <td>QGQST</td>\n",
       "      <td>QGST</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ct</th>\n",
       "      <td>K AO R T</td>\n",
       "      <td>KQRT</td>\n",
       "      <td>KRT</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>emu</th>\n",
       "      <td>IY M Y UW</td>\n",
       "      <td>QMYQ</td>\n",
       "      <td>QMYQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>expo</th>\n",
       "      <td>EH K S P OW</td>\n",
       "      <td>QKSPQ</td>\n",
       "      <td>QKSPQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feb</th>\n",
       "      <td>F EH B Y AH W EH R IY</td>\n",
       "      <td>FQBYQWQRQ</td>\n",
       "      <td>FBYWRQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hz</th>\n",
       "      <td>HH ER T Z</td>\n",
       "      <td>HQTZ</td>\n",
       "      <td>HTZ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>jr</th>\n",
       "      <td>JH UW N Y ER</td>\n",
       "      <td>JQNYQ</td>\n",
       "      <td>JNYQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lb</th>\n",
       "      <td>P AW N D</td>\n",
       "      <td>PQND</td>\n",
       "      <td>PND</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lbs</th>\n",
       "      <td>P AW N D Z</td>\n",
       "      <td>PQNDS</td>\n",
       "      <td>PNDS</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lxi</th>\n",
       "      <td>L K S IY</td>\n",
       "      <td>LKSQ</td>\n",
       "      <td>LKSQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mg</th>\n",
       "      <td>EH M G IY</td>\n",
       "      <td>QMGQ</td>\n",
       "      <td>QMGQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mr</th>\n",
       "      <td>M IH S T ER</td>\n",
       "      <td>MQSTQ</td>\n",
       "      <td>MSTQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mt</th>\n",
       "      <td>M AW N T</td>\n",
       "      <td>MQNT</td>\n",
       "      <td>MNT</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ngo</th>\n",
       "      <td>EH NG G OW</td>\n",
       "      <td>QIGQ</td>\n",
       "      <td>QIGQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sr</th>\n",
       "      <td>S IY N Y ER</td>\n",
       "      <td>SQNYQ</td>\n",
       "      <td>SNYQ</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      sound   encoding vowel_free  freq\n",
       "word                                                   \n",
       "aug             AO G AH S T      QGQST       QGST  <NA>\n",
       "ct                 K AO R T       KQRT        KRT  <NA>\n",
       "emu               IY M Y UW       QMYQ       QMYQ  <NA>\n",
       "expo            EH K S P OW      QKSPQ      QKSPQ  <NA>\n",
       "feb   F EH B Y AH W EH R IY  FQBYQWQRQ     FBYWRQ  <NA>\n",
       "hz                HH ER T Z       HQTZ        HTZ  <NA>\n",
       "jr             JH UW N Y ER      JQNYQ       JNYQ  <NA>\n",
       "lb                 P AW N D       PQND        PND  <NA>\n",
       "lbs              P AW N D Z      PQNDS       PNDS  <NA>\n",
       "lxi                L K S IY       LKSQ       LKSQ  <NA>\n",
       "mg                EH M G IY       QMGQ       QMGQ  <NA>\n",
       "mr              M IH S T ER      MQSTQ       MSTQ  <NA>\n",
       "mt                 M AW N T       MQNT        MNT  <NA>\n",
       "ngo              EH NG G OW       QIGQ       QIGQ  <NA>\n",
       "sr              S IY N Y ER      SQNYQ       SNYQ  <NA>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# with pd.option_context('display.max_rows', None):\n",
    "acronym_residue = df.loc[(df.index.str.len() > 1) & (df.index.str.len() <= 4) & (df.index.str.len() < df.vowel_free.str.len()) & (~df.freq.notna()), ['sound','encoding','vowel_free','freq']]\n",
    "display(acronym_residue)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 422,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(acronym_residue.index, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 423,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "51993"
      ]
     },
     "execution_count": 423,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that many of them are acronyms, which is pronounced as individual letters, we could remove them directly from the cmu dict, the acronyms will be dealed in another independent dico."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 426,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>freq</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a</th>\n",
       "      <td>AH</td>\n",
       "      <td>4863</td>\n",
       "      <td>Q</td>\n",
       "      <td>Q</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aachen</th>\n",
       "      <td>AA K AH N</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>QKQN</td>\n",
       "      <td>QKN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalborg</th>\n",
       "      <td>AO L B AO R G</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>QLBQRG</td>\n",
       "      <td>QLBRG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalto</th>\n",
       "      <td>AA L T OW</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>QLTQ</td>\n",
       "      <td>QLTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aardvark</th>\n",
       "      <td>AA R D V AA R K</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>QRDVQRK</td>\n",
       "      <td>QRDVRK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zuni</th>\n",
       "      <td>Z UW N IY</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>ZQNQ</td>\n",
       "      <td>ZNQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zurich</th>\n",
       "      <td>Z UH R IH K</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>ZQRQK</td>\n",
       "      <td>ZRK</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zweig</th>\n",
       "      <td>Z W AY G</td>\n",
       "      <td>&lt;NA&gt;</td>\n",
       "      <td>ZWQG</td>\n",
       "      <td>ZWG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zydeco</th>\n",
       "      <td>Z AY D AH K OW</td>\n",
       "      <td>0</td>\n",
       "      <td>ZQDQKQ</td>\n",
       "      <td>ZDKQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zygote</th>\n",
       "      <td>Z AY G OW T</td>\n",
       "      <td>7</td>\n",
       "      <td>ZQGQT</td>\n",
       "      <td>ZGT</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>51993 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    sound  freq encoding vowel_free\n",
       "word                                               \n",
       "a                      AH  4863        Q          Q\n",
       "aachen          AA K AH N  <NA>     QKQN        QKN\n",
       "aalborg     AO L B AO R G  <NA>   QLBQRG      QLBRG\n",
       "aalto           AA L T OW  <NA>     QLTQ       QLTQ\n",
       "aardvark  AA R D V AA R K  <NA>  QRDVQRK     QRDVRK\n",
       "...                   ...   ...      ...        ...\n",
       "zuni            Z UW N IY  <NA>     ZQNQ        ZNQ\n",
       "zurich        Z UH R IH K  <NA>    ZQRQK        ZRK\n",
       "zweig            Z W AY G  <NA>     ZWQG        ZWG\n",
       "zydeco     Z AY D AH K OW     0   ZQDQKQ       ZDKQ\n",
       "zygote        Z AY G OW T     7    ZQGQT        ZGT\n",
       "\n",
       "[51993 rows x 4 columns]"
      ]
     },
     "execution_count": 426,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 427,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "sound         AH B R IY V IY EY T IH D\n",
       "freq                                50\n",
       "encoding                    QBRQVQQTQE\n",
       "vowel_free                      QBRVTE\n",
       "Name: abbreviated, dtype: object"
      ]
     },
     "execution_count": 427,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc['abbreviated']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Encoding Statistics"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "encoding length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_string_length_freq(encodings: pd.DataFrame):\n",
