Conlang Morphology Tutorial: Morphemes, Affixes & Alignment

[00:00:01]When you learn a language, your learning  can usually be divided up into three main parts - pronunciation, grammar and words. In order  to become proficient in a language, you need to master all three parts. It follows that in order  to make a language, you need the same. So far in this series, we've looked at pronunciation, or  phonology. Now it's time to begin on grammar.

[00:00:35]"Grammar" is a term used by both linguists and  non-linguists to describe all the bits of language that stick words together. But as linguists, we  more often divide it up into separate fields, namely "morphology", which studies how smaller  elements come together to form words, and "syntax", which studies how whole words combine  to form larger units, like phrases and sentences.

[00:01:06]Even though we do this, it's important to  understand that they aren't strictly separate categories - as you'll see in the next few  episodes, they do often interact with one another and with phonology to form combined fields, such  as "morphosyntax" and "morphophonology". However, they are still useful to distinguish, and this is  the approach taken by most linguistics courses, so I will follow suit in this series, with the  next few episodes focusing on various parts of morphology and related topics and a subsequent  episode dedicated to syntax. It's just worth knowing that a lot of the ideas I'll introduce  in this video are also applicable to syntax.

[00:01:55]Just as phonology focuses on phonemes,  morphology focuses on morphemes. Where a phoneme is an irreducible unit of sound, a  morpheme is an irreducible unit of meaning.

[00:02:13]Morphemes can be classified as either "free"  or "bound". A free morpheme is one which can stand on its own, forming a full word.  For examples, think of any of the basic, unmodified words of English, like "dog",  "sing", "smooth" or "with". Bound morphemes, however, cannot stand alone, relying  on free morphemes to be used. Think the "-ing" in "eating", the "-es" in "wishes" and  the "-er" in "younger". These elements can only be used when attached to host words  - "-ing" could never be used on its own.

[00:03:00]Just as phonemes can have allophones, morphemes  can have allomorphs - different phonological realisations of the same underlying morpheme.  Both free and bound morphemes can have allomorphs.

[00:03:17]We've discussed the allomorphs of the English  plural morpheme in a previous episode - it has different forms after voiced sounds, voiceless  sounds and sibilants. Since these realisations are separate phonemes in English and don't  vary like this in other words with the same sequences of sounds, they're not allophones,  but they do show allomorphic variation.

[00:03:47]A free morpheme with multiple allomorphs  can be seen in the indefinite article "an".

[00:03:54]Before vowels, this takes the form "an", while  before consonants, it loses the N to become "a".

[00:04:04]A few times already in this video and in previous  episodes, I've used words like "noun", "verb", "article", "adjective" and "pronoun".  But what do these mean exactly? Well, they're what we call "parts of  speech" or "word classes". Basically, not all words are created equal - different  ones do different things within a sentence.

[00:04:34]We can categorise them roughly based on "form" and  "function", or "how they look and are internally composed" and "what they do in the sentence and  how they interact with other words". This is, very roughly, the morphology/syntax  distinction we've discussed already.

[00:04:55]In the work of the Ancient Greek grammarians, there were considered to be eight parts of speech  - nouns, verbs, participles, pronouns, adverbs, articles, prepositions and conjunctions.  This system has been expanded and modified over the last two millennia to give the general  classification we use today. But crucially, this system is based on English, Latin and Greek, and  influenced by various other languages of Europe, but it does not work for all languages around  the world. Some languages, for instance, lack adjectives, adverbs or determiners. Others  make only marginal use of certain classes, while others still have parts of speech that  European languages don't. The Wagiman language of northern Australia, for instance, only has a  few true verbs, but has an additional word class, the "coverbs", which must combine with  regular verbs to express their meaning.

[00:06:09]But that aside, let's take a look at some  of the more common parts of speech and the formal and functional behaviours that define them.

