Although every language is mixed to some extent, by virtue of containing loanwords, it is a matter of controversy whether a term mixed language can meaningfully distinguish the contact phenomena of certain languages (such as those listed below) from the type of contact and borrowing seen in all languages. In other words, a "mixed language" is a language that belongs to more than one language family and it is unclear whether there are any mixed languages, i.e. it is unclear to what extent language mixture, can be distinguished from other mechanisms such as code-switching, substrata, or lexical borrowing. 
In 1861, Max Müller denied "the possibility of a mixed language". In 1881, William D. Whitney wrote as follows expressing skepticism regarding the chances of a language being proven a mixed language.
Such a thing as the adoption on the part of one tongue, by a direct process, of any part or parts of the formal structure of another tongue has, so fas as is known, not come under the notice of linguistic students during the recorded periods of language-history. So far as these are concerned, it appears to be everywhere the case that when the speakers of two languages, A and B, are brought together into one community, there takes place no amalgamation of their speech, into AB; but for a time the two maintain their own several identity, only as modified each by the admission of material from the other in accordance with the ordinary laws of mixture; we may call them Aᵇ and Bᵃ, and not AB. … [W]e shall doubtless meet now and then with the claim that such and such a case presents peculiar conditions which separate it from the general class, and that some remote and difficult problem in language-history is to be solved by admitting promiscuous mixture. Any one advancing such a claim, however, does it at his peril; the burden of proof is upon him to show what the peculiar coditions might have been, and how they should have acted to produce the exceptional result; he will be challenged to bring forward some hisotircally authenticated case of analogous results; and his solution, if not rejected altogether, will be looked upon with doubt and misgiving until he shall have complied with these reasonable requirements. 
Wilhelm Schmidt was an important proponent of the idea of mixed languages in the very late 19th and early 20th century.. In the judgement of Thomas Sebeok, Schmidt produced "not a scrap of evidence" for his theory. Margaret Schlauch provides a summary of the various objections to Schmidt's theory of Sprachmischung, by prominent linguistics such as Alfredo Trombetti, Antoine Meillet, and A. Kholodovich.
Despite the old and broad consensus that rejected the idea of a 'mixed language', Thomason and Kaufman in 1988 proposed to revive the idea that some languages had shared genealogy. Meakins, who finds Thomason and Kaufman's account credible, suggests that a mixed language results from the fusion of usually two source languages, normally in situations of thorough bilingualism, so that it is not possible to classify the resulting language as belonging to either of the language families that were its sources.
Despite these recent efforts to rehabilitate mixed languages as an idea, many linguistics remained unconvinced. For example, van Driem rejects one by one each of Thomason and Kaufman's examples as well as those more recently proposed. Most recently, Versteegh rejects the notion of a mixed language, writing that at "no point is it necessary to posit a category of mixed languages." 
Other terms used in linguistics for the concept of a mixed language include hybrid language, contact language, and fusion language; in older usage, 'jargon' was sometimes used in this sense. In some linguists' usage, creoles and pidgins are types of mixed languages, whereas in others' usage, creoles and pidgins are merely among the kinds of language that might become full-fledged mixed languages.
Thomason (1995) classifies mixed languages into two categories. Category 1 languages exhibit "heavy influence from the dominant group's language in all aspects of structure and grammar as well as lexicon" (Winford 171). Category 2 languages show a "categorial specificity of the structural borrowing" or a uniform borrowing of specific categories (Winford).
Mixed language and intertwined language are seemingly interchangeable terms for some researchers. Some use the term "intertwining" instead of "mixing" because the former implies "mixture of two systems which are not necessarily the same order" nor does it suggest "replacement of the either the lexicon or of the grammatical system", unlike relexification, massive grammatical replacement, and re-grammaticalization. The grammar of a mixed language typically comes from a language well known to first-generation speakers, which Arends claims is the language spoken by the mother. This is because of the close relationship between mother and child and the likelihood that the language is spoken by the community at large.
Arends et al. classify an intertwined language as a language that "has lexical morphemes from one language and grammatical morphemes from another". This definition does not include Michif, which combines French lexical items in specific contexts, but still utilizes Cree lexical and grammatical items.
