The main theory dealing with the relationship between thought/cognition and metaphor is George Lakoff’s The Neural Theory of Metaphor (2009). Before this, however, he wrote a book in 1979 with Mark Johnson detailing metaphors and their types titled Metaphors We Live By. In it, they explore conceptual metaphors and the processes that underlie them. They describe conceptual metaphors as one conceptual domain being described in terms of another.
This is slightly different, but may overlap, with literal metaphors — the kind one usually thinks of when thinking about metaphors. Conceptual metaphors encompass a whole domain. For example, argumentation is often talked about with words regarding battle and war (“attack” or “shot down”). Lakoff’s theory revolves around this concept, taking it one step further and saying these domains are embedded within neural processes.
The basics of his theory posit that ideas and concepts are physically encoded and computed by the brain. This is nothing new, as any input from life experience runs through a neural circuitry, but what he proposes is that metaphors themselves have a circuitry. Akin to a critical period for language acquisition, there’s a critical period for metaphor development. At birth, a baby is born with around 100 million neurons that average 2,500 connections, but as the child grows, the number of connections range from 5,000 to 15,000. Connections that are seldom used will be eliminated. Like molding a statue, continuous input of one concept will strengthen certain connections, and lack of input will weaken until they’re shaved away.
An example Lakoff gives in a lecture is the conceptual metaphor “Affection is Heat”. In a society in which the parents cuddle their baby often, cooing and accompanied with loving words, the baby will learn to associate expressions of affection with physical warmth. The neurons processing information of temperature will simultaneously fire with the cells processing language/behavior, and the number of synapses will increase between the two, making the connection integrated and more efficient in firing action potentials (to send electrical signals). This, then, manifests in language like “he is a warm person”, where warm is a metaphor for loving, welcoming, open, affectionate, etc.
If the environment one is exposed to results in solid patterns in the brain, and if those manifest as certain language and behaviors, then we could theorize a backwards flow of information if changing language changes behavior, implicating thought as the middleman. Experiments have been done to identify tangible effects of metaphor on behavior.
Evidence #1 (Gibbs 2012)
To explore whether humans physically embody metaphorical narratives, Psychology Professor Raymond W. Gibbs Jr set up two experiments around the metaphor ‘Relationships are Journeys’, i.e. mapping the development of a relationship in spatial/motion terms.
The first involved forty-eight students at the University of California, Santa Cruz. Participants were randomly split into two groups. Half the participants were presented with two sets of stories, one about a successful relationship and the other about an unsuccessful, but both beginning with the crucial metaphorical statement “Your relationship was moving along in a good direction”. The other half of the participants were presented with the exact same stories, one about a successful relationship and one unsuccessful, but the statement “your relationship was moving along in a good direction” was replaced with the non-metaphorical statement “your relationship was very important to you”.
After reading the two stories assigned to them, all participants had to complete a survey with the following five questions: “Which relationship progressed further?”, “Which relationship was progressing faster at the beginning?”, “Which relationship is progressing faster at present?”, “Which relationship progressed more along a straight line?,” and “In which relationship were the individuals heading in the same direction?”. They answered the question with which story best fit. The results showed that for the metaphorical study group, over 90% selected the successful story to have progressed further (the first question), but only 63% in the non metaphorical group selected the successful story. This pattern holds for the rest of the questions, with results in each question category statistically significant (p < .05), suggesting that one sentence change — one metaphorical sentence — was enough to impact readers’ perception of relationship success. Aside from that one sentence, nothing else in the stories mentioned journeys, the speed, or any progress of the relationship. Inferences, ideally, would be dependent on the reader’s understanding of “moving along”. Gibbs does note that the answers of some of the questions could’ve simply been matching, as “moving along” and “progressing” are arguably synonymous. However, other questions might provide substantial proof for people’s inferences of behavior simply through metaphor.
But will it affect nonlinguistic tasks, i.e. behavior? Gibbs looked at this with his second experiment to identify physical embodiment of inferences about metaphor. Using the same domains of relationships and journeys, he tested 128 participants from the University of California, Santa Cruz. They were blindfolded, but not before being shown a marker (a bright tennis ball forty feet away), and listened to either the metaphorical or non metaphorical set of stories. Half were asked to walk to the ball, the other half asked to imagine walking but stop a stopwatch at the point they’d think they would’ve reached the ball.