[00:06:19]A noun is, generally, a term for a person, place,  thing or idea. It can be placed in relation to a verb to show the entities involved in an action  or state. It can also have properties attributed to it via adjectives and be related to other nouns  with adpositions. In English, nouns can be either singular or plural and can take, or include,  morphemes such as diminutives, denoting a smaller version of something, and agentives, showing the  doer of an action. In other languages, they might also take cases, which show their role in the  sentence, class (suggesting particularities of their behaviour) and definiteness (showing whether  the specific instance of a noun is already known).

[00:07:18]Examples in English are "donkey", "woman", "book",  "darkness", "river", "fun" and "extinction".

[00:07:29]A verb is a word expressing an action or state.  They define what the sentence is about and, along with nouns, are one of the only two word classes  found in all known languages. Functionally, they usually take nouns or pronouns to define  what's happening. They can also be delimited with adverbs to specify their meaning. In  English, they change their form based on person and number to show who's involved, tense  and mood to show if and when it's happening and can combine with other verbs to show further  tense, aspect, mood and voice to describe the nature of the event. In other languages, you may  also encounter features such as evidentiality, to show knowledge of the event and negation, to  show an event as having not happened. English examples are "think", "eat", "sleep",  "want", "describe", "have" and "be".

[00:08:43]Adjectives describe nouns. They can mark  comparison, to compare one noun to another, and in other languages may "agree" with  nouns, mimicking their plurality, case, class and definiteness. Examples in English are  "big", "happy", "yellow", "wooden" and "Canadian".

[00:09:09]Similarly, adverbs describe verbs and adjectives.  They often end in "-ly", especially when derived from adjectives, though many also don't. Examples  are "slowly", "happily", "now", "well" and "here".

[00:09:31]Pronouns are, in some languages, a  subclass of nouns. They stand in for nouns, often to reference something without  repeating or naming it in full. In English, they are similar in function to nouns, but  slightly more restricted. For instance, they usually can't take adjectives, although this  is because technically, they replace whole noun phrases, including a noun and its associated  adjectives and determiners. However, formally, they're a little more elaborate. Unlike nouns,  they vary based on case and their forms aren't as clearly derived as in nouns. English examples are  "you", "we", "her", "what", "no-one" and "those".

[00:10:28]In English, we have a part of  speech known as "prepositions".

[00:10:34]These define the relationship of nouns to other  parts of the sentence. They're mostly simple, not containing any smaller morphemes or able  to have morphemes added to them. However, there are also "complex prepositions", made up  of phrases involving a noun and preposition, which together fill the same role. We call them  "prepositions" because they come before the noun, but in other languages they may come  after, so we term the whole word class "adpositions". Examples are "with",  "in", "to", "regarding" and "in spite of".

[00:11:22]Other common parts of speech are conjunctions,  which join clauses or sentences together, determiners, which specify a noun in some way,  and particles, which don't include or take special morphemes and add grammatical information  to a sentence, rather than physical meaning.

[00:11:46]We can also describe parts of speech as  either "open class" or "close class". This describes how easily new words of that type  can be created. Open classes allow new words to be created easily, while close classes  do not. Open classes are usually lexical; that is, their words have clear, easily  definable meanings. Close classes, however, are usually grammatical; their words are  harder to define and perform functions to bind sentences together. Examples of open  classes are nouns, verbs, adjectives and adverbs. Close classes include pronouns,  determiners, adpositions and conjunctions.

[00:12:46]We'll go into more detail on each of the  parts of speech in the coming episodes, but as a beginner, you're probably best just  sticking to basic parts of speech in English or shown in this list. Try to understand  the basic essence of what each is so you can assign appropriate behaviours to it  when we reach that point in the series.

[00:13:11]In case you haven't noticed, parts  of speech represent the categories of free morphemes. They're the  types of independent word which can occur in a language. But  what about bound morphemes?