Yaron Matras distinguishes between three types of models for mixed language: "language maintenance and language shift, unique and predetermined processes ("intertwining"), and conventionalisation of language mixing patterns". The first model involves the use of one language for heavy substitutions of entire grammatical paradigms or morphology of another language. This is because a speech community will not adopt a newer dominant language, and so adapt their language with grammatical material from the dominant language. Bakker (1997) argues that mixed languages result from mixed populations. Languages "intertwine", in that the morphosyntax (provided by female native speakers) mixes with the lexicon of another language (spoken by men, often in a colonialist context). This appears to have been the case with Michif, where European men and Cree, Nakota, and Ojibwe women had offspring who learned a mixture of French and Cree. The third model "assumes a gradual loss of the conversational function of language alternation as a means of expressing contrast". In other words, language no longer becomes a means of differentiation between two speech communities as a result of language mixing.
Lexical reorientation, according to Matras, is defined as "the conscious shifting of the linguistic field that is responsible for encoding meaning or conceptual representations away from the language in which linguistic interaction is normally managed, organised, and processed: speakers adopt in a sense one linguistic system to express lexical meaning (or symbols, in the Buhlerian sense of the term) and another to organize the relations among lexical symbols, as well as within sentences, utterances, and interaction. The result is a split, by source language, between lexicon and grammar."
A mixed language differs from pidgin and creole languages and code-switching in very fundamental ways. Mixed language speakers are fluent, even native, speakers of both languages. Pidgins, on the other hand, develop in a situation, usually in the context of trade, where speakers of two (or more) different languages come into contact and need to find some way to communicate with each other. This contrasts with mixed languages, where speakers tend to be fluent in one or both of the source languages. Creoles develop when a pidgin language becomes a native language for young speakers. While creoles tend to have drastically simplified morphologies, mixed languages often retain the inflectional complexities of one or both of parent languages.
It also differs from a language that has undergone heavy borrowing, such as Korean, Japanese, and Vietnamese from Chinese (see Sino-Xenic), English from French, or Maltese from Sicilian/Italian. In these cases, despite the heavy borrowing, the grammar and basic words of the borrowing language remain relatively unchanged, with the borrowed words confined mainly to more abstract or foreign concepts. In the case of Maltese, for example, if verbs borrowed from Italian were inflected using Italian inflectional rules rather than Arabic-derived ones, then Maltese would be a candidate for being a mixed language.
Finally, a mixed language differs from code-switching, such as Spanglish or Portuñol, in that, once it has developed, the fusion of the source languages is fixed in the grammar and vocabulary, and speakers do not need to know the source languages in order to speak it. But, linguists believe that mixed languages evolve from persistent code-switching, with younger generations picking up the code-switching, but not necessarily the source languages that generated it.
Yaron Matras distinguishes between three types of models for mixed language: "language maintenance and language shift, unique and predetermined processes ("intertwining"), and conventionalisation of language mixing patterns". The first model involves the use of one language for heavy substitutions of entire grammatical paradigms or morphology of another language. This is because a speech community will not adopt a newer dominant language, and so adapt their language with grammatical material from the dominant language. Bakker (1997) argues that mixed languages result from mixed populations. Languages "intertwine", in that the morphosyntax (provided by female native speakers) mixes with the lexicon of another language (spoken by men, often in a colonialist context). This appears to have been the case with Michif French, where European men and Cree, Nakota, and Ojibwe women had offspring who learned a mixture of French and Cree. The third model "assumes a gradual loss of the conversational function of language alternation as a means of expressing contrast". In other words, language no longer becomes a means of differentiation between two speech communities as a result of language mixing.
Lexical reorientation, according to Matras, is defined as "the conscious shifting of the linguistic field that is responsible for encoding meaning or conceptual representations away from the language in which linguistic interaction is normally managed, organised, and processed: speakers adopt in a sense one linguistic system to express lexical meaning (or symbols, in the Buhlerian sense of the term) and another to organize the relations among lexical symbols, as well as within sentences, utterances, and interaction. The result is a split, by source language, between lexicon and grammar." Most portmanteau language names, such as Franglais and Anglo-Romani, are not mixed languages, or even examples of code-switching, but registers of a language (here French and English), characterized by large numbers of loanwords from a second language (here English and Romani). Middle English (the immediate fore-runner of Modern English) developed from such a situation, incorporating many Norman borrowings into Old English, but it is not considered a mixed language.