His results were consistent with that of the first experiment, but with less significant results. Nonetheless, he found that those who heard the metaphorical successful and unsuccessful stories walked closer to the target than those who heard non metaphorical stories, whether successful or not. In other words, even those hearing the unsuccessful metaphor story walked further than those hearing the unsuccessful non metaphorical story. This implicates the metaphorical statement as the push in closeness towards the target. It’s very possible the phrase “moving along” triggered a spatial mapping within listeners, engaging neural circuitry of linear movement, and impacting physical behavior.
Evidence #2 (Dolscheid et. al. 2013)
Researchers in the Netherlands wanted to look at whether speakers of different languages think differently, even when the language was not in use. They looked at nonlinguistic psychophysical tasks, specifically whether mental representations of musical pitch affected production of it. It also happens that mental representations of pitch in many languages use conceptual metaphor, could be described in terms of target and source domain as ‘Pitch is Space’.
For example, participants in this study were native Dutch and Farsi speakers. Dutch speakers describe pitch similar to English speakers, in terms of highs/lows, at the top and bottom of ranges. Hoog ‘high’ and laag ‘low’ are spatial representations of pitch, more specifically on a vertical axis. Farsi speakers, on the other hand, describe pitch as nazok ‘thin’ or koloft ‘thick’. Nazok also means fragile or breakable, a point relevant later in the discussion.
Dutch and Farsi speakers were divided equally into two conditions: one a height interfering condition, and the other a thickness interference task. Participants sat in front of a screen and saw a vertical line intersected with a horizontal line at different positions with varying thickness. Each line was presented with a tone, and participants were asked to sing the pitch, using the screen as a reference. Speakers were tested in their own languages. There were three practice trials, and none of the instructional material involved mention of space-pitch metaphors. There were a total of 81 line pitch pairings, with horizontal lines intersecting the vertical line at seemingly random points, with 2 seconds given to sing back the pitch after each round, followed by an .500 ms break.
What they found for the Dutch speakers is that their pitch estimates went higher as the horizontal line intersected the vertical line closer to the top. The thickness of the line did not have a significant effect on their sound production (for example, a thin line near the top did not produce a significantly different pitch than a thicker line at a similar intersection point). For the Farsi speakers, it was the opposite — their pitch production reflected the thickness of the horizontal lines, regardless of where they were placed along the the vertical line. Results were significant.
In order to test for interference (and to answer their second question of whether effects were in play when using different languages), researchers trained Dutch speaking participants to recognize pitch in terms of thickness. They also had 81 trials (in addition to 8 practice trials) and were able to produce results statistically indistinguishable from that of Farsi speakers.
What’s interesting was their reverse-thickness training. They had a different group of Dutch speakers learn the Farsi metaphor, but in reverse, with nazok ‘thin’ correlating to a lower pitch and koloft ‘thick’ to a higher pitch. This was to test if acquiring the new metaphor was affirming previous notions of thin and thick in the Dutch language or not. What they found supported their hypothesis — while Dutch speakers were able to learn a new metaphor and have it immediately affect their results in the trials, it was not as close to the first interference task. That means Dutch speakers who learned the Farsi metaphor performed better and similar to Farsi speakers, but those who learned the reverse Farsi metaphor, while performing better than untrained Dutch speakers, did not perform as well as those who learned the metaphor the right way.
This indicates that while metaphor is impacting behavior, this particular metaphor is affirming preexisting notions that aren’t as exhibited as obviously, but exist, perhaps in conjunction to other metaphors.
While we only look at two experimental papers and five experiments, the implications they have are important. First, the Gibbs paper implicates metaphor in impacting behavior. One sentence possibly set off a cascade of processing connecting relationships to a different domain, in this case journey, and that manifested physically. Participants not only mentally viewed the metaphorical stories to have progressed further in a straight line, establishing the relationship between relationships and a physical journey, but also behaved accordingly — those being exposed to the metaphorical condition walked further upon hearing it than those who heard the same story, but devoid of metaphors.