    "\twd_length = encodings.reset_index().apply(lambda x: x.str.len(), axis=0).astype(pd.Int64Dtype())\n",
    "\tdisplay(wd_length.describe(percentiles=[.9,.95,.99]).drop(['count','50%','min']))\n",
    "\n",
    "\twd_length_count = wd_length.melt()\n",
    "\twd_length_count = wd_length_count.pivot_table(index='value', columns='variable', values='value', aggfunc='size')\n",
    "\tplt.style.use('seaborn-v0_8-notebook')\n",
    "\twd_length_count.plot(kind='bar')\n",
    "\t"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- less average encoding length: lazier\n",
    "- less standard deviation of encoding length: evener"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plot_string_length_freq(df[['encoding', 'vowel_free']])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "coincident codes: counts of words with same encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def coincident_stats(encodings: pd.DataFrame, ext = 10):\n",
    "\ttotal = len(encodings)\n",
    "\n",
    "\tmultiplicity = encodings.value_counts()\n",
    "\tmultiplicity.index.name = 'encoding'\n",
    "\tmultiplicity.name = 'multiplicity'\n",
    "\n",
    "\tmultiplicity_freq = multiplicity.value_counts()\n",
    "\tmultiplicity_freq.index.name = 'multiplicity'\n",
    "\tmultiplicity_freq.name = 'freq'\n",
    "\n",
    "\tcoincident_rate = 1 - (multiplicity_freq.loc[1]) / multiplicity_freq.sum()\n",
    "\n",
    "\textremes = multiplicity_freq.loc[ext+1:]\n",
    "\tword_count_with_extreme_encodings = (extremes.index.to_numpy() * extremes.values).sum()\n",
    "\tratio_with_extreme_encodings = np.round( word_count_with_extreme_encodings/total *100 , 2)\n",
    "\n",
    "\tprint(f'coincident_rate: {np.round(coincident_rate * 100,2)}%')\n",
    "\tprint(f'There are {extremes.sum()} encodings have more than {ext} multiplicities, encoding {word_count_with_extreme_encodings}({ratio_with_extreme_encodings}%) words.')\n",
    "\n",
    "\tq = multiplicity.quantile([.8,.9,.95,.99])\n",
    "\tq.index.name = 'quantile'\n",
    "\tq.name = 'multiplicity'\n",
    "\n",
    "\tprint('extreme:')\n",
    "\tdisplay(extremes.to_frame())\n",
    "\tprint('quantile:')\n",
    "\tdisplay(q.to_frame())\n",
    "\n",
    "\textreme_encoding = pd.Series(multiplicity[multiplicity > ext].index)\n",
    "\textreme_encoding.name = 'freq'\n",
    "\textreme_encoding.index.name = 'encoding_length'\n",
    "\tif len(extreme_encoding) > 0:\n",
    "\t\t# display(extreme_encoding)\n",
    "\t\tsn.barplot(x = extreme_encoding.str.len().value_counts().index, y = extreme_encoding.str.len().value_counts())\n",
    "\t# plt.hist(multiplicity,bins=range(min(multiplicity), max(multiplicity)+1,1))\n",
    "\treturn multiplicity, multiplicity_freq, extreme_encoding"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "no coincidence in original dictionary lexicon(of course)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "_ = coincident_stats(df.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "_ = coincident_stats(df.encoding)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "multiplicity, _, extreme_encoding = coincident_stats(df.vowel_free)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "let's taker a careful look at the extreme encodings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "extreme_encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "multiplicity[multiplicity > 10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "with pd.option_context('display.max_rows', None, 'display.max_columns', None):\n",
    "\tdisplay(multiplicity[multiplicity > 10])\n",
    "\tdisplay(df[df.vowel_free.isin(['BLS','FLS'])])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalization\n",
    "\n",
    "We hope that:\n",
    "1. the plurals and the third person single form are uniformly terminated with S(instead of Z)\n",
    "2. the past particles are uniformly terminated with E(instead of T or D)\n",
    "3. the present particles are uniformly terminated with I(this is already done since we encode NG sound by I)\n",
    "\n",
    "you can take these terminations as a grammatical symbol"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "word\n",
       "aardvarks           QRDVQRKS\n",
       "aarhus                  QHQS\n",
       "abacus                QBQKQS\n",
       "abalones            QBQLQNQZ\n",
       "abandonments    QBQNDQNMQNTS\n",
       "                    ...     \n",
       "zones                   ZQNZ\n",
       "zoologists        ZQQLQJQSTS\n",
       "zooms                   ZQMZ\n",
       "zoos                     ZQZ\n",
       "zucchinis            ZQKQNQZ\n",
       "Name: encoding, Length: 13296, dtype: object"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.encoding.loc[(df.index.str.contains('s$', regex=True) == True)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "word\n",
       "abandoned        QBQNDQND\n",
       "abashed             QBQUT\n",
       "abated             QBQTQD\n",
       "abbreviated    QBRQVQQTQD\n",
       "abdicated       QBDQKQTQD\n",
       "                  ...    \n",
       "zigzagged         ZQGZQGD\n",
       "zipped               ZQPT\n",
       "zippered            ZQPQD\n",
       "zoned                ZQND\n",
       "zoomed               ZQMD\n",
       "Name: encoding, Length: 4701, dtype: object"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.encoding.loc[(df.index.str.contains('ed$', regex=True) == True)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "word\n",
       "abandoning       QBQNDQNQI\n",
       "abating             QBQTQI\n",
       "abbreviating    QBRQVQQTQI\n",
       "abdicating       QBDQKQTQI\n",
       "abducting         QBDQKTQI\n",
       "                   ...    \n",
       "zing                   ZQI\n",
       "zippering          ZQPQRQI\n",
       "zipping              ZQPQI\n",
       "zoning               ZQNQI\n",
       "zooming              ZQMQI\n",
       "Name: encoding, Length: 4323, dtype: object"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.encoding.loc[(df.index.str.contains('ing$', regex=True) == True)]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Normalize S-form\n",
    "\n",
    "notice that most of the extreme encodings have the `S` termination, which is due to enormous derivations in plurals/single third person form (we call them S-form). So it's better to spare another grammatical termination for them.\n",
    "\n",
    "To do so, we must first recognize the S-form, there are two necessary criterion (yet not proved to be sufficient, perhaps some risks of mis-recognize?)\n",
    "\n",
    "1. phonetic criterion: end with s|z\n",
    "2. morphological criterion: word differs its lemma by s(es,ies et etc)\n",
    "\n",
    "notice that we don't account for irregular words.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['aardvarks', 'aarhus', 'abacus', 'abalones', 'abandonments', 'abandons',\n",
       "       'abatements', 'abates', 'abbreviates', 'abbreviations',\n",
       "       ...\n",
       "       'zirconiums', 'zircons', 'zithers', 'zlotys', 'zombies', 'zones',\n",
       "       'zoologists', 'zooms', 'zoos', 'zucchinis'],\n",
       "      dtype='object', name='word', length=13272)"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "words_sound_end_with_s = df.loc[df.index.str.contains('s$',regex=True) & df.sound.str.contains('[SZ]$',regex=True)].index\n",
    "words_sound_end_with_s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_suffix(word, lemma):\n",
    "    cnt = sum(w==l for w, l in zip(word, lemma))\n",
    "    return word[cnt:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_s_form(word, non_s_forms: list):\n",
    "\tfor lemmas_per_pos in m(word).values():\n",