[00:13:27]Well, one basic distinction we can make is in  their function. When a bound morpheme attaches onto a free one, we refer to this free morpheme  as either a "root" or a "stem". A root is the most irreducible form of a word, composed of a single  free morpheme. A stem can actually be composed of more than one morpheme, but it's just the lexical  part of a word, without any grammatical elements added on. For example, in the word "brotherhoods",  we can find three morphemes - "brother", "-hood" and the plural "-s". The root of this word  is "brother" - the smallest part of the word with an independent non-grammatical meaning. Onto this,  the morpheme "-hood" is added. This doesn't add grammatical information, but it does modify the  meaning of the word. Adding a morpheme to alter meaning like this is called "derivation". In this  case, no more derivational morphemes are added, so "brotherhood" is the word's stem. However,  we can also add the "-s" on, to form the word "brotherhoods". This morpheme represents  the grammatical category of "plurality", indicating that we're talking about  more than one of the thing in question.

[00:15:05]It's adding to the grammatical status of the word,  but isn't adding actual lexical meaning, as the essence of "brotherhoods" doesn't differ from  that of "brotherhood". This is therefore not a case of derivation, but rather one of "inflection"  - that is, adding grammatical morphemes to a word.

[00:15:30]In this example, we saw both morphemes added  onto the end of the root. But this isn't the only option. Bound morphemes coming after the root are  called "suffixes". If they come before, however, they're "prefixes". Examples in English are  the negative prefixes "un-" and "in-" or "co-", meaning "together". Most English "affixes",  as they're neutrally called, are suffixes, including all grammatical morphemes. But in the  Zulu language of South Africa, this is reversed, with most affixes being prefixes before  the root and only a few suffixes.

[00:16:22]Other languages may be intermediate to this,  using neither more often than the other, such as is the case in the Choctaw  language of the southern United States.

[00:16:34]While prefixes and suffixes are most common, affixes can also occur in places other  than before or after the root. In Tagalog, a language of the Philippines, certain aspects of  verbal morphology are marked through "infixes", which occur inside a word stem, with the  stem being broken up in some predictable way.

[00:17:01]In German, past participles (the forms  of verbs equivalent to English "eaten", "been", "sung" and "gone") are formed with  circumfixes, where morphemes are added both before and after the stem, but where  one or both have no meaning on their own.

[00:17:23]Many languages also show grammatical meaning  by "reduplication", where all or part of a stem is doubled, or sometimes even tripled.  The element which is added in these cases in known as a "duplifix". This can be  seen in the Igbo language of Nigeria, where gerunds (noun-like verbs) are  formed by reduplicating part of the verb.

[00:17:53]Most of the time, languages use some sort  of affix to derive and inflect words. But some use other means. Affixing involves adding a  consistent, unitary morpheme to a root or stem.

[00:18:09]Given that they're just stacked together like  this, we call this "concatenative" morphology.

[00:18:16]However, some languages change the internal  structure or elements of a root. This we call "non-concatenative" morphology. The most famous  example of this is in the Semitic languages, like Arabic, Hebrew and Aramaic, where roots  are not blocks of consonants and vowels like in many other languages, but instead consist of  just consonants. The vowels of the word are then added in at specific places to form derived  and inflected forms. For instance, Arabic has the root "S-L-M" which denotes concepts relating  to safety, peace and surrender. The root never appears in this vowelless form, but all derived  and inflected stems can be traced back to this base. By inserting a short and a long "A" into the  middle, we can form the word "سَلَام" ("salām"), which is a common greeting in the Arab  world. But by inserting other vowels, we get "إِسْلَام" ("ʔislām"), the source of the  English word "Islam", the religion of the Arabs.

[00:19:37]Another derivation makes use of yet another  vowel template, as well as a prefix, to give "مُسْلِم" ("muslim"), the source of  English "Muslim" - a follower of Islam.

[00:19:49]But non-concatenative morphology isn't restricted  to this consonant-root template system of the Semitic languages - we actually also use it  in English! Think about nouns like "mouse", "man" or "foot". To form their plurals,  we don't add "-s" like most other nouns; the vowel in the root gets  changed - "mouse" to "mice", "man" to "men" and "foot" to "feet". Our verbs  do funky things too, with "sing" having past tense "sang" and participle "sung" and "have"  using "had" for both past tense and participle.