Michif derives nouns, numerals, definite/indefinite articles, possessive pronouns, some adverbs and adjectives from French, while it derives demonstratives (in/animate), question words, verbs (in/animacy agreement with the subject/object), and some adverbs/verb-like adjectives from Cree. The Cree components of Michif generally remain grammatically intact, while the French lexicon and grammar is restricted to noun phrases where nouns occur with a French possessive element or article (i.e. in/definite, masculine/feminine, singular/plural). Further, many speakers of Michif are able to identify the French and Cree components of a given sentence, likely from the phonological and morphological features of words. Although the phonological systems of both French and Cree are generally independent in Michif, there is convergence in 1) mid-vowel raising, 2) sibilant harmony, 3) vowel length (e.g. French vowel pairs [i]/[ɪ] and [a]/[ɑ] differ in length as in Cree), and 4) instances where the three nasal vowels /æ̃/, /ũ/, and /ĩ/ occur in the Cree components, although this last point of convergence may be due to Ojibwe influence. Scholars propose that, in the Métis multilingual community, Michif emerged as a need to symbolize a new social identity. The first unambiguous mention of Michif dates to the 1930s.
The Metis of St. Laurent, a tribe of indigenous people in Canada, were made to feel their language was a sign of inferiority by nuns, priests, and other missionaries who insisted that the Metis switch to Standard Canadian French. Because missionaries stigmatized Michif French as an inferior, "bastardized" form of Canadian French, the Metis began to develop a sense of inferiority and shame which they associated with speaking Michif. [Guy], a Metis and native speaker of Michif, claims that the Metis people also used speaking Michif as a way of idenitifying themseleves as a distinct group, and thus the language survived. Nonetheless, it became taboo to speak Michif inter-ethnically.
In an attempt to make students unlearn Michif French, some nuns used a "token-system" in which each student was given ten tokens each week, and for every use of Michif French, a student would have to surrender a token. Students with the most tokens were rewarded with a prize. Overall, this system did not work.
Mednyj Aleut is identified as a mixed language composed of mostly intact systematic components from two typologically and genetically unrelated languages: Aleut and Russian. This mixed language’s grammar and lexicon are both largely Aleut in origin, while the finite verb morphology, a whole grammatical subsystem, is primarily of Russian origin. Nonetheless, there are some syntactic patterns with Russian influence and some Aleut features in the finite verb complex such as, 1) a topic-number agreement pattern, 2) Aleut pronouns with unaccusatives, 3) the Aleut agglutinative tense + number + person/number pattern in one of two alternative past-tense forms. Scholars hypothesize that due to the elaborate Russian and Aleut components of Mednyj Aleut, the Aleut/Russian creoles in which the mixed language arose must have been fluent bilinguals of Aleut and Russian and, therefore, not a pidgin language—that is, “imperfect learning” is usually a feature in the emergence of a pidgin. Furthermore, some code-switching and deliberate decisions likely served as mechanisms for the development of Mednyj Aleut and it is possible that these were motivated by a need for a language that reflected the community’s new group identity.
Ma’a has a Cushitic basic vocabulary and a primarily Bantu grammatical structure. The language also shares some phonological units with languages in the Cushitic phylum (e.g. the voiceless lateral fricative, the voiceless glottal stop, and the voiceless velar fricative that do not occur in Bantu), as well as syntactic structures, derivational processes, and a feature of inflectional morphology. However, few productive non-lexical structures in Ma’a appear derived from Cushitic. Sarah G. Thomason therefore argues for a classification of Ma’a as a mixed language since it does not have enough Cushitic grammar to be genetically related to the Cushitic language. By contrast, Ma’a has a productive set of inflectional structures derived from Bantu. Ma’a also demonstrates phonological structures derived from Bantu—for instance, the prenasalized voiced stops /mb ⁿd ⁿɟ ŋg/, phonemic tones, the absence of pharyngeal fricatives, labialized dorsal stops, ejective and retroflex stops, and final consonants— as well as noun classification, number category, and verb morphology patterns of Bantu. Syntactic and derivational patterns in Ma’a vary between Cushitic and Bantu origins—some Ma’a constructions used, such as genitive and copula constructions, are both from Cushitic and Bantu. These observations, in view of additional language contact cases like Asia Minor Greek, Anglo-Romani, and Mednyj Aleut, suggest that Ma'a arose as a product of massive interference from a Bantu language via intense cultural pressure on a Cushitic-speaking community.