    "\t\tfor lemma in lemmas_per_pos:\n",
    "\t\t\tsuffix = get_suffix(word, lemma)\n",
    "\t\t\tif suffix in ['s','es','ies', 'ves', 'ces', 'ses', 'ices', 'zes']:\n",
    "\t\t\t# if 's' in suffix :\n",
    "\t\t\t\treturn True\n",
    "\tnon_s_forms.append(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 13272/13272 [00:00<00:00, 31704.99it/s]\n"
     ]
    }
   ],
   "source": [
    "non_s_forms = []\n",
    "for word in tqdm(words_sound_end_with_s):\n",
    "\tis_s_form(word, non_s_forms)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1820"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(non_s_forms)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "some rare s-forms we exclude them"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'n': {'pons'}}"
      ]
     },
     "execution_count": 121,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m('pontes')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "simple verification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(False, True)"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'produces' in non_s_forms, 'bliss' in non_s_forms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {},
   "outputs": [],
   "source": [
    "s_form_candidates = words_sound_end_with_s.to_list()\n",
    "for word in non_s_forms:\n",
    "\ts_form_candidates.remove(word)\n",
    "\n",
    "s_forms = pd.Index(s_form_candidates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['aardvarks', 'abalones', 'abandonments', 'abandons', 'abatements',\n",
       "       'abates', 'abbreviates', 'abbreviations', 'abdicates', 'abductions',\n",
       "       ...\n",
       "       'zirconiums', 'zircons', 'zithers', 'zlotys', 'zombies', 'zones',\n",
       "       'zoologists', 'zooms', 'zoos', 'zucchinis'],\n",
       "      dtype='object', length=11452)"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s_forms"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "reencoding s-forms by `A`-termination"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.loc[s_forms, 'encoding'] = df.loc[s_forms, 'encoding'].str.replace('[SZ]$', 'A', regex=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "aardvarks           QRDVQRKA\n",
       "abalones            QBQLQNQA\n",
       "abandonments    QBQNDQNMQNTA\n",
       "abandons            QBQNDQNA\n",
       "abatements         QBQTMQNTA\n",
       "                    ...     \n",
       "zones                   ZQNA\n",
       "zoologists        ZQQLQJQSTA\n",
       "zooms                   ZQMA\n",
       "zoos                     ZQA\n",
       "zucchinis            ZQKQNQA\n",
       "Name: encoding, Length: 11452, dtype: object"
      ]
     },
     "execution_count": 129,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc[s_forms, 'encoding']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Normalize past particle"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "words terminates with `-ed` consists also a large part of english vocabulary. Likewise, we need to recognize the past particle from all `-ed` form. The criterion is:\n",
    "\n",
    "1. phonetic criterion: end with t|d\n",
    "2. morphological criterion: word differs its lemma by ed(or d)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['abandoned', 'abashed', 'abated', 'abbreviated', 'abdicated',\n",
       "       'abducted', 'abed', 'abetted', 'abhorred', 'abided',\n",
       "       ...\n",
       "       'yielded', 'yoked', 'zapped', 'zed', 'zeroed', 'zigzagged', 'zipped',\n",
       "       'zippered', 'zoned', 'zoomed'],\n",
       "      dtype='object', name='word', length=4701)"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "words_sound_end_with_t_or_d = df.loc[df.index.str.contains('ed$',regex=True) & df.sound.str.contains('[TD]$',regex=True)].index\n",
    "words_sound_end_with_t_or_d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_past_particle(word, non_past_particles: list):\n",
    "\tfor lemmas_per_pos in m(word).values():\n",
    "\t\tfor lemma in lemmas_per_pos:\n",
    "\t\t\tsuffix = get_suffix(word, lemma)\n",
    "\t\t\tif 'd' in suffix:\n",
    "\t\t\t\treturn True\n",
    "\tnon_past_particles.append(word)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({'v': {'fee', 'feed'}}, {'v': {'need'}})"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m('feed','v'), m('need','v')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 4701/4701 [00:00<00:00, 25226.03it/s]\n"
     ]
    }
   ],
   "source": [
    "non_past_particles = []\n",
    "for word in tqdm(words_sound_end_with_t_or_d):\n",
    "\tis_past_particle(word, non_past_particles)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "446"
      ]
     },
     "execution_count": 142,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(non_past_particles)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'embed' in non_past_particles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 146,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'feed' in non_past_particles"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "some common word may be the past particle of a rare word, e.g. `fee-feed`, so there are some particular case need to be treated separately."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {},
   "outputs": [],
   "source": [
    "non_past_particles.append('feed')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "reencoding past particle by 'E'-termination"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [],
   "source": [
    "past_particle_candidate = words_sound_end_with_t_or_d.to_list()\n",
    "for word in  non_past_particles:\n",
    "\tpast_particle_candidate.remove(word)\n",
    "\n",
    "past_particles = pd.Index(past_particle_candidate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['abandoned', 'abashed', 'abated', 'abbreviated', 'abdicated',\n",
       "       'abducted', 'abetted', 'abhorred', 'abided', 'abolished',\n",
       "       ...\n",
       "       'yelped', 'yielded', 'yoked', 'zapped', 'zeroed', 'zigzagged', 'zipped',\n",
       "       'zippered', 'zoned', 'zoomed'],\n",
       "      dtype='object', length=4254)"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "past_particles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.loc[past_particles, 'encoding'] = df.loc[past_particles, 'encoding'].str.replace('[TD]$', 'E', regex=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['vowel_free'] = remove_vowels(df.encoding)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "word\n",
       "a                Q\n",
       "aachen         QKN\n",
       "aalborg      QLBRG\n",
       "aalto         QLTQ\n",
       "aardvark    QRDVRK\n",
       "             ...  \n",
       "zuni           ZNQ\n",
       "zurich         ZRK\n",
       "zweig          ZWG\n",
       "zydeco        ZDKQ\n",
       "zygote         ZGT\n",
       "Name: vowel_free, Length: 51950, dtype: object"
      ]
     },
     "execution_count": 155,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['vowel_free']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Manually correct exceptions"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Type letters, 2-grams, 3-grams by themselve  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "sound         HH AY\n",
       "freq               \n",
       "encoding         HQ\n",
       "vowel_free       HQ\n",
       "Name: hi, dtype: object"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.loc['hi']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [],
   "source": [
    "rare_2_grams =  df.loc[(df.index.str.len() == 2) & (~df.freq.notna()), ['encoding','freq']].drop('hi')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 443,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(rare_2_grams.index, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.loc[(df.index.str.len() <= 3), 'vowel_free'] = df.index[(df.index.str.len() <= 3)].str.upper()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### reexaminate stats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>word</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>7.930279</td>\n",