[00:20:38]Importantly, not just any change  to the stem is non-concatenation, only those cases where certain phones change  in predictable or semi-predictable ways. The past tense form of English "go", for example, is  "went", which is not a case of non-concatenation, but rather one of "suppletion". Suppletion is  where a certain form of a word is not related to the reference form, but rather started out  as a separate word, then got co-opted as, say, a plural, past tense or diminutive. Suppletion  is incredibly rare as a d regular strategy of derivation or inflection, but crops up in many,  or most, languages as an irregularity. In English, we also see it in the plural "people"  compared to singular "person" and the comparative and superlative "better"  and "best" compared to the usual "good".

[00:21:50]So, now we've looked at what morphemes are and  where they go, but there's still the question of how many you need. As it turns out, this is  another place we see variation. In some languages, words can have many inflectional morphemes  added to them, while others don't allow any.

[00:22:14]This forms a spectrum. Languages with a low  morpheme-to-word ratio are termed "analytic".

[00:22:23]These languages don't use many affixes per word,  if any, and rely instead on syntax to convey this information. English is an analytic language.  We don't have much in the way of inflection, with our nouns only showing plurality and  possession and our verbs showing only tense, person, number and mood, with other forms being  shown through long strings of separate verbs.

[00:22:54]Note that this is only relevant for inflection  and not derivation. English has a rich range of derivational morphemes available, but only  a few inflections, so it remains analytic.

[00:23:09]Languages which are very analytic, to the  point of having no inflectional morphology, are called "isolating" languages. The Yoruba  language of Nigeria and surrounding parts of West Africa is isolating, as any  grammatical functions are filled by separate words and placed in certain  positions relative to the main headword.

[00:23:35]On the other hand, some languages mark far  more categories on their words. These are called "synthetic languages". Synthetic languages  have two subtypes - fusional and agglutinative.

[00:23:53]In agglutinative languages, each grammatical  category has its own morpheme and these are stacked together separately. Turkish belongs  to this class. Most morphemes denote just one thing and are stacked together at the end  of the word. In fusional languages however, many grammatical categories are condensed  into just a few morphemes. In Hindi, there's a single suffix conveying  the first-person singular masculine future indicative form of a verb, giving that  morpheme five categories combined into one.

[00:24:39]The line between fusional and agglutinative isn't  a hard one. Even Turkish uses a single morpheme for both person and number in its verbs and the  fusional language Persian forms its imperfect tenses by adding a separate prefix to the  perfective form. And just as the extreme of the analytic side is an isolating language, a highly  synthetic language can be called a "polysynthetic" language. These may be fusional or agglutinative,  but are generally more agglutinative. An example language is Greenlandic, in which entire sentences  can be expressed by a single inflected verb.

[00:25:30]Next, let's talk about alignment. This is a topic  that's important for several parts of speech, both in morphology and syntax. It'll  underline several of the things we talk about in the next few episodes, so it's  important to get a basic understanding of now.

[00:25:52]We've seen that nouns and verbs can  work together to make a sentence. But nouns aren't all the same in how they  do this. We can define two main types of sentence - those where the verb  requires two noun phrases and those where it requires only one. We call these  "transitive" and "intransitive" respectively.

[00:26:17]Transitive verbs generally denote an "agent" ("A")  doing something to a "patient" ("P") (think "the girl eats the cake", "David sees John" or "my cat  killed God"). Intransitive verbs generally show something happening to an "experiencer" ("S"), or  the experiencer doing something without affecting anything else. Think "the nurse runs", "the  snake dies" or "the Moon rises". It is actually, as ever, a little more complicated than  this - if you want to challenge yourself, look up "semantic roles" (I'll leave a link in the  description), but just stick with this for now.

