14.6 Semantic change

Change in meaning

Change in the meanings of words is often called semantic shift or semantic drift. As with most language change, semantic shifts usually involve some intermediate stage of variation, in which a word can be used with both the older and the newer meanings. When this occurs, the word is said to be polysemous, and the situation is said to be a case of polysemy. Over time, the older meaning may become obsolete, leaving only the newer meaning, which itself may ultimately be replaced by an even newer meaning sometime later.

Shift in scope

One of the ways that the meaning of a word can shift is in the range of concepts it includes. For example, a word’s meaning could shift from some specific concept to a larger, more general category that includes the original concept along with others. This is called widening, also known as generalization, extension, or broadening. The English word arrive underwent widening in its history. Early Middle English ariven was a borrowing of the Norman French word ariver, both with the meaning ‘to reach land after a long journey at sea’, ultimately from Latin ad rīpam ‘to the shore’. The meaning of ariven quickly widened, and its Modern English descendent arrive can now include any destination, any mode of transportation, and any travel time.

The widening of ariven to arrive is depicted in Figure 14.11, which shows how the more specific older meaning of arriving to shore by boat after a long voyage (delineated by the smaller oval with a light grey border) is contained within the more general newer meaning (the larger oval with a black border), which now also includes many other meanings that were not part of the older meaning, such as arriving to a campsite after driving all day, arriving at the top of a mountain after climbing for a few hours, and arriving by a short parachute jump into a football stadium.

Genericization (discussed in Section 14.4 as a type of neologism) can also be treated as a special case of widening, with the meaning of a brand name like dry ice or zipper widening from referring to one specific company’s product to all similar products made by any company. Other examples of widening from English are given in (1) and from other languages in (2).

Semantic shift in scope can go in the other direction, from a general meaning to a more specific meaning. This is called narrowing, also known as specialization or restriction. The English word deer underwent narrowing in its history. In Early Old English, dēor could refer to many kinds of animals, particularly mammals, but it narrowed throughout Old English and Middle English to refer only to wild mammals, especially those that were hunted. By Modern English, the meaning had narrowed even further to just members of the Cervidae family, especially red deer, roe deer, and similar animals. We can see evidence of the older meaning in related words from other Germanic languages where this narrowing did not happen, such as German Tier, Dutch dier, and Norwegian dyr, which all mean ‘animal’.

The narrowing of dēor to deer is depicted in Figure 14.12, which shows how the more specific newer meaning of the Cervidae family (delineated by the smaller circle with a black border) is contained within the more general older meaning (the larger circle with a light grey border), which also included many other meanings (faded images) that are no longer part of the newer meaning, such as pigs, donkeys, squirrels, etc.

Other examples of narrowing from English are given in (3) and from other languages in (4).

Metonymy

Semantic shift can also cause a word to refer to something not even included in the original meaning at all. One common way this occurs is through metonymy, which is when the shift makes use of some sort of direct association between two different meanings. For example, in many monarchies, the current monarch and their corresponding government are often referred to as the Crown, because of the physical crown that a monarch often wears, especially during ceremonial duties. Note that this is not widening or narrowing, because a monarch is a human being, while a crown is an inanimate object, so they are fundamentally different types of objects that do not form a natural set. However, because the monarch and their crown often occur together, they have a direct association that can result in metonymy. Other examples of metonymy from English are given in (5) and from other languages in (6).

Synecdoche is a specific type of metonymy that involves some kind of shift in meaning involving a whole object or concept and some individual inherent piece of that whole. If the meaning begins by referring to an individual piece and then shifts to refer to the entire whole is sometimes called pars pro toto ‘part for whole’. Pars pro toto synecdoche affected the Old English glæs, which originally referred to the hard transparent substance. Very early, glæs also referred to a drinking container made of glass by pars pro toto synecdoche, since glass is an inherent component of such containers. By Modern English, the plural glasses could similarly refer to spectacles, which also have glass as a component (in the lenses), so this was also pars pro toto synecdoche. Nowadays, drinking vessels and spectacle lenses can be made from other substances (e.g. plastic), but they can still be referred to as glasses due to widening.

The reverse type of synecdoche is totum pro parte ‘whole for part’. This is a very common type of synecdoche for names of countries and cities to be used to refer to prominent subparts. For example, Canada technically refers to the entire country, but through totum pro parte synecdoche, it can also be used to refer just to the Canadian government (as in Canada negotiated a trade deal with the European Union), to Canadian national sports teams (as in Canada qualified for the World Cup in 2022), or to other notable subparts of the country.