       "      <td>6.837421</td>\n",
       "      <td>4.642964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.452292</td>\n",
       "      <td>2.371894</td>\n",
       "      <td>1.509648</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90%</th>\n",
       "      <td>11.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95%</th>\n",
       "      <td>12.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>7.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99%</th>\n",
       "      <td>14.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>20.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>12.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          word  encoding  vowel_free\n",
       "mean  7.930279  6.837421    4.642964\n",
       "std   2.452292  2.371894    1.509648\n",
       "90%       11.0      10.0         7.0\n",
       "95%       12.0      11.0         7.0\n",
       "99%       14.0      13.0         9.0\n",
       "max       20.0      18.0        12.0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_string_length_freq(df[['encoding', 'vowel_free']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 446,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "coincident_rate: 18.0%\n",
      "There are 269 encodings have more than 10 multiplicities, encoding 3634(7.0%) words.\n",
      "extreme:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>freq</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>multiplicity</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>53</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              freq\n",
       "multiplicity      \n",
       "11              53\n",
       "10              52\n",
       "12              40\n",
       "14              30\n",
       "13              29\n",
       "15              18\n",
       "16              10\n",
       "17               6\n",
       "18               5\n",
       "19               5\n",
       "20               4\n",
       "21               3\n",
       "22               2\n",
       "25               2\n",
       "34               1\n",
       "24               1\n",
       "26               1\n",
       "27               1\n",
       "28               1\n",
       "30               1\n",
       "31               1\n",
       "32               1\n",
       "33               1\n",
       "35               1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "quantile:\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>multiplicity</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>quantile</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.80</th>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.90</th>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.95</th>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.99</th>\n",
       "      <td>8.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          multiplicity\n",
       "quantile              \n",
       "0.80               1.0\n",
       "0.90               2.0\n",
       "0.95               3.0\n",
       "0.99               8.0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x550 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "multiplicity, _, extreme_encoding = coincident_stats(df.vowel_free)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 447,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "encoding_length\n",
       "0       FLQ\n",
       "1       FLA\n",
       "2       BRQ\n",
       "3       BLA\n",
       "4       KLQ\n",
       "       ... \n",
       "212    PRTQ\n",
       "213     SKL\n",
       "214     PLI\n",
       "215     STI\n",
       "216     KRT\n",
       "Name: freq, Length: 217, dtype: object"
      ]
     },
     "execution_count": 447,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "extreme_encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 448,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "encoding\n",
       "FLQ     35\n",
       "FLA     34\n",
       "BRQ     33\n",
       "BLA     32\n",
       "KLQ     31\n",
       "        ..\n",
       "PRTQ    11\n",
       "SKL     11\n",
       "PLI     11\n",
       "STI     11\n",
       "KRT     11\n",
       "Name: multiplicity, Length: 217, dtype: int64"
      ]
     },
     "execution_count": 448,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multiplicity[multiplicity > 10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.to_csv('./dict_en.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df= pd.read_csv('./dict_en.csv', sep = ',', index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>freq</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>a</th>\n",
       "      <td>AH</td>\n",
       "      <td>4863.0</td>\n",
       "      <td>Q</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aachen</th>\n",
       "      <td>AA K AH N</td>\n",
       "      <td>NaN</td>\n",
       "      <td>QKQN</td>\n",
       "      <td>QKN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalborg</th>\n",
       "      <td>AO L B AO R G</td>\n",
       "      <td>NaN</td>\n",
       "      <td>QLBQRG</td>\n",
       "      <td>QLBRG</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aalto</th>\n",
       "      <td>AA L T OW</td>\n",
       "      <td>NaN</td>\n",
       "      <td>QLTQ</td>\n",
       "      <td>QLTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>aardvark</th>\n",
       "      <td>AA R D V AA R K</td>\n",
       "      <td>NaN</td>\n",
       "      <td>QRDVQRK</td>\n",
       "      <td>QRDVRK</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    sound    freq encoding vowel_free\n",
       "word                                                 \n",
       "a                      AH  4863.0        Q          A\n",
       "aachen          AA K AH N     NaN     QKQN        QKN\n",
       "aalborg     AO L B AO R G     NaN   QLBQRG      QLBRG\n",
       "aalto           AA L T OW     NaN     QLTQ       QLTQ\n",
       "aardvark  AA R D V AA R K     NaN  QRDVQRK     QRDVRK"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.vowel_free = df.vowel_free.str.lower()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.loc[:,['vowel_free','freq']].to_csv('./dict_en.yaml', sep ='\\t')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Convenient abbr for common words use"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 458,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>sound</th>\n",
       "      <th>freq</th>\n",
       "      <th>encoding</th>\n",
       "      <th>vowel_free</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>word</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>able</th>\n",
       "      <td>EY B AH L</td>\n",
       "      <td>2940</td>\n",
       "      <td>QBQL</td>\n",
       "      <td>QBL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age</th>\n",
       "      <td>EY JH</td>\n",
       "      <td>3253</td>\n",
       "      <td>QJ</td>\n",
       "      <td>AGE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>air</th>\n",
       "      <td>EH R</td>\n",
       "      <td>3492</td>\n",
       "      <td>QR</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>all</th>\n",
       "      <td>AO L</td>\n",
       "      <td>4718</td>\n",
       "      <td>QL</td>\n",
       "      <td>ALL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>also</th>\n",
       "      <td>AO L S OW</td>\n",
       "      <td>4553</td>\n",
       "      <td>QLSQ</td>\n",
       "      <td>QLSQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>and</th>\n",
       "      <td>AH N D</td>\n",