It should be noted that participants were all English-speaking and from one locale, and the stories were in English, this study alone couldn’t prove the broader theory of linguistic relativity. English speakers are able to view the trajectory of one thing, in this case relationships, in terms of spatial notions and that affects their behavior. Less obvious but just as interesting was the first experiment in that even the way individuals viewed the success of a relationship was affected by the metaphor. Despite the first story being a success, it was viewed as more successful (92%) when described with the metaphorical statement than compared in the story without it (63%). This is especially interesting when considering the impact metaphors have in rhetoric, from daily life to politics. Metaphors, of course, exist for a reason — whether to understand an unfamiliar idea in terms of something more familiar, or to add flare to the mundane, or to imbue life experiences with imagery and evoke imagery usually only sight can. The experiment elucidated the degree of impact metaphors can have on viewing something as common as a relationship and how successful it is, almost exaggerating it unconsciously.
The second experimental paper is where specific linguistic relativity theories were addressed. The first experiment established that a ‘Pitch is Space’ metaphor affected participants ability to produce the pitch — Dutch speakers who saw a vertical line intersected closer to the top instinctively (or abided by what’s intuitive to them) produced higher pitches, and Farsi speakers abided thickness. When participants were using their own language and making natural inferences from their languages’ metaphors, they behaved accordingly. The question then was how they would behave if they learned a new metaphor?
Participants weren’t given a long time to learn the metaphor and were tested on it immediately — Dutch speakers were able to reproduce results rivaling native Farsi speakers in that time. This could support a language-influences-thought hypothesis when using a certain language (thinking-for-speaking), because if they hadn’t and results were closer to native Dutch speakers, that would mean their understanding of pitch metaphor crosses languages. But it didn’t. There wasn’t little interference either: the speakers were primed to behave as Farsi speakers, that when the language is in immediate use and once speakers understand what the terms mean, they are able to perform tasks in line with how the language frames them.
However, this isn’t as clear cut, as the researchers didn’t end there. They preformed another experiment in which Dutch speakers were trained with Farsi terms, but in reverse, and made many more errors. This suggests that as different as Farsi and Dutch seem, they may both have other terms that mirror similar ideas. In English, for example, most would agree that thin, not thick, is closer to the term fragile (in Farsi, the terms for both are the same), which is closer to high, whereas low is deep, at the bottom, heavy, and related to thick, not thin. This semantic chaining is not uncommon and culture probably plays a role in how people view these terms — the former chain of thin-fragile-high could be attributed to stereotypically feminine qualities (women are fragile and have high voices), and the latter masculine (low voices, bigger, etc). These factors are difficult to parse out and would need multiple studies to determine whether or not culture is the reason for certain affirmations of metaphor.
Perhaps the easiest explanation, for now, is that metaphors can affect our thinking-for-speaking: that the language we speak affects the way we think and in turn our affects our behavior when that language is in use. One could also theorize, due to the experiments in the second paper, that while metaphors differ across cultures and languages, communities may have lesser known metaphors that when learning a second language are affirmed, speeding up the process of metaphor acquisition, as seen with Dutch in acquiring Farsi terms.
Criticisms
One paper from Hungarian researcher and Linguistics professor Zoltán Kövecses lists a few criticisms of Lakoff’s theory, including but not limited to the issue of methodology, schematicity, and culture.
The first critique looks at three levels of metaphors, and how much of conceptual metaphor researchers examine their own mental lexicons to state definitively as to what the metaphor type is, rather than looking at real corpora and instances of usage, as usage changes rapidly overtime.
The second is that while conceptual metaphors compare two domains, not everything within one domain can be conceptualized in terms of the other. Like in the ‘Argument is War’ metaphor, one could say “he shot down her argument” but it would be strange to say “he gunned down her statement”.
The issue of culture is one that comes up ad nauseum, but as discussed in the last portion of the discussion, Lakoff’s initial part of the theory, of environment building circuitry in our brain is, not well understood. Exactly what factors and behaviors that play into a feedback loop is difficult to identify individually, and most ways to test them would be unethical.
Perhaps first would be to establish a neural relationship between two domains. In theory, single-unit-recordings could provide insight as to whether triggering one concept sets off a electro-physiological response in another node or neuron that processes the source domain than some other random domain. Unsure if this has been done before, but could be the first step towards understanding metaphor, Lakoff’s theory, and then zooming out, on the question of linguistic relativity.