Other examples of synecdoche from English are given in (7) and from other languages in (8), with pars pro toto examples in (7a, 8a) and totum pro parte examples in (7b, 8b).

Metaphor

Another type of shift to a meaning not included in the original meaning is metaphor, which is a shift based on similarity in form or function between concepts that do not otherwise have a relationship. This differs from the direct association needed for metonymy, because concepts involved in metonymy are typically encountered together (though not obligatorily so): monarchs are often found or depicted with their crowns, prayers in some religions are counted on a string of beads, etc. However, metaphor connects concepts that do not have any such ordinary direct association.

For example, the English verb grasp ‘hold (in a physical sense)’ has undergone a shift by metaphor to have the additional meaning ‘understand’, as if the brain is mentally holding a concept in a similar way to how hands can physically hold an object. But you can hold something (physically grasping it) without understanding it (metaphorically grasping it), and vice versa, so they are not directly associated. The connection between the two uses of grasp is an abstract and indirect metaphor, rather than concrete and direct metonymy.

Other examples of metaphor from English are given in (9) and from other languages in (10). Note the different types of metaphor demonstrated by these examples: physical similarity in form (9a, 10a) and more abstract conceptual similarity (9b, 10b).

Shift in quality

Another common type of semantic shift is a shift in positive or negative aspects of a meaning. Elevation or amelioration (or sometimes melioration) is a shift to a more positive meaning. This means elevation could change a neutral meaning to a positive one or a negative meaning to a neutral or positive meaning. The change could be even smaller, from positive to very positive or from negative to less negative. As long as the new meaning is more positive in some way than the original, the shift is elevation.

For example, Modern English eager descends from Middle English egre, which was a borrowing from French that originally meant ‘angry, acidic’. We can see this original meaning reflected in related words like acid, acrid, and acerbic, which all ultimately come from Latin ācer ‘sour, sharp’. Over time, the negative quality of the meaning of eager was lost, and it now has the more positive meaning ‘enthusiastic’.

Other examples of elevation from English are given in (11) and from other languages in (12). Note the different types of elevation demonstrated by these examples: neutral to positive in (11a, 12a) and negative to positive in (11b, 12b).

The quality of a meaning can shift in the opposite direction, so that the new meaning is more negative than the original. This kind of shift is called degeneration, pejoration, or deterioration. As with elevation, degeneration could involve a drastic change from positive to negative or a smaller change, such as from positive to less positive. For example, Old English cnafa originally referred to any young man or boy. Over time, this meaning underwent degeneration and acquired increasingly negative meanings, first ‘young male servant’ and then eventually ‘dishonest person, villain’.

Other examples of degeneration from English are given in (13) and from other languages in (14). Note the different types of degeneration demonstrated by these examples: neutral to negative in (13a, 14a) and positive to negative in (13b, 14b).

In some cases, a word or expression with a negative meaning may begin to be avoided in favour of a different expression, which is called a euphemism, a kind of taboo avoidance (see Section 2.2). Many concepts are prone to euphemism: lack of intelligence, sexuality, bodily functions, etc. The topic of death is particularly uncomfortable, so there are many kinds of euphemisms (both solemn and humorous) for death and dying across languages, such as English pass away and kick the bucket, Polish spaść z rowerka ‘fall off a bicycle’, Finnish heittää lusikka nurkkaan ‘throw your spoon in the corner’, Yorùbá kpakpòdà ‘change states’, Q’anjob’al max mutz’eloq ‘closed one’s eyes’, and Mandarin 去世 (qùshì) ‘leave the world’.

In some cases, an entire concept is so stigmatized that any euphemism for it may undergo degeneration and eventually be replaced by a different euphemism, which itself may eventually be similarly replaced, and so on. This kind of chain of repeated replacement of euphemisms is called a euphemism cycle (Taylor 1974) or sometimes a euphemism treadmill. For example, spaces used for elimination of bodily wastes have been called by a variety of names throughout the history of English, many of them the result of euphemistic use of other words, such as Old English gang, which comes from the verb gangen ‘to walk’. Eventually, some of these euphemisms begin to be perceived as too crass, requiring new euphemisms. This happened with the word toilet (from French toilette ‘little cloth’), which originally referred to cloth used for carrying or protecting clothes. After a while, a new wave of euphemisms came about (bathroom, restroom, washroom), which in turn triggered more euphemisms later (ladies/mens room, facilities).