       "      <td>4878</td>\n",
       "      <td>QND</td>\n",
       "      <td>AND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>any</th>\n",
       "      <td>EH N IY</td>\n",
       "      <td>4613</td>\n",
       "      <td>QNQ</td>\n",
       "      <td>ANY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>are</th>\n",
       "      <td>AA R</td>\n",
       "      <td>4788</td>\n",
       "      <td>QR</td>\n",
       "      <td>ARE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>area</th>\n",
       "      <td>EH R IY AH</td>\n",
       "      <td>3820</td>\n",
       "      <td>QRQQ</td>\n",
       "      <td>QRQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>art</th>\n",
       "      <td>AA R T</td>\n",
       "      <td>2728</td>\n",
       "      <td>QRT</td>\n",
       "      <td>ART</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>away</th>\n",
       "      <td>AH W EY</td>\n",
       "      <td>3290</td>\n",
       "      <td>QWQ</td>\n",
       "      <td>QWQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>best</th>\n",
       "      <td>B EH S T</td>\n",
       "      <td>3552</td>\n",
       "      <td>BQST</td>\n",
       "      <td>BST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>body</th>\n",
       "      <td>B AA D IY</td>\n",
       "      <td>3399</td>\n",
       "      <td>BQDQ</td>\n",
       "      <td>BDQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>but</th>\n",
       "      <td>B AH T</td>\n",
       "      <td>4748</td>\n",
       "      <td>BQT</td>\n",
       "      <td>BUT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>can</th>\n",
       "      <td>K AE N</td>\n",
       "      <td>4683</td>\n",
       "      <td>KQN</td>\n",
       "      <td>CAN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cent</th>\n",
       "      <td>S EH N T</td>\n",
       "      <td>3214</td>\n",
       "      <td>SQNT</td>\n",
       "      <td>SNT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city</th>\n",
       "      <td>S IH T IY</td>\n",
       "      <td>4010</td>\n",
       "      <td>SQTQ</td>\n",
       "      <td>STQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cost</th>\n",
       "      <td>K AA S T</td>\n",
       "      <td>3472</td>\n",
       "      <td>KQST</td>\n",
       "      <td>KST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cut</th>\n",
       "      <td>K AH T</td>\n",
       "      <td>2054</td>\n",
       "      <td>KQT</td>\n",
       "      <td>CUT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>data</th>\n",
       "      <td>D EY T AH</td>\n",
       "      <td>3955</td>\n",
       "      <td>DQTQ</td>\n",
       "      <td>DTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>day</th>\n",
       "      <td>D EY</td>\n",
       "      <td>4251</td>\n",
       "      <td>DQ</td>\n",
       "      <td>DAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>did</th>\n",
       "      <td>D IH D</td>\n",
       "      <td>4383</td>\n",
       "      <td>DQD</td>\n",
       "      <td>DID</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>due</th>\n",
       "      <td>D UW</td>\n",
       "      <td>3369</td>\n",
       "      <td>DQ</td>\n",
       "      <td>DUE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>duty</th>\n",
       "      <td>D UW T IY</td>\n",
       "      <td>2085</td>\n",
       "      <td>DQTQ</td>\n",
       "      <td>DTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>end</th>\n",
       "      <td>EH N D</td>\n",
       "      <td>3925</td>\n",
       "      <td>QND</td>\n",
       "      <td>END</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>even</th>\n",
       "      <td>IY V IH N</td>\n",
       "      <td>4318</td>\n",
       "      <td>QVQN</td>\n",
       "      <td>QVN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ever</th>\n",
       "      <td>EH V ER</td>\n",
       "      <td>3072</td>\n",
       "      <td>QVQ</td>\n",
       "      <td>QVQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fact</th>\n",
       "      <td>F AE K T</td>\n",
       "      <td>3894</td>\n",
       "      <td>FQKT</td>\n",
       "      <td>FKT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>far</th>\n",
       "      <td>F AA R</td>\n",
       "      <td>3685</td>\n",
       "      <td>FQR</td>\n",
       "      <td>FAR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>felt</th>\n",
       "      <td>F EH L T</td>\n",
       "      <td>2302</td>\n",
       "      <td>FQLT</td>\n",
       "      <td>FLT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>few</th>\n",
       "      <td>F Y UW</td>\n",
       "      <td>3850</td>\n",
       "      <td>FYQ</td>\n",
       "      <td>FEW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>find</th>\n",
       "      <td>F AY N D</td>\n",
       "      <td>3590</td>\n",
       "      <td>FQND</td>\n",
       "      <td>FND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>for</th>\n",
       "      <td>F AO R</td>\n",
       "      <td>4848</td>\n",
       "      <td>FQR</td>\n",
       "      <td>FOR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>form</th>\n",
       "      <td>F AO R M</td>\n",
       "      <td>3970</td>\n",
       "      <td>FQRM</td>\n",
       "      <td>FRM</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>free</th>\n",
       "      <td>F R IY</td>\n",
       "      <td>3387</td>\n",
       "      <td>FRQ</td>\n",
       "      <td>FRQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>from</th>\n",
       "      <td>F R AH M</td>\n",
       "      <td>4778</td>\n",
       "      <td>FRQM</td>\n",
       "      <td>FRM</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gas</th>\n",
       "      <td>G AE S</td>\n",
       "      <td>2537</td>\n",
       "      <td>GQS</td>\n",
       "      <td>GAS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>get</th>\n",
       "      <td>G EH T</td>\n",
       "      <td>3920</td>\n",
       "      <td>GQT</td>\n",
       "      <td>GET</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>got</th>\n",
       "      <td>G AA T</td>\n",
       "      <td>2652</td>\n",
       "      <td>GQT</td>\n",
       "      <td>GOT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>had</th>\n",
       "      <td>HH AE D</td>\n",
       "      <td>4728</td>\n",
       "      <td>HQD</td>\n",
       "      <td>HAD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hand</th>\n",
       "      <td>HH AE N D</td>\n",
       "      <td>3747</td>\n",
       "      <td>HQND</td>\n",
       "      <td>HND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hard</th>\n",
       "      <td>HH AA R D</td>\n",
       "      <td>2205</td>\n",
       "      <td>HQRD</td>\n",
       "      <td>HRD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>has</th>\n",
       "      <td>HH AE Z</td>\n",
       "      <td>4708</td>\n",
       "      <td>HQA</td>\n",
       "      <td>HAS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>held</th>\n",
       "      <td>HH EH L D</td>\n",
       "      <td>3531</td>\n",
       "      <td>HQLD</td>\n",
       "      <td>HLD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>help</th>\n",
       "      <td>HH EH L P</td>\n",
       "      <td>3078</td>\n",
       "      <td>HQLP</td>\n",
       "      <td>HLP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>her</th>\n",
       "      <td>HH ER</td>\n",
       "      <td>4658</td>\n",
       "      <td>HQ</td>\n",
       "      <td>HER</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>him</th>\n",
       "      <td>HH IH M</td>\n",
       "      <td>4538</td>\n",
       "      <td>HQM</td>\n",
       "      <td>HIM</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>his</th>\n",
       "      <td>HH IH Z</td>\n",
       "      <td>4763</td>\n",