Media Lengua, also known as Chaupi-shimi, Chaupi-lengua, Chaupi-Quichua, Quichuañol, Chapu-shimi or llanga-shimi,[nb 1] (roughly translated to "half language" or "in-between language") is a mixed language that consists of Spanish vocabulary and Ecuadorian Quichua grammar, most conspicuously in its morphology. In terms of vocabulary, almost all lexemes (89%), including core vocabulary, are of Spanish origin and appear to conform to Quichua phonotactics. Media Lengua is one of the few widely acknowledged examples of a "bilingual mixed language" in both the conventional and narrow linguistic sense because of its split between roots and suffixes. Such extreme and systematic borrowing is only rarely attested, and Media Lengua is not typically described as a variety of either Quichua or Spanish. Arends et al. list two languages subsumed under the name Media Lengua: Salcedo Media Lengua and Media Lengua of Saraguro. The northern variety of Media Lengua, found in the province of Imbabura, is commonly referred to as Imbabura Media Lengua and more specifically, the dialect varieties within the province are known as Pijal Media Lengua and Anglas Media Lengua.
Scholars indicate that Media Lengua arose largely via relexification mechanisms. Pieter Muysken suggests that the social context in which the language emerged as an intralanguage involved a presence of “acculturated Indians” that neither identified with traditional, rural Quechua nor with urban Spanish cultures. This is an instance of a language developing from a need for “ethnic self-identification”.
Light Warlpiri, seen as a form of Warlpiri by speakers, derives verbs and verbal morphology largely from Australian Kriol, while nouns are largely from Warlpiri and English and nominal morphology from Warlpiri. Light Warlpiri likely developed as an intralanguage via code-mixing between Warlpiri and either Kriol or English. This code-mixing conventionalized into Light Warlpiri, which is now learned by Lajamanu children as a first language, along with Warlpiri, although Light Warlpiri is often produced first and used in daily interactions with younger speakers and adults within the Lajamanu community. Light Warlpiri is considered a new language for several reasons: 1) Light Warlpiri speakers use an auxiliary verb-system of that older Warlpiri speakers do not while code-mixing, 2) elements are distributed differently in Light Warlpiri than in code-mixing varieties of older Warlpiri speakers, 3) Light Warlpiri is a native language, which indicates stability of the language, and 4) grammatical structures and lexical items from each source language occur consistently in Light Warlpiri.
Gurindji Kriol exhibits a structural split between the noun phrase and verb phrase, with Gurindji contributing the noun structure including case-marking, and the verb structure including TAM (tense-aspect-mood) auxiliaries coming from Kriol. In this respect, Gurindji Kriol is classified as a verb-noun (V-N) mixed language. Other examples of V-N mixed languages include Michif and Light Warlpiri. The maintenance of Gurindji within the mixed language can be seen as the perpetuation of Aboriginal identity under massive and continuing cultural incursion.
Both Cappadocian Greek and Cypriot Maronite-Arabic are cases of extreme borrowing—the former from Turkish and the latter from Greek. The remaining Greek dialects of Asia Minor display borrowing of vocabulary, function words, derivational morphology, and some borrowed nominal and verbal inflectional morphology from Turkish. Cypriot Arabic largely shows borrowing of vocabulary, and consequently Greek morphosyntax. These cases, including Ma'a, differ socially from Michif and Mednyj Aleut because they have evolved out of intense language contact, extensive bilingualism, and a strong pressure for speakers to shift to the dominant language. Nonetheless, neither language has an entire grammar and lexicon that is derived from a single historical source and in each case the linguistic group achieves fluent bilingualism. The social context in which they arose largely distinguishes them from pidgins and creoles and, for some scholars, identifies them closely with mixed languages.
The Kaqchikel-K'iche' Mixed Language, also known as the Cauqué Mixed Language or Cauqué Mayan, is spoken in the aldea of Santa María Cauqué, Santiago Sacatepéquez, Department of Sacatepéquez in Guatemala. A 1998 study by the Summer Institute of Linguistics (SIL) estimated speaker population at 2,000. While the language's grammatical base is from K'iche', its lexicon is supplied by Kaqchikel.