[00:27:11]All languages need a way of distinguishing the  agent from the patient, so speakers can tell who is doing the action to whom. "The girl eats the  cake" is very different to "the cake eats the girl". Languages have a variety of different  ways of doing this, which we'll explore more in the coming episodes. As you can see, in English  we use the order of the constituents. Swapping the order swaps the meaning. In other languages,  word order is less important. In German, you can arrange the sentence either way around and  still know which noun is filling each role. That's because German relies partly on noun morphology  to distinguish A from P. Other languages instead use verb morphology. In Greenlandic, there  are different forms of the verb according to both the agent and the patient. If these  are swapped, a different suffix must be used.

[00:28:22]But how does this relate to the  intransitive experiencer? Well, if a language is going to mark the role  of the nouns in transitive sentences, it generally likes to keep things symmetrical  and do the same for intransitive sentences, especially since you can sometimes turn  transitive sentences intransitive or vice versa - keeping some element of marking is  helpful for this conversion. So we end up with these three roles and find that in different  languages, they are grouped in different ways.

[00:29:01]In English, we treat the experiencer of  intransitives the same way we do the agent of transitives. Since our main marking method  is word order, we put both the A and the S roles before the verb, while only the patient  comes after. Our verbs also use morphology to agree with these elements. In the present  tense, we can add an "-s" suffix to the verb if the agent or experiencer (which together we  call the "subject" of the verb) is third-person singular, while if it's a pronoun "I", "you",  "we", "they" or a plural noun, it doesn't do this. Languages which behave like English in this  way are said to have a "nominative" alignment.

[00:29:57]But, as you might expect, there are other  options. Basque is what's known as an "ergative" language. It uses the same marking for  the experiencer as the patient. This stems from the idea of the experiencer being affected by the  action, rather than instigating it. In English, the patient of the verb "kill" and the experiencer  of "die" are filling the same role, but we structure our sentences nominatively. In Basque,  this relationship is more obvious, as exactly the same marking is used for the patient and the  experiencer. The agent is treated independently.

[00:30:47]Some languages play on the potential double  function of the experiencer. These are called "split" languages. In some instances, a  nominative pattern is used, while in others, an ergative one occurs. The reasons for using  one pattern or the other vary between languages.

[00:31:10]Sometimes it is related to whether the  experiencer is more active or passive, but sometimes it's down to grammatical or  pragmatic conditions. Hindi is an example of this, where the marking system of the  noun depends on the "aspect" of the verb.

[00:31:30]Other systems are rarer. One is with "direct"  languages, where no distinction is made between agent, patient and experiencer. In word order,  this may mean there's no fixed word order, while in terms of morphology, it  simply indicates a lack of this.

[00:31:53]In "tripartite" systems, A, P  and S are all marked separately.

[00:32:01]Each may have distinctive morphology, or  there may be special word order for each one.

[00:32:09]One of the coolest systems is probably  "direct-inverse" or "hierarchical" alignment, where the marking depends on where the  participants fall on a hierarchy of importance.

[00:32:23]If a constituent is of a certain slot relative to  another, they're always marked the same, but there may be an additional marked to show that the roles  are reversed. This is the system used by Ojibwe.

[00:32:39]The rarest system is that of "transitive"  languages, where the agent and patient are marked the same, but the experiencer is marked  separately. Given that this doesn't actually provide extra information about the roles, since  you can already tell if a verb is transitive or not, not many languages use it. We mainly see  it when a language which historically used one system is changing over to another, or where there  is some other reason to distinguish the cases.

[00:33:16]Now, a lot of the time, languages use  the same alignment across the board, but some have different systems depending on  the type of marking. For instance, Nepali uses ergative noun morphology, but nominative verb  morphology. Verbs show the same marking for the agent and experiencer, but nouns show the same  marking for the patient and the experiencer.

[00:33:46]One thing that's actually very common is for one  or more marking strategies to use a direct system, while others use something else. This is because  you usually don't need all three systems to show the same thing, so it can be simplified by  ignoring this. Mandarin uses a nominative-style word order, with agent and experiencer coming  before the verb and the patient after. However, it uses no verb or noun morphology to show this, so  could be said to have direct alignments for these.