       "      <td>HQA</td>\n",
       "      <td>HIS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>into</th>\n",
       "      <td>IH N T UW</td>\n",
       "      <td>4568</td>\n",
       "      <td>QNTQ</td>\n",
       "      <td>QNTQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>its</th>\n",
       "      <td>IH T S</td>\n",
       "      <td>4633</td>\n",
       "      <td>QTS</td>\n",
       "      <td>ITS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>just</th>\n",
       "      <td>JH AH S T</td>\n",
       "      <td>4159</td>\n",
       "      <td>JQST</td>\n",
       "      <td>JST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>kind</th>\n",
       "      <td>K AY N D</td>\n",
       "      <td>2848</td>\n",
       "      <td>KQND</td>\n",
       "      <td>KND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>land</th>\n",
       "      <td>L AE N D</td>\n",
       "      <td>3771</td>\n",
       "      <td>LQND</td>\n",
       "      <td>LND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last</th>\n",
       "      <td>L AE S T</td>\n",
       "      <td>4015</td>\n",
       "      <td>LQST</td>\n",
       "      <td>LST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>law</th>\n",
       "      <td>L AO</td>\n",
       "      <td>4197</td>\n",
       "      <td>LQ</td>\n",
       "      <td>LAW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>left</th>\n",
       "      <td>L EH F T</td>\n",
       "      <td>3665</td>\n",
       "      <td>LQFT</td>\n",
       "      <td>LFT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>let</th>\n",
       "      <td>L EH T</td>\n",
       "      <td>3405</td>\n",
       "      <td>LQT</td>\n",
       "      <td>LET</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>list</th>\n",
       "      <td>L IH S T</td>\n",
       "      <td>2179</td>\n",
       "      <td>LQST</td>\n",
       "      <td>LST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lost</th>\n",
       "      <td>L AO S T</td>\n",
       "      <td>2058</td>\n",
       "      <td>LQST</td>\n",
       "      <td>LST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>low</th>\n",
       "      <td>L OW</td>\n",
       "      <td>3238</td>\n",
       "      <td>LQ</td>\n",
       "      <td>LOW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>man</th>\n",
       "      <td>M AE N</td>\n",
       "      <td>4241</td>\n",
       "      <td>MQN</td>\n",
       "      <td>MAN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>many</th>\n",
       "      <td>M EH N IY</td>\n",
       "      <td>4398</td>\n",
       "      <td>MQNQ</td>\n",
       "      <td>MNQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>may</th>\n",
       "      <td>M EY</td>\n",
       "      <td>4653</td>\n",
       "      <td>MQ</td>\n",
       "      <td>MAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>men</th>\n",
       "      <td>M EH N</td>\n",
       "      <td>4053</td>\n",
       "      <td>MQN</td>\n",
       "      <td>MEN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mind</th>\n",
       "      <td>M AY N D</td>\n",
       "      <td>2955</td>\n",
       "      <td>MQND</td>\n",
       "      <td>MND</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>most</th>\n",
       "      <td>M OW S T</td>\n",
       "      <td>4463</td>\n",
       "      <td>MQST</td>\n",
       "      <td>MST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>must</th>\n",
       "      <td>M AH S T</td>\n",
       "      <td>4368</td>\n",
       "      <td>MQST</td>\n",
       "      <td>MST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>new</th>\n",
       "      <td>N UW</td>\n",
       "      <td>4603</td>\n",
       "      <td>NQ</td>\n",
       "      <td>NEW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>next</th>\n",
       "      <td>N EH K S T</td>\n",
       "      <td>3364</td>\n",
       "      <td>NQKST</td>\n",
       "      <td>NKST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>non</th>\n",
       "      <td>N AA N</td>\n",
       "      <td>2910</td>\n",
       "      <td>NQN</td>\n",
       "      <td>NON</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>nor</th>\n",
       "      <td>N AO R</td>\n",
       "      <td>2801</td>\n",
       "      <td>NQR</td>\n",
       "      <td>NOR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>not</th>\n",
       "      <td>N AA T</td>\n",
       "      <td>4808</td>\n",
       "      <td>NQT</td>\n",
       "      <td>NOT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>now</th>\n",
       "      <td>N AW</td>\n",
       "      <td>4433</td>\n",
       "      <td>NQ</td>\n",
       "      <td>NOW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>off</th>\n",
       "      <td>AO F</td>\n",
       "      <td>3825</td>\n",
       "      <td>QF</td>\n",
       "      <td>OFF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>oil</th>\n",
       "      <td>OY L</td>\n",
       "      <td>2730</td>\n",
       "      <td>QL</td>\n",
       "      <td>OIL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>old</th>\n",
       "      <td>OW L D</td>\n",
       "      <td>3975</td>\n",
       "      <td>QLD</td>\n",
       "      <td>OLD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>once</th>\n",
       "      <td>W AH N S</td>\n",
       "      <td>3354</td>\n",
       "      <td>WQNS</td>\n",
       "      <td>WNS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>one</th>\n",
       "      <td>W AH N</td>\n",
       "      <td>4713</td>\n",
       "      <td>WQN</td>\n",
       "      <td>ONE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>only</th>\n",
       "      <td>OW N L IY</td>\n",
       "      <td>4573</td>\n",
       "      <td>QNLQ</td>\n",
       "      <td>QNLQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>open</th>\n",
       "      <td>OW P AH N</td>\n",
       "      <td>3110</td>\n",
       "      <td>QPQN</td>\n",
       "      <td>QPN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>our</th>\n",
       "      <td>AW ER</td>\n",
       "      <td>4518</td>\n",
       "      <td>QQ</td>\n",
       "      <td>OUR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>out</th>\n",
       "      <td>AW T</td>\n",
       "      <td>4543</td>\n",
       "      <td>QT</td>\n",
       "      <td>OUT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>over</th>\n",
       "      <td>OW V ER</td>\n",
       "      <td>4458</td>\n",
       "      <td>QVQ</td>\n",
       "      <td>QVQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>own</th>\n",
       "      <td>OW N</td>\n",
       "      <td>4173</td>\n",
       "      <td>QN</td>\n",
       "      <td>OWN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>part</th>\n",
       "      <td>P AA R T</td>\n",
       "      <td>4343</td>\n",
       "      <td>PQRT</td>\n",
       "      <td>PRT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>past</th>\n",
       "      <td>P AE S T</td>\n",
       "      <td>2841</td>\n",
       "      <td>PQST</td>\n",
       "      <td>PST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pay</th>\n",
       "      <td>P EY</td>\n",
       "      <td>2907</td>\n",
       "      <td>PQ</td>\n",
       "      <td>PAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>per</th>\n",
       "      <td>P ER</td>\n",
       "      <td>4278</td>\n",
       "      <td>PQ</td>\n",
       "      <td>PER</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>plan</th>\n",
       "      <td>P L AE N</td>\n",
       "      <td>2967</td>\n",
       "      <td>PLQN</td>\n",
       "      <td>PLN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>post</th>\n",
       "      <td>P OW S T</td>\n",
       "      <td>2139</td>\n",
       "      <td>PQST</td>\n",
       "      <td>PST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>put</th>\n",