A special kind of limited elevation occurs when a slur is reclaimed (see Section 2.3 for discussion). The original negative meaning is reversed to positive, but typically only when the reclaimed slur is used by the targeted group. As time goes on, the elevated usage may expand beyond the targeted group and become standardized usage.

Shift in intensity

Semantic shifts can also affect the relative intensity of a meaning. Hyperbole (a.k.a. exaggeration or overstatement) is a semantic shift in which the newer meaning has a weaker intensity than the original meaning. For example, Old English bana originally mean ‘murderer’, but its Modern English descendent bane now means ‘any source of harm or misfortune’, which is generally not as extreme as death. Other examples of hyperbole from English are given in (15) and from other languages in (16).

Hyperbole can lead to semantic bleaching, which is when the lexical content is eroded, which in the extreme can lead to grammaticalization (see Section 14.5 for discussion of grammaticalization). Many English words having to do with reality or literal truth have undergone semantic bleaching to become intensifiers, a kind of degree word (see Section 6.5 for discussion of degree words). For example, Middle English verray ‘true’ was borrowed from Old French verai ‘true’ (the etymon for Modern French vrai ‘true’). Due to semantic bleaching, Middle English verray eventually shifted Modern English very which is now only an intensifier with no necessary meaning of truth (otherwise, very fake and very false would be illogical).

This same process has happened many times to many other English words, such as literally, which has been undergoing semantic bleaching for over two hundred years, since at least the late 1700s. Its original meaning in Middle English was ‘ verbatim, not figuratively’. However, this general semantic area is particularly prone to semantic bleaching, and just like verrai, really, actually, truly, genuinely, veritably, and many other similar words, literally is now commonly used as a semantically bleached intensifier to make a statement stronger.

This intensifier usage of literally is often flagged as prescriptively incorrect in a figurative context, such as literally dying of laughter, because it superficially appears that literally means ‘figuratively’, which is the opposite of its etymological meaning. However, literally is acting as a generic intensifier that can be used in both literal and figurative contexts, in the same way that very and really can. For example, both literally laughing and laughing and literally dying of laughter mean something like ‘laughing a lot’, with literally intensifying the overall meaning. It just happens to be intensifying a literal meaning in one and a figurative meaning in the other. The claim that literally cannot be used in a figurative context because of its original meaning is an etymological fallacy (see Section 14.1 for discussion of etymological fallacies). Interestingly, the same people who complain about literally dying of laughter typically do not make the same complaints about really dying of laughter, even though both are the result of the same kind of semantic shift. They have accepted the shift for an entire set of words except one, a common oversight in etymological fallacies.

Litotes or understatement is the opposite of hyperbole, with the newer meaning having a stronger intensity than the original meaning. For example, Middle English gruccen originally mean ‘to complain or grumble’, but its Modern English descendent grudge now means ‘persistent bitterness or resentment’, often held over a longer period of time than a mere complaint. Other examples of litotes from English are given in (17) and from other languages in (18).

Note that any given semantic shift might be a combination of different types. For example, the shift in meaning of English giddy (15a) is not just an example of hyperbole (from stronger to weaker), but also elevation (from negative to positive) and perhaps also a bit of metaphor due to similarity in outward behaviour (confusion, unpredictability, lack of restraint, etc.).

Regular semantic change

Semantic shifts normally do not have the same kind of Neogrammarian regularity that we find for sound change (see Section 14.3 for discussion of regular sound change). Instead, semantic shifts are usually sporadic in nature, affecting only one word at a time, rather than affecting an entire group of synonyms or all other words with similar types of meanings, all at once, in the same way.

It makes sense why regular semantic shift should be rare. Imagine if narrowing and elevation changed the meaning of all words for people to refer only to royalty, so that woman would shift to mean ‘queen’ only, boy would shift to mean ‘prince’ only, person would shift to mean ‘royal person’ only, etc. While this kind of change is not problematic for a single word, having it happen regularly to every word of the same type would be a huge inconvenience for communication. We would need further changes to allow us to continue efficiently talking about all women, boys, people, etc., regardless of royal status. This is not technically impossible, but it is implausible due to its impracticality.

That said, there are some cases of semantic shifts that could reasonably be classified as regular in some sense. Chamorro (a Malayo-Polynesian language of the Austronesian family, spoken on Guam and in the Northern Mariana Islands) has a set of four directional terms that have all shifted together as a set (Solenberger 1953, Borja et al. 2006, Chung 2020). Based on a variety of evidence, the four Chamorro directional terms seem to have originally referred to directions relative to the sea and coast: lågu ‘seaward’, håya ‘inland’, kåttan ‘along the coast, with the sea on the left’, and luchan ‘along the coast, with the sea on the right’, as shown for Guam in Figure 14.13. Note how the directions referred to by these terms (in green) correspond to different cardinal directions (north, south, east, and west, in dark red) depending on location.