[00:34:26]The final general thing you need to know is  "glossing". We won't look too much at this in this video - I'll leave some links in the description  and I'll do a full video on the topic at some point in the future, but you do need the basics  for the coming videos, so let's take a quick look.

[00:34:48]When we're looking at languages and their  structure, it's hard to tell what's going on if we don't speak that language,  which, usually, we don't. So we note down the meaning and function of every morpheme  under the language sample - a kind of "running, in-text translation" that also shows all  the important information. This is called a "gloss" and is something you've probably  seen quite a lot already in this series.

[00:35:19]The basic principle is to translate every lexical word and give a functional description of  every grammatical morpheme. We do this using a set of standardised abbreviations  which are written in capital letters.

[00:35:37]Let's gloss the English sentence "The man eats his  chips". We start by translating (or in this case, just rewriting) all the lexical words - "man",  "eat" and "chip". Note that we use the unmarked form of "eat" and "chip" - what's known as the  "lemma" form. All the other words and morphemes are grammatical, so we leave these for now.  Next, let's look at the word "the". This is the definite article. It's filling the function of  making the noun definite - referring to a certain individual. This is the only role it's filling  and it's use doesn't preclude any other functions, so we gloss it just as "definite", the  abbreviation for which is "DEF". Next, let's do the same with "his". This is the third-person  singular masculine possessive pronoun, showing the chips as belonging to the man. This morpheme  shows four grammatical functions - third person (i.e. not involving me or you"), singular  (denoting possession by in individual, not a group), masculine (denoting possession  by a male individual) and possessive (showing ownership or holding of the chips). Combinations  of person and number occur often enough that we can combine them into one abbreviation. "Third  person singular" is denoted by either "3SG" or just "3S". "Masculine" is "M" and "possessive" is  "POSS". Since these are all in a single morpheme, we combine them using dots without spaces, so the  full gloss for this word is "3SG.M.POSS". Now, onto the bound morphemes. The two most obvious of  these are the "-s"s on "eats" and "chips". These are filling different roles. On the verb, it's  showing a third-person singular agent and the present tense, while on the noun, it's showing  plurality. However, these are separate morphemes to their host word, so we shouldn't join them  on with dots. Instead, we use dashes. A dash indicates a morpheme break, a dot indicates the  same morpheme. We see both in "eats". In reality, there is actually a bit more going on with this  word - it shows more than just the categories we've discussed, but for simplicity, let's stick  with this. Ideally, all categories should be marked in a gloss, but you can usually omit  some if they're not relevant to a discussion.

[00:38:52]So, that's it, right? Well, not quite. "Man" does  actually show something, even though we can't see it. When we use this word, it's obvious we mean  only one person. Two or more would be "men". We can therefore add an extra gloss here, showing  that this morpheme also codes the singular. "Man" is a special case, because the morpheme itself  shows the singular. But other nouns are different, as they use a separate morpheme. If we had the  word "chip", we'd know it was only one because of the absence of the plural suffix. We can therefore  suppose an invisible morpheme with no sound, which exists just by virtue of another's absence.  We call these "zero-morphemes" and denote them using an "O" with a line through it. We can  then gloss this just the same as any other.

[00:40:00]When working with other languages,  we usually put the gloss in between the sentence and it's translation. We also  often separate out morphemes with dashes, so we can see where the morpheme boundaries  are. If you want to get to grips with glossing, the standard system uses the "Leipzig glossing  rules", which are linked in the description.

[00:40:26]So, that brings us to the end of this episode.  I realise this isn't a very practical episode, but hopefully you can start thinking about  how you might want your conlang to behave.

[00:40:39]If you're following along, feel free  to create some temporary words using your phonology and phonotactics and start  sketching out some of what we've discussed in this video. The next episodes will go into  various topics on specific parts of speech, so until next then, thank you for watching.