       "      <td>P UH T</td>\n",
       "      <td>3461</td>\n",
       "      <td>PQT</td>\n",
       "      <td>PUT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>red</th>\n",
       "      <td>R EH D</td>\n",
       "      <td>2319</td>\n",
       "      <td>RQE</td>\n",
       "      <td>RED</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>run</th>\n",
       "      <td>R AH N</td>\n",
       "      <td>2098</td>\n",
       "      <td>RQN</td>\n",
       "      <td>RUN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>saw</th>\n",
       "      <td>S AO</td>\n",
       "      <td>2457</td>\n",
       "      <td>SQ</td>\n",
       "      <td>SAW</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>say</th>\n",
       "      <td>S EY</td>\n",
       "      <td>3910</td>\n",
       "      <td>SQ</td>\n",
       "      <td>SAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sea</th>\n",
       "      <td>S IY</td>\n",
       "      <td>2256</td>\n",
       "      <td>SQ</td>\n",
       "      <td>SEA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>see</th>\n",
       "      <td>S IY</td>\n",
       "      <td>4488</td>\n",
       "      <td>SQ</td>\n",
       "      <td>SEE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>self</th>\n",
       "      <td>S EH L F</td>\n",
       "      <td>2817</td>\n",
       "      <td>SQLF</td>\n",
       "      <td>SLF</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>sent</th>\n",
       "      <td>S EH N T</td>\n",
       "      <td>2278</td>\n",
       "      <td>SQNT</td>\n",
       "      <td>SNT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>set</th>\n",
       "      <td>S EH T</td>\n",
       "      <td>3920</td>\n",
       "      <td>SQT</td>\n",
       "      <td>SET</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>she</th>\n",
       "      <td>SH IY</td>\n",
       "      <td>4608</td>\n",
       "      <td>UQ</td>\n",
       "      <td>SHE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>six</th>\n",
       "      <td>S IH K S</td>\n",
       "      <td>2831</td>\n",
       "      <td>SQKS</td>\n",
       "      <td>SIX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>son</th>\n",
       "      <td>S AH N</td>\n",
       "      <td>2370</td>\n",
       "      <td>SQN</td>\n",
       "      <td>SON</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>tax</th>\n",
       "      <td>T AE K S</td>\n",
       "      <td>3144</td>\n",
       "      <td>TQKS</td>\n",
       "      <td>TAX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ten</th>\n",
       "      <td>T EH N</td>\n",
       "      <td>2261</td>\n",
       "      <td>TQN</td>\n",
       "      <td>TEN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>test</th>\n",
       "      <td>T EH S T</td>\n",
       "      <td>2802</td>\n",
       "      <td>TQST</td>\n",
       "      <td>TST</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>the</th>\n",
       "      <td>DH AH</td>\n",
       "      <td>4888</td>\n",
       "      <td>XQ</td>\n",
       "      <td>THE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>told</th>\n",
       "      <td>T OW L D</td>\n",
       "      <td>2725</td>\n",
       "      <td>TQLD</td>\n",
       "      <td>TLD</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>too</th>\n",
       "      <td>T UW</td>\n",
       "      <td>3950</td>\n",
       "      <td>TQ</td>\n",
       "      <td>TOO</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>top</th>\n",
       "      <td>T AA P</td>\n",
       "      <td>2151</td>\n",
       "      <td>TQP</td>\n",
       "      <td>TOP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>true</th>\n",
       "      <td>T R UW</td>\n",
       "      <td>3028</td>\n",
       "      <td>TRQ</td>\n",
       "      <td>TRQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>two</th>\n",
       "      <td>T UW</td>\n",
       "      <td>4558</td>\n",
       "      <td>TQ</td>\n",
       "      <td>TWO</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unit</th>\n",
       "      <td>Y UW N AH T</td>\n",
       "      <td>2190</td>\n",
       "      <td>YQNQT</td>\n",
       "      <td>YNT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>upon</th>\n",
       "      <td>AH P AA N</td>\n",
       "      <td>4363</td>\n",
       "      <td>QPQN</td>\n",
       "      <td>QPN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>use</th>\n",
       "      <td>Y UW S</td>\n",
       "      <td>4348</td>\n",
       "      <td>YQS</td>\n",
       "      <td>USE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>used</th>\n",
       "      <td>Y UW Z D</td>\n",
       "      <td>4328</td>\n",
       "      <td>YQZE</td>\n",
       "      <td>YZE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>very</th>\n",
       "      <td>V EH R IY</td>\n",
       "      <td>4453</td>\n",
       "      <td>VQRQ</td>\n",
       "      <td>VRQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>view</th>\n",
       "      <td>V Y UW</td>\n",
       "      <td>3201</td>\n",
       "      <td>VYQ</td>\n",
       "      <td>VYQ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>want</th>\n",
       "      <td>W AA N T</td>\n",
       "      <td>3243</td>\n",
       "      <td>WQNT</td>\n",
       "      <td>WNT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>war</th>\n",
       "      <td>W AO R</td>\n",
       "      <td>3801</td>\n",
       "      <td>WQR</td>\n",
       "      <td>WAR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>was</th>\n",
       "      <td>W AA Z</td>\n",
       "      <td>4818</td>\n",
       "      <td>WQS</td>\n",
       "      <td>WAS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>way</th>\n",
       "      <td>W EY</td>\n",
       "      <td>4263</td>\n",
       "      <td>WQ</td>\n",
       "      <td>WAY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>went</th>\n",
       "      <td>W EH N T</td>\n",
       "      <td>3063</td>\n",
       "      <td>WQNT</td>\n",
       "      <td>WNT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>who</th>\n",
       "      <td>HH UW</td>\n",
       "      <td>4628</td>\n",
       "      <td>HQ</td>\n",
       "      <td>WHO</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>why</th>\n",
       "      <td>W AY</td>\n",
       "      <td>3304</td>\n",
       "      <td>WQ</td>\n",
       "      <td>WHY</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>yet</th>\n",
       "      <td>Y EH T</td>\n",
       "      <td>3629</td>\n",
       "      <td>YQT</td>\n",
       "      <td>YET</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>you</th>\n",
       "      <td>Y UW</td>\n",
       "      <td>4743</td>\n",
       "      <td>YQ</td>\n",
       "      <td>YOU</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            sound  freq encoding vowel_free\n",
       "word                                       \n",
       "able    EY B AH L  2940     QBQL        QBL\n",
       "age         EY JH  3253       QJ        AGE\n",
       "air          EH R  3492       QR        AIR\n",
       "all          AO L  4718       QL        ALL\n",
       "also    AO L S OW  4553     QLSQ       QLSQ\n",
       "and        AH N D  4878      QND        AND\n",
       "any       EH N IY  4613      QNQ        ANY\n",
       "are          AA R  4788       QR        ARE\n",
       "area   EH R IY AH  3820     QRQQ        QRQ\n",
       "art        AA R T  2728      QRT        ART\n",
       "away      AH W EY  3290      QWQ        QWQ\n",
       "best     B EH S T  3552     BQST        BST\n",
       "body    B AA D IY  3399     BQDQ        BDQ\n",
       "but        B AH T  4748      BQT        BUT\n",
       "can        K AE N  4683      KQN        CAN\n",
       "cent     S EH N T  3214     SQNT        SNT\n",