However, these directional terms apparently underwent semantic shift after contact with Europeans, who more commonly used words for the cardinal directions (north, south, east, and west). Since Hagåtña (originally known as Agana) was the major population centre at the time, the corresponding cardinal direction meanings were fixed based on Hagåtña, which has an east-west coastline. So the local meaning of lågu ‘seaward’ corresponded to the European notion of ‘north’, while håya ‘inland’ corresponded to south, and so on.

As Chamorros from Hagåtña moved to other parts of Guam, they brought their new cardinal meanings of the directional terms with them. Thus, in Malesso’ (originally Merizo), the original ‘seaward’ meaning of lågu corresponded to a southwestern direction, but it was supplanted by the new Hagåtña meaning ‘north’ (19a), essentially, narrowing from ‘seaward’ to ‘seaward with respect to Hagåtña’, which was then reinterpreted as cardinal ‘north’. A similar shift happened for the other three directional terms (19b–d).

Essentially, the collective meanings of all of the Chamorro directional words rotated in Guam from local relative directions to global cardinal directions (Figure 14.14). The older meanings are still available for some speakers, so there is some variation.

Interestingly, the words shifted differently on different islands, but the underlying nature of the shift was the same. On Saipan, the major population centres at the time of European contact were Garapan and Chalan Kanoa, which both have a roughly north-south coastline. This meant that lågu ‘seaward’ corresponded to a western direction in those villages, so it eventually shifted to mean ‘west’ for everyone on Saipan, and the meanings of the other three directional words shifted accordingly (Figure 14.15).

The result is a superficially inexplicable mismatch in modern meanings of the Chamorro directional terms between the two islands (20).

To understand this pattern, we have to know the relevant history, most importantly, where the relevant population centres were when the Chamorros encountered the concept of cardinal directions. By understanding the history, we can see how the meanings of these words were originally fully consistent across the islands (‘seaward’, ‘inland’, etc.), and they rotated in a regular way to align with the relevant population centres at the time (Hagåtña on Guam, Garapan and Chalan Kanoa on Saipan). Because the population centres faced different cardinal directions on the different islands, the resulting modern meanings are different. Importantly for the question of regular semantic shift, the structural arrangement of the directions did not change, and the meanings were affected as an entire set. So if someone is facing luchan, they always have lågu on their right, håya on their left, and kåttan behind them, no matter which part of which island they are on, and no matter whether they are using the original or newer meanings.

Check your understanding

References

Berlin, Brent. 1972. Speculations on the growth of ethnobotanical nomenclature. : Language in Society 1(1): 51–86.

Borja, Joaquin Flores, Manuel Flores Borja, and Sandra Chung. 2006. Istreyas Mariånas: Chamorro. Saipan: Estreyas Marianas Publications.

Brachet, Auguste. 1868. Dictionnaire étymologique de la langue française. Paris: Bibliothèque d’Éducation, J. Hetzel et Cie.

Chung, Sandra. 2020. Chamorro grammar. Santa Cruz: University of California. http://dx.doi.org/10.48330/E2159R

Derksen, Rick. 2008. Etymological dictionary of the Slavic inherited lexicon. Leiden Indo-European Etymological Dictionary Series. Volume 4. Leiden: Brill.

Derksen, Rick. 2015. Etymological dictionary of the Baltic inherited lexicon. Leiden Indo-European Etymological Dictionary Series. Volume 13. Leiden: Brill.

Ishiyama, Osamu. 2019. Diachrony of personal pronouns in Japanese: A functional and cross-linguistic perspective. Current Issues in Linguistic Theory 344. Amsterdam: John Benjamins.

Olko, Justyna. 2015. Language encounters: Toward a better comprehension of contact‑induced lexical change in colonial Nahuatl. Politeja 12(6): 35–52.

Penny, Ralph. 2002. A history of the Spanish language. 2nd edition. Cambridge: Cambridge University Press.

Solenberger, Robert R. 1953. Recent changes in Chamorro direction terminology. Oceania 24(2): 132–141.

Taylor, Sharon Henderson. 1974. Terms for low intelligence. American Speech 49(3/4): 197–207.

Zhang, Hongming. 1998. Chinese etyma for river. Journal of Chinese Linguistics 26(1): 1–47.