       "city    S IH T IY  4010     SQTQ        STQ\n",
       "cost     K AA S T  3472     KQST        KST\n",
       "cut        K AH T  2054      KQT        CUT\n",
       "data    D EY T AH  3955     DQTQ        DTQ\n",
       "day          D EY  4251       DQ        DAY\n",
       "did        D IH D  4383      DQD        DID\n",
       "due          D UW  3369       DQ        DUE\n",
       "duty    D UW T IY  2085     DQTQ        DTQ\n",
       "end        EH N D  3925      QND        END\n",
       "even    IY V IH N  4318     QVQN        QVN\n",
       "ever      EH V ER  3072      QVQ        QVQ\n",
       "fact     F AE K T  3894     FQKT        FKT\n",
       "far        F AA R  3685      FQR        FAR\n",
       "felt     F EH L T  2302     FQLT        FLT\n",
       "few        F Y UW  3850      FYQ        FEW\n",
       "find     F AY N D  3590     FQND        FND\n",
       "for        F AO R  4848      FQR        FOR\n",
       "form     F AO R M  3970     FQRM        FRM\n",
       "free       F R IY  3387      FRQ        FRQ\n",
       "from     F R AH M  4778     FRQM        FRM\n",
       "gas        G AE S  2537      GQS        GAS\n",
       "get        G EH T  3920      GQT        GET\n",
       "got        G AA T  2652      GQT        GOT\n",
       "had       HH AE D  4728      HQD        HAD\n",
       "hand    HH AE N D  3747     HQND        HND\n",
       "hard    HH AA R D  2205     HQRD        HRD\n",
       "has       HH AE Z  4708      HQA        HAS\n",
       "held    HH EH L D  3531     HQLD        HLD\n",
       "help    HH EH L P  3078     HQLP        HLP\n",
       "her         HH ER  4658       HQ        HER\n",
       "him       HH IH M  4538      HQM        HIM\n",
       "his       HH IH Z  4763      HQA        HIS\n",
       "into    IH N T UW  4568     QNTQ       QNTQ\n",
       "its        IH T S  4633      QTS        ITS\n",
       "just    JH AH S T  4159     JQST        JST\n",
       "kind     K AY N D  2848     KQND        KND\n",
       "land     L AE N D  3771     LQND        LND\n",
       "last     L AE S T  4015     LQST        LST\n",
       "law          L AO  4197       LQ        LAW\n",
       "left     L EH F T  3665     LQFT        LFT\n",
       "let        L EH T  3405      LQT        LET\n",
       "list     L IH S T  2179     LQST        LST\n",
       "lost     L AO S T  2058     LQST        LST\n",
       "low          L OW  3238       LQ        LOW\n",
       "man        M AE N  4241      MQN        MAN\n",
       "many    M EH N IY  4398     MQNQ        MNQ\n",
       "may          M EY  4653       MQ        MAY\n",
       "men        M EH N  4053      MQN        MEN\n",
       "mind     M AY N D  2955     MQND        MND\n",
       "most     M OW S T  4463     MQST        MST\n",
       "must     M AH S T  4368     MQST        MST\n",
       "new          N UW  4603       NQ        NEW\n",
       "next   N EH K S T  3364    NQKST       NKST\n",
       "non        N AA N  2910      NQN        NON\n",
       "nor        N AO R  2801      NQR        NOR\n",
       "not        N AA T  4808      NQT        NOT\n",
       "now          N AW  4433       NQ        NOW\n",
       "off          AO F  3825       QF        OFF\n",
       "oil          OY L  2730       QL        OIL\n",
       "old        OW L D  3975      QLD        OLD\n",
       "once     W AH N S  3354     WQNS        WNS\n",
       "one        W AH N  4713      WQN        ONE\n",
       "only    OW N L IY  4573     QNLQ       QNLQ\n",
       "open    OW P AH N  3110     QPQN        QPN\n",
       "our         AW ER  4518       QQ        OUR\n",
       "out          AW T  4543       QT        OUT\n",
       "over      OW V ER  4458      QVQ        QVQ\n",
       "own          OW N  4173       QN        OWN\n",
       "part     P AA R T  4343     PQRT        PRT\n",
       "past     P AE S T  2841     PQST        PST\n",
       "pay          P EY  2907       PQ        PAY\n",
       "per          P ER  4278       PQ        PER\n",
       "plan     P L AE N  2967     PLQN        PLN\n",
       "post     P OW S T  2139     PQST        PST\n",
       "put        P UH T  3461      PQT        PUT\n",
       "red        R EH D  2319      RQE        RED\n",
       "run        R AH N  2098      RQN        RUN\n",
       "saw          S AO  2457       SQ        SAW\n",
       "say          S EY  3910       SQ        SAY\n",
       "sea          S IY  2256       SQ        SEA\n",
       "see          S IY  4488       SQ        SEE\n",
       "self     S EH L F  2817     SQLF        SLF\n",
       "sent     S EH N T  2278     SQNT        SNT\n",
       "set        S EH T  3920      SQT        SET\n",
       "she         SH IY  4608       UQ        SHE\n",
       "six      S IH K S  2831     SQKS        SIX\n",
       "son        S AH N  2370      SQN        SON\n",
       "tax      T AE K S  3144     TQKS        TAX\n",
       "ten        T EH N  2261      TQN        TEN\n",
       "test     T EH S T  2802     TQST        TST\n",
       "the         DH AH  4888       XQ        THE\n",
       "told     T OW L D  2725     TQLD        TLD\n",
       "too          T UW  3950       TQ        TOO\n",
       "top        T AA P  2151      TQP        TOP\n",
       "true       T R UW  3028      TRQ        TRQ\n",
       "two          T UW  4558       TQ        TWO\n",
       "unit  Y UW N AH T  2190    YQNQT        YNT\n",
       "upon    AH P AA N  4363     QPQN        QPN\n",
       "use        Y UW S  4348      YQS        USE\n",
       "used     Y UW Z D  4328     YQZE        YZE\n",
       "very    V EH R IY  4453     VQRQ        VRQ\n",
       "view       V Y UW  3201      VYQ        VYQ\n",
       "want     W AA N T  3243     WQNT        WNT\n",
       "war        W AO R  3801      WQR        WAR\n",
       "was        W AA Z  4818      WQS        WAS\n",
       "way          W EY  4263       WQ        WAY\n",
       "went     W EH N T  3063     WQNT        WNT\n",
       "who         HH UW  4628       HQ        WHO\n",
       "why          W AY  3304       WQ        WHY\n",
       "yet        Y EH T  3629      YQT        YET\n",
       "you          Y UW  4743       YQ        YOU"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "with pd.option_context('display.max_rows', None):\n",
    "\tdisplay(df.loc[(df.freq > 2000) &(df.index.str.len() <=4) & (df.vowel_free.str.len() > 2)])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Further reductions:\n",
    "\n",
    "continue reducing the encoding length and the multiplicity (although sometimes we need to trade off between these two objectives)\n",
    "\n",
    "This will be accounted for lazy/fuzzy pronounciation and be implemented by the users and not directly in the encoding.\n",
    "\n",
    "For example we can reduce approximants whose property can be considered as an interpolation between a vowel and a consonant. Sometimes it's not easy to tell them from a continuous speech flow and usually omitted by lazy speakers.\n",
    "\n",
    "There are several phonological circumstances for approximants in English:\n",
    "`A` for approximant, `C` for consonant, `V` for vowel\n",
    "\n",
    "1. At the beginning `A-V` (`A-C` is not possible)\n",
    "2. At the end `-A`\n",
    "3. `V-A-C`\n",
    "4. `V-A-V`\n",
    "5. `C-A-V`\n",
    "\n",
    "- for semi-vowels `Y,W`, we identify the following circumstances as reducable: 2, 3, 4, 5\n",
    "- for approximant `R,L`, we identify the following circumstances as reducable: 2, 3\n",
    "\n",
    "and even more radical reduction: e.g. nasal elision\n",
    "- for nasal `M,N,NG`: 2, 3(condition to certain trailing consonant)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9.13 ('base')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "88279d2366fe020547cde40dd65aa0e3aa662a6ec1f3ca12d88834876c85e1a6"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}